"8

@: .

i ne Appiication o t M o a n
 O Brasil, nos ultimos anos, teve o privilegio de ser
o palco de diversos projetos relevantes de obras
subterrâneas, entre os quais se destacam: os túneis e
cavernas subterrâneas da Usina Hidroelétrica de Serra
da Mesa, o projeto de cavernas subterrâneas (para a
Petrobrhs) para estocagem de GLP em Siío Sebastião,
túneis e estações subterrâneas para sistemas
metroviarios em Brasilia, Sâo Paulo e Rio de Janeiro, e
túneis para vias expressas em diversas cidades, como é
o caso da Linha Amarela no Rio de Janeiro. e do Rodo
Anel em 590 Paulo.
O CBMR (Comite Brasileirode Mecânica das Rochas),
em uma pmmg8o conjunta com o CBT (Comitê Brasileiro
de Túneis) e a ABMS (Associação Brasileira de Mecânica
dos Solos e Engenharia Geotécnica), organizou o curso
do Dr. Evert Hoek, na semana anterior ao 24' Congresso
Mundial de Túneis ("Tunnels and Metropolises?, da ITA
(International Tunnelling Association), realizado em São
Paulo (Abril de 98). O curso práticu "Engenharia de Rochas
e a AplicaçSo de Tecnicas Modernas ao Projeto e
Construção de Obras Subterrâneas" atraiu um número
elevado de participantes, e possibilitou a publica@o das
suas notas de aula no formato de livro.
O CBMR e o CBT tem a honra de disponibilizar para
a comunidade geotécnica brasileira e internacional esta
publicaç%o,que certamente contribuirá para aprimorar a
qualidade tkcnica do projeto e construção de obras
subterrâneas a nlvel global.

CBMR
(Comitê Brasileiro de Mecânica das Rochas)
CBT
(Comite Brasileiro de Túneis)
ABMS
(Associagão Brasileira de Mecânica dos Solos
 Notes from a short course by
Dr. Evert Hoek

April, 22 t o 24, 1998

Sao Paulo, SP
Brazil

ORGANIZATION .
Roberto Kochen
&
Paulo Cella

PROMOTION
(CBMR) Brazilian Rock Mechanics Cornmittee
(CBT) Brazilian Tunnelling Committee
(ABMS) Brazilian Society for Soil Mechanics and Geotechnical Engineering
 Organization
ROBERTOKOCHEN
&
PAULOCELLA

Grafic design and derktop publisliiiig

DANILO HENR~QUE CARVALHO
PATRICIA BUGLIAN
PRODUIORESASSOCIADOS
(011) 814 5444

Covei pliotos of
S ú n u DA MESA
Uiiderground Powei Plant
FUKNAS CENTRAIS EL~TRICU
 Contents
Tlie course presented a series of case Iiistories, eacb designed to bring out specific aspects of the
application of inodern rock engineering techniques to tunnel design. While many fundamental issues are
discussed, they are presented from a practical point of view for use by geologists and engiiieers who are
concerned with real tunnelling problems. The case histories, drawn froin Dr Hoeli's veiy wide international
experience, covered the range from extreinely poor quality rocks to hard jointed rod< inasses and iiicluded
very shallow tunnels as well as very deep tunnels. Site investigation and rock mass classification techriiques
are reviewed and the estimatioii of rock mass properties is discussed. Special attention is given to the design
of the excavations associated with hydroelectric projects and underground metro developmeiits.

Foreword
Brazil, in the last years, had tlie privilege of heing the host for inajor tunnelling projects, lilce: tlie
underground caverns of Serra da Mesa Hydroelectric power plant, Petrobrás planned oil storage caveriis in çXo
Sebastião, tunnels for subway systems in Brasilia, São Paulo and Rio de Janeiro, atid freeway tun~ielsin inany
cities, with emphasis for the Yellow Line in Rio de Janeiro, and the Ring Road Express Link in São Paulo.

CBMR (Brazilian Rodc Mechanics Committee), in a joint promotion with CBT (Brazilian Tunnelling
Coinmittee), had the privilege of hosting Dr. Evert Hoelc, in the week before tlie 24th World Tunnel Coii-
gress ("Tunnels and Metropolises") of the International Tunnelling Association, held in São Paulo (April 98).
-
The short course I b c k Engineering The Application of Modem Techniques to Underground Desipn",
attracted a record iiumber of participapts, iii terins of courses organized hy CBMR, and made pqssible the
publication of these notes authored by Dr. Evert Hoek.

n i e Brazilian Rock Mechanics Coinmittee and the Brazilian Tunnelling Coinmittee have the privi-
lege of inaking available to the Rock Mechanics and Tunnelling Cominunity this puhlication, whicli we liope
will help advance the proficiency leve! of tunnel design Gorldwide.

Fernando O. Franciss - President, CBMR (Brazilian Rodi Mechanics Coininittee)

Argiiniro A. Ferreira - President, CBT (Brazilian Tunnelling Committee).
ABMS (Brazilian Society for Soil Mechanics and Geotechnical Engineering).
 Preface

Increasingly sophisticated tools are becoming available to the rock engineer and yet many
fundamental design questions remain - how does one decide on the acceptable risk of an open pit
slope failure, the amount of deformation that can be tolerated in the foundation of an arch dain,
the allowable loads in the reinforcing cables in an underground cavern or whether a tunnel should
be constructed using drill and blast methods or by tunnel boring machine?
Questions such as these, which are of great practical significance in civil and mining engi-
neering, are addressed in these notes which are based on a number of case histories - each carefully
chosen to illustrate the concepts and practical approaches used. The tools and techniques covered
range from simple 'back of the envelope' estimates to the most sophisticated numerical models
currently available. The discussion includes the selection of appropriate input data, use of sensitiv-
ity studies and probabilistic approaches and the iterative use of obsenred performance to calibrate
and refine models during design and constructioil.
These notes have been prepared as material to be used during courses and also to provide a
reference volume after the completion ofthe courses.'It is emphasised that these are notes and that
the volume is not a formal text. It has not been and will not be published in its present forin and
the contents will be revised from time to time to meet the needs of particular audiences.
Readers are encouraged to send their comments, corrections, criticisins and suggestions to me
at the address given below. These contributions will help me to improve the notes for future courses.

44/- Di Eveit Hoek

Evert Hoek Consultiiig Engineer Inc
3034 Edgeinoi~tBoulevard
P O Box75516
North Vancouver, B C
Caiiada V7R 4x1

Far +1 604 980 3517

Plioiie +I 604 988 3064
Eiiiail: eliaek@ibrn.net
 Dr Evert Hoek

Dr Hoek was born in Zimbabwe and graduated in mecliaiii-

cal engineering from the University of Cape Town in South Africa
in 1955. For his post-graduate work he specialised in experimental
stress analysis and obtained a master's degree for his worlc on three-
dimensional photoelastic techniques. In 1958 he jointed tlie South
African Council for Scientific and Industrial Research and became
involved in the application of stress analysis techniques to the
study of rock stress problems in deep leve1 gold mines. He was
awarded a PhD in 1965 by the University of Cape Town for a
thesis on brittle fracture in rock.
In 1966 he accepted an invitation froin the Imperial Col-
lege of Science and Technology, one of the colleges of the University of London, to establish an
interdepartmental centre of research and teaching in rock mechanics. He was appointed Professor
of Rock Mechanics in 1970 and awarded a DSc in Engineering by London University in 1975.
In 1975 he moved to Canada as a Principal of Golder Associates, an international geotechnical
consulting firm. In 1987 he accepted the post of Industrial Research Professor of Rock Engineering
iil tlie Department of Civil Engineering at the University óf Toronto.
Dr Hoelc returned to a full time consultin~practice in Vancouver in 1993. He is now an
v .

-
indeoendent consultant and is a member of a number of consultinrr boards and a technical review
consultant on severa1 inajor civil and mining engineering projects. He lias recently worlted on
projects in Canada, Greece, India, Venezuela, Chile, Hong Kong, Indonesia, Australia and the
Philippines.
He has published numerous technical papers and three boolts. His inost recent book on roclt
support for hard rock excavations, co-authored with Professors P.K. Kaiser and W.F. Bawden, was
~ublishedin January 1995. Amongst the awards which he has received are the E. Bunvell Award
from the Geological Society of America (1979), Elected a Fellow of the Royal Acadeiny of Eilgi-
neering, UK (1982), Rankine Lecturer, British Geotechnical Society. (1983), The Gold Medal of tlie
Institution of Mining and Metallurgy, UK, (1985), The Muller Award, International Society of
Rock Mechanics (1991), William Smith Medal, Geological Society ofLondon, (1993), the award of
an honorary DSc in Engineering by the University of Waterloo, Canada (1994) and presentation of
the Glossop Lecture to the Geological Society in London (1998).
 Message from the sponsors
The puhlication of this book was sponsored hy the following Brazilian consulting compatiies:

VETEC Engenharia
VETEC Engenharia is proud to contribute to the puhlication of Dr. Hoek's hook, wliicli will surely
improve current knowledge oftunnel design and construction. Tunneling techniques that cai1 increase safety,
and enahie cost reductions, will he increasingly important in the near future, as well as environmental pres-
ervation features.
VETEC has dealt with these important aspects (safety, cost and enviionment) in the design of 6,000
meters of tunnels for São Paulo Ring Road Link, with excavation cross sections of up to 200 square ineters,
without precedents in Brazil. These same aspects were considered in the design of 5,740 ineters of tunnels at
Ferrovia d o Aço (Steel Railway Line), in Brazil.
Beirig aware that respect for nature and mankind will he the great challenge of Civil Engineering in
tlie nesr future, VETEC congratulates Prof. Hoek for this outstanding hook.

VENCE Engenharia e Empreendimentos

The publication of this book by Prof. Hoek will be of great value for professionals dedicated to h1111lei
design and construction. It is not only another sonrce of technical inforrnation of high quality, bnt also a
work of great merit, reflecting the outstanding knowledge of tbe author regarding underground works.
VENCE Engenharia knows that pressure for cost reductions, the current awareness regarding envi-
ronmental preseivation, and new theoretical design methdds require an ever increasing use of modern engi-
neering techniques. These techniques must comprise the latest technological advances available ori a global
scale, now spreading fast through world integration.
Nevertheless tlie new resources rnade available hy computing and telecomrnuiiication advances, leai-ning
new engineering tecliniques has to b~ done using classical procedures. It is through short courses. seminars,
congresses and, iiiaiuly, by technical publications, that learning takes place. For this reason, VENCE is
proud to spotisor thç publication of the current edition of this book, and wishes that it's inission of spread-
ing knowledge will he performed in an efficient way.
Civil Engineering Design, Constmction Supervision, Highways, Transport Plaiining, Water and Sew-
age Systems, Stmctural and Foundation Pesign, Environment, Automatic Control of Traffic.
 Table of Contents

Development of rock engineering 1

Introduaion
Rackbursts m d elastic theory
Discontinuous rock masses
Engineering rock inechmics
Geological data collection
Laboratory testing of rock
Rock mass classification
Rock mass strength
in situ stress ineasurements
Giouiidwatei problems
Rock reinforcernent
Excavation rnethods in rack
Analytical tools
Concl~isions

When is a rock engineering design acceptable

Introduction
Landslides in reservoirs
Deforrnation of rock slopes
Structtirai failures in rad<masses
Excavations in weak rock
Factor of safety
Probabilistic analyses

Rock mass classification

Introduction
Engineeiing rack mass classification
Terzaglii's rod( rnass classification
Classifications involving stand-up time
Rock quality designatian index (RQD)
Rock Stmcture Rating (RSR)
Geamechanics Classification
Modificatioris to RMR for miiiing
Rock Tunnelling Quality Index, Q
Using rock mass classification systems

Shear strength of discontinuities

Introduction
Sliear strength of planar surfaces
Shear strength of rough surfaces
Bairon's estimate of shear strength
Field estimares of JRC
Field estimares ofJCS
Influence of scale onJRCandJCS
Shear strength of filled discontinuities
Influence of water pressure
Instantaneous cahesion and friction
5 Structurally controlled instability in tunnels
5.1 Introduction
5.2 Identification of potential wedges
5.3 Support to contra1 wedge failure
5.3.1 Rock boltinp wedges
5.3.2 Sliotcrete support forwedges
5.4 Consideration of excavation sequeiice
5.5 Appiication of probability theory

The Rio Grande project - Argentina

Iritrod~~ction
Tailrace mnnel support
Support for power cavern
Discussion of support desigii and costs
Analysis using UNWEDGE program
Input Data
Input of excavation cross-sedion
Determination ofwedge geometry
Installation and analyris of rockbolts

A slope stability problem in Hong Kong

Introduction
Description of problem
Limit equilibrium models
Estiinates of shear strengtli
Estimate of eartliquake acceleration
Analysis of mobilised shear strengtli
Decision on shorr-term stability of the Sau Mau Ping slope
Evaluation of long-terin remedial measures
Final decision on long t e m remedial works

Factor of safety and probability of failure

Introduction
Sensitivity studies~
An intraduction to probability theory
Probability of failure

Analysis of rockfall hazards

Introduction
Mechanics of rockfalls
Possible ineasures whidi could be taken to reduce rockfall hazards
Identification of patential rockfall problems
Reduction a f eiiergy levels associated with excavation
Physical restraint of rockfalls
Rockfall Hazard Rating System
Slope Height
Ditcli Effcctiveness
Average Vehicle Risk (AVR)
9.4.4 Percent of Decirion Sight Distance
94.5 Roadway Width
9.4.6 Geologic Character
9.4.7 Block Size or Quantity of Rockfall Per Event
9.4.8 Climate and Presence ofWzter oii Siope
9.4.9 Rockfall History
9.5 Risk analysis of rockfalls oii higliways
9.5.1 RHRS rating for Argillite Cut
9.5.2 Risk analysis for Argillite Cut
9.6 Comparisan between assessed risk and acceptable risk
9.7 Conclusions

10 In situ and induced stresses 119

10.1 Introduction
10.2 In situ stresses
10.3 The World stress map-
10.4 Developing a stress measunng programme
10.5 Aiialysis of induced stresses
10.6 Numerical methods of stress analysis
10.6.1 Boundary Element Method
10.6.2 Finite element and firiite differeiice inethods
10.6.3 Distinct Element Method
10.6.4 Hybrid approaches
10.6.5 Two-dimensianal and three-dimeiisiorial models
10.6.6 Stress analysis using the program PHASE2
10.7 Practical example of two-dimensional stress analysis
10.7.1 Analyris a f top headiiig stability
10.7.2 Analysis of complete excavatian
10.73 Conclusion

11 Rock mass properties

11 1 Introduction
11.2 Generalised Hoek-Brown criterioli
11.3 Intact kock properties
11.4 Influence of sample size
11.5 Geological strength Iridex
11.6 Mohr-Coulamb parameters
11.7 Deformation modulur
11.8 Post-failure behaviour
11.8.1 Very good quality hard rock masses
11.8.2 Average quality rock mass
11.8.3 Very poor quality rock mass
11.9 Reliahility of rock mass strength estimates
11.9.1 Inputparameters
11.9.2 Output parameters
11.9.3 Slope stability calculation
11.9.4 Tuniiel stability calculations
11.9.5 Conclusions
11.10 Practical exainples of rock inass property estimates
11.10.1 Massive weak rock
11.10.2 Massive strong rock inasses
11.10.3 Aveiage quality rock mass
11.10.4 Poor quality rock mass at shallow deptli .
11.10.5 Poar quality rock mass under higli stress
11.10.6 Slope stability consideiatioiis

12 Tunnels in weak rock

12.1 Introduction
12.2 Deforination arouiid an advancing tuiiiiel
12.3 Tuniiel defarmatioii'aiialysis
12.3.1 Definition of failure criterioii
12.3.2 Analysis of tunnel behaviour
12.4 Dimensianless plats of tunnel deformatioii
12.5 Estimares of support capacity
12.6 Practical exarnple
12.6.1 Estimate of rock mass properties

13 Large Powerhouse caverns in weak rock

13.1 Introd~~ction
13.2 Rock rnass strengtli
13.3 Iii situ stress conditioiis

-
13.4 Piilar size between excavations
13.5 Problems iii using a coiicrete ardi iii weak rock
13.6 Crane bearns
13.7 Choice a f cnvern sliapes
13.8 Influence of joints and bedding planes
13.9 Design of reinforcernent
13.9.1 Estimating support pressures
13.9.2 Design of rockbolt and cable support
13.9.3 Use of sliatcrete linings
13.9.4 Suppart installation sequeiices
13.10. Excavation rnethods
13.11. Caverii instrumeiitation
13.12. Suminary and conclusions

14 Rockbolts and cables

14.1 Introductioii
14.2 Rockbolts
14.2.1 Mechanically anciiored rockbolts
14.2.2 Resiii anchored rockbolts
14.3 Dowels
14.3.1 Grouted dowels
14.3.2 Friction dowels or 'Split Set' stabilisers
14.3.3 'Swellex' dowels
14.4 Load-deformation characteristics
14.5 Cables
14.5.1 Boiid strengtli
14.5.2 Grouts aiid groutiiig
14.5.3 Cabie iiistallation
14.5.4 Cables fbr slope reinforcement

15 Shotcrete support
15.1 Introduction
15.2 Shotcrete techiloloby
15.2.1 Dry nlix sliotcrete
15.2.2 Wet iiiir shotcrete
15.2.3 Steel fibre reiiiforced micro silica shotcrete
15.2.4 Mesh reinforced sliotcrete
15.3 Shotcrete applications
15.4 Design of shotcrete support

16 Blasting damage in rock

16.1 Introductioii
16.2 Historical perspective
16.3 Blastiiig dainage
16.4 Damage cantrol
16.5 Blastiiig design arid caiitrol
16.6 Conclusion
 Development of rock engineering

1.1 lntroduction
We tend to think of rode mechanics as a modern engineering discipline and yet, as early as 1773,
C0~ilomhincluded results of tests on rodes from Bordeaux in a paper read before the French Acadeiny in
Paris (Coulomb (1776), Heyinan (1972)). French engineers started construction o f t h e Pdnama Canal in 1884
ancI this taslc was taken over by the US Ariny Corps ofEngineers in 1908. In the half century betweeii 1910
anc1 1964,60 slides were recorded in cuts along the canal and, although these slides were not analysed in rodc
me chanics terms, recent workby the US Corps ofEngineers (Lutton et al(1979)) shows that these slides were
predominantly controlled by structural discontiiluities and that modern rock mechanics concepts are hlly
applicable to the analysis of tliese failures. In discussing the Panama Canal slides in his Presidential Address
to the first international conferente oii Soil Mechanics and Foundation Engineering i11 1936, Karl Terzaghi
(Terzaglii (1936), Terzaglii and Voight (1979)) said 'The catastrophic descent of the siopes of tlie deepest cut
of the Panaina Canal issued a warning that we were overstepping the liinits of our ability to predict tlie
consequences of our actions ....:
In 1920Josef Stini started teaching 'Technical Geology' at the Vienna Technical Uiliversity and hehre
he died in 1958 he Iiad published 333 papers and bool<s (Muller (1979)). He founded the jourilal Geulu~yieziizd
Bauzuteren, the forerunner of today's journal Rock Mechanics, and was probably the first to einphasise the
importante of structural discontinuities on the engineering behaviour of rocle masses.
Otlier notable scientists and engineers from a variety of disciplines did some interesting worlc o11 rock
behaviour during the early part of this century. von Karmin (1911), King (1912), Griggs (1936), Ide (1936),
and Terzaghi (1945) all worked on the failure of rock materiais. I n l 9 2 1 Griffitli proposed his theory of brittle
material failure and, in 1931 Buclcy started using a centrifuge to study the failure of mine models under
simulated gravity loading.
None ofthese persons would have classified theinselves as rodc mechanics engineers - the title had not
been invented at that time - but all ofthem made significant contributioils to the fundamental basis of the
subject as we lznow it today. I have made n o attempt to provide an exhaustive list of papers related to rock
inechanics which were puhlished before 1960 but the references given above will show that iinportant devel-
opments in the subject were taking place well before that date.
The early 1960s were very important in the general developinent of rodz mec!iailii> world-wid~because
a niimber of catastrophic failures occurred whicli cleariy dernonstyated that, in rodc as well as i11 soil, 'we were
over-stepping the lirnits of our ability to predict the coiisequences ofour actions' (Terzaghi and Voiglit (1979)).
In Deceinber 1959 the foundation of the Malpasset concrete ardi dain in France failed and the
resultingflood Icilled ahout 450 people. In October 1963 about 2500 people in the Italian town of Longarone
were killed as a result of a landslide generated wave which overtopped the Vajont dain. These two disasters
Iiad a major iinpact on rodc mechanics in civil engineering and a large number of papers were writteil on tlle
possible causes of the failures (Jaeger (1972)).
Rock Engiiieering: Course notes t,y Evert Hof

Figure 1.13: TheViijont

dam during ~mpaundng
of the reiervair In the
midde distance. in the
centre of the picture, ii
Maunt Tor w t h the
unstabie i o p e visble a i
a white i c a r on the
mountain i d e above
the waterline.

Figure l . l b : During the

fliling of the Vaiont
reietvaii the toe of the
siope on Maunt Toi wai
subrnerged and this
precipitated a slide. The
rnound af debris from
the side isvisibe n the
central part of the
photagraph. The very
rapid desienr of the
iiide material dliplated
the watei in the
reservoir causing a 100
m high wave to ouertop
the dam wall. The dam
itself, asibe in the
foreground, was argely
undamaged.
 Chapter pment of rock enginee

Figure 1 . 1 ~The
: tawn
of Longõrane, located
downstream of the
Vajont dam. b e b e tlie
Mount Toi failure in
October 1963.

Figure 1.ld: The

remam of the town ai
Longarone after the
fload cauied by the
Overtopping of the
Vajont dam as a resut
of the Mount Tac
failure. More than 2000
persons weie killed in
this flood.
 neering Course notes I3y Evert Hoj

In 1960 a coa1 mine at Coalbrook in South Africa collapsed with the loss of 432 lives. This eveiit was
responsible for the initiation of ai1 intensive research prograinme which resulted in major advances i11 rlie
inethods used for designing coa1 pillars (Salamon and Munro (1967)).
The formal developrnent of rock mechaiiics as ali engineering discipline iri its own right dates froin this
period in the early 1960s and I will attempt to review these developineiits in the following cliapters of these
notes. I consider myself extremely fortunate to have been intimately involved in the subject since 1958. I have
also been fortuiiate to have been in positions which required extensive trave1 and wliich have hrought me into
personal coiitact with rnost ofthe persons with whom the developrnent of modem rode inechanics is associated.

1.2 Rockbursts and elastic theory

Rodcbursts are explosive failures of rocli which occur when very high stress coiicentrations are in-
duced around underground openiiigs. The problein is particularly acute in deep level inining i11 hard brittle
rock. Figure 1.2 shows the dainage resulting from a roclcburst i11 an underground mine. The deep level gold
mines in the Witwatersrand area in South Africa, the Kolar gold mines iii India, the iiiclíel mines centred on
Sudbury in Canada, the mines i11 the Coeur d'Alene area in Idaho in the USA and the gold :nines in tlie
Kalgoorlie area in Australia, are amongst the mines which liave suffered froin roclcburst probleins.
As early as 1935 the deep level nickel mines iiear Sudbury were experiencing rocl<burstproblerns and a
report on these problems was prepared by Morrison in 1942. Morrisoii also worked on rodiburst probleins i11 t l ~ e
Kolar gold fields in Iildia and describes some of tliese problems in his boolc, APhiíosophy ofGround Control(l976).
Early work o11 roclcbursts in South African gold mines was reported by Gane et al (1946) aiid a
summary of rockburst researcli up to 1966 was presented by Cook et al(1966). Worli on tlie seisinic locatioii
of rocl<burstsby Cook (1963) resulted in a significant improvement of our understanding of the meclianics
of rockhiirsting and laid the foundations for the microseisinic monitoring systeins which are now coinrÍoi1
iri inines witli roclcburst probleins.

Figure 1.2: The reiuts

of a rockbuiit in an
urldrrgmund <nineiii
brittle rock subjerted to
very high stressei.
 A cbaracteristic of almost a11 ro&bursts is that they occur in highly stressed, brittle rode. Conse-
quently, tlie analysis of stresses induced around underground mining excavations, a key in the geiieration of
rockbursts, can be dealt with by means of the theory of elasticity. Much of the early work in roclc mechanics
applied to inining was focused o n the problem of rockbursts and this worlc is dominated by theoretical
solutions which assume isotropic elastic rock and which make n o provision for the role of structural
discontinuities. In the first edition ofJaeger and Cook's book Fundamcntak ofRock Mechanics (1969), men~ioii
ofstructural discontinuities occurs o n about a dozen ofthe 500 pages of the booli. This coiniiient does hot
iinply criticisin of this outstanding bòok but it illustratesthe doininance of elastic theory in tlie approach to
rocli mechanics associated with deep-leve1 mining problems. Boola by Coates (1966) aiid by Obert aiid
Duva11 (1967) reflect the sarne emphasis o n elastic theory.
This einphasis on the use of elastic theory for the study of rodc inechanics problems was particularly
strong in the English speaking world and it had both advantages and disadvantages. The disadvantage was that
it ignored the critica] role of structural features. The advantage was that the tremendous concentration of effort
on tliis approach resulted in advances which may not have occurred if the approach had been more general.
Many mines aqd large civil engineering projects have benefited from this early work in the application of
elastic theory and most of the modem underground excavation design methods have their origins iii this worlc.

1.3 Discontinuous rock masses

Stini was one of the pioneers of rocli mechanics in Europe and lie emphasised tlie iinportance of
structural discontinuities in controlling the hehaviour ofrock masses (Muller (1979)). Stini was invoived in a
wide range ofnear-surface civil engineering works and it is iiot surprising that his einphasis was oii tlie role of
discontinuities since this was obviously the dominant problem in ill his worlc. Similarly, the text book by
Talobre (1957), reflecting the French approach to rodi mechanics, recognised the role of structure to a inucli
greater extent than did the texts ofJaeger and Cook, Coates and Obert and Duvall.
A major impetus was given to this work by the Malpasset dam failure and the Vajotit disaster ineii-
tioned earlier. The outstanding worli by Londe and his co-workers in France (Londe (1965)), Londe et a1
(1969,1970)) and by Wittke (1965) and John (1968) in Germany laid the foundation for the three-diineii-
sional structural atialyses which we have availahle today. Figure 1.3 shows a wedge failure coiltrolled by two
intersecting structural features in the betich of an open pit mine.

1.4 'Engineering rock mechanics

Civil and mining eiigineers have been building stmctures o n or in rock for centuries and the prin-
ciples of engineering in rock have been understood for a long time. Rock mechanics is merely a formal
expression ofsome of these principies and it is only during the past few decades that the theory aiid practice
in this subject have come together in the discipline which we know today as rock inechanics. A particularly
important event in the developtnent of the subject was the merging of elastic theory, which domiiiated tlie
English language literature on the subject, with the discontinuum approach of the Europeans. Tlie gradual
recognition that rock could act both as an elastic material and a discontinuous mass resulted in a inuch rnore
rnature approach to the subject than had previously heen the case. At the same time, the subject borrowed
tecliniques for dealing with soft roda and clays from soil mechanics and recognised the importante of
viscoelastic and rheological behaviour in materials such as salt and potash.
I should poiiit out tliat significant work on rock mechanics was being carried out in countries sucli as
Russia, Japan and China during the 25 years covered by this review but, due to language differences, this
workwas almost unknown in the English language and European rocli inechanics centres aiid aiinost noiie of
it was incorporated into the literature produced by these centres.
Rock Engir urse notes t

Figure 1.3: A wedge

faiure contraled by
inteiiectina rtructural
features n the rork
mass farming the benih
of an open pit mine
 01 Develc ock englnec

1.5 Geological data collection

The corner-stone of any practical rock mechanics analysis is the geological data base upon wliich the
definition of rock types, stmctural discontinuities and material properties is based. Eveii tlie inost sopliisti-
cated analysis can become a ineaningless exercise if tlie geological information upoii wliich it is based is
inadequate or inaccurate.
Methods for the collection ofgeological data have not changed a great-deal over the past 25 years and
í e is still n o acceptable substitute for the field mappiiig and core logging. There have been some advaiices
tlie equipment used for such loggiiig and a typical exainple is the electro~iiccompass illustrated i11 Figure
. The emergente of geological engineering or engiileering geology as recognised uiiiversity degree counes
has been an iinportant step in the developmeilt o i rod< mechanics. These courses traiii geologists to be
specialists in the recognition and interpretation ofgeological information which is signiticant iii eiigiiieeritig
design. These geological eiigiueers, foilowing in the traditioii started by Stini i11 the 1920s, play an increas-
:ly important role iii modern rock engineering.
Once the geological data liave been collected, computer processing of this data cai1 he of great assis-
ice iii plotting the inforination and iii the interpretation of statistically significant trends. Figure 1.5 illus-
trates a plot of contoured pole concentrations and corresponding great circles produced by tlie program
DI PS1 developed at the Universiry ofToroiito.
Surface aiid down-hole geophysical tools and devices such as horehole caineras have been available
severa1 years and thèir reliahility and usefulness has gradually iinproved as electronic compoiients aiid
in:inuiàcturing techniques have beeii iinproved. However, current capital and opeiating costs of these tools
arc: high and these iãctors, togetlier with uncertainties associated with the interpretatioii of the infbrinatioii
ob taiiied iioin them, have terided to restrict their use in rock engineering. It is probable that the use of thesc
11s will becorne more widespread iii years to come as furtlier developineiits occur.

1.6 Laboratory testing of rock

There 1x1s always been a teiideiicy to equate rock mechanics witli laboratory testing of rock speciineiis
and hence lahoratory testiiig has played a disproportionately large role in the subject. Tliis does iiot iinply
that laboratory testing is not iinportant but I would suggest that only about 10 to 20 perceiit of a well
balanced rock ineclianics prograik should be allocated to laboratory testing.
Laboratory testiiig techniques have been borrowed froin civil and rnechaiiical eiigiiieeriiig and have
nained largely unaltered for tlie past 25 years. An exceptioii lias heeti the developiiietit of servo-controlled
ff testiiig inachines which pertnit the determination of the complete stress-straiii curve for rocl<s. Tliis
inforination is important in tlie design ofunderground excavatioiis since tlie properties of the iãiled rod<
surroundiilg the excavations have a significaiit iiifluence upoii the stability of tlie excavations.

1.7 Rock mass classif ícation

A major deficiency of laboratory testing of rode specimens is that the speciinens are liinited i11 size
d therefore represent a very sinall and highly selective sample of the rock inass from whicli they were
noved. In a typical engineering project, the sainples tested iii tlie laboratory represeiit only a very sinall
ction of one percent of the volume of tlie rod< inass. 111 addit?oii, siiice only tliose speciineiis which
rvive tlie collection and preparation process are tested, the resiilts of these tests represeiit a Iiighly biased
nple. How then cai1 these results be used to estimate tbe properties of tlie in situ roclc mass?
111 ali atteinpt to provide guidai~ceon the properties of rock inasses upon wliicli tlie selectioti of
~unilelsiipport systems can be based, a number of rodc inass classificatioii systems have beetl developed.

~ailablefiamRorsiencelnc..31 BalsamAve..Taronto,Ontaro.CanadaM4E385 tel: 1-416~698-8217fan. 1-416-698-0908ema11:io~wdreOroiiience

com
lering. Course notes b!

Figure 1.4: A Ciar

eiectronic geologia1
compais rnanufaitured
by FW. Breihapt n
Gerrnanv

Figure 1.5: Plot of

itiuctural features using
the program DIPS.
 Typical of these classifications are those puhlished by Bieniawslu (1973, 1974) and by Barton, Lien and
Lundi: (1974). These classifications include information on the strength of tlie iiitact rod< material, tlie spaciiig,
er and surface properties of the stmctural discontinuities as well as allowances for the influence of subsurface
dwater, in situ stresses and the orientation and inclination of doininant discontinuities. Figiire 1.6 repro-
a chart, published by Barton (1989), which can be used for selecting different types of support for uiider-
d excavations on the basis of the rodi inass classification publislied by Barton, Lien aiid Liiiide (1974).
These rock mass classificatioii systeins Iiave proved to be very useful practical eiigineering tools, not
only 1->ecausetliey provide a starting point for the design of tuniiel support but also because tliey force users
to examine the properties of the roclz mass iii a very systeinatic inanner. The engineering judgeinents which
can bt:inade as a result of tlie fainiiiarity and understanding gaiiied from this systeinatic study are probably
as useh11as any of tlie calculations associated with the classificatioii systems.

1.8 Rock mass strength

One of the major problems confronting desigiiers of engineering stmctures in rock is that of estiinat-
ing the strength of the rod< inass. This rock mass is usually inade up of an interlocliing inatrix of discrete
'.'--LI. These blocl<s may have been weathered to varying degrees aiid the contact surfaces between the
; may vary from cleaii and fresli to clay covered and slickensided.
Determination of the strength of an in situ rock inass by laboratory type testing is generally not practical.
: this strength mnst be estimated froin geological observatiotis and froin test results o n individual rode
pieces or roclz surfaces which have been removed from the rodi mass. This question has beeii discussed exten-
sively by Hoelz and Browti (1980) who used the results of theoretical (Hoek (1968)) and model studies (Brown
(1970), Ladanyi and Ardiainhault (1970)) and the limited arnount of available streiigth dava, to develop ai1
~iricalfailure criterion for jointed rodz masses. Hoelz (1983) also proposed tliat the rod< mass classificatioii
ems, described in the previous section of this paper, can be used for estiinating tlie rod< inass consiants
lired for this empirical failiire criterion. Practical applicatioii of this failure criterion in a nuinher of eiigi-
nee ring projects has shown that these estimates are reasonably good for disturbed rock inasses but that, in
tighitly interlo&ed undishlrbed rock inasses such as those whidi inay be eiicountered in tunnelling, tlie esti-
mal:ed strength values are too low. Further work is required to improve tlie Hoelz-Browii aiid otlier failure
critisia for jointed rock masses atid some ideas on this topic are discussed in a later chapter in thesc notcs.

Figure 1.6: Chart foi

pielimnaly seection of
rupport foi underqround
excavations on the baiis
"nneiling Quality
,x Q published by
i, L~enand Lunde
) The Excavatan
.upport Ratio ESR
depends upon the
iaton for which the
erground exravation
hai been deiiqned.
After Barton (1989).
 Rock Engii urse notes I

1.9 In situ stress measurements

Tlie stability of deep uiiderground excavations depends upon the strength of the rode mass surround-
ing the excavations aiid upoii the stresses induced in this rock. These induced stresses are a function of tlie
shape of the excavations aiid the in situ stresses which existed before the creatioii of tlie excavations. Tlie
inagiiitudes of pre-existiiig iii situ stresses have been found to vary widely, depending upoii the geological
Iiistoiy of tlie rock inass iii whicli they are measured (Hoelc and Brown (1980)). Theoretical predictioiis of
these stresses are coiisidered to be uiireliable and, hence, measurement of the actual in situ stresses is iieces-
sary for inajor undergrouiid excavation desigii. A phenoinenon which is frequeiitly obserued i11 inassive rock
subjected to Iiigli iii situ stresses is 'core disking', illustrated in Figure 1.7.
Duriiig early site investigatioiis, when no underground access is available, the only practical inethod for
ineasuriiig iii situ stresses is hydrofiacturiiig (Haimson (1978)) in which the hydraulic pressure required to open
existing craclcs is used to estiinate in situ stress levels. Once underground access is available, overcoriiig tecli-
niques for iu situ stress measurement (Leeinan and Hayes (1966), Worotniclci and Walton (1976)) cai1 he used
and, provided that sufficient care is talcen ~n executing the ineasureinents, the results are usually adequate for
desigii purposes. One of the instruinents used for in situ stress measurement is illustrated in Figure 1.8.

1.10 Groundwater problems

Tlie presence of large volumes of groundwater is an operatioiial problem in tunnelliiig but water
pressures are geiierally not too serious a problein in uiiderground excavation eiigineeriiig (Hoek (1982)).
Exceptioiis 3re pressure tuiiiiels associated with liydroelectric projects. In these cases, inadequate confiniiig
stresses clue to insufiicieiit depth ofburial of the tunnel can cause serious probleins in the tunnel and in tlie
adjacent slopes. The steel liniiigs for tliese tui~iielscan cost severa1 thousand dollars per inetre aiid are
frequeiitly a critical factor in the design of a hydroelectric project. Tlie installatioii of a steel tunnel liniiig is
illustrated i11 Figure 1.9.

Figure 1.7: Disking of a

150 rnm core a i grante
as a reiult of higli in i i u
stres5ès
 ick engineering

Groundwater pressures are a major factor in all s l n p stability problerns and an understanditlg of the
role of subsurface groundwater is an essential requirement for any meaningful slope design (Hoek and Bray
I, Brown (1982)).

While the actual distributions ofwater pressures in rock slopes are probably much more coinplex than
tlie sitnple distributions normally assumed i11 slope stability analyses (Freeze and Clierry (1979)), sensitivity
,iies
.
based upon these simple assuinptions are generally adequate for the design of drainage systeins
isur and Kaufman (1962)). Monitoriiig of groundwater pressures by rneans of piezoineters (Brown (1982))
ie rnost reliable means of establishing the input paraineters for these groundwater rnodels aiid iõr check-
upon tlie effectiveness of drainage rneasures
In the case of dams, forces generated by the water acting o n the upstream face of tlie dain and water
pre:isures generated in the foundations are critical in the assessment of the stability of t l ~ edarn. Estiinates ot
the water presstire distribution in the fo~indationsand of the influente of grout and drainage curtains upon
this distribution have to be made with cate since they have a significant impact upon the overail darn aild
found:ation design (Soos (1979)).
The major advances which have been rnade in the groundwater field during the past decade have beeii in
rnr uiiderstanding ofthe transport of pollutants by groundwater. Because of the urgency associated with nuclear
and toxic waste disposal in industrialised countries, there lias been a concentratioin ofresearcll efiòrt in tliis field
and advances have been irnpressive. The results of this researdi do not have a direct iinpact on conventioiial
M O technical engineering but there have been inany indirect benefits froin tlie development ofiiistruineiitation
'mputer software which can be applied to both waste disposal and geotechnical problerns.

.
-' j.8: A e l l for
iiurlng the in iitu
a1 itress iieid in a
ais. developed in
tralia (Woratnicki
.id Walton 19761.
 Rock Engiineering Course notes I

Figure 1.9: initalation

05 iteel lining in a
preiure tunnel n a
iiydroeletrn prolecr.

1.11 Rock reinforcement

Safety during construction and long term stability are factors which have to be considered hy the
desigilers of excavations in rod<.It is not unusual for tliese requirements to lead to a need for the installatioil
of soine forrn of rock support. Fortunately, practical developments i11 this field have been significant during
the pasi 25 years and today's rock engineer has a wide choice of support systems (Hoek and Brown (1980),
Hoek (1982), Farrner and Shelton (1980)).
In tunnellirlg, there is still an iinportant role for steel sets and concrete lining,in dealing with very
poor gronnd but, in slightly better giound, tlie use of coinbinations of rockbolts and shotcrete has becoine
very coinmon. The use of long untensioiied grouted cables in underground rnining (Clifford (1974), Fuller
(1983), HLIII~ arld Aslcew (1977), Brady and Brown (1985)) has been a particularly iinportant innovation
which has resulted in significant improvements in safety and mining costs in rnassive ore bodies. The lessoiis
learned froin these inining systems have been applied nith considerable success i11 civil eiigineering and tlie
use of unteiisioned dowels, installed as close as ~ossibleto tlie advancing face, has many advanrages in liigh
speed tunnel construction. Tlie use oiuiitensioned grouted cables or reinforcing hars lias also proved to be
a very effective aiid ecoiiomical technique in rock slope stabilisation. This reinforceinent is iiistalled progres-
sively as tlie slope is benched downward and it is vety effective in Icnitting tlie roclc inass togetlier aiid
preveiiting tlie initiation of ravelling.
The desigii of.rod< support systeins tends to be based upon einpiricai rules geiierated Srom experi-
eilce (Lang (1961), Endersbee aild Hofto (1963), Cording, Hendron and Deere (1971)) aiid curreritly avail-
able analytical models are not very reliable. Soine iiiteresting theoretical inodels, which provide a very
clear understaiidiiig of tlie inechanics of rock support iii tunnels, have been developed iii recent years
(Rabcewicz (1969), Daeinaii (19771, Brown et a1 (1983), Brown and Bray (1982)). Tliese rnodels h:ive to be
used with caution when designiiig actual tuiiiiel support since they are based upon very siinple assuinp-
tions and rock conditions underground inay vary from these assuinptions. Witli the developinent of
poweriul iluinerical models such as that descrihed by Lorig and Brady (1984), inore realistic and reliable
support desigi~swill eventually become possible but it will he severa1 years before tbese inodels cai1 be
iised as desigil tools.

1.12 Excavation methods in rock

As pointed out earlier, the strength of joiiited roclc masses is very dependeilt upon the iilterloclcing
betweeii iiidividual rock pieces. This interlocl<iiigis easiiy destroyed and careless blasting duriiig excavation
is oiie of the inost coininon causes of underground excavatioii instability. The followiiig quotatiori is takeii
froin a paper hy Holmberg and Persson (1980):
The innoient rock mass is ojcn blamedjar insufficierit stabili& thnt i1 actual!y the result ofrough nnd ciirelcss
binsting. Where noprecautions have bem takcn to avoid blasting damage, no kno7a/ed& g t h c real stabilily qfthc i~ndij-
tctrbed rock cax begainedjom looking nt the rcmaining roch zmiall What onc scer arc the sad reinnins ofzohaf cuuldhave
been a perfcciiy safc and stable rock&cc.
Techniques for controlling blast damage in rock are well-known (Svanholm et a1 (1977), Laiigefors and
Kihistroin (1963), Hagaii (1980)) but it is soinetimes difficult to persuade owners aiid coiltractors tliat tlie
application of these techniques is worthwhile. Experience i11 projects in wliicli carefully controlled blastiiig
has beeii used generally sliows that the amount ofsupport can be reduced sigiiificantly and tliat tlie overall
cost of excavation and support is lower tlian in the case of poorly blasted excavatioiis (Hoek (1982)). Ex-
ainples ofpoor aiid good quality hlasting in tunnels are illustrated iii Figures 1.10 and 1.11.
Machiiie excavatioii is a technique which causes very little disturbance to the rodc surrouiiding ali
underground eicavation. A wide range of tunnelliiig inadiines have heeii deveioped over the past 25 years
and these machines are now capable of worlcing in almost all rode types (Rohbins (1976), McFeat-Sinith
(1982)). Further developinentof these machines can be expected and it is probable that macliine excavation
will play a much inoreiinportant role in future tunnelling than it does today.

1.13 Analytical tools

Analytical inodels have always played an important role in rock inechanics. The earliest inodels date baclc to
closed forin solutions such as that fbr calculating the stresses surroundiiig a circular liole i11 a stressed plate
published by Kirsch i11 1898. The development of the compiiter iii the early 1960s inade possible tlie use of
iterative numerical techniques such as finite element (Clough (1960)), houndary eleinent (Croucli aiid Starfield
(1983)), discrete element (Cundall (1971)) and combinations of these inetliods (von Kiininelinaiin et al
(1984), Lorig and Brady (1984)). Tliese liave become almost universal tools in rock inechanics.
Rock Engineering. Course notes b,Y Evert Hoel

Figure 1.11: an
exarnple a i good
blaiting in a tunnel.

Figure 1.10: sn
exarnple a i poor
blaiting in a tunnel.
 Chapter í11 Develop

The computer lias also made it inuch more convenient to use powerful liinit equilibriuin ineiliods
i (1979), Brown and Ferguson (19791, Shi aiid Goodrnati (19811, Warburton (1981)) and prohabilistic
appro;iclies (McMahon (1971), Morriss and Stoter (1983), Priest and Brown (1982), Read aiid Lye (1983)) for
roclc iriechanics studies.
The recent adveiit of the micro-computei and the rapid developinents whicli liave taken place iii
..cnsive hardware have brought us close to the era of a coinputer o n every professioiial's desi<.Tlie power
iese macliines is transforming our approach to rock inechanics analysis since it is iiow l>ossible to per-
1 a large nuinber ofsensitivity or probabilistic studies in a fractioii of the tiiiie wliicli was required foi a
ie aiialysis a few years ago. Given the inherently inhoinogerieous nature of rock masses, sucli seiisitivity
studie:s enabie us to explore the intluence ofvariations in the valne of each itiput pararneter arid to b:ise our
engint:ering judgeinents upoii the rate of change in the calculated value rather than oii a single answer.

1.14 Conclusions
:r the past 25 years, rock mechanics has developed into a mature subject which is iiiuilt on a solid founda-
ofgeology and engineering mechanics. Individuais drawn froin inany different disciplines liave contrib-
LILC" ' -LI3 tliis subject aiid have developed a wide range of practical tools atid tecliniques. Tliere is still a great

deal o f roorn for developinent, iiiriovation and improvement in almost every aspect of tlie suliiject and it is
a field wliicli will continue to provide exciting challenges for inaiiy years to come.
 When is a rock engineering design acceptable

2.1 Introduction
When is a design in rocl<engineering acceptable? The aim of the following text' is to deinonstrate tliat
tliere are iio simple universal iules for acceptability nor are there standard factors of safety which can be used
to guarantee tliat a rock structure will be safe and that it will perform adequately. Each design is unique and
the acceptability of the structure has to be considered in terms of the particular set of circumstances, rock
types, design loads and end uses forwbich it is intended. The responsibility of the geotechnical engineer is to
find a safe and economical solution which is compatible with all the constraints which apply to the project.
Such a solution should he based upon engineering judgeinent guided by practical and theoretical studies
such as stability or deformation analyses, if and when these aiialyses are applicable.
Tables 1 to 4 suininarise some ofthe typical problems, critical parameters, analysis rnethods and accept-
ability criteria which apply to a number of different rodi engineering struchlres. These exainples have been
drawn iioin iny own consulting experience and I make no claiins that tliis is a coinplete list nor d o I expect
readers to agree with all oftlie i t e m which I have included under the various headings. The purpose of present-
ing these tables is to deinonstrate the diversity of problems and criteria whidi have to l-ie considered aiid to
emphasise tlie dangers of attempting to use standard factors of safety or other acceptability criteria.
In order to amplify soine of the items included in Tables 1 to 4, several case histories will be discussed
i11 terins of the factors which were considered and the acceptability criteria which were used.

2.2 Landslides in reservoirs

The presence ofunstahle slopes in reservoirs is a major concern for tlie designers of dains for hydro-
electric and irrigation projects. The Vajont iàilure in 1963 alerted the engineering community of the danger
of uiiderestimating the potential for the inobilisation of existing landslides as a result of submergence of the
slide toe during impounding of the reservoir.
During the construction of the Mica and Revelstolie dams o n the Columbia River in British Coluin-
bia, Canada, several potential slides were investigated. Two of these, the Dowiiie Slide, a 1.4 billion cubic
inetre ancient rodi slide, and Dutchman's Ridge, a 115 million cubic rnetre potential rocli slide, were given
special atteiltion because of the serious consequences which could have resulted from failure ofthese slides
(Imrie (1983), Lewis and Moore (1989), Iinrie, Moore and Enegren (1992)).
n i e Downie Slide and Dutcliinan's Ridge are located in steep, narrow, V-shaped sections ofthe Colurnbia
River valley which has been subjected to several episodes of glaciation. The bedrodc at these sites consists inaiiily
of Pre-Cambrian para-gneisses and scbists within or on tbe fringe ofthe Shuswap Metamorpliic Coinplex. In both
cases, the potential slide planes, determiiied by diamond drilling aiid slope displacement moiiitoring, are rela-
thely flat-lying outward-dippiug tectonic faults or shears which daylight i11 tlie hase ofthe river valley.

'Based upon the text ai the Muller lecture presented at the 7th Congress of the International Saciety for Rock Mechanirr held in Aachen, Gerrnany in
Septernber 1991
 'hen 1s a roc k engineerin g design ,icceptable
ipter 02 ~ni
.
P
n
,5
+
'i:
 m e notes
o * z :
-
m
C .g . z' 3 2
r
6
-&-oE
S < q
r + . '2 b * - c6 .=
< e2 =El -* k
2
= :L:
o
n ,c :. 4
"& r
u2 .F o
. ,
j :r: v v 1; .g
3 m
m c .-o 5 z
.;
E*
-
,.,.. -e
;L ~ . o < - r - "5 Z j
s 8
0 -
'I ;,.ZI°E;
.=o
a
b
x
y

E
'-
o,;
1=
LdY*n$ " "c-.
*u ?. ;C<. r-
.
. .e .;$a;-
9
.ta-:
<
.o
..
.-
O *;"r
2
Fz
'r;I:n,Ém z O ~ i
E=e:; 4 r
5
t
bgEri,,-
2$$,g2*2
Lo ~ ~< Eo P. E ~-.6o:.~Qc-Ego :õ
h,,-
pa :zcb
2z j 9 5 , 0 T
-.cm o Z.5
w :*%E:$:;,:
s - -. . .
y E Q . 2 u u ç u * . i s u u
á2"6.?'.* 5<ègfdZ 1a:!=..=.
-c-Tz
. . . . . . . . . . . . . .z;$;$.;
ài 2. z~E~g .z .. ;~$ wài?::
~. u~ i~i 23 . :Zz ~ ~.i iE Z O
. ~ Y ~ Ô Lz EOP = . W O
. . .
6zg?g%
 Ipoqlaw
' k a ' a u 4! =iemdo Fna!'Zaia!"'ad 'nrw 'aBr8ea1 on!i>eo!pei 'lerodrip airem
!-,a,""" a"!." piel",le> a'? 31"s qzqr
01 se& 0s L l s i r w v i d d e i04 a l q s r u!eulii a,nal) paaeuiep 1301 "4: u! u o ! i q ! ~ i q p iaiempunoi) i=a~2nu punoislapun
p, I,, aiia io&!l!qeqwd PYF
Z9" ""PU 'aF!un PuC SIauun: 'n)c9S ' s u a ! l ~ f i e ~ =a q l au!punol~ns
s t ~ qqanwqi ~l,qi">!ued 'ra!ipolan pue &q!l!ud
.i pxeai,u!
.IF!'3"" an!i>eop=i 40 u a d r u e n I!W!I o? rui.ued wl, ialempunoi3 .sem ~ U I u! ==im lewlaiilpue n l ! ~u l i .! B ~ sua!lene,x.! ~ ~ ~ ~ ~
iapio u! wle auauu!einm ia:r!ue> qszm o,4, leuiiaqr kq pue uq.ninn ' s e u 7302 aqi u! saina %I ~ u ! q,o, ~ ~ ~ ~ , ~ ~ ~
a q i q B n o r q ? ~ ~ a ~ a ~ ~ i w e ~ p up3,npu! ~ ~ B ~uriu.e,.dilsp!o ~ a ~ ~ ~ puE saisais l.ai -eaJ 1ewn11r JO q @ u a ~ i rJ-US pue iyi )O au!lled. p,npu!
~ o l L ~ s u a ~ w s i ! un ab! ni p e ~ q e i d = ~UV
~= oI pmn .ie w~lrut el ,!u
i ni N h!l!qea~'ad 'uo:*u!~u! ' u o l e l u a ! ~ ~ . illeuiaqi n/puc nasi,
T Y ~ P ! P J ~ uB!ap ~ l ~ ! > u o z 'nllem pus 4eai uiane, a q i io)
'8u!%selq pus Bu!ll!ip 40 &!lenb pazup!
.1x?>
o i io!iuara s! nu*wa,cldr!p 40 Bu"ox!uo~ nuauial!nbai uoddns mleui!va g pmn aie - s r r a i i r na!$ u! j o ""1
pue sxLleue le>!iaunu ~q pu!uiiaiap ai= PYe''Ya"s uoddnr Pue uo!ieneix3 . ap~"z!uaeuipue 5ain2e.ij '!
.az!I!qeir
s e m q > o ~ a q u!
1 s u a u a x l d r ! p aqa i e q i pue uo!iennxa uiane, 4.3 a B r p q x s Ag paxpu! 7 = ! 1 i ! I l~,nr,nm )o B"!.F~s
parsinianorou r! uoddnr i q 1 aeq1 'uoddn. nuaui=e~dr!ppue - 3 i i ~ 'neui poi aqa 1301 aq: uodn Bu!puadap ' s e l u
psiiosu! 4 ~ l o ~ a u ou w z q SE^ ainl!e~40 u! r88p.u ~e!luazod Ile I a uo!iez!gms!fi p v r "! D1Wze4 l w u ~ n i i s 4 0q l s u a ~ l sJ e a v 1302 j o a n l ! q A E ~ pSu r
i u a w aqi i e q i +rn!pu! s!apoiu ln!iawnu ~o!iew!ui~aup a q i 204 p m "e ~ p o q q a uI.,! Pu' UO!IFU!I3*! 'uo!ael~*!~o04 UO P
+I ~I!IY~ rn aap.u~u!p!lr
-0leuz i- anb!vqw& umua[aid ~ p ! ~ a q d ~u! 40 uoraiua!Jo Pua i
uaqm p q p e ! u~!ap e1q4da,,e ni au!llr, uqip Q A U ~
'@"lF" 'fu!! 40 .! BY!IoI!"ou< a".Wa3Fld.!0 nususi!nbai 'ruis:rKs IiOdd"9
'IF!I~~EW 3 3 9
'730, p>"!o[
.uqgnnxl w d d n r p u t &4.>rr 4o i o ~ ? q'ainl!e4 40 40 s?uanbar ua!lsllauu! puç &pede3 .
4 a w aq1 "o -pnir =!iiawcied 10) pxn sir 4"s pawddns Llairnb "! qaYuna nalleils
i
iq p x o d n 4.11") ai* ~ a qwojsq puoddnr .ua!linnn
pue payauap! eq plnoqr y ~ l n q asprn a a p w ~n!i!i=)o d l c u e wn!q!~!nba 1!u>!1 Bu!rnp Bu!ire~qpur au!il!ip 40 13!lt"b +Peu! $0 au!llaAuun
pue l=?a!i=s! azurnbm uyvlleasu! w d d n s 'laUUna s q i Bu!punaiinr nriu y=oi 'uo!irnnxa p ua!iiiua!io pue adeqs .. ~ l n x ~( rai nj u n i i r
. q q q pur n s p a m ¶u!lle) Oz '91 u! 9%- le!iua:od I!. 40 u?l=z!l?nr!n .nciu7,0, 8 U ! v ~ a l u ! 4 pul)ap
PYI Su!p!p 20) 47 pa>nplnoqs ,auauia>m4 Pue uYoiEu!uilaaaP i q 1 i04 ~n ai= rpoqlaui aqi u! n i n ~ t a )leinun,~. 40 qouaiar T 0 ! 4 x> a"!P!ls
~ ~ ~pur
- u ! i i ~ o r u 3 ~ a q i a u ! p n p u ! ~ & a ~ . >I ~r r~ ~! ~Y~(~I ~O~ Y~ L ~ O Y ~ ~ ! Y ~ X ~ Y O ! ~ieaqr ! A ~ ~ ~ S.ua!i.iua!io.
I F ~uqiru!pu! m B u ! ~ ~ =uia A ! i p k ! ~ t ~ 9
'iolxvm YO!U~IZ~YO~ j o v d ~ c SE" PDA aqi
iuzvodw! ue r! ruo!ieuilo43p 40 au!101!uoN '9uevKr paddnr 'ria""", 7301 1105
u! s u i m a x l d r ! ~a i P i u ! w 4 pue uoddns ! uod
' S I E ~ ~ ~ P " r!
,e "o!,eI puc i
'uqisuiiopp Iuapuadapui!i lo Bu!!l=mr4o i04 amanhas uo!lellsiru! ptie k p e d e ? +u!w
.a=u3nbar pup pq.u
.i uolenmq =
-d"4! a ~ n r o ~ao-,**
p n a i e se l s u u s aq140 ainrap 104 p u S ! a p -iaiap o i rpoqiau ~ n ! ~ a u i niau w i o j p m q a
.
-nl,o, LF~Y.W! Bu!zwnk 'BU!II~MS
i q irnu n i n v n i a r ~suiazu!p u i rau!q>eui Bqsn r a í l r u e UODFI~IY! uoddns-yx8
-P*JO 4"I"'!~~d 'l?!~aiod Bu!ll=s smaaas psmpu!
su!llauunl .p?al a(q~ida,>r ue e? ainrq, y2ai aqi u! riuauia2rldr!p aiqr .
.ninieaj f e i n u n l s lenp!n 4 P+W>n r! q@ua'Jr
1!"!10i P V F n E u 1301 s q l aZ!l!qeE 01 1 U V q w d pue S ~ Y O Zainl!q )o i u a w av!wlaiap
-!ynraq plnoqruoddn5pallqru!~o&pedaj o i rpoqiaui lm!,awnu B U ! B.K~EYF ~ -!P"! J0
Pue s e u 4301 40 qOua11s a'n1!e4 1'08
'ruo!ldunne u B p p sq> y r q , a i lo!iuana -rq@ual Bu!u!l ~ w i au!uriaiap
r o i sinmid . r m b d imd+9,pKq
r! Bu!u!~ IWV ~i40 v u a PICI~~I~> aql ~!I~DIPA~ =!LU~YLP UInUI!~.Pu U. -313. "! ,.iemp"noi9 . -=n-Jd "1 qauun1 a i n r n i d
ir ~ l y U O qU! BY!M a~n-d ,!lneipL~ l r d ~ u ! ~ WIIY!Y!Y
d uaaqaq Y ~ ! ~ z ~ w o ~ B.~!$
.,", ieuraw ueleui~p
' m r a ~ =d ! u i ~ d pMI h n q i ! IYO!I
~ iauuni m
n,,u.,A
*i! pu~~u!u!l . xaol O: a n au!u!~ ~ laair
. e l + d ~1-4 $1.1 10 IYO!lel>dO 3!117301p(q 0013. PYO B Y O /SYO!Wal)O
~ l a y ( l ~ u em l l Ç .1," su!puno'in. aqa "! 4O B"!lvnq ='"xdn8
~ p d 104 h p q =!IL~ u n u ! x e u sup '!qdeiBodol 'rlauunl
, ? r ledpu!id uinu!u!w o i lpuuni u!
r1 uzq1 ral r! i(= aq1 u! salas l=d!=u"d =iein>>e u o i j =moi !auunl alnrraid Buale ,!lncipAq iunui!xew . p u ! l i i i i = u m i o psu!lun
unu!u!w s q l m q m p i ! n b > r! Bu!u!l 1)9s rqidap i a m uinui!u!u j a uo!)eu!ui.i+<] ~ 0 1 )i a q e a ! an!ra>x3
 iapter 02 liYhen 1s a rcick engineering design acceptable

Based on thorough investigation and monitoring programs, British Coluinbia EIydro aiid Power ALI-
thority (BC Hydro) decided that remedial measures had to be talcen to iinprove the srability of botli the
Downie Slide and Dutchman's Ridge. These remedial measures consisted of-draiiiage diailageadits exteiiding withiii
andlor behiild the failiire surfaces and supplemented by drainholes drilled from chambers excavated along
the adits. Worl< o n the Downie Slide was carried out in the period 1977 to 1982 (which included a 3 year
observatioti period) and work o n Dutchman's Ridge was carried out from 1986 to 1988.
A section tlirough Dutchrnan's Ridge is given in Figure 2.1 and this shows the water ievels in the slope
before reservoir filling and after reservoir filling and the construction of the drainage systein. Figure 2.2
shows contours of reduction in water Levels as a resiilt of the installatiotl of the drainage systein w11icl-i'
consisted of 872 m of adit and 12,000 in of drainhole drilling. Note that the drawdowil area oii tlie right
harid side of the poteritial slide was achieved by long boreholes froin the end of the draiiiage adit braiicli.
Comparative studies of the stability of the slope section showii in Figure 2.1, based upon a fictor ofsakty
of1.00 for the slope after reservoir filling but before implementation of the draiiiage systein, gave a fictor of'safeiy
of 1.06 for the drained slope. This 6010 improvement in factor of safety may iiot seein veiy significant to thc
designes of srnall scale rodc and soil slopes but it was considered acceptabie in this case for a nuinber of reasoiis:

The factor of safety of 1.00 calculated for the undrained slope is based upon a 'baclc-analysis' of observed
slope behaviour. Provided that the same method of analysis and shear strength parameters are iised for the
stability analysis ofthe same slope with different groundwater conditioils, the ratio of the factors of safety is
a very reliahle indicator of the change in slope stability, even ifthe absolute values of the factor ofsafety are
not accurate. Consequently, the degree of uncertaiiity, wliich has to be allowed for in slope designs where
no baclc-analyses have beeil perforined, can be eliininated and a lower factor of safety accepted.
The groundwater Irvels in the slope were reduced by draiiiage to lower tlian tlie pre-reservoir coiiditions
and the stability o f t h e slope is at least as good if not better then these pre-reseivoir coiiditions. Tliis
particular slope is considered to have withstood severa1 significant earthquakes duritlg the 10,000 years
since the last episode of glaciation which is rcsponsible for tlie preseiit valley shape.

. iz00

'.
ntial riide boundary

Water leve1 before 800

rerewoir filling

- 600
Water leve1 after
rercrvoir filling
and drainage L-

2.1: 5eition through Dutihrnan'i Rdge showing potental Figure 2.2: Contouri at watei leve ieduction (in melrei) as a
rface and water leveli before and after drainage. iesult a i the implementation of drainage in Dvtihrnan's Rdge
22 Rock Engineering

3. Possibly the inost significant indicator of an improvement in stability, for botli the Downie Slide and
Dutchman's Ridge, has been a significant reduction in the rate of down-slope movement wliicli has been
inonitored for the past 25 years. In the case of the Downie Slide, this inovement has practically ceased. At
Dutchinan's Ridge, the moveinents are significaiitly slower and it is anticipated that they will stahilize
when the drainage system has been i11 operatioii for a few.more years.

2.3 Deformaton of rock slopes

In a slope i11 which the rock is jointed but where there are no significant discontinuities dipping out of
the slope which could cause sliding, deformation and failure of the slope is controlled by a coinplex process
of hlocl< rotatioil, tilting and sliding. In an extreine case, where the rod< mass consists of near vertical joiilts
separating columns of massive rock, toppling inovement and failure may occur.
Figure 2.3 is a section through part of tlie power tunnel for tlie Wahleach liydroelectric project in
British Colurnbia, Canada. A break in tlie steel liiiing in this power tuiinel occurred in Januaiy 1989 and it is
thouglit tliis break was caused by a slow down-slope gravitational inoveinent caused by blocl< rotations
withiil a iiear-s~irfacezone of loosened jointed rode.
The Wahleach project is located 120 kin east ofVancouver and power is generated froin 620 in of Iiead
betweeii Wahleacli Lake and a surf~~ce powerhouse located adjacent to the Fraser River. Water flows through
a 3500 in loiig three inetre diameter uiilined upper tunnel, a rodc trap, a 600 in two metre diameter coilcrete
encased steel lined sliaft incliiied at 48" to the horizontal, a 300 in long lower tunnel and a 485 in long
surface penstodc to tlie powerliouse.
The tuiitiels were excavated mainly in granodiorite which varies from highly fractured and inoderately
weatliered in the upper portioiis of the slope to moderately fractured and fresli iii botli the lower portioiis of
the slope aiid helow tlie liighly fractured mass. Two main joint sets occur in the rock inass, oile set strikiiig
parallel to tlie slope and the other perpendicular to it. Botli dip very steeply. Average joint spacings range
f'rorn 0.5 to 1 in. A few joiiits occur sub-parallel to the ground suriàce and these joints are inost well devel-
oped in the gronnd surfice adjacent to tlie incliiied sliaft. Thorongh iinvestigations failed to reveal aiiy
significant shear zoi~esor faults orieiited i11 a direction conducive to sliding.

.r and
Zone o f fractured

' .
relaxed rack

-, / Surne shaft
Steel lining break
Buckles in
steel Iiner

- Access adit

Repiacernent conduit
L
Valve house
 Tlie toe of tlie slope is buried heneath colluvial and fan deposits from hvo creeks which have incised
tlie Fraser Valley slope to form tlie prominence in wliich the inclined shaft was excavated. Tliis protninence
is crossed by several linear troughs which trend aloiig the ground surface coiitours aiid are evidetice of
previous down-slope inoveinent of the proininence. Mature trees growing iii these troughs itidicate a history
of inoveineiit of at least several Iiuiidred years (Moore, Imrie and Balier (1991)).
Tlie water coiidnit operated without incident between the initial filliiig iii 1952 and May 1981
when lealcage was first noted f r o m t h e upper access adit located near the intersectioii o f the inclined shaft
and tlie upper tuniiel (see Figure 2.3). This leakage stopped when two drain pipes einbedded i11 the coii-
crete baclifill beneath tlie steel lining were plugged at their upstream ends. Large holes had been eroded i11
tliese drainage pipes where they were not encased in concrete and it was concluded that this corrosion was
responsible for tlie lealiage. This conclusioil appeared to be valid utitil 25 January, 1989 whei? a mucli
larger water flow occitrred.
Investigations in the dewatered t~iniielrevealed a 150 inin wide circuinferential teiisioii crack iii tlie
I liuing of the upper tunnel, about 55 in from its iiitersection with the iiiclined sliaft. In additioti, eight
~pressioiialhuckle zones were found in the upper portion of tlie incliiied shaft. Subsequetit investiga-
..AS revealed that approximately 20 inillion cubic inetres of rocli are involved i11 dowii-slope creep wliicli,
duriiig 1989-90, ainounted to several centiinetres per year and which appears to be ongoing. This dowil-slope
creep appean to he related to a process of hloclirotation rather than to any deep seated sliding as was thc casc
at hotli tlie Downie Slide and Dutchinan's Ridge.
While discrete eleinent tnodels inay give some indication of the overall inechanics of this type of
slope deforination, tliere is n o way in which a factor of safety, equivalem to that for sliding fdilure, can be
calculated. Consequently, iii decidiiig upon the remedial measures to be iinpleineiited, other factors Iiave to
Lie taken iiito coiisideration.
After tliorougli study hy the BC Hydro and their coiisultants, it was decided to coiistruct a replace-
tnent conduit consisting of an uiiliiied sliaft aiid tuiit~elsection and a steel lined section where tlie rock covcr
is insufficient to contain the interna1 pressure in the tunnel. Tliis replacernent conduit, illustrated iii Figure
2.3, will reinove tlie steel iined portiotis o f t h e system f?om zones in whicli large displaceineiits are likely to
occur in the future. This in turn will ininimise the risk of a rupture oftlie steel lining which rvo~iidiiiject higli
pressure water into the slope. It was agreed that such high pressure water lealiage could be a cause for
iiistability of tlie overall slope. Further studies are beiiig uiidertalien to deterinine whether additioilal draiii-
age is required iii order to provide further safeguards.
Caref~iltneasurernents o f the displaceineiits in the inclined shaft, the lengtli of tlie steel lining cans as
npared with the original specified lengths and the opening of the tensile crad<in tlie upper portioii of tlie
:I iined tunnel, provided an overall picture of tlie displacements iii the rodi inass. n i e s e observed dis-
:etnents were coinpared with displacement patteriis coinputed by ineans o f a nutnber of n~iinericalstud-
usiiig both contiiluurn and discrete elerneiit inodels and the results of tliese studies were used iii decidiiig
upon the locatioti of the replaceineiit conduit.
In addition to tlie construction of tliis replacement conduit to re-route tlie water away iiom tlie upper
aiid potentially unstahle part of the slope, a comprehensive displacement atid water pressure inonitoriiig
:em lias heeii iiistalled aiid is being inonitored by BC Hydro (Baker (1991), Tatcliell (1991)).

2.4 Structural failures in rock masses

111 slopes, foundations and shallow underground excavations in hard rodi, failure is frequeiitly con-
trolled by the presence of discoiitinuities such ast'aults, shear zones, bedding planes aiid joints. Tlie intersec-
tion of these structural features can release bloclis or wedges which can fall o r slide froin the surface of the
avation. Failure o f t h e iiltact rocli is seldorn a problem iii tliese cases where deformatioii and fiilure are
 eering Coi

caused by sliding along individual discontinuity surfaces or along liiies ofiiltenection ofsuriàces. Separation
of planes and rotation of blocks and wedges can also play a role in the deformation and failure process.
An aiialysis ofthe stability ofthese excavations depends priinarily upoii a correct interpretation of the
structural geological conditions in the rock mass followed by a study of the blod<s and wedges which cai1 be
released by the creatioil of the excavation. Identification and visualisatioil of these blocl<sand wedges is by f i r
the most important part of this analysis. Analysis of tlie stability of the bloclc and wedges, and ofthe reinforc-
ing forces required to stabilize thein, is a relatively simple process once this identification lias been carried out.
The Rio Grande Pumped Storage Project is located in the Province of Córdoba in the Republic of
Argentina. Four reversible pump-turbines operating at an average head of 170 m give the project a total
iiistalled capacity of 750 MW. These turbines are installed in a 25 m span, 50 in high, 105 in long cavern at
an average depth of 160 in .
The rod< iii wliich the utiderground excavations are situated is a inassive tonalitic gneiss of excellent
quality (Ainos et a1 (1981)). The gneiss lias an average uniawial compressive streiigth of 140 MPa. The inaxi-
muin principal stress, determined by overcoring tests, is 9.4 MPa and is almost horizontal and orierited
approxiinately norinal to the cavern axis. In massive rocks, this 15:l ratio of uniaxial strength to inaxiinuin
principal stress is unlikely to result in any significant failure in the rocl<and this was coiifirmed by tluilierical
stress analyses (Moretto (1982), Sarra Pistone and de1 Rio (1982)). The principal type ofinstabilitywhich had
to be dealt with in the underground excavations was that of potentially unstable blocbs and wedges defined
by intersecting structural features (Haminett and Hoek, (1981)). In one sectioii of the cavern, the axis of
whicli is oriented i11 tlie direction 158-338, four joint sets were mapped and were found to have tlie followiiig
dip/dip direction values:

Table 5. Dip and dip direction values for joints in one location in the Rio Grande cavern
N. Dip Dip dir. Comments
131 ~nfrequentlyoccurng joints
264 shear joint set
226 shear jont set
4 50 345 tension joint set

Figure 2.4: Penpeitive view a i Rio Grande power cavern showing potentally unitable wedges in the roof, sidewalls. bench and floor.
 acceptable

Figure 2.4 is a perspective view of the Río Grande power cavern showing typical wedges which cai1 be
ined in tlie roof, sidewalls, bench and floor by joint sets 2 , 3 and 4. These figures represent tlie maxiinum
,,;sible sizes ofwedges which cai? be formed and, during construction, the sizes of the wedges were scaled
down iii accordance with average joint trace lengths measured in the excavation faces. In Fig~ire2.4 it is
evident that the roof and the two sidewall wedges were potentially unstable and that they iieeded to be
)-'-5ilised.
L*, This stabilisation was achieved by the placement of tensioned and grouted rockbolts which were
ins talled at each stage of the cavern excavation. Decisions o n the number, length aiid capacity of the rodibolts
wei.e made by on-site geotechnical staff using limit equilibriuin calculations based upon the voluine ofthe
weijges defined by the ineasured trace lengths. For those wedges which involved sliding on one plane or
aiong the line of intersection of two planes, rockbolts were installed across these planes to briiig the sliding
factor ofsafety of the wedge up to 1.5. For wedges which were free to fall frotn the roof, a factor ofsafety of
2 was used. This factor was calculated as the ratio of the total capacity of the holts to tlie weight of tlie wedge
ali<i was intended to account for uncertaiiities associated with the bolt installation.
The floor wedge was o f n o significaiice while tlie wedges in the bench at tlie base of tlie upstreain wall
wei.e stabilised by dowels placed in grout-filled vertical holes before excavation of the lower benches.
Early recognition of the potential instability problems, identification and visualizatioti of tlie wedges
whicli could be released and tlie installation of support at each stage of excavation, before the wedge bases
were fully exposed, resulted in a very effective stabilisation program. Apart from a miniinal amount of inesli
and shotcrete applied to areas of intense jointing, n o other support was used in the power cavern which Ilas
:rated without any signs of instability since its completion in 1982.

2.5 Excavations in weak rock-

In contrast to tlie structurally controlled failures iii strong rodc discussed in tlie previoiis section, tliere
are inaiiy cases where tunnels and caverns are excavated i11 rodi inasses which are wealc as a result of inteiisc
iointing or because the rock material itself has a low strength. Roclis such as sliales, inudstoi~es,siltstones,
~ hyllites
: aiid tuffs are iypical wealc roclcs in which even moderate i11situ stresses are likely to induce failure in
tlie ro& surrounding underground.excavations.
Progressive failure of this type, which can occur in the rodi surrounding an underground excavatioii
i weali rode inass, is a difficult analytical problem and tliere are no siinplenuinerical inodels nor iàctor of
:ty calculations which can be used to define acceptable liinits to this failure process. Judgeinent on tlie
:quacy of a support design has to be based upon an evaluation of a number of factors sucli as the magni-
le and distribution of deformations in the rock and the strisses induced in support eleiiieiits such as
6Luutedcables, steel sets or concrete linings. This design process is illustrated by means of ali exainple.
The Mingtan pumped storage project is located in the central region of the island ofTaiwan and
utilizes the 400 m head difference between the Sun Moon Lake and the Shuili River to generate up to I600
IVI W at times of pealc demand. The power cavern is 22 m wide, 46 in high aiid 158 in long and a parallel
I"'"'

transibrmer hall is 13 in wide, 20 in liigli and 17 m long. Tlie caverns are 45 in apart and :ire located at a
deptti of 30 in below surface in the steep left bank of the Shuili river (Liu, Clieng and Chang (1988)).
The rock mass consists of weathered, interbedded sandstones, siltstones and shales dippiilg at about
35" to the horizontal. The Rock Mass Ratings (RMR) (Rieniawski (1974)) aiid T~~niielling Quality Index Q
(Barton, Lien aiid Lunde (1974)) and approxiinate shear strength values for the various coinponeilts ofthe
rodz inass are given in Table 6 below.
Weali beds of siltstone, up to 2 m thick, appear to have ca~iseda concentration of shear inoveinents
du1ring tectoiiic activity so that fault zones have developed parallel to the bedding. The cominon feature
ob:ierved for all these faults is the presence of continuous clay filling with a thidiness varying froin a few mm
'"0 inni. The cavern axis is intentionally oriented at right angles to the strilce of these faults.
 eering. Coiirse notes t

Table 6. Rock mas8 classifications and approximate friction angles 4and cohesive strengths c f o r the rock
mass in which the Minqtan
. power cavern is excavated
Rock type RMR 4 4' degrees c' MPa
Jointed sandstone 63-75 12-39 50 1 .O
Bedded sandstone 56-60 7-31 45 0.8
Fautle or shears 10-33 0.1-1.1 30-40 0.15-0.3

Tlie tneasured in situ stresses iii the rodc mass surrounding the caverii are approxitnately

Maxitnuin principal stress (horizontal) , ~ =MPa

o , ~ ~ ,10.9
Miniinum principal stress (vertical) o,,~ =~7.5~ ,M Pa

The inaxitnum principal stress is oriented parallel to brdding and nortnal to tlle caverii axis.
The results of a simple analysis of tlie zones of oventress in the rock inass surroutiding the Miiigtati
underground excavations are presented in Figure 2.5. The crosses in this figure are directions aiong whicli
shear failure is predicted in the bedded sandstone while the horizontal lines indicate slip aloiig the faults
(assuming c = 0.15 MPa, $ = 30"). While it is not claimed that this analysis adequately represents the com-
plex process of M u r e itiitiatioti and propagation o n the inclined faults and in the jointed roclc mass sur-
rouiiding tlie caverns, it does give sufficient inforinatioii to suggest that tlie cavern can be stabilised by
meaiis of a pattern of long grouted cahles supplemented by a layer ofshotcrete o11 the excavatioii surfice.
The design of the reinforcing cal-iles was based upon experiente and precedent practice. Figures 2.6
aiid 2.7 give the letigths of roci<bolts and cables in the roof atid sidewalls of some typical large powerhouse
caverns i11 wealc roclc inasses. Plotted oti the same graphs are empirical relatioilships suggested by Barton
(1989) for bolt and cable lengths for underground powerhouses.

r (SIip bedding faults

""

(IFZJ~ 1I Power cavern

1 bedded sandstone

Figure 2.5: Predcted ronei of over-itreii aong faulti and in bedded ianditonesurrounding the power iavern and tianiformer hnll of the Mingtan Proleri
 hen 1s a roc k enqineeri icceptable

Bedding faults ofsigiiificant thickness which were intersected in the roof of the cavern were treated by
;high pressure water jets to remove theclay and then filling the cavities with non shrink ceinentitious
,,,,,Lar (Cheng (1987), Moy and Hoek (1989)). This was followed by the installation o f 5 0 toiine capacity
unteiisioned grouted cables fioin a drainage galiery 10 in above the cavern roof iri arder to create a pre-
reinforced rocli inass above the cavern. All of this worli was carried out froin constructioii adits before tlie
' I contract for the cavern excavatioil coininenced.
Duriiig benching down in the cavem, 112 tonne capacity teiisioned and grouted cables were iiistalied o11a 3
3 in grid in tlie sidewalls. Tlie final layout of the cables in the rock surrounding tlie power cavern and tlie
foriner hall is illustrated in Figure 2.8. Five metre long grouted rodíbolts were installed as required at the ceiitrc
of tlie squares formed hy the cahle face plates. A 50 mm layer of steel fibre reinforced inicro-silica shotcrete was
applied within 5 to 10 m of th&,.This shotcrete was later built up to a thidrness of150 inin on the roof aild upper
sidewalls and 50 mm on the lower sidewalls where itwould evenhlaiiy be incorpocited into the concrete iòundatioiis.
A l<eyeleinent in the decisioii malíing process on the adequacy ofthe support systein was a inonitor-
ind aiialysis process wliich involved the following steps:
Iisplaceinents iii the rocli surrounding the excavations monitored by means of coilvergence arrays aiid
:xtensoineters, soine of whicli had been installed froin constructioii galleries hefore excavatioii of tlie
:averns commenced.
Uumerical modelling of each excavation stage iising non-linear multiple-inaterial inodels. Tlie inaterial
?roperties used i11 the models of the early excavation stages were adjusted to ohtaiii the best inatch
between predicted and ineasured displacements.
Predictioii of displacements and support loads during future excavation stages and adjustineiit of support
capacity installation and pre-tensioning to control displacements and cable loads.
Measurement of displaceinents and cable loads (usiiig load cells on selected cables wliich had been de-
bonded) and comparison between measured and predicted displacements and cable loads.
Iiistallation ofadditional cables a r adjustment of cable loads to coiitrol unusual displacemeiits or sup-
port loads.

Waldeck I I .
'Okuyorhina

i"1
ò . Hongrin/
Minptan (beddcd randrtone)@
Nevada tert rite

- .
Rangipo (bah.)
Drakensbor~(boltr)
/
/
P o r t a ~ eMountain (boltr)
Poatina (balts)

2.6: Lengthi a i rockbolti and iablei uied for roaf

in some large caverns in weak rock. Equationi defin-
d n e s were iuggeited by Bartan (1989).
 eering Coi

The aim ofthis program was to maintain as uniform a displacement pattern around the excavations as
possible and to keep the loads on the cables at less than 45% of their yield load. The intermediate rodcbolts
and the shotcrete were not accounted for in the nuinerical modeliing siiice it was assumed that their role was
confined to supporting the rock iminediately adjacent to the excavations and that the overall stability was
controlled hy the 10 to 15 m long grouted cables.
Figure 2.8 shows the combination of materials used in analysing one section of the cavern, assuining
that the bedding faults could be represented by horizontal layers in the mo-dimensioilal model. In order to
inatch the measured and predicted displacements in the roclí mass, it was found that a 2.5 in thick zone of
softened and wealcened material had to be wrapped around the excavations to account for blast dainaged
material (achieving good blasting results was difficult in this interbedded rock).
In Figure 2.9, the predicted and measured displaceinents along six extensoineters installed in t l ~ e
power cavern sidewalls are compared. The overall agreement is considered to be acceptable. Maximuin
sidewall displacements were of the order of 100 mm at the mid-height of the upstream wall, adjacent to one
of the inajor faults. Elsewhere, displacements were of the order to 25 to 46 inin.
Figure 2.10 shows the results of monitoring at seven stations along the axis of tlie power caveril.
Before excavation of the cavern commenced, extensometers were installed at each of these stations from a
drainage gallery above the roof arch and from construction galleries as shown in the upper part of Figure
2.10. In addition, load cells were installed o n cables adjacent to soine o t t h e extensotneters.
Rapid responses were recorded in all extensometers and load cells as the top headiiig passed under-
neath them. Further responses occurred as the haunches of the cavern arch were excavated and as the first
bench was reinoved. As can be seen from the plots, after this rapid response to tlie initial excavation stages,
the displacetnents and cable loads became stahle and showed very little tendency to increase witli time. Tlie
difference in the magnitudes of the displacements and cable loads at different statioiis can be reiated to the
proximity of the inonitoring instrumeiits to faults in the rock above the caverii arch.

d Power cavern
I

Sandrtone 2 : E=6x1@ MPa, b=50'. c=l MPa

Figure 2.8: Layaut of cabes used ta iupport the rock surroundlng the
powei cavern and the transformei hal in the Mingtan pumped storage
praject. The loiation and propertiei of the roik vnts represent those
uied n the numerical analyisof faiure. deformation and iabie loadng
n a lypica vertcal ieiton.
Figure 2.9: Campariion between caculated and meaiured displace~
mentsalongiixexteniometeninitalled in theidewalliof theMingtan
power cavern.
 apter 02 Vi ing design acceptable

Drainage gallery

A Extensometer

Construction galleris

T o p heading
~ ~ H a u n c h e i Benching

Dec '87 Dec '88 Dec '89

re 2.10: Surface dsplacements and cable loadi measured at seuen stationi along lhe power cavern axii.

The rapid load acceptaiice and the modest loadiiig of'the cables together with tlie control of the
~lacementsin the rock mass were the goals o i the support design. Measureinents ohtaiiied froin the
msometers aiid cable load cells iiidicate that these goals liave. heen met.

2.6 Factor of safety

The four case histories, discussed in previous sections, have heen presetited to demonstrate h a t a
vaiiety of criteria have to be cotisidered in deciding upon the adequacy of a rode structure to perform its
design objectives. This is tme for any design in roclc since the performance of eacli stmcture will he uniquely
dependent upon tlie particular set of rock coiiditions, design loads and interided eiid use.
In one group of structures, traditional designs liave been hased upon a 'factor of safety' against slid-
ing. These stmctures, which include gravity and fill dams as well as rode and soil slopes, a11 involve the
potential for sliding along well defiiied failure surfaces. The factor of safety is defined as the factor hy wliich
tlie sliear strength parameters inay he reduced in arder to bring the slope (or dam foundation) iiito a state of
limiting equilibrium (Morgenstern (1991)). The numerical value of the factor ofsafety chosen for a particular
design depends upon the leve1 of confidence which the designer has in the shear streiigth parameters, tlie
groundwater pressures, the location of the critical failure surface and the magnitude of the extertial driving
fn-c acting upon the stmcture.
 Rock Engir irse notes b

Figure 2.11 illustrates a set of hypothetical distribution curves representing the degree of uiicertainty
associated with available inforination o n sliear strength parameters and disturbing stresses for different stages
in the desigii of a rock or soil stnicture. The factor of safety is defined as AIB where A is the ineaii of the
distribution of shear strength values and B is the mean of the distribution of driving stresses. For the purpose
of this discussion, the same factor of safety has been assumed for all tliree cases illustrated.
During preliminary design studies, the amount of information available is usually vely limited. Estiinates
of the shear strength of the rock or soil are generally based upon the judgement of an experienced engineer or
geologist which inay be supplemented, in some cases, by estimares based upon rocl<inass classifications or simple
index tesu. Siinilarly, the disturbing forces are not lmown with very inuch certainty since the location of the
critical failure surface will not Iiave been well defined and the magnitude ofexternally appiied loads inay not have
been established. In the case of dain design, the magnitude of the probable maximum flood, wliich is usually
based upon probabilistic analysis, frequently remains ill defined until very late in the desigii process.
For this case, the range of both available shear strength and disturbing stresses, which have to be
considered, is large. If too low a factor of safety is used, there may be a significant probability of failure,
represented by the section where the distribution curves overlap in Figure 2.11. Iii order to ininiinise this
failure probability, a high value for the factor of safety is sometimes used. For example, in the 1977 editioii of
tlie US Bureau of Reclamation Engineering Monograph on Design Criteria for Concrete Arch and Gravity
Dams, a factor of safety of 3.0 is recommended for normal loading conditions when 'only liinited inforina-
tioii is available oii the strength parameters'. This value can be reduced to 2.0 when the strength parameters
are 'deterinined by testing of core sainples frorn a field investigatioli program or by past experieiice'.
Duritig detailed design studies, the amount of iiiformation available is usually significantly greater
than in the preliiniiiary desigii stage discussed above. A coinprehensive program of site investigatioils aiid
laboratoly or in situ shear strength tests will normally have been carried out and tlie exteriial loads acting o11
the structure will have been better defined. In addition, studies of the grouiidwater flow and pressure distri-
butions in the roclc inass, together with inodifications of these distributioiis by grouting and draiiiage, will
usually have beeii carried out. Consequently, the ranges of shear strengtli and driving stress values, which
have to be considered in the design, are smaller and tlie distribution curves are more tiglitly constrained.

Value A

Figure 2.11: Hypatherical distribution curves iepresenting the degree o i uniertainty arsocated with information an driving irressei and iheai
itrengths at different itagei in the design o i a structure such a i a dam ioundation.
 i k enyineer iny desiyn ,icceptable

Tlie case histories of the Downie Slide and Dutchinan's Ridge, discussed earlier, are good exainples of
designs based upon back-analyses. In botli of these cases, very extensive site investigations and displaceinent
moiiitoriiig had established the location of the critical failure surfaces with a higli degree of certainiy Careful
monitoring of the groundwater iii the slopes (256 piezometer measuring points were installed i11 Dutchinan's
Ridge) Iiad defiiied the water pressures in the slopes and tlieir fluctuations over several years. Soine sliear
t-+;ng on fault inaterial recovered from cores was carried out hut, more iinportantly, tlie inobilized shear
igth along the potential failure surfaces was caiculated by back-analysis, assuiniiig a iàctorol-safety of
for existing conditions.
Figure 2.11 illustrates the hypothetical distrihution curves for the range of values for shear streiigth
aiid driving stresses for the case of a structure in which an existing failure has been carefully back-analyzed.
Depending upon the degree of care which has heen taken witli this baclc-analysis, these curves will he very
tightly constrained and a low factor of safety cai1 he used for the desigu of the reinedial works.
This discussion illustrates tlie point that different fictors of safety may be appropriate for differeiit stages
in the design of a rock stnicture. This difference is priinarily dependent upoii the leve1 of confidente which the
designer has in the values of shear strength to be iiicluded in the analysis. Hence, a critical questioii which arises
a\I of these cases is the determination or estimation of the shear strength along the potential sliding surface.
paper on the streiigth of rockfill inaterials, Maraclii, Chan aiid Seed (1972) suininarize this problein as
~ws:'No stability analysis, regardless of how intricate and theoretically exact it inay be, can be useiiil for
gii if an iiicorrect estimatioii of the shearing strength of tlie constnictioii tnaterial has been made'.
Except in siinple cases involving hoinogeneous soils or planar coiitinuous weak seains, deterininatioii
of tlie shear strength along potential sliding surfaces is a notoriously difficult probiem. This is particularly
true of tlie determination of the cohesive coinponent, c', o € the commonly used Mohr-Couloinb fàiliire
3n. Lahoratoiy test specimens tend to he too small to give representative results while in situ tests are
It aiid expensive and, uiiless carried out with very great care, are liable to give unreliable results.
For failure surfaces whidi involve sliding on rough a r undulating rock surfaces such as joints or bedding
, the metliodology proposed hy Barton (1976) is appropriate for estiinating tlie overall sliear streiigth of
tential sliding surface. This involves adding a ineasured a r estiinated roughness coinponent to tlie basic
lal strength which can he determined on sawn and polislied laboratory shear test specirnens.
For heavily jointed rode inasses in which there are iio dotninant weakness zoiies sucli as faults or shear
a crude estimate of the shear strength of the rock mass cai] be obtained by ineans of tlie use of rock
rnass classification systems as proposed by Hoelc and Brown (1988).
In all cases, a greater reliance can be placed upon the frictional componeiit, @, of the Molir-Couloinb
.,~- ~.
bIlCdi S.trength equation and extreme care has to be taken in the estiination of the cohesive strength, c'. Wliere
~

no reli able estimates of this value are available from carefully conducted shear tests or froin hacl<-analysisof
existin g failures, it is pmdent to assume a cohesive strength of zero for any stability analysis iiivolviiig
stnlctlires such as dam foundations.
In the design of fill and gravity dams there is a teiideiicy to move away from the high factors of safety
of 2 or 3 wliich have been used in the past, provided that care is talcen in choosing seiisihls conseivative sliear
strength paraineters, particularly for continuous wealc seams in the foundations. An t.......pIe of tlir range of
; ofsafety wliich can be used in the design of eartli a r roci<till dains is given in Tahle 7.

2.7 Probabilistic analyses

The uncertainty associated with the properties ofgeotechnical inaterials and the great care whicli 1x1s to
be takt:n in selecting appropriate values for analyses has prompted several authors to suggest that the traditional
deterniinistic methods ofslope stability analyses should be replaced hy probabilistic inethods (Priest and Brown
,McMahon (1975), Vaninarcke (1980), Morriss and Stoter (1983), Read and Lye (1983)).
 trse notes b

Table 7: Factors of safetv for different loadina conditions in the desian of earth and rockfill dams.
Loadina condition S.F. Remarks
End of construction porewater pressures 1.3
in the dam and undissipated porewater
pressures in the foundation. No reservoir loading.
Reservoir at full supply level with steady state 1.3 Possiby the most critical
seepage in the dam and undissipated (even if rare) condition.
end-of-construction porewater pressures
in the foundation.
Reservoir at full supply levei with 1.5 Critical t o design.
steady state seepage.
Reservoir at probable maximum flood level 1.2
with steady state seepage conditions.
Rapid reservoir drawdown from full supply 1.3 Not significant in design. Failures very
level to minimum s u ~ p l vlevei. rare and, if they occur, usually shallow.

One branch of rodi mechanics in which probabilistic analyses have been accepted for many years is that of
tlie design ofopen pit mine slopes. This is hecause open pit planners are familiar with the coiicepts ofrisli analysis
applied to ore grade and metal price fluctuations. Probabilistic methods are used in estiinating tlie economic
viability ofvarious options in ckveloping an open pit mine and so it is a small step to incorporate the prohability
of a geotechnical fdilure into the overall risk assessment of the mine. The mine planner lias the clioice of reduciiig
the probability of iàilure by the installation of reinforcement, reduciiig the angle of the slope or accepting tliat
failure will occur and providing for extra equipment which inay be needed to clean up the failure. Since the iniiie
is usually omied and operated by a single company and access to the mine benches is restricted to trained
personiiel, acceptiiig a risk of failure and dealing with the consequences on a routine basis is a viable option.
O n the otlier hand, the einotioiial impact of suggesting to tlie pnhlic tliat there is a fiilite risk of
failure attaclied to a dain design is such that it is difficult to suggest the replacement of the standard factor of
safety design approach with one which explicitly states a probability of failure o r a coefficient of reliability.
The current perception is that the factor of safety is more ineaningful than the probability offailure. Eveii if
this were not so, there is still the problein of deciding wliat probability of failure is acceptable for a rod<
structure to which the general public has access.
- In spite of these difticulties, there does appear to be a slow but steady trend in society to accept tlie
concepts of risli analysis more readily than has been the case in the past. The geotechnical coinmunity has an
obligation to take note of these developments and to encourage the teacliing and practical use of probabilis-
tic as well as deterininistic techiiiques with the aim ofremoving the cloak ofmystery which surrouiids the use
of these inethods.
Fortuiiately, tliere is a coinproinise solution which is a forin of risk analysis used intuitively by most
experienced engineen. This is a parametric analysis in which a wide range of possibilities are considered iii a
conventional deterininistic analysis in arder to gain a 'feel' for the sensitivity of the design. Hence, the factor of
safety for a slope would be calculated for botli filly draiiied and fully saturated groundwater conditions, for a
range of frictioii angles and cohesive strengths covering the full spectmin which could be anticipated for tlie
geological conditions existing on the site, for externa1 forces ranging froin zero to the inaximuin possihle for
that slope. The availability of user-friendly microcomputer software for most forms of limit equilibrium analy-
sis means that tliese parametric studies can be carried out quid<ly and easily for inost designs.
 Rock m a s s classification

3.1 lntroduction
During the feasibility and preliininaty design stages of a project, when vety little detailed inforination
.L- rodiinass and its stress and hydrologic characteristics is available, the use of a rode mass classificat'ion
"1, LUC

scheme can be of consideralle benefit. At its simplest, this inay involve using the classification sclieine as a
check-list to ensure that all relevaiit inforination has beeii considered. At tlie other end of the spectruin, one
01 1n011:rocli mass classification scheines can be used to huild u p a picture of the coinposition and character-
~sticsof a rock inass to provide initial estiinates of support requireinents, and to provide estiinates oftlie
strengtli and deforination properties of the roclz mass.
It is important to uiiderstand that the use of a rode mass classificatioil scheine does not (aud caiinot)
repiace some of the inore elaborate design procedures. However, the use of these design procedures requires access
to relatively detailed information on i11 situ stresses, rodi inass properties and planned excavation sequence, none
ofwhich may he available at an early stage in the project As this inforination becomes available, the use ofthe
m"l, <-, rss dassification schemes should be updated and ~isedin conjunction wiih Site specific analyses.

3.2 Engineering rock mass classification

Rock inass classification scheines have been developing for over 100 years since Ritter (1879) at-
teinpted to forinalise an empirical approach to tunnel design, in particular for deteriniiiing support require-
ments. Wliile the classification schemes are appropriate for their original applicatioii, especially if used within
the bounds of the case histories from which tliey were developed, considerable caution inust be exercised in
ying rock inass classifications to other rock engineering probleins.
Suininaries of some iinportaut classification systems are presented in this chapter, and althougli every
npt has been inade to present all of the pertinent data froin the original texts, there 2 e nuinerous notes
anu comments which cannot be included. The interested reader should make eveiy effort to rrad rhc cited
references for a full appreciation of the use, applicahility and limitations of each system.
Most of the multi-parameter classification schemes (Widcham et al (1972) Bieniawslzi (1973, 1989) and
)n et al (1974)) were developed from civil engineering case histories in which all of the components of tlie
enpiiieering geological character of the rock mass were included. In underground liard rock inining, Iiowevcr,
espei:ially at deep levels, rodz mass weathering and the influente ofwater usually are not significant aiid inay be
ignored. Diiferent classification systems place different emphases on the various paraineters, and it is rccoin-
.~,,..ded
,me,> that at least m o methods be used at any site during the early stages of a projcct.
 ieering Coi

3.2.1 Terzaghi's rock mass classification

Tlie earliest reference to the use of rodc mass classification for the design of tuniiel support is i11 a
paper by Terzaglii (1946) in which the rock loads, carried by steel sets, are estimated o n tlie basis of a descrip-
tive classificatioii. While n o usefiil purpose would be served by including details of Terzaghi's classificatioii
i11 this discussion o n the design of support, it is interesting to examine the rodc mass descriptions included i11
his original paper, because he draws attentioii to those characteristics that domiiiate rodc mass behaviour,
particuiarly iii situations where gravity coilstitutes the dominant driving force. The clear and concise defini-
tions aiid the practical comments included in these descriptions are good examples of the type of eiigineer-
ing geology informatiou, which is most useful for eugineering design.
Terzaglii's descriptions (quoted directly from his paper) are:

- Intact roclc contains iieither joints nor hair cradcs. Hence, if it breaks, it breaks across sound roclc. On
accouiit of the injury to the roclc due to blasting, spalls inay drop off the roof severa1 hours or days after
blasting. Tliis is lmown as a spalling condition. Hard, iiitact roclc may also be encountered i11 the popping
conditioii involving the spontaneous and violent detachmènt of rodc slabs froin the sides or roo6
Stratijied roclc consists of individual strata with little o r n o resistance against separatioii along the bound-
aries betweeri the strata. The strata may or may not be weakened by transverse joints. In sucli rock the
spalling condition is quite coinmon.
Moderatcly joiztcd rodc contains joints and hair cradcs, hut tlie bloclcs between joiiits are locally growii
together or so intimately interlodced that vertical walls do not require lateral support. Iii roclcs of this
type, both spalling and popping conditioiis may be encountered.
Blochy and scamy rodc consists of cliemically intact or alinost iiitact rock fiaginents wliich are eiitirely
separated koin eacli other and imperfectly interlocked. In such roclc, vertical walls inay require late-
ral support.
Crushcd but cliemically intact rock has the character of crusher run. If inost or all of the fragineiits are as
small as fiiie sand grains and n o recementation has taken place, crushed rodc below the water table
exhibits the properties of a water-hearing sand.
Squcezing rock slowly advances into the tunnel without perceptible volume increase. A prerequisite for
squeeze is a high percentage of rnicroscopic and sub-microscopic particles of micaceous ininerals or clay
rninerals witb a low swelling capacity.
Szoclling rodc advances into the tunnel chiefly on account of expansion. The capacity to swell seems to Iiie
limited to tliose roda that contaiii clay minerals such as montmorillonite, with a high swelling capacity.

3.2.2 Classifications involving stand-up time

Lauffer (1958) proposed that the stand-up time for an iinsupported span is related to the quality of the
rock mass in which the span is excavated. In a tunnel, the unsupported span is defined as the span of the
tunnel or the distatice hetween the face and the nearest support, if this is greater thaii the tunnel span.
Lauffer's original classification has since been modified by a number of authors, notably Pacher et a1 (1974),
and now forins part of the general tunnelling approach known as the New Austrian Tunnelling Method.
The significance of the staiid-up time concept is that an increase iii the span of the tunnel leads to a
significant reduction i11 the time available for the installation of support. For example, a small pilot tuniiel
may he successfully constructed with minimal support, while a larger span tunnel in the saine rock mass may
not be stable without the immediate installation of substantial support.
The New Austrian Tunnelling Method includes a number of techniques for safe tuiinelling i11 roclc
conditions i11 which the stand-up time is limited before failure occurs. These techiiiques iiiclude the use of
sinaller headings and benching or the use of multiple drifts to forin a reinforced ring inside which the bullc of
 fication

the tunnel cai1 he excavated. These techniques are applicable in soft rodci such as shales, pliyllites aiid inudstones
inwhich the sq~~eezing and swelliiig probleins, described hy Terzaghi (see previous section), are lilcely to occur. Tlie
techniques are also applicable when tunnelling in excessively brolien rodi, hut great care should be t a k n iii
atternpting to apply these techniques to excavations in hard ro& in wliicli different failure mechaiiisins occur
In designing support for liard rodi excavations it is pmdent to assume that the stability of tlie rode inass
surrounding.the excavation is not time-dependent. Hence, if a stnicturally defined wedge is exposed i11 tlie roof of
ai1 excavation, it will fall as soon as the rocli supporting it is removed. This can occur at tlie time of tlie I-ilastor
during tlie sulirsequeiit scaliiig operation. Ifit is required to Iieep such a wedge in place, or to euliaiice the inargiii
ofsafety, it is essential that the support be installed as early as possible, preferably before tlie rodi supporting thc
hill wedge is reinoved. O n the other limd, in a Iiighly stressed rodi, failure will generally he iiiduced by somc
chaiige iii tlie stress field surro~indiiigtlie excavation. The failure may occur gradually and inanifest itself as
spalling or siahbing or it inay occur suddenly iii the forin of a rocli burst. In either case, tlie support design inust
take into account the change in tlie stress field rather than the 'stand-up' time of the excavatioii.

3.2.3 Rock quality d e s i g n a t i o n i n d e x (RQD)

Tlie Roclc Quality Desiglation index (RQD) was developed by Deere (Deere et al 1967) to provide a
quantitative estiinate of rock mass quality from drill core logs. RQD is defined as the percentage of intact core
pieces longer than 100 min (4 indies) in tlie total length ofcore. Tlie core sliouid be at least NW size (54.7 inin
or 2.15 inches iii diaineter) and should be drilled with a double-tuhe core barrel. Tlie correct procedures for
ireinent of the leiigtli of core pieces and the calculatioii ofRQD are suininarised in Figure 3.1.
Palinstrotn (1982) suggested that, wlien n o core is available hut discontiiiuity traces are visihie i11
e exposures or exploration adits, the RQD may be estimated froin lhe nuinber of discontiriuities per
unitvoluine. The suggested relatioiiship for clay-ffee rock masses is:

where J,,is tlie suin of the niimber of joiiits per uiiit length for all joiiit (discotitin~iity)sets liiiowii as the
volumetric joiiit count.
RQD is a directionally dependent parameter and its value may change significaiitly, dependiiig upon
the borehole orientation. Tlie use of the volnmetric joint count can be quite iiseful in reducing tliis direc-
tional dependeiice.
RQD is intended to represent the rock mass quality in situ. Wlien using diainond drill core, care inust
be talien to eiisure tliat fractures, which have been caused by handling or the drilliiig process, are identified
and igiiored when determiniiig the value of RQD, When using Palmstroin's relatiorisliip for exposurc inap-
ping, hlast induced fractures should not he iiicliided when estiinating J,,.
Deere's RQD lias beeti widely used, particularly iii Nortli Ainerica, for tlie past 25 years. Cordiiig aiid
(1972), Merritt (1972) and Deere and Deere (1988) have attempted to relate RQD to Terz~ghi'srock
load factors aiid to roclibolt requirements in tuiinels. I11 the context of tliis discussion, the inost iinportaiit
use ofRQD is as a coinponent of the RMR and Qrock inass classificatiotis covered later in this cliapter.

3.2.4 Rock Structure Rating (RSR)

Wicl<hain et ai (1972) described a quantitative method for describing the quality of a roclc inass and
for selecting appropriate support on the basis of their Roclc Structure Rating (RSR) classificatioii. Most of tlie
case histories, used in the development of this systein, were for relatively sinal1 tunnels supported by incaiis
ofsteel sets, although historically this system was the tirst io malie reference to shotcrete support. 111spite of
this liinitation, it is worth examining the RSR system in some detail since it deinonstrates thc logic involved
in developiiig a quasi-quantitative rock mass classification systein.
 ourse notes

Total length of wre run = 200 CmS

C Length of core pieces > 10 cm length

RQD = x 100
Tatal lengt11 of core run

38+17*20+35 x100 = 55%

RQD =
200

Figure 3.1: Praiedure for rneaiurernent and calculation of ROD (After Deere, 1989).

The sigiiificance of tlie RSR systein, in the context of this discussion, is that it introduced tiie con
cept of rating each of the components listed below to arrive at a iluinerical value of RSR = A B C. + +
1. ParameterA, Geoloo: General appraisal of geological structure on tlie basis of:
a. Roclz type origin (igneous, metamorphic, sedimentary).

2. ~ a r a m e t c Geomctry:
~~, Effect of discontinuity pattern withrespect to the direction of the tunnel drive oii
the basis oi:
a. Joint spacing.
h. Joint orieiitation (strike aiid dip).
c. Direction of tutinel drive.
3. Parameter C: Effect of groundwater inflow and joint condition oii the basis of:
a. Overall rock inass quality on the basis of A and B combined.
b. Joint coiiditioii (good. fair, poor).
c. Ainount ofwater inflow (in galloiis per minute per 1000 feet of tunnel).
The value of A + B = 46 and this means that, for joints of fair conditioil (slightly weathered and
altered) and a inoderate water inflow of between 200 and 1,000 gallons per minute, Table 3.3 gives tlie rating
for C = 16. Hence, the final value of the rock structure rating RSR = A B C = 62. + +
 c rnass class

Note that the RSR classification used Imperial uiiits and that these
units have been retained in this discussioii.
Three tables froin Wiclcham et al's 1972 paper are reproduced iii
Tahles 3.1, 3.2 and 3.3. These tables can be used to evaluate the
Drivc witli dio rating of each of these parameters to arrive at the RSR value (inaxi-
inum RSR = 100).
For example, a hard metamorphic rock wiiich is slightly folded or
faulted has a ratiiig of A = 22 (from Table 3.1). The rodc rnass is
moderately jointed, with joints strilcing perpendicular to tlie tunnel
x i s which is being driven east-west, and dipping at between20" aiid
50". Tahle 3.2 gives the rating for B = 24 for driviiig with dip (de-
Drive against d i p fined in tlie inargin sketch).

A iypical set of predictioii curves for a 24 foot diaineter tunnel are given iii Figure 3.2 which shows
that, for the RSRvalue of 62 derived above, the predicted support would be 2 inches of shotcrete and 1 iiich
diameter roclcboits spaced at 5 foot centres. As indicated in the figure, steel sets would be spaced at more
than 7 feet apart and would iiot be considered a practical solution for the support of this tunnel.
For the same size tunnel in a rock mass witb RSR = 30, the support could be provided by 8 WF 31
steel sets (8 incli deep wide flange I section weigliing 31 Ib per foot) spaced 3 feet apart, or by 5 inches o i
shotcrete and 1 incli diarneter rockbolts spaced at 2.5 feet centres. In this case it is probable tliat the steel set
soliition would be clieaper and more effective than the use of rodcbolts and shotcrete.
Altliough the RSR classification systein is not widely used today, Wicl<liain et al's worlc played a sigiiifi-
role in the developrnent of the classification schemes discussed in the remaining sectioiis of this ciiaptrr

3.3 Geomechanics Classification

Bieniawski (1976) published the details of a roclc inass classificatioii called tlie Geoineclianics Classifi-
n or the Roclc Mass Ratiiig (RMR) system. Over the years, this system has beeii successively refined as
inore case records liave been exainined and the reader should be aware that Bieniawski has inade significaiit
changes in the ratings assigned to different parameters. The discussion which follows is based upoii tlie 1989
version of the classification (Bieiiiawslu, 1989). Both this version and the 1976 version will be used in Cliap-
ter8 which deals with estimating the strength of rockmasses. Thé followingsix parameters are used to classify
a rodi mass using the RMR system:.

1. Uniaxial compressive strength of roclc material.

2. Rode Q a l i t y Designation (RQD).
3. Spacing of discontinuities.
4. Condition of discontinuities.
5. Groundwater conditions.
6. Orientation of discontinuities.

In applying this classification system, the rock 1nass.i~divided into a number of structurai regions aiid
region is classified separately. The houndaries of the structural regions usually coincide with a major
tural feature such as a fault or with a change in rock type. In some cases, significant changes in discon-
ty spacing or characteristics, withiii tlie saine rodc type, may necessirate the division of the roclc rnass
a number of small stnictural regions.
 eering. COLirse notes b

Figure 3.2: RSR support estmater for a 24 ft. (7.3 m) diameter circular tunnel. Note that rockbolti and shntciete are generally used together (After
Wirkham et a1 1972).

Table 3.1: Rock Structure Ratina: Parameter A: General area aeoloav.

Basic Rock Type
Hard Medium Soft Decomposed Geolog~calStructure
Igneous 1 2 3 4 Sliqhtly Moderately Intensively
Metamorphic 1 2 3 4 Folded or Folded or Folded or
Sedimentary 2 3 4 4 Massive Faulted Fauted Faulted
T Y P 1~ 30 22 15 9
T Y P2~ 27 20 13 8
T Y P3~ 24 18 12 7
T Y P4~ 19 15 10 6

Table 3.2: Rock 8tructure.Ratinq: Parameter B: Joint pattern, direction of drive.

Strike i t o Axis Strike II t o Axis
Direction of Drive Direction of Drive
60th I With Dip I Aqainst Dip Either direction
Dip of Prominent Jointsa Dip of Prominent Joints
Averaqe ioint spacina Flat Dippinq Vertical Dipping Vertical Flat Dipping Vertical
1 . Very closely jointed, < 2 in 9 11 13 10 12 9 9 7
2. Closely jointed, 2-6 in 13 16 19 15 17 14 14 11
3. Moderately jointed, 6-12 in 23 24 28 19 22 23 23 19
4. Moderate t o blocky, 1-2 f t 30 32 36 25 28 30 28 24
5. Blocky to rnassive, 2-4 ft 36 38 40 33 35 36 24 28
6. Massive. > 4 f t 40 43 45 37 40 40 38 34
 mass class~fication

Table 3.3: Rock Structure Ratinq: Parameter C: Groundwater, ioint condition.

Sum of Parameters A +B
13-44
I 45 - 75

Anticipated water inflow Joint Condition

gpm/1000 ft of tunnel Good Fair Poor Good Fair Poor
None 22 18 12 25 22 18
Slight, < 200 gpm 19 15 9 23 19 14
Moderate, 200-1000 gprn 15 22 7 21 16 12
Heavy, > 1000 gpm 1O 8 6 18 14 1O
Qip: flat: 0-20"; dipping: 20-50"; and vertical: 5 0 - 9 0
Joint condition: good = tight or cemented; fair = slightly weathered or altered; poor = severely weath-
ered, altered or open

Tlie Rock Mass Rating systein is presented in Table 3.4, giviiig the ratings for each of tlie six parain-
eters listed above. These ratings are suinined to give a value ofRMR. Tlie following exainple illustrates tlie
use ofthese tables to arrive at an RMR value.
A tunnel is to be driveii througli a sliglitly weathered graiiite.with a dominant joint set dipping at
6 0 agaiiist the direction of tlie drive. Index testing and logging of diainond drilled core give typical Poiiit-
load strength iudex values of 8 MPa and average RQD values of 70%. The sliglitly rough and slightly
weathered joints with a separation of < 1 inin, are spaced at 300 min. Tunnelling conditioiis are antici-
pated to be wet.
The RMR value is deteriniiied as follows:
Table Item . Value Ratina
3.4: A I Point load index 8 MPa
3.4: A.2 RQD 70%
3.4: A.3 Spacing of discontinuities 300 rnm
3.4: E 4 Condition of discontinuities Note 1
3.4: A.5 Groundwater Wet
3.4: B Adjustment for joint orientation Note 2
Total 59

I. For slightly rough aiid altered discontinuitysurfaces with a separation o f < 1 inrn, Table 3.4.A.4 gives
aratingof25. When more detailed inforrnation is available, Table 3.4.E can be used to obtaiii a more refined
rating. Hence, in tbis case, the rating is tlie sum of: 4 (1-3 m discontinuity length), 4 (separation 0.1-1.0 inin),
3 (sliglitly roiigli), 6 (tio infilling) and 5 (slightly weathered) = 22.
Note.7. Table 3.4.F gives a descriptioii ofrFair' for the conditions assumed where the tuiinel is to be driveii
against tlie dip of a set ofjoiiits dipping at 60". Using tliis description for 'Tunnels and Mines' in Table 3.4.B
gives an adjustment rating of - 5 .
Bieniawslci (1989) publislied a set of guidelines for the selectioii of support in tuiinels iii rock Cor wliicli
tlie v;iliie ofRMR lias been determined. These guidelines are reproduced in Tdbie 3.5. Note that tliese guide-
liiies Ilave been piiblished for a 10 in spaii horseshoe shaped tunnel, constmcted using drill and blast niethods,
'd< mass subjected to a vertical stress < 25 MPa (equivalent to a deptli below surface of <900 in).
 Rock Engirieering. Coi~ r s enotes k

For the case coiisidered earlier, with RMR = 59, Table 3.5 suggests that a hlnnei could be excavated by top
heading and bench, with a 1.5 to 3 m advance in the top heading. Support should be installed afler each blast and
the support should I-ie placed at a maxiinum distance of 10 in froin the face. Systeinatic rodc holting, usiiig 4 in
loiig 20 inin diaineter fiilly grouted bolts spaced at 1.5 to 2 in in the crown and walls, is recoinmended. Wire inesli,
with 50 to 100 rriin of shotcrete for the crown and 30 min of shotcrete for the walls, is recommeiided.
The value o f R M R of 59 indicates tliat the rock mass is o n the boundary between tlie 'Fair roclc' and
'Good roclc' categories. In tlie initial stages of desigii and coiistruction, it is advisable to utilise tlie support
suggested for fair roclc. If the coiistructioii is progressing well with n o stability probleins, aiid tlie support is
perforiniiig very well, then it sliould be possible to gradually reduce the support requirernents to those
indicated for a good rodc mass. In addition, if the excavation is required to be stable for a short ainount of
time, then it is advisable to try the less expensive and extensive support suggested for good rodc. However, if
tlie rodc inass surrounding the excavation is expected to undergo large inining induced stress cliaiiges, then
more substantial support appropriate for fair rock should be installed. This exainple indicates that a great
deal of judgeinent is needed in the applic&ioii of rock rnass classificatioii to support desigii.
It sliould he iioted that Table 3.5 has not had a major revision since 1973. In many inining aiid civil
eilgineering applications, steel fibre reinforced shotcrete may be considered in place ofwire inesh and sliotcrete.

3.4 Modifications to RMR for mining

Bieniawski's Roclí Mass Rating (RMR) system was originally based upon case liistories drawii
froin civil engiiieeriiig. Coiisequently, the mining industry teiided to regard the classificatioii as soine-
what coiiservative and severa1 modifications have been proposed in order to malíe the class'ification
more relevant to mining applications. A compreiiensive suminary of tliese modiiications was compiled
by Bieniawslci (1989).
Laubsclier (1977, 1984), Lauhscher and Taylor (1976) and Laubscher and Page (1990) have described a
Modified Roclí Mass Rating system for inining. This MRMR system talces the basic RMR value, as defined by
Bieniawslíi, aiid adjusts it to account for in situ and induced stresses, stress changes aiid the effects of blasting
aiid weathering. A set ofsupport recoininendations is associated with the resultiiig MRMR value. Iii usiiig
Laubsclier's M R M R systein it sliould he borne iii mind tliat many of the case histories ~ipoiiwhicli it is based
are derived froin caviiig operatioiis. Origiiially, blodc caving in asbestos mines in Africa forined the basis for tlie
inodificatioiis but, subsequently, other case histories from around the world have been added to the database.
Cummings et al(1982) aiid Kendorslci et a1 (1983) have also inodified Bieiiiawslíi's RMR classificatioii
to produce tlie MBR (inodified basic RMR) system for mining. This systein was developed fõr bloclc caving
operations iii tlie USA. It involves the use of different ratiiigs for the original parameters used to determine
tlie value of RMR aiid the subsequent adjustment of the resulting MBR value to allow for blast dainage,
induced stresses, structural features, distance from the cave'front and size of tlie caving block. Support
recommendations are presented for isolated or development drifts as well as for the final support ofiiitersec-
tions and drifts.

3.5 Rock Tunnelling Quality Index, Q

O n the basis o f a n evaluatioii of a large number ofcase histories ofuiidergrouiid excavations, Bartoii
et al (1974) of tlie Nomegian Geotechnical Institute proposed a Tunnelling Qoality Iiidex (a
for tlie deter-
ininatioii of rocb inass characteristics and tunnel support requirements. The iiumerical value of the i n d e x Q
varies on a logaritliinic scale from 0.001 to a maxiinuin of 1,000 and is defined by:

Q = -RQD x-xi J J""

r Cnr (3.2)
 k rnass cias

where
RQD is the Rod< Quality Designation
, is tlie joint set number
J, is the joint rougliness number
, is the joiiit alteration number
J, is the joiiit water reductioii factor
SRF is the stress reduction factor
In explaining tlie ineaning of tlie paraineters used to deterinine the value o f a Barton et al (1974) offer tlie
following comments:

The first quotient (RQDX,), representing the stmcture of the rock inass, is a crude ineasure of the
block or particle size, with the two extreine values (100/0.5 and 10/20) differiiig by a fictor of400. If tlie
quotient is iiiterpreted in units ofcentitnetres, the extreme 'particle sizes' of 200 to 0.5 cin are seen to be
crude but fairly realistic approxiinatioiis. Probably tlie largest blod<s should be several tiines this size and the
smallest fraginents less than half the size. (Clay particles are of course excluded).
The second quotient (J,/J,) represents the roughness and frictional characteristics of the joiilt walls ar
ling materials. This quotient is weighted in favour of rough, unaltered joints in direct coiitact. It is to be
:pected that such surfaces will be close to peak strength, that they will dilate stroiigly when sliearrd, aiid
ey will therefore be especially favourable to tuniiel stability.
When rock joints have thin clay mineral coatings and fillings, the strengtli is reduced sigiliticaiitly.
Nevertheless, rod<wall contact after small shear displaceinents have occurred inay be a very iinportant iãctor
for preserviiig the excavation from ultimate failure.
Where no rock wall contact exists, the conditions are extremely unfavourable to tuiinel stahility. Tlie
-iction angles' (given in Tabie 3.6) are a little below the residual strength values for inost clays, aiid are
xsibly down-graded by the fàct that these clay bands or fillings may tend to consolidate during sliear, at
ast if norinal consolidation or if softeiiing and swelling has occurred. The swelling pressure of montinoril-
nite may also be a iãctor here.
Tlie tliird quotieiit Vf,!SRF) consists of m o stress paraineters. SRFis a measure of: 1) looseniiig load
the case of an excavatioii through shear zones and clay bearing rock, 2) rock stress in coinpeteiit rock, and
2 1 squeezing loads in plastic iiicoinpetent rocl<s.It can be regarded as a total stress parameter. The paraiiieter
Iraisa measure ofwater pressiire, which lias ali adverse effect on the shear strength ofjoiiits due to a reductioii
in effective norinal stress. Water inay, in addition, cause softeiiing and possible out-wasli i11 the case of elay-
fil led joints. It has proved iinpossihle to combine these two parameters in terrns of iiiter-block efkctive
S t ress, because paradoxically a liigh value of effective normal stress may sometiines signiiji less stahle colidi-
tii,ns than a low value, despite the higher shear strength. The quotient (I?#/SRF) is a coinplicated einpirical
ij,ctor describing the 'active stress'.
It appears tliat the rock tunnelling quality Q c a n iiow be considered to be a function o f o n l y three
irameters which are crude measures of:
BIok size (RQD43
2. Inter-block shear strength V!J,)
3. Active stress uT,/sRn
Undoubtedly, there are several other paraineters whicli could be added to iinprove tlie accuracy of tlie
classificatioiisystein. O n e of-thesewould be the joint orientation. Although inany case records iilclude the
iiecessay inforinatioii on structural orientatioii in relation to excavation axis, it was not fouiid to be the
important general parameter that inight be expected. Part of the reason for this may be that tlie orieiitatioiis
iany types ofexcavations can be, and normally are, adjusted to avoid the maxiinum effect ofuiif;?vournbly
 Rock Engir me notes k

Table 3.4: Rock Mais Ratina Svstem (After Bieniawski 1989).

Rafing 6 5 1 O
1"liliing (gouge) Nane Hard liling c 5 rnm Hard liling > 5 mrn Soltfilling s 5 mm Sofffilling > 5 mm
RaI8ng 6 4 2 2 O
Wealherlng Unweathered Slghlly wealhered Moderalely Highiy weathered Decarnposed
Ratlngs 6 5 weafheied 1 O
I I I I
F. EFFECT OF DISCONTINUITY STRIKE AND DIP ORIENTATION INTUNNELLINGlf
Stiike perpendicular to tunnel a x s
-
Di8ve wllh dip - Dip 45 - 90° Drive with dip - O p 2 0 - 45"
Strike parallel ta tunne ax!:
-
Very lavourable Favaurable
Dlp 45 - 90'
ver) favaugable
I -
Drive aganstdtp - Dip45-s0~ Drive againn dip - ~ i p 2 0 - 4 s D i p 0 2 0 - Imerpecfive of slrike"
~aii Unfavourable ~air

'Some condilionr are mutualy exclusive. For exarnple, ii infilling is present. lhe raughness 01 lhe surtace wlll be overshadawed by fhe influente
aflhe gouge. ln such cases use A.4 dredly.
" Madifiedafter Wickham e i al (1972).
 :mas8 class

Table 3.5: Guidelines for excavation and support of 10 m span rock tunnels in accordance with the RMR
system (After Bieniawski 1989).
Rock mais class Excavation Rock bolts Shotcrete Steel sets

fully grouted)
I - Very good rock Full face, Generally no support required except spot bolting.
RMR: 81-100 3 m advance.
- %od rock Full face. Locally, bolts in 50 m m in None.
61-80 1-1.5 m advance. crown 3 m long, crown where
Complete support spaced 2.5 m with required.
20 m from face. occasiona wire mesh.
lll - Fair rock Top heading Systematic bolts 50-1 00 m m in None
RMR: 41 -60 and bench 4 rn long, spaced crown and
1.5-3 m advance 1.5 - 2 rn in crown 30 m m in sides.
in top heading. and walls with
Cornrnence support wire mesh
after each blast. in crown.
Complete support
10 m from face.
oor rock Top heading Systernatic bolts 100-150 m m in Light t o
RMR: 21-40 and bench 4-5 m long, spaced crown and medium ribs
1.0-1.5 madvance 1-1.5 m in crown 100 m m in sides. spaced 1.5 m
in top heading. and walls with where required.
Insta11 support wire rnesh.
concurrently with
excavation, 10 m
from face.
'ery poor rock Multiple drifts Systematic bolts 150-200 m m M e d u m to
c 20 0.5-1.5 m advance 5-6 m long, spaced in crown, heavy ribs
in top heading. 1-1.5 m in crown 150 m m in spaced 0.75 m
Install support and walls with sides, and with steel
concurrently with wire mesh. 50 m m on face. lagging and
excavation. Shotcrete Bolt invert. forepoling if
as soon as possible required.
after blastinq. Close nvert.

oriented major joints. However, tliis choice is not available iii the case of tunnels, aiid more tliaii liali the
case records were in this category. The parainetersJa, J, and J,, appear to play a more iinportant role thaii
orientatioti, because the number ofjoint sets determines the degree of freedom for blocl<inoveinent (if aliy),
and the frictional aiid dilational cliaracteristics can v a y more than the down-dip gravitational coinpoiient of
unfa~ lourably oriented joints. Ifjoint orientations had been included the classificatioii would have heen less
gener-al, and its essential siinplicity lost.
Tahle 3.6 gives the classification ofindividual parameters used to obtain the Tunnelling @aIity Index
g i u r . a ro& mass. Tlie use ofthis table is illustrated in the followiilg exainple.
 Lock Engineenng Cour

A 15 in span crusher chamber for an underground mine is to be excavated in a norite at a depth of 2,100 in
below surface. The rodc mass contains two sets ofjoints controlling stability. niese joults are undulating, rough and
unweathered with very minor surface staining. RQD values range 6om 850in to 95% and laboratory tesu oti core
sainples of intact rock give an average uniaxial compressive strength of 170 MPa. The principal stress directions are
approxiinatelyvertical and horizontal and die magnitude o€the horizontal principal stress is approxirnately 1.5 tiines
that of the vertical principal stress. The rodc inass is locally dainp but there is no evideiice of flowing water.
The numerical value ofRQD is used directly in the calculation ofQand, for this roclc inass, an average
value of 90 will be used. Table 3.6.2 sliows that, for two joint sets, the joint set nuinber,J,= 4. For rough or
irregular joints which are undulating, Table 3.6.3 gives a joint roughness numher ofJ,= 3. Table 3.6.4 gives
rhe joint alteration numher,Ja = 1.0, for unaltered joint walls with surface staining only. Table 3.6.5 shows
that, for ai1 excavation witli ininor inflow, the joint water reduction factor,Jw= 1.0. For a depth below surface
o i 2,100 in the overburden stress will be approximately 57 MPa and, in this case, the inajor principal stress
o,=85 MPa. Since the uniaxial compressive strength of the norite is approxiinately 170 MPa, this b'w e s a
ratio of ?/o,=2. Table 3.6.6 shows that, for coinpetent rock with rock stress probleins, this value of q / u ,
can he expected to produce heavy rode burst conditions aiid that the value of SRFshould lie between 10 and
20. A value o f S R F = 15 will he assumed for this calculation. Using these values gives:

In relating the value of the indexQto the stability and support requirements of underground excava-
tions, Barton et al (1974) defined an additional parameterwhich they called the EqniuuloztDimmsion, D<,of
the excavation. This dimension is obtained by dividing the span, diameter or wall height of the excavatioil
by a quantity called the Excuvution SxpportRutio, ESR. Hence:
Excavation span, diameter or heighi (m)
D, =
Excavation Support Ratio ESR
The value ofESR is related to the intended use of the excavation and io the degree ofsecurity wliich
is deinanded of the support system installed to maintain the stability of the excavation. Barton et al (1974)
suggest the following values:

Excavatioii category ESR

A Temporary mine openings. 3-5
Permanent mine openings, water 1.6
tuniiels for hydro power (excludiiig
high pressure penstocks), pilot tunnels,
drifts and headings for large excavations.
C Storage rooms, water treatrnent 1.3
plaiits, miiior road and railway
tunnels, surge chambers, access tunnels.
D Power stations, major road 1.0
and railway tunnels, civil defence
chambers, portal intersections.
E Underground nuclear power 0.8
stations, railway stations, sports
and public facilities, factories.
 The cnisher station discusse<i above falls into the category of perinanent mine openiiips aiid is as-

,=
signed an excavation support ratio ESR = 1.6. Hence, for an excavation span of 15 ni, the equivalent diinen-
. .
sion, L 15/1.6 = 9.4.
The equivalent diinension, DE, plotted against the value o f a is used to define a iiuinher of support
categories in a chart published iii the original paper by Barton et al (1974). This chart lias recently beeii
i i n r I 3 t r :d by Griinstad and Barton (1993) to reflect the increasing use of steel fibre reiiiforced shotcrete in

;round'excavation support. Figure 3.3 is reproduced froin this updated chart.

From Figure 3.3, a value o f D r o f 9.4 and a value ofQof4.5 places this crusher excavatioii iii category
ich requires a pattern of rockbolts (spaced at 2.3 in) and 40 to 50 min of uiireinforced shotcrete.
Because of the inild to h e a ~ yrodc burst conditions which are anticipated, it inay be prudeiit to
aesrress the rod< in the walls of this cmslier chamber. This is achieved by using relatively heavy productioii
blasting to excavate the chainber aiid oinitting the smooth blasting usually used to triin tlie final walls of ali
excava tion such as an ~indergroundpowerhouse at shallower depth. Caution is recoininended iii the use of
destre!;s blasting and, for critical applications, it may be advisable to seek the advice of a blasting specialist
before embarl<ing oii this course of action.
Loset (1992) suggests that, for rocls with 4 <Q< 30, blasting damage wili result i11 the creatioii of new
'joints' with a consequeiit local reduction in the value of Q f o r the rock surrounding the excavation. He
suggests that tliis can be accounted for by reducing the RQD value for the blast damaged zoiie.
Assuining that tlie RQD value for the destressed rock around tlie cnisher chainber drops to 50010, the
resultingvalue ofQ=2.9. Froin Figure 3.3, this value o f a for an equivaleiit diinensioii, D, of 9.4, places tlie
excavaitiou just inside category (5) which requires rockbolts; at approximately 2 in spaciiig, aiid a 50 inin thicl<
layer cfsteel fibre reinforced shotcrete.
Bartoii et ai (1980) provide additional information o11 roci<bolt length, maxiinuin uiisupported spaiis
and roof support pressures to suppleinent the support recoinmendations published iii tlie origiiial 1974 paper.
The length L ofrockbolts can be estiinated from the excavation width B and the Excavatioii Support
R a t i n ESR:

+0.15 B
-
ESR
The maxiinum unsupported span can be estimated froin:
Maxirnum span (unsupported) = 2 E S R P (3.4)
Based upon analyses of case records, Griinstad 2nd ~ a r t o ' n(1993) suggest that the relationship he-
the value o f Q a i i d the permaiient roof support pressure Pro0,isestiinated froin:

3.6 Using rock mass classification systems

The m o most widely used rock mass classificatioiis are Bieniawski's RMR (1976, 1989) and Barton et
ai's Q(1974). Both methods incorporate geological, geometric and design/engineering parainetea iii arriviiig
at a quantitative value of their rock mass quality. The siinilarities betweeii RMR aiid Qstem tioni tlie use of
identical, or verysiinilar, parameters in calculating tlie final rodcmass quality ratiiig. The differences between
the systems lie in the different weightings given to similar parameters and in tlie use of distiiict paraineters in
nm- *r the other scheine.
 Rock Engineering COL

Table 3.6: Claisification of individual parameters used in the Tunnelling Quality Index Q (After Barton et al 1974).
DESCRIPTION VALUE NOTES
1. ROCK QUALITY DESIGNATION RQD
A. Very poor O 25 - 1. Where RQD is reported or measured as
5 10 (including O), a nominal value of 10
B. Poor 25 - 50 is used t o evaluate Q.
C. Fair 50 - 75
D. Good 75 - 90 2. RQD intervas of 5, iie. 100, 95, 90 etc.
are sufficienty accurate.
E. Exceent 90 - 100
2. JOINT SET NUMBER I"
A. Massive, no or few joints 0.5 - 1.0
B. Onejo~ntset 2
C . One joint set plus random 3
D. Twojoint sets 4
E. Two joint sets plus random 6
F. Three jont sets 9 1. For intersections use (3.0 x i )
G. Three j o n t sets pus randam 12
H. Four or more joint sets, random, 15 2. For portais use (2.0 x<)
heavily jointed, 'sugar cube', etc.
J. Crushed rock. earthiike 20
3. JOINT ROUGHNESS NUMBER 1
~,
a. Rock wall contact
h. Rock wall contact before 10 cm shear
A. Discontinuous jaints 4
B. Rough and irregular, undulating 3
C. Smooth undulating 2
D. Sckensided undulating 1.5 1. Add 1 .O if the mean spacing af the reevant
joint set is greater than 3 m.
E. Rough or irregular, planar 1.5
F. Smooth, planar 1.O
G. Slickensided, planar 0.5 2. i = 0.5 can be used for planar, sickensided
joints having iineations, provided that the
c. N o rock wall contact when sheared lineations are oriented for minimum stiength.
H. Zones containing cay minerais thick 1.O
enough to prevent rock wall contact (nom~nal)
i. Sandy, gravely or crushed zone thick 1 .O
enough t o prevent rock wall contact (nominal)
4. JOINT ALTERATION NUMBER I, $r degrees
a. Rock wall contact (approx.)
A Tightly healed, hard, non-softening, 0.75 1. Vaues of #r, the residual
friction angle, impermeable
filling are intended as an
approximate guide t o the
rnineraogcal propeities of the
B. Unaltered joint wais, surface staining only 1.O 25 - 35 alteration products, f present.
C. Slighty aitered joint walls, non-softening mineral 2.0 25 - 30
coatings, sandy particles, clay-free
d~sintegratedrock, etc.
 &r03 Rock mass classif ication

Table 3.6: icont'd.) Ciassification of individual oarameters used in the Tunnelino Ouaitv Index O (After Barton et al 1974).
DESCRIPTION VALUE NOTES
4. JOINT ALTERATION NUMBER 1 $r degrees
a. Rock w a l l contact (~PP~ox.)
D. SIty-, or sandy-clay coatings, small clay- 3.0 20 25-
fradion (non-softening)
E. Softening or low-friction clay mineral coatings, 4.0 8-36
i.e. kaoiinite, mica. Also chlorite, talc, gypsum
and graphite etc, and small quantities of swelling
-
clays. (Discontinuous coatings. 1 2 m m or iess)
b. Rock w a l l contact before 10 c m shear
F Sandy particies, clay-free, dis~ntegratngrock etc. 4.0 25 - 30
G. Stronqly over-consolidated, non-softening 6.0 16-24
nineral fiilings (continuous < 5 m m thick)
u m or low over-consolidation, softening 8.0 12 - 16
nineral filiings (continuous < 5 rnm thick)
i . ,wriing clay fiings. i.e. rnontmorillonite, 8.0 - 12.0 6 - 12
(continuous < 5 rnm thick). Values of Ja
depend on percent of sweliing clay-siie
particies, and access to water.
c. No rock w a l l contact w h e n sheared
K. Zones or bands of dsintegrated or crushed 6.0
L. rock and clay (see G, H and J for clay 8.0
M. conditions) 8 . 0 - 12.0 6-24
N. Zones or bands of silty- or sandy-ciay, srnall 5.0
clay fraction, non-softenng
O. Thick continuou8 zones or bands of clay 10.0 - 13.0
P & R . (see G.H and 1 for ciay conditions) 6.0 - 24.0
5. IOINT WATER REDUCTION jw approx. water pressure (kgf/cmz)
A. Dry excavation or minor inflow i e . y 1.O < 1.0
< 5 lüm locali
0. Medium inflow or pressure, occasional 0.66 1.O - 2.5
outwash of joint fillings
C. Large inflow or high pressure in 0.5 2.5 - 10.0 1. Factars C t o F are crude
cornpetent rock with unfilled joints estimates; increasek
if drainage instaled.
D Large inflow or high pressure 0.33 2.5 - 10.0
E Exceptionaliy high inflow or pressuie 0.2 - 0.1 > 10 2. Specia probems caused by
at blasting, decaying with time ice formationare not considered.
F Exceptionallv hiah i n f o w or wressure
- -
0.1 0.05 > 10
~ ~

6. STRIiSS REDUCTION FACTOR SRF

a. VVeaknesszones intersecting excavation, which m a y
caiuse loosenins o f rock mass w h e n t u n n e l i s excavated
A. Mull ipe occurrences of weakness zones rontaining clay or
allydisintegrated rock, very loose surrounding rock any depth) 100 1. Reduce these vauesof SRF
by 25 - 50% but o n y if the
reevant shearzones influente
do not interseiithe excavation
 Irse notes b

Table 3.6: (cont'd.) Classfication of individual oararneters used in the Tunneiiino Oualitv Index O (After Barton et ai 1974).
DESCRiPTION VALUE NOTES
6. STRESS REDUCTION FACTOR SRF
-
a. Weakness zones intersectins excavation, which m a v
cause loosening o f r o c k mass w h e n t u n n e l i s excavated
B. Slngle weakness zones containing clay, or chemically dis- 5.0
tegrated rock (excavation depth < 50 m)
C. Single weakness zones containing clay, or chemicaily dis-
tegrated rock (excavation depth > 50 rn)
D. Mutiple shear zones in competent rock (clay free), ioose
surrounding rock (any depth)
E. Single shear zone in cornpetent rock (clay free). (depth of
excavation 50 rn)
F Single shear ione in rompetent rock (clay free). idepth of
excavation > 50 m)
G. Loose open joints, heavily jointed or 'sugar cube', (any depth)
H. Low stress, near surface > 200 > 13 when 5<o,/o,fl 0, reduce oC
J. Medium stress -
200 10 13 - 0.66 t o 0.80~ando, t o 0.80, When
o,/ol > 10, reduce g and q
K. High stress, very tight structure 10 - 5 0.66 - 0.33 t o 0 . 6 g a n d g = unconfined
(usually favourable t o stability, rnay compressive strength, and a,
= tensile strength (point load)
and a, and o,are the major
L. Mild rockburst (massive rock) 5 - 2.5 -
0.33 0.16 and minor principal stresses.
M . Heavy rockburst (massive rock) < 2.5 < 0.16 3. Few case records availabie
where depth of crown below
c. Squeezing rock. plastic f l o w o f incompetent rock surface 18less than span wdth.
under influente o f h i g h rock pressure Suggest SRFincreasefrom 2.5
t o 5 for such cases (see H).
N. Mild squeezing rock pressure
O. Heavy squeezing rock pressure
d. Swelling rock. chemical swelling activity depending
o n presence of w a t e r
P. Mild swelling rock pressure
R. Heavv swellinq rock pressure

ADDITIONAL NOTES O N THE USE OF THESE TABLES

When making estimates of the rock mas8 Quality (0). thefollowing guidelines should befollowed in addition t o the notes listed in
the tables-
1 When borehole core is unavailable, RQD can be estimated from the number of joints per unit volume. in which the number of
joints per metre for each joint set are added. A simpe relationship can be used t o convert this number t o RQD for the case of
-
clay free rock masses: RQD= 115 3.3 Jv(approx.), whereJv= total number of joints per m3(O < RQD < 100 for 35 > 1" > 4.5).
2. The parameteri representing the nurnber of joint sets will often be affeited by foliation, schistosity, slaty ceavage or bedding
etc. l i strongly developed, these parallel 'joints' should obviously be counted as a completejoint set. However, if there are few
'joints' visible, or if only occasionai breaks in the core are due to these features, then it will be more appropriate t o count them
as 'random' joints when evaluating 1".

then a second, less favourably oriented joint set or discontinuity may sometimes be more sibnificant, and its hgher value of jk,
should be used when evaluating 0. The value of //I a in fact relate to the surface most likely t o allow failure ta initate
should
 nass classification

Table 3.6: (cont'd.) Classification of individual parameters used in the T u n n e l n g Q u a t y Index Q (After Barton et a 1974).
DESCRIPTION VAI I i F NOTF<

4 Whe n a rock rnass contains clay, the factor SRF appropriate to oosening loads shouid be evaluated. ln such cases the
streriqth of the intact rock is of little interest. However, when iointins is minimal and clav is comoletelv absent. the itrenoth
of th
.
.
-A
.

reduction factor evaluaton,

cks that deteriorate when exposed to moist or saturated conditions

Figure 3.3: Ertimated suppan categorles based on rhe tunnelling quality index Q (After Grimstad and Barton 1993).

RMR uses compressive strength directly whileQonly considers strength as it relates to in situ stress in
competent rock. Both schemes deal with the geology and geometry of the rock mass, but in slightly different
ways. Both consider groundwater, and both inclnde some component of rock material strength. Some esti-
mate of orientation can he incorporated into Q u s i n g a guideline presented by Barton et al (1974): 'the
parameters J, and JcE should ... relate to the surface most likely to allow failure to initiate.' The greatest
difference betweeil the two systems is the lack of a stress parameter in the RMR system.
When using either of these methods, two approaclies can be taken. One is to evaluate the rock mass
specifically for the parameters included in the classification methods; the other is to accurately characterise
the rock mass and then attribute paraineter ratings at a later tiine. The latter method is recommended since
ir gives a full and complete description of the rock mass which can easily be translated into either classifica-
tion index. If ratingvalues alone had been recorded during mapping, it would be almost impossihle to carry
out verification studies.
 ieering Co
Rock Eng~r

Figure 3.4: Histagrams showing variations in RQD, i ,I, and I* for a dry jainted sandstone under 'medium' stress ianditioni. reproduced from fieid
notes prepared by D r N. Eaiton.
 mass classif ication

In many cases, it is appropriate to give a range ofvalues to each parameter in a rock mass classification
and to evaluate the significance of the final result. An example of this approach is given in Figure 3.4 which
is reproduced from field notes prepared by Dr. N. Barton on a project. 111this particular case, the rock mass
is dry and is subjected to 'medium' stress conditions (Table 3.6.6.K) and hence Jm = 1.0 and SRF = 1.0.
Histograins showing the variations in RQDJ,, J, andJ,, along the exploration adit mapped, are presented in
this figure. The average value o f Q = 8.9 and the approximate range o f Q i s 1.7 <Q< 20. The average value
ofQcan be used in choosing a basic support system while the range gives ao indication of the possihle
adjustments which will be required to meet different conditions encountered during construction.
A further example of this approach is given in a paper by Barton et al (1992) concerned with the
design of a 62 m span underground sports hall in jointed gneiss. Histograms of all the input paraineters for
the Qsystem are presented and analysed in order to determine the weighted average value of Q
Carter (1992) has adopted a similar approach, but extended his analysis to include the derivation of a
probability distribution function and the calculation of a prohability of failure in a discussion on the stabil-
ity of surface crown pillars in ahandoned metal mines.
niroughout this chapter it has been suggested that the user of a rock mass classification scheme
shniild check that the latest version is being used. An exception is the use of Bieniawski's RMR classification
:kmass strength estimates (discussed in Chapter 8) where the 1976 version as well as the 1989 version
:d. It is also worth repeating that the use of two rock mass classification scheines is advisable.
 Shear strength of discontinuities

4.1 Introduction
AI1 rock masses contain discontinuities such as bedding planes, joints, shear zones and faults. At
shallow depth, where stresses are low, failure of the intact rock material is minimal and the behaviour of the
ro& mass is controlled by sliding o n the discontinuities. In order to analyse the stability of thissystem of
individual rock blocks, it is necessary to understand the factors tliat control the shear strength of the
discontinuities whicli separate the blocks. These questions are addressed in the discussion that follows.

4.2 Shear strength of planar surfaces

Suppose that a number of samples of a rocl< are obtained for shear testing. Each sample contains a
througb-going bedding plane that is cemented; in other words, a tensile force would have to be applied to
the two halves of the specimen in order to separate them. The bedding plane is absolutely planar, having
n o surface irregularities or undulations. As illustrated in Figure 4.1, in a shear test each specimen is sub-
jected to a stress o,,normal to the bedding plane, and the shear stress T, required to cause a displacement
6, is measured.
The shear stress will increase rapidly until the peak strength ir reached. This corresponds to the sum of
the strength of the cementing material bonding the two halves of the bedding plane together and tlie fric-
tional resistance of the matching surfaces. As the displacement continues, the shear stress will fall to some
residual value that will then remain constant, even for large shear displacements.
Plotting the peak and residual shear strengths for different normal stresses results i11 tlie m o lines
illustrated in Figure 4.1. For planar discontinuity surfaces the experimental points will generally fall along
straight lines. The peak strength line has a slope of $I and ai1 intercept of c on the shear strength a i s . The
residual strength line has a slope of$,.
The relationship between the peakshear strength and the normal stress oncan be represented by the
Mohr-Coulomb equation:
T =C + o,,tan 4 (4.1)
where L is the cohesive strength of the cemented surface and
$ is the angle of friction.
 trength of (

shear displacement 6

w shear stress

normal stress o"

I

displacement 6 normal stress 7,

Figure 4.1: Shear teiting a i diirontinuitiei

In the case of the residual strength, the cohesion c has dropped to zero and the relationship between
$, and 0" can he represented by:
7 , = ontan ar (4.2)
where a, is the residual angle of friction.
This example has been discussed in order to illustrate the physical meaning of the term cohesion, a
soil inechanics term, which has been adopted by the rod< mechanics community. In shear tests o11 soils, the
stres:E levels are generally an order of magnitude lower than those involved in rod< testing and the cohesive
strenigth of a soil is a result of the adhesion ofthe soil particles. In rocl<mechanics, true cohesioii occurs when
cem1inted surfaces are sheared. However, in many practical applications, the term cohesion is used for con-
venii:nce and it refers to a mathematical quantity related to surface roughness, as discussed in a later section.
Cohesion is sirnply the intercept o n the t axis at zero normal stress.
The básic friction angle 9,is a quantity that is fundamental t o the understanding of the shear streilgth
of d iscontinuity surfaces. This is approximately equal to the residual friction angle $, hut it is generally
mea:iured by testing sawn or ground rock surfaces. These tests, which can be carried out on surfaces as small
as 5(I inin x 50 mm, will produce a straight line plot defined by the equation:

A typical shear testing machine, which can be used to determine the basic friction angle $, is illus-
tr:ited in Figures 4.2 and 4.3. This is a very simple machine and the use of a rnechanical lever arm ensures that
th e normal load o n the specimen remains constant throughout the test. This is an important practical con-
SI(leration since it is difficult to maintain a constant normal load in hydraulically or pneumatically con-
e*, dled systems and this makes it difficult to interpret test data.
 urse notes I

Figure 4.2: Diagrarnrnatc reition through ihear machine used by Hencher and Rirhaidi (1982).

Figure 4.3: Shear rnarhine

of the type uied by Henchei
and Richards (1982) for
rneasurernent af the shear
rtrength of sheet joints in
Hang Kong granite
 )4. Shear st rength of d

Note that it is important that, in setting up the specimen, great care has to be taken to ensure that tlie
shear surface is aligned accurately in order to avoid the ileed for an additional angle correction.
Most shear strength determinations today are carried out by determining the hasic friction angle, as
described above, and theil making corrections for surface roughness as discussed in the following sections of
this chapter. In the past there was more einphasis o n testing full scale discontinuity surfaces, either in the
laboratory os in the field. There are a significant number of papers in the literature of the 1960s and 1970s
describing large and elaborate in situ shear tests, many of which were carried out to determine the shear
strength ofweak layen in dam foundations. However, the high cost of these tests together with the difficulty
of interpreting the results has resulted in a decline in the use of these large scale tests and they are seldoni
seen today.
The author's opinion is that it makes hoth economical and practical sense to carry out a number of small
scale laboratory shear tests, usiilg equipment such as that illustrated in Figures 4.2 and 4.3, to determine tlie basic
fiiction angle. The roughness component which is then added to this hasic friction angle to give the effective
friction aiigle is a number which is site specific and scale dependem and is hest obtained by visual estiinates in the
iield. Practical techniques for making these roughness angle estimates are described on the followiilg pages.

4.3 Shear strength of rough surfaces

A natural discontinuity surface in hard rodc is never as smooth as a sawn or groutid surface of tlie
type used for determining the basic friction angle. The undulations and asperities o n a natural joint sur-
face have a significant influence o n its shear behaviour. Generally, this surface roughness iiicreases the
shear strength of the surface, and this strength increase is extremely important in terms of the stability of
excavatiotis in rock.
Patton (1966) demonstrated this influence hy means of an experiment in which he carried out shear
tests on 'saw-tooth' specimens such as the one illustrated in Figure 4.4. Shear displacement in these speci-
inens occurs as a result of the surfaces moving up the inclined faces, causing dilatioil (ali iticrease in voluine)
of the specimen.
The shear strength of Patton's saw-tooth specimens can he represented hy:

where a, is the basic friction angle of the surface and

i is tlie angle of the saw-tooth face.

4.4 Barton's estimate of shear strength

Equation (4.4) is valid at low normal stresses wliere shear displacement is due to sliding along the
inclined surfaces. At higher normal stresses, the strength of the intact material will be exceeded and the teeth
will tend to break off, rreulting in a shear strength behaviour wliich is more closely related to the intact
material strength than to the frictional characteristics of the surfaces.
Wliile Patton's approach lias the merit of heingvery simple, it does not reflect the reality that clianges in
sheat sttength with increasing normal stress are gradual rather than abmpt. Banon aild his co-workers (1973, 1976,
1977,1990)studied the behaviourofnatural rockjoints and have proposed that equation (4.4) can be re-written as:

.r = a, tan $h
i3
+ JRC logio -

where JRC is the joint roughness coefficient and

JCS is the joint wall compressive strength
 Rock Engineering: Cc

failure of
normal stress o, intact rock

shear stress r shearing on saw-

tooth suriaces

normal stress a,,

Figure 4.4: Patton's experiment an the rhear strength of saw~toothsperimeni.

4.5 Field estimates of JRC

The joint roughness coefficient JRC is a number that can be estiinated by comparing the appearance
of a discontinuity surface with standard profiles published by Barton and others. One of the most useful of
these profile sets was published by Barton and Choubey (1977) and is reproduced in Figure 4.2.
Tlie appearance ofthe discontinuity surface is compared visually with the profiles shown and the JRC
value corresponding to the profile which most closely marches that of the discontinuity surface is chosen. In
the case of small scale laboratory specimens, the scale of the surface roughness will he approximately the
same as that of the profiles illustrated. However, in the field the length of the surface of interest may be
severa1 inetres or even tens of metres and the JRC value must be estimated for the full scale surface.

4.6 Field estimates of JCS

Suggested methods for estiinating the joint wall compressive streilgth were published by the ISRM
(1978). The use of the Schmidt rebound hammer for estiinating joint wall compressive strength was proposed
by Deere and Miller (1966), as illustrated iii Figure 4.7.

4.7 Influente of scale on JRC and JCS

O n the hasis of extensive testing ofjoints, joint replicas, and a review of literature, Barton and Bandis
(1982) proposed the scale corrections for JRC defined by the following relationship:

whereJRC", and L. (length) refer to 100 mm laboratory scale sarnples andJRC", and L,, refer to i11 situ blocic sizes.
Because of the greater possibility ofweaknesses in a large surface, it is liliely that the average joint wall
compressive strength UCS) decreases with increasing scale. Barton and Bandis (1982) proposed the scale
corrections for JCS defined by the following relationship:

where J C S and L. (length) refer to 100 mm laboratory scale samples and JCSx and L , refer to in situ block sizes
 ies

re 4.5: Roughnesr profilei 2nd correiponding JRC valuer (After Bartori and Choubey 1977).
 Rock Engineering Coi

Straight e d g e

Asperity amplitude - mm
1
tet- Length of profile & -m
- t
~

Length of profile -m

Figure 4.6: Alternative method for estimating JRCfrom measurements of Surface ioughness amplitude from a itraight edge (Barton 1982).
 trength of I

Average dispersion of strength

for most rocks - MPa

O
1 4
1O
I I I
20
, ,
30
, ,
40
,
; ,
50 60
I q
8 * I I I
O 1O 20 30 40 50 60 I+
L
L
O
i
i

i
i
10
i
l

i
,
20
i
,

,
30
,

,
,

,
,
40
,
,

,
50
,

,
,
60 6'
o 10 20 30 40 50 60 V
Schmidt hardness - Type L hammer

Figure 4.7: Ertimate of joint wall cornpressive strengrh from Schmidt hardness.
 ,erIng Cour

-
4.8 Shear strenath of filled discontinuities
The discussion presented in the previous sections has dealt with the shear strength of discontii~uitiesin
which rod<wallcontact occurs over the entire length of the surface under consideration. This shear strength - can be
reduced drastically when part or all of the surface is not in intiinate contact, but covered by soft filling material
such as clay gouge. For planar surfaces, such as bedding planes in sedimentary rode, a thin clay coating will result
in a significant shear strength reductioil. For a rough or undulating joint, the filling thickness has to be greater than
the ainplitude of the undulations before the shear strength is reduced to that of the filling material.
A comprehensive review of the shear strength of filled discontinuities was prepared by Barton (1974) and
a summary of the shear strengths of typical discontinuity fillings, based on Barton's review, is given in Table 4.1

Table 4.1: Shear stienqth of fled discontinuities and f i i n q material! (AfteiBarton 1974).
Rock Deicription Peak Peak Residual Residual
c' (MPa) ,V c' (MPa) V
Basalt Clayey baialtic breccia, wide vaiiatioii 0.24 42
from clay to basalt conteiit
Bentoiiite Bentoiiite seam iii chalk 0.015 7.5
Tliiii layers 0.09-0.12 12-17
Trianial tests 0.06-0.1 9-13
Bentoiiitic rhale Trianial teits 0-0.27 8.5-29
Direct sliear testr 0.03 8.5
Clays Over-coiisolidated. slipr, 0-0.18 12-18.5 0-0.003 10.5-16
joints aiid miiior sliears
Clay sliale Triaxial tests 0.06 32
Stratification surfacea O 19-25
Coa1 inensure ro&s Clay rnyloiiite seams, 10 to 25 mrn 0.012 16 O 11-11.5
Dolornite Altered shale bed, 0.04 14.5 0.02 17
f I50 inm thi&
Diorite, graiiodioiite aiid porpliyiy Clay gouge (2% ciay, P I = 17%) O 26.5
Graiiite Clay filled faults 0-0.1 24-45
Sandy loam fault filling 0.05 40
Tectonic shear zone, schistose 2nd brokeii
gianites, disintegrated rock and gouge 0.24 42
Greyacke 1-2 mm clay in bedding planes O 21
Limestoiie 6 m m clay layei O 13
10-20 m m clay fillings 0. 1 13-14
<1 min clay filling 0.05-0.2 17-21
Limentoiie, mar1 aiid ligiiites lnterbedded lignite layers 0.08 38
Lignitelmarl contact O. 1 10
Lirnestoiie Marlaceous jaiiits, 20 m m tliick O 25 O 15-24

Ligiiite Layer behveeii lignite and clay 0.014-.O3 15-17.5

Montrnorilloiiite 80 m m seains ofbentonite (mont- 0.36 14 0.08 11
Bentonite clay morilloiiite) clay iii clialk 0.016-.O2 7.5-11.5
Scliists, qiiartzites 100-15- m m thick clay filling 0.03-0.08 32
and siliceous scliists Stiatification witli thiii clay 0.61-0.74 41
Stratification witli thick clay 0.38 31
Slates Fiiiely laminated and altered 0.05 33
Quartz 1 kaoliii 1 pyiol~isite Remoulded trianial tertr 0.042-.O9 36-38
 ength of di,icontinuitie

Where a significant thiclu~essof clay or gouge fillings occurs in rodc masses and where the shear
strength of the filled discontii~uitiesis likely to play an important role in the stability of the rock inass, it is
strongly recommended that samples of the filling be sent to a soil mechanics laboratory for testing.

4.9 Influente of water pressure

When waterpressure is present in a rock mass, the surfaces of the discontinuities are forced apart and the
normal stress o,,is reduced. Under steady state conditions, where there is sufficient time for the water pressures
in the rod<mass to reach equilihrium, the reduced normal stress is defined by q'= - nj, where ir is the water (o
pressure. The reduced normal stress 4' is usually called the effective normal stress, and it can he used in place
ofthe normal stress term sn in all of the equations presented in previous sections of this chapter.

4.10 Instantaneous cohesion and friction

Due to the historical developinent of the subject of roclc mechanics, maily of the analyses, used to
calculate factors of safety against sliding, are expressed in terms of the Mohr-Couloinb cohesion (c) and
friction angle ($j, defined in Equation 4.1. Since the 1970s it has been recognised that the relatioiiship
between shear strength and normal stress is more accurately represented by a non-linear relationship such as
that proposed hy Barton (1973). However, because this relationship (e.g. Equation 4.5) is not expressed in
terms of c and $, it is necessary to devise some means for estimating the equivalent cohesive strengths and
angles of friction from relationships such as those proposed by Barton.
Figure 4.8 gives definitions of the instantaneous cohesion c, and the instantancousj?iction aiigle 4; for a
normal stress of qs.These quantities are given by the intercept and the inclination, respectively, of the
tangent to the non-linear relationship hetween shear strength and normal stress. These quaiitities inay be
used for stability analyses in which the Mohr-Coulomb failure criterion (Equation 4.1) is applied, provided
that the normal stress 4, is reasonably close to the value used to define the tangent point.
In a typical practical application, a spreadsheet program can be used to solve Equation 4.5 2nd to
calculate the instantaneous cohesion and frictionvalues for a range ofnorinal stress values. A portion ofsuch
aspreadsheet is illustrated in Figure 4.9.

o" normal stress o,

Figure4.8: Definition of instantaneous cohesion c, and instantaneoui friction angle B, for a non-linear failuie iriterion.
 Rock Eng~neering COL

Barton shear failure criterion

Imput parameters:
Basic friction angle (PHIB) - degrees 29
Joint roughness coefficient (IRC) 16.9
Joint compressive strength (JCS) 96
Minimum normal stress (SIGNMIN) 0.360

I I
I
Normal Shear dTAU Friction Cohesive

I stress
(SIGN)
MPa
strength
(TAU)
M Pa
dSIGN
(DTDS)
angle
(PHI)
dearees
strength
(COH)
MPa

Cell formulae:
SIGNMIN = IO"(LOG(JCS)-((&)-PHIB)/JRC))
TAU = TAN((JRC*LOG(LCSlSlGN)+PHIB)*PI()I180)-
(JRCILN(1O))*(TAN((JRC*LOG(JCS/SIGN)+PHIB)*P~()/~~O)"~+~)*PI()/I
80
Figure 4.9: Prnrout of
DTDS = TAN((JRC"LOG(JCSlSlGN)+PHIB)*PI()/l80)-(JRCILN(1 O)) ipreadsheet cells and
*(TAN((IRC*LOG(JCS/SIGN)+PHIB)*P1()/180~2+1)*PI()/l8O formulae uied to calculate
shear strength, initanta-
PHI = ATAN(DTDS)*180/PI() neous fricton angle and
COH = TAU-SIGN-DTDS instantaneous cahesion for
a range of norma itressei.

Note that equation 4.5 is not valid for o

= O and it ceases to have any practical meaning for $, JRC +
loglouCS/4) > 7P. This limit can be used to determine a minimum value for o#.An upper limit for qsis
given by o=JCS.
In the spreadsheet shown in Figure 4.9, the instantaneous friction angle @,for a normal stress of ql,
has been calculated from the relationship

The instantaneous cohesion is calculated from:

ti= 7 - ontan @; (4.10)
In choosing the values of cj and $, for use in a particular application, the average normal stress q acting on
the discontinuity planes should be estimated and used to determine the appropriate row in the spreadsheet.
For many practical problems in the field, a single average value of o"will suffice but, where critica1 stability
problems are being considered, this selection should be made for each important discontinuity surface.
 Structurally controlled instability in tunnels

5.1 Introduction
In tunnels excavated in jointed rod< masses at relatively shallow depth, the inost cominon types of
failure are those involving wedges falling from the roof or slidiiig out of the sidewalls of the openings. These
wedges are formed hy intersectiiig stmctural features, such as hedding planes and joints, whicli separate the
rocl<mass into discrete but interlocl<edpieces. When a free face is created hy the excavation of the opeiiiiig,
the restraint from the surrounding rock is removed. O n e or more of these wedges can fall or slide from the
surface if the hounding planes are continuous or rock bridges along the discontinuities are hroken.
Unless steps are talcen to support these loose wedges, the stability of the bad<and walls of tlie opening may
deteriorate rapidly. Each wedge, whidi is allowed to fall or slide, will cause a reduction in the restraint and the
interlocling of the ro& mass and this, in tum, will allow other wedges to fali. This failure process will continue
until natural arching in the rock mass prevents further unravelling
. or until the opeiiing is iùll of fallen material.

The steps which are required to deal with this prohlem are:
1. Determination of average dip and dip direction of significam discontinuity sets.
2. Identification of potential wedges which can slide or fall from the had< or walls.
3. Calculation of the factor of safety of these wedges, depending upon the mode of failure.
4. Calculation ofthe amount ofreinforcement required to bring the factor ofsafety of individual wedges up
to an acceptahle level.

5.2 Identification of potential wedges

The size and shape of potential wedges in the rock mass surrounding an opening depends upoii tlie
size, shape aiid orientation of the opening and also upoii the orientation of the signiticaiit discontiiiuity sets.
The three-dimensional geometry of the problem necessitates a set of relatively tedious calculations. While
these can he performed hy hand, it is far more efficient to utilise one of the computer programs which are
available. One such program, called UNWEDGE1, was developed specifically for use in undergrouiid liard
rod<mining and is utilised in the following discussion.

5 program s available from Rocscience nc., 31 Balsarn Ave., Toronto. Ontario, Canada M4E 385 tel: 1-416-698-8217, fax: 1-416-698-0908
cmaml: iomniare~rocsciencecom
 Rock Eng~r

Consider a roclí mass in which three strongly developed joint sets occur. The average dips and dip
directions of these sets, shown as great circles in Figure 5.1, are as follows:

Joint set dip" dip direction"

Figure 5.1: An q u a l area lower hemisphere plot of great circies representing the aveiage dip and dip direit~onsof three discontinuty iets in a rock
mass Also shown, as a chain dotted ine. i8 the trend of the axis of a tunnel excavated in t h s rock m a s i The tunnel plunge is marked with a cross

It is assumed that all ofthese discontinuities are planar and continuous and that the shear strength ofthe
surfaces can be represented by a friction angle @ = 30" and a cohesive strength of zero. These shear strength
properties are very conservative estimates, but they provide a reasonahle starting point for most analyses of this
type. A more detailed discussion on the shear strength of discontinuities is given in Chapter 4.
A tunnel is to be excavated in this rock mass and the cross-section of the rainp is given in Figure 5.2.
The axis of the tunnel is inclined at 15" to the horizontal or, to use the terminology associated with structural
geology analysis, the tunnel axis plnnges at 15'. In the portion of the tunnel under consideration in this
example, the'áxis runs at 25' east of north or the trend of the axis is 025".
The tunnel axis is shown as a chain dotted line in the stereonet in Figure 5.1. The trend of the axis is
shown as 0254 measured clockwise from north. The plunge of the axis is 1S0and this is shown as a cross on
the chain dotted line reprqsenting the a i s . The angle is measured inwards from the perimeter ofthe stereonet
since this perimeter represents a horizontal reference plane.
The three structural discontinuity sets, represented by the great circles plotted in Figure 5.1, are en-
tered into the program UNWEDGE, together with the cros's-section of the tunnel and the plunge and trend
of the tunnel axis. The program then determines the location and dimensions of the largest wedges which
can be formed in the roof, floor and sidewalls of the excavation as shown in Figure 5.2.
The inaximum number of simple tétrahedral wedges which can be formed by three discontinuities in the
rodc mass surrounding a circular tunnel is 6. In the case of a square or rectangular tunnel this number is reduced
to 4. For the tunnel under consideration in this example, the arched roof allows an additional wedge to form,
giving a total of f i e . However, this additional wedge is very small and is ignored in the analysis which follows.
 Chapte lity in tunnc

Note that these wedges are the largest wedges which can be formed for the given geometrical condi-
tions. Tlie calculation used to deterinine these wedges assumes that the discontiiiuities are ubiquitous, i11
other words, they can occur anywhere in the rock mass. The joints, bedding planes and other structural
features included in tlie analysis are also assumed to be plaiiar and continuous. These conditions mean that
the aualysis will always fiiid the largest possible wedges which can form. This result can generally be cousid-
ered conservative since the size of wedges, formed in actual rock masses, will be liinited by the persistence
and the spacing of the structural features. The program UNWEDGE allows wedges to Le scaled down to
more realistic sizes if it is considered that maximum wedges are unlikely to forin.
Details of the four wedges illustrated in Figure 5.2 are given in the following table:
Wedge Weight -tonnes Failure mode Factor of Safety
Roof wedge 13 Falls O
Side wedqe 1 3.7 Slides o n JlIJ2 0.36
Side wedge 2
Floor wedqe
3.7
43
Slides o n J3
Stable -
0.52

The rooi'wedge will fall as a resnlt of gravity loading and, because of its shape, there is no restraiiit
ie three bounding discontiiiuities. This means that the factor of safeiy of the wedge, once it is released
by exc;mation of the ramp opening, is zero. In some cases, sliding on oiie plane or along the line of intersec-
tian of :mo planes may occur in a roof wedge and this will result in a finite value for the factor of safety.
The two sidewall wedges are 'cousin' images ofone another in that they are precisely the saine shapr
but di:sposed differently in space. Consequently, the weights of these wedges are identical. The factors of
safety are different since, as showii in the table, sliding occurs o n different surfaces in the two cases.
The floor wedge is completely stable and requires no further consideration.
Tlie prograin UNWEDGE is intended for use in situatio& where the in situ stresses are low and
their influence can be neglected without the introduction ofsignificant errors. These are the condi-
n which wedge failnres are most prevalent in hard roclc inasses.

-2: Wedger formed in the roof, floor and iidewails a i a ramp excavated in a jointed rock rnasr. in which the averagr d p and d p drecton of
dominant strucfural teatures are debned by the great circles plotted in Figure 5.1
 Rock Engir

Where high in situ stress levels occur in blocky rock masses, the factors of safety predicted by the
program UNWEDGE can be incorrect. In the case oftall thin wedges, the in situ stresses will tend to clamp the
wedges in place and the calciilated factor of safety will be too low. O n the other hand, for shallow flat wedges,
the calculated factor ofsafety may be too high since the high in situ stresses may force the wedge out. For most
practical tunnellingsituations these errors are not significant and can be compensated for by an adjustment of
the factor of safety. For research into failure mecbanisms and for some site appiications in which the influence
of in sim stresses is critical, for example large caverns, a more sophisticated inethod of analysis inay be required.

5.3 Support to control wedge failure

A characteristic feature ofwedge failures in blocky rock is that veiy little movement occurs in the rode
inass before failure of the wedge. In the case of a roofwedge that falls, failure can occur as soon as the base of
the wedge is fully exposed by excavation of the opening. For sidewall wedges, sliding of a few millimetres along
one plane or the line of intersection of two planes is generally sufficient to overcome the pealz strength of tliese
surfaces. This dictates that movement along the surfaces must be minimised. Consequently, the support system
has to provide a 'stiff' response to moveinent. This means that mechanically anchored rodzbolts need to be
tensioned while fully grouted rockbolts or other continuously coupled devices can be left untensioned.

5.3.1 Rock bolting wedges

For roofwedges the total force, which should be applied by the reinforceinent, should be sufficieiit to
support the full dead weight of the wedge, plus an allowance for errors and poor quality installation. Hence,
for the roofwedge illustrated in Figure 5.3, the total tension applied to the rock bolts or cables should be 1.3
to 1.5 x W, giviiig factors of safety of 1.3 to 1.5. The lower factor ofsafety would be acceptable in a teinporary
mine access opening, such as a drilling drive, while the higher factor of safety would be used in a more
perrnanent access opening such as a highway tunnel.
When the wedge is clearly identifiable, some attempt should be made to distribute the support elements
uniformly about the wedge centroid. This will prevent any rotations which can reduce the factor of safety.
In selecting the rodz bolts or cable bolts to be used, attention must be paid to the length and location of
these bolts. For grouted cable bolts, the length L*"through the wedge and the length L, in the rodí behind the
wedge should both be sufficient to ensure tliat adequate anchorage is available, as shown in Figure 5.3. In the case
of correctly grouted bolts or cables, these lengths should generally be about one metre. Where there is uncertainty
about the quality ofthe grout, longer anchorage lengths should be used. When mechanically anchored bolts with
face plates are used, the lengths should be sufticient to ensure that enough rock is available to distribute the loads
from these attachments. These conditions are automatically cbecked in the program UNWEDGE.

Figure 5.3: Rockbalt support rnechanirns for wedgei in the raof and sidewalli of tunnels
 Chapti rolled instat)ility in tunr

In the case ofsidewall wedges, the holts or cables can he placed in such a way that the shear strength
ofthe sliding surfaces is increased. As illustrated in Figure 5.3, this means that more bolts a r cables are placed
to cross the sliding planes than across the separation planes. Where possihle, these bolts a r cahles should be
inclined so that the angle 0 is between 15" and 3 P since this inclination will induce the highest shear
resistance along the sliding surfaces.
The program UNWEDGE includes a number of options for designing support for underground
excavations. These include: pattem bolting, from a selected drilling position or placed normal to the excava-
tion surface; and spot bolting, in wbich the location and length of the bolts are decided by the user for each
installation. Mechanically anchored bolts with face plates or fully grouted bolts or cables can be selected to
provide support. In addition, a layer of shotcrete can be applied to the excavation surface.
Figure 5.4 shows the ro& bolt designs for the roof wedge and one of the sidewall wedges for the
tunnel excavation example discussed earlier. For the roof wedge, three 10 tonile capacity inechaiiically aii-
chored rock bolts, each approximately 3 m long, produce a factor of safety of 1.63. The sidewall wedge,
which only weighs 3.7 tonnes, requires only a single 10 toniie roclc bolt for a factor of safety of 4.7. The
position of the collar end of the bolt sliould be located for ease of drilling.

5.3.2 Shotcrete support for wedges

Shotcrete can be used for additional support ofwedges in bloclv ground, and cai1 be very effective if
applied correctly. This is because the base of a typical wedge has a large perimeter aud hence, even for a
relatively thin layer of shotcrete, a significant cross-sectional area of the material has to be punched through
before the wedge can fail.
Consider the example illustrated in Figure 5.2. The base of the roof wedge (shown cross-hatched in
tlie upper left hand diagram) has a perimeter of 16.4 m. A layer of shotcrete 50 min thick will inean tliat a
toti11cross-sectioiial area of0.8 m 5 s available to provide support for the wedge. Assuming a relatively modest
slie ar strengtli for the shotcrete layer of 2 MPa (200 tonnes/mz) means that a wedge weighing 164 tonnes cai1
>ported. In the case of the tunnel excavation discussed earlier, tbe wedge weighs 13 tonnes and heilce
nm thick layer of shotcrete would give a high ultimate factor of safety.
It is important to eilsure that the shotcrete is well bonded to the rodc surface in order to prevent a
tion in support capacity by peeling-off of the shotcrete layer. Good adhesion to the rock is achieved by
ng the rock surface, using water only as feed to the shotcrete machine, before the shotcrete is applied.
The difficulty in using shotcrete for the support of wedges is tliat it has very little strength at the time
Aication and a period of severa1 days is required before i$ full strength can be relied upon. Siiice
:s require immediate support, the use of shotcrete for short terin stabilisatioil is clearly inappropriate.

re 5.4: Rock bolting design for the roof wedge and one of the sidewall wedges in the tunnel euarnple discussed earlier
 Rock Engi ~ursenotes

However, if a ininiinal number ofrock bolts are placed to ensure that the short term stability of the rockmass
is talcen care of, a layer of shotcrete will provide additional long terili security.
111 very strong rock with large wedges, the use of shotcrete is wasteful since only that shotcrete cover-
ing the perimeter of the wedge is called upon to provide any resistance. The ideal application for shotcrete is
in more closely joitited rock masses such as that illustrated in Figure 5.5. In sucb cases wedge failure would
occur as a progressive process, starting with smaller wedges exposed at the excavation surface and gradually
worlciiig its way hack into the ro& mass. 111 these circumstances, shotcrete provides very effective support
and deserves to he much more widely used than is currently the case.

5.4 Consideration of excavation sequence

As lias been einphasised severa1 times in this chapter, wedges tend to fall or slide as soon as tliey are
fuily exposed in an excavated face. Consequently, they require iinmediate support in order to ensure stabil-
ity. Placing this support is ali important practical questioil to be addressed when working iii bloclcy ground,
which is prone to wedge failure.
When the stmctural geology ofthe rockmass is reasonably well understood the prograin UNWEDGE
can he used to investigate potential wedge sizes and locations. A support pattern, which will secure these
wedges, cai1 then he designed and rod<bolts cai1 be installed as excavation progresses.
When dealing with larger excavations such as caverns, underground cmsher chambers or shaft sta-
tions, the prohlein of sequential support installation is a little simpler, since tliese excavations are usually
excavated in stages. Typically, in an underground crusher chamber, the excavation is started witli a top
heading which is then slaslied out hefore the remainder of the cavern is excavated by bencbing.
The rnargin sketcli shows a large opening excavated in four stages with rock bolts or cables installed at
each stage to support wedges, which are progressively exposed in the roof and sidewalls of the excavation.
The length, orientation and spacing ofthe bolts or cables are chosen to ensure that eacli wedge is adequately
supported before it is fully exposed in the excavation surface.

Figure 5.5: Ravelling

of srnall wedgei in a
cloieiy iointed rock mais.
Shotcrete can provide
effective support in such
rock massei.
 Chapi Wlled instaibility in tunnels E

, I I
/
,
', i Tap heading ,, '
I--

,, ,Slash heading , '

I I
\

a Second beiich

When dealing with large excavations of this type, the structural geology of the surrounding rod<inass
will have been defined from core drilling or access adits and a reasonable projection of potential wedges will
be available. These projections can be confirmed by additional mapping as each stage of the excavatioii is
completed. The prograin UNWEDGE provides an effective too1 for exploring the size 2nd shape ofpoteii-
tia1 wedges and the support required to stabilise them.
The inargin slcetch shows a situation iii which the support design is based upoii the largest possible
;es wliich can occur in the roof and walls of the excavation. Tliese wedges can sometimes form in rode
es with very persistent discontinuity surfaces such as bedding planes in layered sedimentary rocks. In
y metainorphic or igneous rocks, the discontinuity surfaces are not continuous and the size of the
; e s that can form is limited by the persistence of these surfaces.
The program UNWEDGE provides severa1 options for sizing wedges. One of the inost commonly inea-
sured lengths in structural mapping is the length of a joint trace on an excavation surface and one of the sizing
options is based upon this trace length. The surface area of the base of the wedge, the voluine of the wedge and the
apex height of the wedge are all calculated by the program and all of these values can I-ie edited by the user to set
a scale for the wedge. This scaling option is very important when using the program interactively for designing
support for large openings, where the maximum wedge sizes become obvious as the excavation progresses.

5.5 Application of probability theory

Tlie program UNWEDGE has been designed for the analysis of a single wedge defined by three
secting discontinuities. LVhile this is adequate for many practical applications, it does not provide any
ties for selecting the three inost critical joints in a large discontinuity population nor for aualysing the
ber aiid location of wedges, which can form along the length of an opening such as a drive.
Early attempts have been made by a number of authors, including Tyler et al (1991) and Hatzor and
Goodinan (1992), to apply probability theory to these problems and some promising results have been
obtained. The analyses developed thus far are not easy to use and cannot be considered as design tools.
Hoa.ever, these studies have shown the way for future developinent of such tools and it is anticipated that
:rful and user-friendly inethods of probabilistic analysis will be available within a few years.
 The Rio Grande project - Argentina

6.1 Introduction
The Rio Grande pumped storage project is located on the Rio Grande river near the town of Santa
Rosa de Calainucita in the Province of Cordoba in Argentina. It has an installed capacity of 1000 MW and
provides electrical storage facilities for the power grid and, in particular, for a nuclear power plant about 50
kin away from Rio Grande.
The project is owned by Agua y Energia Electrica, the principal Argentinean electrical utility organisatioii.
Preliininary feasibility studies were carried out by the owner and these were followed by detailed design studie;
by Studio G. Pietrangeli of Roine. Tbe scheme was partly financed by Italy and some of the constructioii was
done by Condote de Agua, an Italian contractor. Golder Associates were involved itl the desigti aiid supervisioti
of support installed to control the stability of most of the major underground excavations.
The inain underground facilities are located in massive gneiss of very good quality. The upper reser-
voir is impounded behind a rockfill dam and water is fed directly from the intakes down twin penstocks
which then bifurcate to feed into the four pump-turbines. These turbines, together with valves and the
control equipinent, are housed in a large underground cavern with a span of 25 m and a height of 44 m.
Draft tubes froin tlie turbines feed into twin tunnels which, with a down-streain surge shaft, form the
surge control system for this project. The twin tunnels join just downstreain of the surge tanlc and discharge
into a single tailrace tunnel with a span of 12 m and height of 18 m. This tailrace tunnel is about 6 Icm long
aiid was constnicted by a f ~ ~ lfacel - drill-and-blast top heading, with a span of 12 in and height of 8m,
followed by a 10 in bencliing operation. A view of the top heading is given in Figure 6.1.

6.2 Tailrace tunnel support

Because of the excellent quality of the gneiss, most of the underground excavations did not require
support and minimal provision for support was made in the contract documents. Assessment of underground
stability and installation of support, where required, was done on a 'design-as-you-go' basis w h i d ~proved to be
vety effective and economical. Recent reports from site, many years after the start of construction and commis-
sioning of the plant, show that there have been no problems with rod<falls or underground instability.
Decisions o n support were made on the hasis of inspection of the excavated faces by a resident team
of geotechnical engineers. Where the appearance of the face indicated that a zone of heavily jointed rocl<,
usually associated with faulting, was being entered, the top heading was reduced to a 6 m spaii by 8 m high
pilot tunnel to limit the volume of unstable roclc which could be released from the roof. This pilot tuiinel
was large enough to accoinmodate the seven-boom jumbo, as illustrated in Figure 6.4, but small enough to
limit the size of roof falls to manageable proportions. Bolting from inside the pilot heading was used to pre-
support the potentially unstable wedges and blocks in the roof.
 )ter 06 The Rio Grand~

~~ ~ ~~~ ~

dril-and-blait and. because

o1 the exiellent buality of
the rnasiive ynesi. wai
largey unsupported

Figure 6.2: ~ e r h a n i i a l i y
anchared rackbolts of the
type uied on the Rio
Grande praject. Theie bolti
weie teniioned to 70% of
their yeld laad upon
initallation and then a t a
later stage. were
re-tensioned and
fully yiauted.
 eering Co[irse notes h~yEvert Hoc

Figure 6.3: A wedge failure

in the iaof of tlie top
heading of the RIO Grande
tairace tunnel.

In the case of tlie tailrace tunnel, which is itself a large excavation, the support comprised inechani-
cally anchored and cement grouted rockbolts as illustrated in Figure 6.2, with mesh reinforced shotcrete
where required. These holts were generally installed to control the type ofwedge failure illustrated in Figure
6.3. Iii tlie case of particularly large wedges, calculations of tlie factor of safety and support requirements
were carried out on a programinable calculator, usiiig an early version of the prograin UNWEDGE.
 pter 06 The Rio Grancle project - Argentina

6.3 Support for power cavern

A cross-section of the power cavern is given in Figure 6.5 and this figure includes the Tive main
excavation stages for the cavern. Careful mapping of signiticant struckral features in the roof and walls of
the central access drive at the top of the cavern provided information for estimating potentially uiistable
blocl<sand wedges which could form in the roof of the cavern. Figure 6.6 illustrates a nuinber of such wedges
in one section of the cavern roof. At each stage of the cavern excavatioii, long rod<bolts (up to 10 rn leiigth)
were installed to stabilise wedges or blocics which Iiad been deteriniiied as being potentially unstable.

Figure 6.4: A 6 rn wide

headng drven ahead of the
tunne face ta permt pre-
reinforcernent of potentially
unitable wedgei in the
roof. The ieven-boom
jurnbo li ieen warklng in
the heading.

Figure 6.5: Cavern profile

and encavaton stagei.
 rse notes b1í Evert Hoel

Figure 6.6: A plan a i the tracei a i geologial features mapped in part of the cavern roof The shaded areai represem potentaly unitable wedger
requiring renforiement.

Because gneiss has usually undergone some tectonic deforination during its geological history, projec-
tion of structural features froin visible exposures tends to he an imprecise process. Consequently, the poteil-
tially unstable blocl<s and wedges had to be reassessed after each excavation step revealed new information.
The structural plan illustrated in Figure 6.6 Iiad to be inodified inany times during excavation and that
shown is the final plan prepared after the full cavern roof Iiad been exposed.
A general view of the cavern excavation is given in Figure 6.7. This photograph was taken when the
bulkof the cavern had been coinpleted and only a few benches in the bottom of the caverii remained to he
excavated. The enlarged top of the cavern is to accoinmodate tlie overhanging crane that is supported o11
columns froin the cavern floor. An alternative design for this cavern would have beeii to support the crane
on concrete beams anchored to the walls as is commonly done in good quality rod<.
 pter 06 Th e Rio Grant Argentina

Figure 6.7:A view of the

25 m ipan Rio Grande
pawer iavern during
exiavation of the lower
benihei.

6.4 Discussion of support design and costs

Apart from rockbolts installed to control isolated structurally controlled hlocks and wçdges i11 the
roofand sidewalls and some areas of closely jointed ro& which were shotcreted, the cavern was unsupported.
While this was successf~~l for this particularproject, it is n o t the approach which should geneially be used for
a critical excavation such as an underpround powerhouse.
 The dainage resulting froin even a small rocl<fall in such a cavern is out of ali proportion to the
savings acliieved by eliininatiiig pattern rocl<bolting and full shotcrete lining. Hence, in additioii to the
rocl<bolts installed to control stmctural instability, as described earlier, I would recoininend a normal
pattern of 25 mm diameter, 5 m long bolts (20010 of the excavation span) ori a 2.5 in grid. In addition, I
woiild recoininend the placement of 50 mm of fibre-reinforced micro-silica shotcrete over the entire roof
and upper sidewalls o f t h e cavern. Based on current north American costs, this additional support, iiivolv-
iiig approxirnately 600 rockbolts and about 300 m3 ofshotcrete, would have cost approximately US $200,000.
In terms of the overall project cost and the increased long-term security in the cavern, this would normally
be regarded as a good investinent.
In contrast, coiisider the 6 I<inlong tailrace tunnel in which the consequences of a sinal1 rocbfdll are
ininirnal. Assume that a pattern of 4 in long bolts on a 2 in grid (say 10 bolts per sectioii) aiid a 50 min
siiotcrete thickiiess Iiad been specified for the roof and upper sidewalls of the tailrace tunnei. This would
involve 30,000 holts and 5,400 m3 ofshotcrete at a total cost approachiiig US$ 5 millioii. This example
illustrates the need to give careful considerition to the function and rislcs associated with eacli undergrouiid
excavation hefore deciding upon tlie support system to be used.

6.5 Analysis using UNWEDGE program

UNWEDGE, described in the previous chapter, is ,a user-friendly inicro-coinputer prograin wliicli
can be used to analyse the geoinetty and the stabilityofwedges defined by intenectingstructural discontinuities
in the rock inass surroundiiig an underground excavation. The analysis is based upon tlie assumption tliat
the wedges, defiiled by three intersectiiig discontinuities, are subjected to gravitational loading only. In otlier
words, the stress field in the rod< mass surrounding the excavation is not taken into account. While tliis
assumption leads to soine inaccuracy in tlie analysis, it geneially leads to a lower factor ofsafety than that
whicli would occur if the in situ stresses were taken into account.
The application of the program UNWEDGE to the analysis of a potentially unstable wedge iii tlie
Rio Grande cavern is illustrated in the following discussion.

6.5.1 Input Data

The dips and dip directions of a number of planes (maximum 20) can be entered directly into tlie
pop-up table which appears when the 'Input data' option is chosen or this information can be eritered in the
form of a DIPS file. Once the data has been read into the prograin, the great circles representing tlie
discontinuities are displayed o n the screen as illustrated in Figure 6.8 and the user is prompted to select the
tiiree jointplanes to be included in tlie analysis. Once the information on these planes has been entered, tlie
unit weight of tlie rock and the shear strengths of the joints are entered. Finally, the water pressure actiiig on
the joint surface is entered. In most cases, the default water pressure of O will be cliosen but the user may
checl<tlie sensitivity of the wedge to pore water pressure by entering appropriate values.
In tlie case of the rode mass surroundingthe Rio Grande Cavern, the dips and dip directions of the
following four sets of joints are included in Figure 6.8:

1 50/131 infrequently occurring joints

2 88/225 shear joint set
3 85/264 shear joint set
4 50/345 tension joint set

Joints 2, 3 and 4 forin the central wedge illustrated in Figure 6.6 ánd these joints have been included
i11 the analysis which follows.
 , Rio Grande

Figure 6.8: Great circles representingfaurjoint setswhich oiiur in the rock Figure 6.9: ia-ordlnates used t o defne the profie of
m a s iuriaunding the RIO Grande cavern imported a i a D P S fle. the cavern.

6.5.2 Input of excavation cross-section

In setting up this analysis, the co-ordinates shown i11 Figure 6.9 were used to define the cavem profile. These
co-ordinates must be entered sequentially and must form a closed figure. n i e profile is formed froin straight line and
arc segments and a suficient number of co-ordinates should be entered to ensure that a smooth protile is generated.

6.5.3 Determination of wedge geometry

Dependingupon the shape ofthe cross-section, a maximuin ofsix wedges can be formed with three intersect-
ing joint planes. Selecting the 'View Wedges' option initiates thècalculation which determines the shape and size of
;e wedges. The two wedges formed on the cavem end walls cai1 be viewed by activating the 'View Ends' optioii.
Figure 6.10 shows the wedges formed i11 tlie case of the Rio Grande power cavem for the tliree joint platles 2,
id 4 defined in Figure 6.8. The right hand bar contains information on the weight of each of these wedges, tlie
ire mode and the calculated factor ofsafety. Obviously, the most dangerous wedge in this sit~~ation is tlhe wedge
ned in the roofwhile the wedge formed i11 the floor is stable and need not be considered furtiier in this a~ial~sis.

6.5.4 Installation and analysis of rockbolts

The program UNWEDGE automatically determined the largest wedge which can occur in the rock
mass adjacent to the excavation profile. In the case of the roof wedge, shown i11 Figure 6.10, tlie wedge
extends over the full 25 m span of the cavern and weighs 11,610 tonnes. While, in çxceptional circuiii-
:s, such wedges may occur, the limited extent ofjoints in maiiy roclc masses will restrict the size of the
:s to much smaller dimensions than those determined by UNWEDGE for the large excavations. As
ated in Figure 6.6, the span ofthe central roofwedge (measured normal to the cavern a i s ) M about 5.5
,,Athe trace length of joint number 4 (50/345) is approximately 10 m.
 Rock Enm

R ~ O
crands POW.~ cluern SELECTWEDGE
Wedgc il 1
11618 TonneE
R a * s t e % o"
J1 JZ
88,225 85/264
S.F.=8.14
NO Uedsc *2
151 I o n n e s
UEDCE Slides on
32
FORIIED 85,264
S.F.i8,85
Uedge " 3
NO
UEDCE
3 ÍOAMED

Wedge 4
8 9 2 6 Tonnes
*
Ued9e s t a b l c
S . F , = +INF

Medre n 5
631 Tannes
Slider 0"
52 53
85,264 58,345
S.F.zO.51
Uedse L 6
8.8 T o n n c s
Uedso fllls
4 6 E.Fi=B.BR

> Selcct v i e r <EESCI to aihorf>

Figure 6.10: Peripectve view of the wedges formed in the rack masi iurrounding the Rio Grande power cavern

Figure 6.11: Perspectiveview of roof wedge in the Rig Grande cavern roof. The size af this wedge has been defined by iening the trace length of the
501345 joint to 10 m.
 rrgentina

When one of the wedges is selected for more detailed ailalysis a 'Size' optioii is displayed and tliis
allows the user to define the size of the wedge in terms of the area of the face on the excavation surface, the
volume of the wedge, the height of the apex of the wedge or the length of one of the joint traces. I11 this case
a trace length of 10 m is entered for the joint defined by 50/345 and the resulting wedge is illustrated i11
Figure 6.11. This wedge weighs 255 tonnes and will require abont seven 50 tonne capacity h l l y grouted
cables to givr a factor of safety of about 1.5 which is considered appropriate for a cavern of this q p e .
UNWEDGE allows the user to add a layer of shotcrete and calculates the factor of safety increase as
a result of such an addition. Since tlie shotcrete can only be added once the surface of the wedge is fully
exposed it is not taken into account in calculating the support required to stabilise the wedge. The increase
insafetyfactorwhich occurs after the shotcrete has set can be regarded as a long term bonus and it does allow
the user to choose a sliglitly lower factor of safety for the immediate support of the wedge.
 A slope stability problem in Hong Kong

7.1 lntroduction
In the early 1970s a series oflandslides occurred in Hong Kong as a result ofexceptionally heavy rains.
These slides caused some loss of life and a significant amount of property damage. Consequently, an exten-
sive review was carried out on the stability of soil and rock slopes in the Territory.
During this review, a rode slope on Sau Mau Ping Road in Kowloon was identified as being potentially
nnstahle. The stability ofthis particular slope was critical because it was located immediately across the road
from two blodcs of apartments, each housing approximately 5,000 people.
Figure 7.1 gives an general view down Sau Mau Ping Road, showing the steep rock slopes on the left
and the apartment blocks on the right.
The concern was that a major rock slide could cross the road and damage the apartment blocks. In
order to decide upon whether or not the residents of the two apartment blocks should be evacuated, the two
questions which required an immediate response were :
1. What was the factor of safety of the slope under normal conditions and under conditions which could
occur during an earthquake or during exceptionally heavy rains associated with a typhoon.
2. What factor of safety could he considered acceptable for long ter'm conditions and what steps would be
required in order to achieve this factor of safety.

7.2 Description of problem

The rock mass in which the slope adjacent to the Sau Mau Ping Road was cnt is an unweathered
granite with exfoliation or sheet joints similar to those illustrated in
Figure 7.2. These joints are parallel to the surface of the granite and the spacing between successive
joints increases with increasing distance into the rock mass. Undercutting of these sheet joints can cause a
rodc slide such as that illustrated in Figure 7.3.
During excavation of the original slopes for the Sau Mau Ping Road, a small rock slide was induced by
blasting.+he surface on which this failure occurred is illustrated in Figure 7.4. Blasting, such as that used in
civil construction in an urban environmeut, does not impose very large loads on rock slopes and it can be
assuined that the factor of safety of the slope was close to unity.
The potentially unstable slope under consideration is visible in the background of this photograph. It
is obvious from this photograph that the sheet joint surface continues under the potentially unstable slope.
Hence, from the evidence of the small scale failure, it can be deduced that the factor of safety of the slope in
question is not very high.
 Chapter 07 A slc problem i rI Hong Kon

Figure 7.1: Aview down

- ~ Mau
u Ping Road in
Kowlaon rhawing
trnent biocki across
oad from the steep
iock slopei.

Figure 7.2: Sheet ~ o n t i n g

n granite. These featurei,
sometimes refeired ta as
'anion skin' jointi, are the
result of exfoliation
processes duiing cooling
of the granite.
 Rock Englni e e r i n g Coi

Figure 7.3: A major rock rlide on a hlghway rauied by the under-
cutting of sheet j o n t i in a granite ilape. In hard rocks such as yran-
ite, failure can aciur very suddenly if the factor af rafety af the
ilope icloie to 1. Ariie in groundwater levels during a heavyitorm
ar c e jacking in winter may be rufficient t o induce failure.
Figure 7.4: The faiure iurface defned by a iheet jont iurfare on
which a small slide oiiurred during blarting of the original cut ilope
for IheSau Mau Piny Road. The potentialiy unitableiiope under c o n ~
sideration i viribie in the back-ground.

The geoinetry of the slope is illustrated in Figure 7.5 which shows a 60 m high slope with three 20 m high
benches. The overall slope angle is 50" and the individual bench faces are inclined at 70" to the horizontal. An
exfoliation joint surface dips at 35" and undercuts the slope as shown in the figure. The slope face strilíes parallel
to the underlying exfoliation surface and hence the slope can he analysed by means of a two-dimensional model.
Tension cracks are frequently observed behind the crest of slopes which have a fictor of safety of iess
than about 1.2. These cracks are dangerous in that they allow water to enter the slope at a particularly critical
location. Unfortunately, in the case ofthe Sau Mau Ping slope, recently cultivated marlíet gardens located on
the top of the slope made it impossible to determine whether or not such tension cradcs were present and
heilce it was decided to carry out two sets of analyses - one with and one without tension craclci. These
analyses were carried out for both the overall slope and for individual benches.

7.3 Limit equilibrium models

At the time of this investigation, n o rock mechanics facilities existed in Hong Kong and n o diainond
drilling or laboratory testing had ever been carried out on the granitic rodís in which this slope had been
excavated. Consequently, the problem was tackled on the basis of a crude forin ofrisií analysis, using simple
 Chapler 07 A slc problern tn Hong Kon~

analytical models to predict the response o f the slope to a range of possible conditions. The two models are
defined in Figure 7.6 and Figure 7.7.

The Syinbois and dimensions used in these models are as follows:

Svrnbol Parameter Dimensions
Factor of safety against sliding along sheet joint Calculated
Height of the overall slope or of each bench 60 m or 20 m respectvely
Angle of slope face, rneasured from horizontal 50"
Angle of failure surface, measured from horizontal 35"
Depth of tension crack Calculated (rn)
Depth of water in tension crack or on failure surface Variable (rn)
Horizontal earthquake acceleration 0.08 g (proporton of g)
Unit weight of rock 0.027 MN/m3
Unit weight of water 0.01 MNlm3
Weight of rock wedge resting on failure surface Calculated (MN)
Base area of wedge Calculated (rn2)
Uplift force due to water pressure on failure surface Calculated (MN)
Horizontal force due t o water in tension crack Calculated (MN)
Cohesive strength along sliding surface Variable (MNlm2)
Friction angle of sliding surface Variable (degrees)
Force applied by anchor system (if present) Specified (MN)
Inclination of anchor, anti-clockwise from normal Specified (deqrees)

Note that this i s a two-dimensional analysis and these dimensions refer to a 1 metre thicb slice tlirough the
slope. I t i s also important to recognise that this analysis considers only force equilibriuin and assumes that
a l i forces pass through the centroid o f the wedge. In otlier words, moment equilibriuin i s n o t considered
in this analysis. While this i s a simplification o f the ach-al situation depicted in Figure 7.6 and Figure 7.7,
the erron introduced are n o t considered to be significant, given the uncertainty o f the otlier input data
used in these analyses.

Figure 7.5: Geornetry aaurned for the two-dirneniional analysis a i the Sau Mau Ping Road slope
 Rock E n g i rieering Coi

where

Figure 7.6: Factor of Safety ialiuation for a ilope with no tension crack
 Cha ty problem In Hong KOI

Water pressure

where

A=-H - 2
sin

figure 7.7: Factor of Safety ialculation for a slope with a water-filled tension irack.
 Rock Engini?erfng Cou

In Figure 7.7, the depth z of the tension crack is calculated by equation 7.6. This equation is ob-
tained hy minimising equation 7.5 with respect to the tension crack depth z (Hoek and Bray 1974). This
minimisation is carried out for a dry slope and the accuracy of equation 7.6 decreases as the water depth in
the tension crack increases. However, for the purposes of this analysis, the estimate given by equation 76
is considered acceptable.

7.4 Estimates of shear strencith

O n e of the inost critica1 steps in any limit equilibriuin analysis is the deterrnination or tlie estirna-
-
tion of the shear strength . parameten
. (c and @)for the surface along which it is anticipated that sliding will
take place. 111the case ofthis slope o n Sau Mau Ping Road, n o information on shear strength was available
at the time o f t h e initial studies and so estimates had to be made on the hasis ofpuhlished inforination foi
similar rocks.
Hoek and Bray (1974) published a plot, reproduced in Figure 7.8, of cohesive strengths and friction
angles for rod<s and soils, based upon the results of published back analysis of slope failures. Superimposed
on this plot is an elliptical zone which encompasses the estimated range of shear strength for sheet joints in
unweathered granite. In choosing this range it was considered that the friction angle probahly ranges frorn
30" for very smooth planar surfaces to 45" for rough or partly cemented surfaces. The cohesive strength c is
more difficult to estimate and the range of 0.05 to 0.2 MPa was chosen on the basis of the results of back-
analyses of slope failures, plotted in Figure 7.8.

UndiSlYrbed h a m m ~ h
maises w h no major
siivCII"I1pa"ems
dfpping lawams slope

Ertimated mnge 0f rhear Uodisluibed ioc!, mssrsr

otrenglh for aheet joints wilh a few ~ l m ~ l ~ m s
in unweathered granite dipping fowsrds siow

joinled hardioch
di31uibed by blasling
o i encesa loading
o
• Wealherdsaff roch or

o * C1.q
0.0 * t* , * , .o , , SOil
Saod
O 6 10 15 20 25 30 35 O 45 50

Fiiction ange + - degreer

Figure 7.8: Reta tionihip between friction angles and coheiive strengths mobilised at faiiure of sopei in varous materiais. The plotted pointi were
obtained from published inforrnation from the baik analyis of slope faiiures. (After Hoek and Bray 1974).
 y problern I n Hong KOI

Some readers may be surprised that a cohesive strength has been assumed for joint surfaces wliicb obvi-
ously Iiave no tensile strength or 'stiduness' as would be found in a clayey soil. In fact, this assuined cohesive
strength is defined by ihe intercept, on the shear strength axis, of a tangent to a curvilinear Mohr envelope. Tliis
curvature is the result of the interlocking of asperities on the inatching surfaces of the joints and tlie iiicrease in
shear strength given hy this interlodang plays a cnicial role in thc stability of slopes such as that uiider consider-
ation in this chapter. A h l l discussion on the shear strength of discontinuities in rod<masses is given in Chapter 4.

7.5 Estimate of earthquake acceleration

Hong Kong is not considered a highly seismic region but relatively miiior earthqual~sare not uii-
knowii in the region. Consequently, it was felt that some allowance should be inade for tlie possible iiitlu-
ence of earthquake loading on the stability of the Sau Mau Ping slope.
The traditional method of incorporating the acceleration induced by earthqualces or large blasts in
slope stability analyses is to add an outward force aW to the forces acting o n the slope (see Figure 7.6 and
Figure 7.7), where a is the acceleration as a proportion of g, the acceleration due to gravity. Tliis 'pseudo-
static' forin of analysis is Icnown to be very conservative but, in tlie case of the Sau Mau Piiig slope, this
conservatism was not considered to be out ofplace.
In discussion with local engineers and geologists, the consensus opinion was that tlie horizoiital
acceleration which could be induced by a 10 year return period earthqualce i11 the region would be approxi-
mately 0.08 g. This value was used in all of the sensitivity analyses discussed i11 tlie following sectioiis.

7.6 Analysis of mobilised shear strength

One inethod for assessing the stability of slopes is to calculate the shear strength that wo~ildbe
mobiilised at failure and to compare this strength with the shear strength which is available along the failure
surfas:e. In the case of the Sau Mau Ping slope, this was done by suhstituting F= 1 in equations 7.1 and 7.5
and r;olving for the cohesive strength c and the friction angle I). The results of this aiialysis are plotted in
-.
:ure 7.9. The estimated range of available shear strength (froin Figure 7.8) is also shown on tliis ploi.
Figure 7.9 shows that only two of the cases analysed resuit in conditions where the sliear streiigtli
~bilisedat failure falls within the estimated range of available shear strength. These two cases are desig-
ted 2 and 4 and they are for fully saturated slopes, with and without tension cradcr.

7.7 Decision on short-term stability of the Sau Mau Ping slope

From the results of the sensitivity study described above it was concluded that instability of tliis particu-
slope could occur if the slope was fully saturated and subjected to earthquake loading. Typlioons occur
,era1 times every year in Hong Kong and the intensity of'precipitation during these events is certainly suffi-
nt to saturate the slopes. As discussed earlier, minor earthquakes do occur in the regioii but they are tiot very
frequent. Consequently, the chance of siinultaneous saturation and earthqude loading was considered to be
sm all aiid in was concluded that tliere was no serious short-term threat ofinstability of the Sau Mau Ping slope.
In discussion with the highway authorities in Hong Kong, the following decisions were inade:
No evacuation of the residents of the two apartment blocks, located across the street froin tlie slope iii
question, would be carried out.
Horizontal drainage holes would be drilled into the slope face to penetrate tlie potential failure surface in
an attempt to reduce uplift pressures in the slope.
Piezoineters would be installed in holes drilied froin the top of the slope. These piezometers would be
ineasured iegularly during periods of significant rainfall and the road would be ciosed to traffic if water
levels rose to levels decided by the engineers responsible for the project.
 Rock Englneering Cot

4. An iiivestigation would he carried out into the most effective remedial ineasures to stahilise the slope for
the long-term.

Figure 7.10 shows the drilling of the horizontal drain holes into the slope face and Figure 7.1 1 shows
the drilling of the vertical holes into which the piezometers were installed. These piezometers were moni-
tored for the next few years, while preparations for the final stabilisation of the slope were made, and the
road was closed to traftic on two occasions when water levels were considered to be dangerously high.

7.8 Evaluation of long-term remedial measures

While the short-term threat of instahility was considered to he small, the longer-term stahility of the
slope was considered to he unacceptahle and a study was carried out to evaluate various options for stahilising
the slope. It was agreed that a factor of safety of 1.5 was required to ineet long term requirements. The
following alternatives were considered:

1. Reducing the height of the slope.

2. Reducing the angle of the slope face
3. Drainage of the slope.
4. Reinforceinent of the slope.

The limit equilihrium inodels defined in Figure 7.6 and Figiire 7.7 were used for this evaluation aiid
the results are plotted in Figure 7.12.

1 Estimated ran!?e of rhear

strenoth for sheet joints Legend'

1. Overall slope with dry

tension crack (&=O j
Overall slope with wa-
m ter-fiiled tension crack
5 015 (2, = 2)
Overall s l o ~ ewith no
c
u
////I tension crack. drv (H,=
.-a 0)
Overall siooe with no
g oia ////i ' tenslon crack. saturated
v
(H". = H)
Sinole
" bench with drv
tension crack iz ,-" = OI,
6 Single bench with wa-
ter-filled tension crack
(I,= Z)
7 Single bench with no
o 00
O 5 10 15 20 25 30 35 40 45 50

Fnction angle $ - degrees

Figure 7.9: Cornparsan of the ihear'srrength mobilised by faiiure under vaiious condittoni wth rhe eaimated ihear itiength avalabie on iheet
laints n unweathered granite.
 pter 07. A ! ty problern In Hong Kor

Figure 7.10: Drlling horzonta drain haies into the fareof one of the Figure 7.11: Driling vertrai diamond core holei n t a the Sau Mau
benches of the Sau Mau Png ilope. Ping ilope. These holes were uied for geoterhncal investgatan
purpaieiand alio foi theinitalation of piezometen in therockrnaii

Leyend:
1. Reduslion in slope
heiaht H foi sooe w8th
teniion Crack
2. Reduclion in slope
beight n for sopa with
notenrion crack
3. Reduction of slope face
angie v, foi slope wilh
tenrion cmck
4 Reducfion i" sope face
anyle v,for slope with
na tenrion crack
5 orasnage of riope witn
iension ciack
6 . Drainage af slope with
na teniian crack
7. Renfarcemeni 01 rope
With tension cr8ck
8. Reinfarcemenr of rlope

Figure 7.12: Evaluaton of remedial optioni to inireare the stabilrty of the slape.
 Rock Engln eering Cou

In calculating the factors of safety shown in this figure, the shear strength was inaintained constant
and was defined by c = 0.10 MPa and I$ = 35'. Similarly, an earthquake acceleration of n = 0.08 g was used
for ali the analyses. The percentage change refers to the ratios of slope height, slope angle and water depth to
the origiiial dimensions defined in Figure 7.5:
In the case of the reinforcement options, the percentage change refers to the ratio of anchor force T to
the weiglit of the wedges (24.8 MN for the slope witli the tension cracli and 28.6 M N for the slope with no
teiision cradc). The anchor inclination was kept constant at O =I$ = 35". This anchor inclination gives tlie
ininimum anchor load for a dry slope and it can be determined by ininimising equations 7.1 or 7.5 with
respect to O.
Tlie curves presented in Figure 7.12 show clearly that some remedial measures are much more effec-
tive than others and it is worth exainining each of the options in turii.
Curves 1 (slope with tension crack) and 2 (slope without tension cradc) show that reduction of the slope
height is not ai1 effective solution to the problem. In order to achieve tlie required factor of safety of 1.5,
the slope height would have to he reduced by 50%. If this solution were to be adopted, it would be more
practical to excavate the entire slope since inost of the volume of the rock to be excavated is contained in
the upper half of the slope.
Cume 3 (slope with teiision cradc) sliows that reduction of the slope angle is a very effective remedial
measure. The required factor of safety of 1.5 is achieved for a reduction of less than 25% of the slope
angle. In other words, a reduction of the overall slope face angle froin 50° to 37.5" would achieve tlie
desired resiilt. This finding is generally true and a reduction in the face angle of a slope is usually an
effective reinedial step. In the case of slopes under construction, using a flatter slope is always one of tlie
prime choices for achieving greater stability.
Curve 4 (slope without tension crad<) is an aiioinaly and demonstrates that calculations can sometimes
produce nonsense. The reduction in factor ofsafety shown by this curve is a result of the reduction in tlie
weight of the sliding block as the face angle is reduced. Since the water pressure on the sliding surface
reinains constant, the effcctiveshess acting o n the sliding surface decreases and hence the frictional coinpo-
nent of the resisting forces decreases. When a very thin sliver of roclc remains, the water pressure will float
it offthe slope. The problem with this analysis iies in the assuinption that the blodi is coinpletely imper-
meable and that the water remains trapped beneath the failure surface. Iii fact, the block would brealc LI^
long before it floated aiid hence the water pressure acting on the failure plane would be dissipated.
Curves 5 and 6 show that drainage is not a very effective option for eitherof the slope inodels cotisidered.
In neither case is a factor ofsafety of 1.5 achieved. This is sometliing of a surprise since drainage is usually
one of the most effective and economical remedial measures. The reasons for tlie poor perfortnance of
drainage in tliis case is due to the combination of tlie geometry of tlie slope and the shear strength of the
failure surface.
Cumes 7 and 8 show that, for both slope models considered, slope reinforcement hy means ofrockboits
or cables can be an effective remedial measure. The anchor force required for a factor of safety of 1.5
would be about 100 toniies per inetre of slope length for the slope with n o tension crak.

7.9 Final decision on long term remedial works

The two most attractive options for long term reinedial worlis on this slope are reinforcement by
ineans of cables or bolts or reduction of the slope face angle. The first option was finally rejected hecause of
the high cost and because of the uncertainty ahout the long terin corrosion resistance of reinforceinent
which could be placed in the slope. This latter concern may not have heen justified but, considering the very
poor qiiality of some of the construction in Hong Kong at the time of this study, it was decided that the risk
was not worth talcing.
 y problem In Hong Kor

The option finally chosen was to reduce the slope face angle down to 3 5 O by excavating the entire
blockresting on the failure surface and hence removing the problein entirely. Since good quality aggregate is
always required in Hong Kong it was decided to work this slope face as a quarty. It took severa1 years to
organise this activity and, during this time, the water levels in the slope were monitored by means ofpiezom-
eters. Although the road was closed twice during this period, n o major problems occurred and tlie slope was
finally excavated badc to the failure plane.
 Factor of safety and probabilíty of failure

8.1 Introduction
How does one assess the acceptability of an engineering design? Relying on judgeinent alone can lead
to one of the t i o extreines illustrated in Figure 8.1. The first case is economically unacceptable wliile tlie
example illustrated in the lower drawing violates all normal safety standards.

8.2 Sensitivity studies

The classical approach used in designing engineering stnictures is to consider the relationship be-
tween tlie capacity C (strength or resisting force) of the eleinent and the demand D (stress or disturbing
force). The Factor of Safety of the stnicture is defined as F= C/D and failure is assurned to occ~irwhen F is
less thail 1.
Rather than base an engineering design decision on a single calculated factor of safety, an approach
which is frequently used to give a more rational assessment of the risks associated with a particular design isto
carry out a sensitivity study. This involves a series of calculations in which each significant parameter is varied
systematically over its maximum credible range in arder to determine its influence upon the factor of safety
This approach was used in the analysis of the Sau Mau Ping slope in Hong Kong discussed i11 the-
previous cliapter. It provided a useful means of exploring a range of possibilities and reachiiig practical
decisions on some difíicult problems. O n the following pages this idea ofsensitivity studies will be extended A

to the use of prohability theory aild it will be shown that, even with very limited field data, practical, useful
inforination can be obtained froin an analysis of probahility of failure.
i

Figure 8.1: Rockbalting aiteinativei nvalving individual judgement. (Drawings based upan a ranoon in a brochure on rockfals published by the
Department of Mnes a i Western Auitrala.)
 and Probat

8.3 An introduction to probability theory

A complete discussion on probability theory exceeds the scope ofthese notes aiid tlie techniques discussed
on the followingpages are intended to introduce the reader to the subject and to give an indicatioii of the power
of these techniques in engineering decision malcing. A more detailed treaunent of this subject will be found in a
book by Harr (1987) entitled Reliability-baed design in civil engneering. A paper on geotechnical applications of
probability theory entitled 'Evaluating calculated risk in geotechnical engineering' was published by Whitman
(1984) and is recommended reading for anyone with a serious interest i11 this subject. Pine (1992), Tyler et al
(1991), Hatzor and Goodman (1993) and Carter (1992) havepuhlished papers on the application of probability
theory to the analysis ofproblems encountered in underground miniiig and civil engineering.
Most geotechnical engineers regard the subject of probability theory with doubt and suspicion. At
least part of the reason for this mistrust is associated with the language which has been adopted by those who
specialise in the field of probability theory and risk assessment. The following definitions are given in an
attempt to dispel some of the mystely which tends to surround this subject.

Random vuriables: Parameters such as the angle of friction of rocl< joiiits, the uniaxial coinpressive
strength of rock specimens, the inclination and orientation of discontinuities in a rock mass aiid tlie iiiea-
sured in situ stresses i11 the rocksurrounding an opening do not have a siiigle fxed value but may assume any
number ofvalues. There is n o way of predicting exactly what the value of one of these parameters will be at
any given location. Hence these parameters are described as random variables.

Probability distribution: A probability density function (PDF) describes the relative lilcelihood that a
random variable will assume a particular value. A typical probability density function is illustrated opposite.
In this case the random variable is continuously disuibuted (i.e., it can take on all possihle values). The area
uuder the PDF is always unity.
An alternative way of presenting the same information is in the form of a cumulative distribution
function (CDF), which gives the probability that the variable will have a value less than or equal to the
selected value. The C D F is the integral of the corresponding probability density function, i.e., the ordinate
at x, o11 tlie cumulative distribution is the area under the probability density function to the left oix,. Note
the f(x) is used for the ordinate of a PDF while FJx) is used for a CDF.
One of the most common graphical representations of a probibility distribution is a liistograin in
which the fraction of all observations falling within a specified interval is plotted as a bar above that interval.
Datuanalysis: For many applications it is not necessary to use all of the information contaiiied in a distribu-
tion function and quantities summarised only by the dominant features of the distribution inay be adequate.

J, íy>

1 ,. .,

Probability density
function (PDF)
./- Xi
Curnulative distribution
function (CDF)
I
 Rock Engtneering COL

The ramplemean orexpectedvalne orfrst moment indicates the centre of gravity of a probability distribu-
tion. A typical application would be the analysis of a set of results x,,x, ........,x from uniaxial strength tests
carried out in tlie laboratory. Assuming that there are n individual test values x , the meauX is given hy:

The rampleuanances2or thesecondmomcntabontthemean o f a distribution is defined as the mean ofthe

square of the difference between the value of x; and the mean value 2 . Hence:

Note that, theoretically, the denominator for calculation ofvariance ofsamples should be n, not ( n - 1).However,
for a tinite number of samples, it can be shiwn that the correction factor nl(n-I), known as Bessei's correction,
gives a better estimate. For practical purposes the correction is only necessary when the sample size is less than 30.
The srandard dmiiation s is given by tlie positive square root of the variance I. In the case of the
commonly used normal distribution, about 680l0 of the test values will fall witliin an interval defined by the
rncan? onestandarddmiiationwhile approximately 95% of all the test results will fall within the range defined
by the mean i two rtundard deviationr. A small standard deviation will indicate a tightly clustered data set
while a large standard deviation will be found for a data set in which there is a large scatter about tbe mean.
The coeflcient ojvariation (COV) is the ratio of the standard deviation to the mean, i.e. COV = s / i .
COV is dimensionless and it is a particularly usehil measure of uncertainty. A small uncertaiiity would typically
be represeiited by a C O V = 0.05 while considerable uncertainty would be indicated by a COV = 0.25.

Normal distribution: The normal or Gaursian distribution is the most common type of probability
distribution function and the distributions of many random variables conform to this distribution. It is
generally used for probabilistic studies in geotechnical engineering unless tliere are good reasons for selecting
a different distribution. Typically, variables which arise as a suin o f a nuinber of random effects, iloile of
which dominate the total, are normally distributed.
The prollem of defining a normal distribution is to estimate the values of the governing parameters
which are the tnie mean (p)and true standard deviation (a).Generally, the best estimates for these values are
given by the sample mean and standard deviation, determined from a number of tests or observations.
Hence, from equations 8.1 and 8.2:

It is important to recognise that equations 8.3 and 8.4 give the most probable values of p and 0 and
iiot necessarily the true values.
Obviously, it is desirable to include as many samples as possible iil any set of observations but, in
geotechnical engineering, there are serious practical and financia1 limitations to the amount of data which
can be collected. Consequently, it is often necessary to make estimates on the basis ofjudgement, experiente
or from comparisons with results publislied by others. These difficulties are often used as an excuse for not
using probabilistic tools in geotechnical engineering but, as will be shown later in this chapter, useful results
,.
can still be obtaiiled froin very limited data.
Having estimated the mean p and standard deviation a , the probability density function for a normal
distribution is defined by:
 of Safety :ind Probabi Iity of failur

-
As will be seen later, this range o f - < x < =- can cause problems when a normal distribution is used as a basis
for a Monte Carlo analysis in which the entire range of values is randomly sampled. This can give rise to a
few very small iiumbers (sometimes negative) and very large numbers which, i11 certain analyses, can cause
numerical instability. In order to overcome this problem the normal distribution is sometimes tmncated so
that only values falling within a specified range are considered valid.
There is n o closed form solution for the cumulative distribution function (CDF) which must by
found by numerical integration.

Other distributions: In addition to the commonly used normal distribution there are a number of
alternative distributions which are used in probability analyses. Some of the most useful are:

Beta diriributions (Harr, 1987) are very versatile distributions which can be used to replace almost any of
the common distributions and which do not suffer from the extreme value problems discussed above
because the domain (range) is bounded by specified values.
Exponentialdistributioni are sometimes used to define events such as the occurrence of earthquakes or
roclibursts or quantities such as the length ofjoints in a rodi mass.
Lognomzaldishibutions are useful when considering processes such as the crushing of aggregates in which the
final particle size results from a number of collisions of particles of many sizes moving in different direc-
tions with different velocities. Such multiplicative mechanisms tend to result in variables which are logiior-
mally distributed as opposed to the normally distributed variahles resulting from additive mechanisins.
Weibuidistributions are used to represent the lifetime of devices in reliability studies or the outcome of
tests such as point load tests on rock core in which a few very high values may occur.

It is no longer necessary for the person starting out in the field ofprobability theory to iinow and understand
the mathematics involved in all of these probability distributions since commercially available software
programs can be used to carry out many of the computations automatically. Note that the author is not
advocating the blind use of 'black-box' software and the reader should exercise extreme caution is using such
software without trying to understand exactly what the software is doing. However there is no point in
writing reports by hand if one is prepared toLspend the time learning how to use a good word-processar
correctly and the same applies to mathematical software.
One of the inost useful software packages for probability analysis is a program called BestFitl wliich
has a built-i11library of 18 probability distributions and which can be used to fit any one of these distribu-
tions to a given set of data or it can be allowed automatically to determine the ranlung of the fit of all 18
distributions to the data set. The results from such an analysis can be entered directly into a companion
program called @RISK which can be used for risk evaluations using the techniques described below.

'BeitFitfor Windowiand itsiornpanion program @RISKfar Micraioit Excel of Lotus 1-2-3 (for Windowior Macint0ih)are availablefrom the Paliiade
Corporation, 31 Deiker Road, Newiield, New Yoik 14867, USA. Fax nurnber 1 607 277 8001.
 Rock Engil urse notes I

Sampliiig techniques: Consider a problem in which the factor of safety depends upon a number of
random variables such as the coliesive strength c, the angle of friction f and the acceleration a due to earth-
quakes or large blasts. Assuming that the values of-these variables are distributed about their ineans iu a
inaniier whjch can be described by one ofthe continuous distribution functions such as the normal distribu-
tion described earlier, the problem is how to use this information to deterinine the distributioii of factor of
safety values and the probability of failure.
The Monte Carlo method uses random or pseudo-random nuinbers to sample from probability dis-
tributioiis and, if sufficiently large iiuinbers of samples are generated and used in a calculation such as that
for a factor of safety, a distribution of values for the end product will he generated. The terin 'Monte Carlo'
is believed to have been introduced as acode word to describe this hit-and-miss technique used during secret
worl< on the development of the atoinic bomb during World War I1 (Ilarr 1987). Today, Monte Carlo tech-
niques can he applied to a wide variety of problems involving random beliaviour and a number of algo-
rithms are availahle for generating random Monte Carlo samples from different types of input probability
distributions. With highly optimised sofmare programs such as @RISK, problems involving relatively large
samples can be run efficiently on most desktop or portable computers.
The Latin Hypercube sampliilg technique (Imam et al(1980), Startzman and Watterbarger (1985)) is a
relatively receiit development which gives comparable results to the Monte Carlo technique but with fewer
samples. The method is based upon stratified sampling with random selection within each stratuin. Typically
an analysis using 1000 samples obtained by the Latin Hypercuhe technique will produce comparable results
to ali analysis usiug 5000 samples ohtained using the Monte Carlo method. Both techniques are incorpo-
rated in the prograin @RISK.
Note that both the Monte Carlo and the Latin Hypercube techniques require that the distribution of
a11 the iiiput variables should either be known or that they be assumed. When n o information on the
distribution is available it is usual to assume a normal o r a truncated normal distribution.
The Generalired Point Estimate Method; developed by Roseiiblentli (1981) and discussed in det.ail by
Harr (1987), can be used for rapid 'calculation of the mean and standard deviation of a quantity such as a
factor of safety which depends upon random behaviour of input variables. Hoek (1989) discussed the appli-
cation of this technique to the analysis of surface crown pillar stability while Pine (1992) has applied this
technique to the analysis of slope stability and other mining problems.
To c~lculatea quantity such as a factor of safety, iwo point estiinates are inade at one standard devia-
tion on either side of the mean (@a) from each distribution representing a random variable. The factor of
safety is calculated for every possible combination of point estimates, producing 2" solutions where n is the
nuinber of randorn variables involved. The mean and the standard deviation of the factor of safety are tlien
calculated froin these 2* solutions.
Wliile this technique does not provide a full distribution of the output variable, as do the Monte
Carlo and Latin Hypercube methods, ir is very simple to use for problems with relatively few randorn
variables and is useful when general trends are being investigated. When the probability distribution func-
tion for the output variable is Bnown, for example, from previous Monte Carlo analyses, the mean and
standard deviation values can be used t o calculate the complete output distribution .

8.4 Probability of failure

In the case of the Sau Mau Ping slope problem the factor of safety of the overall slope with a tension
crack is defined by:
1. Fixed dimensions:
Overall slope height H=60m
Overall slope angle v=
, 50.
 Chapte

Failure plane angle

Unit weight of rock
Unit weight of water

% 2. Random variables .Mean values

Friction atigle on joint surface @ = 35"
Cohesive strength ofjoint surface c = 10 tonnes/m2
Depth of tension crack z= 14m
Depth of water in tension crack zqo=z/2
Ratio of horizontal earthquake
to gravitational acceleration a = 0.08

Figure 8.2 illustrates the layout of a Microsoft Excel spreadsheet with plots of the probability
distribution functions of the random input variables and of the calculated factor of safety.It is worth
discussing each of the plots in detail to deinonstrate the reasoning behind the choice of tlie prohalility
distribution functions.
1. Friction angle $- A truncated normal distribution has been assumed for this variable. The mean is assuined
to be 35" which is the approximate centre of the assumed shear strength range illustrated in Figure 7.8.
The standard deviation of 5" implies that about 68% of the friction angle values defined by the distribu-
tion will lie between 30" and 4P. The normal distribution is tmncated by a minimum value of 15" and a
maximum value of 60" which have been arbitrarily chosen as the extreme values represented by a smooth
slidensided surface and a fresh, rough tension fracture.
2. Cohesivestrengthc - Again using the assumed range ofshear strength values illustrated in Figure 7.8, a value of
10 tonnes/in%as been chosen as the mean cohesive strength and the standard deviation has been set at 2
t o n n e d m h n the basis of this diagram. In order to allow for the wide range of possible cohesive strengths
the minimum and maximum values used to tmncate the normal distribution are O and 25 toimes/tn2
respectively. Those with experience in the interpretation of laboratory shear strength test results may argue
that tlie friction angle @ and the cohesive strength c are not independent variables as has been assumed iii
this analysis. This is because the cohesive strength generaily drops as the friction angle rises and vice versa.
The program @RiSK allows the user to define variables as dependent but, for the salce of simplicity, the
friction angle 4 and the cohesive strength chave been kept independent for this analysis.
3. Enxion crack depthz - Equation 7.6, defining the tension crack depth, has been derived by minimisation
of equation 7.5. For the purposes of this analysis it has been assumed that this value of z (14 m for the
assumed conditions) represents the mean tension crack depth. A truncated normal distribution is as-
sumed to define the possible range of tension crack depths and the standard deviation has been arbi-
trarily chosen at 3 m. The miniinum tension crack depth is zero but a value of 0.1 m has been chosen
to avoid possible numerical problems. The maximum tension craclc depth is given by z = H(1-tan$) /
tanilr/= 24.75 m which occurs when the vertical tension craclc is located at the crest of the slope.
4. Wate~depthz,~in tensioncradi - The water which would fill the tension craclc in this slope would come froin
direct surface run-off during heavy rains. In Hong Kong the heaviest rains occur during typhoons and it
is likely that the tension crack would be completely filled during such events. The probability of occur-
reuce of typhoons has been defined by a tmncated exponential distribution wbere the meanwater depth
is assumed to be one half the tension crack depth. The maximum water depth cannot exceed the tension
era& depth z and, as defined by the exponential 'distribution, this value would occur very rarely. The
minimum water depth is zero during dry conditions and this is assumed to be a frequent occurrence.
 Rock E n gineering:
~ Cc

overall Sau Mau Ping S l o p with awrter-fi1l.d tension craek

Analysis DI

Fbred quanfi6es C ~ I c ~ I a fqusnfi6es

ed
üverall sbpa hdght H=W m zcalc= 14.01 m
Overall slop angle psil= 50 degrees A= m.18 sq.m
Failure plane angie psip = 35 dcgrees W = 2392.38 tonnes
Una W h t o1 r a k gammar = 2.6 Vcu.m U = 192.89 tmnes
Una wight ofwaler gammaw = 1 Vcu.m V = 11.57 tannes

Factor o1 Safety = 1.33

Rendom variables
araw Mean std. dev. Mln. MW. Dlstr. Assumed distribunon
Fridion q i e phl 35.00 5.W 15.03 W.W 35.03 Truncatednormai
Cohasivs strength coh 10.03 2.W 0.W 25.00 10.W Tmnatednormai
TensiM crack depth z 14.01 3.W 0.10 24.75 14.01 Tnrncatedmrmai
D q t h of waler m 7.W 0.W 14.01 4.81 Twnc. ugionential
Eaithquake acc. alpha 0.08 0.00 0.16 0.05 Trunc. q n e n t i a i

15 25 35 45 55
-
Fnctian angle degrees o 5 10 15 20
-
Coherion 1onneslsq.m

C
m

O 5 1O 15 20 25 O 5 10 15
-
~ensloncrack depth rnebei -
Waterdepth in cack rnelrps

p~

Figure 8.2: Spreadsheet for @RiirLatin Hypercube analysiiof Sau Mau Ping siopewith distributionsof random input variables and the probabilitydeniity
function for the cacuiated factor of safew. The probabiltyoffailure.ihown bythe daik region for Fcl,ir approximately 7% for theasumed condtioni
 and Probatiility of failu

Note that the water depth z , ~is defined in terms of the tension crad; depth z which is itself a random
variable. In calculating z the program @RISK first samples the truucated norinal distribution defining z
and then combines this value with the information obtained from sainpling the truncated exponential
distribution to calculate z*,,.
5 . Ratioofhor&ontalearthquakeaccelerahon tograuitationalaccelp/ahona - T h e frequent occurrence of earthquakes
ofdifferent magnitudes can be estimated by means of an exponential distribution which suggests that large
earthquakes are very rare while small ones are very common. In the case of Hong Kong local wisdoin
suggested a 'design' horizontal acceleration of 0.08g. In other words, this leve1 of acceleration could be
anticipated at least once during the operating life of a civil engineering structure. A rough rule of thuinb
suggests that the 'maximum credible' acceleration is approximately twice the 'design' value. Based upon
these very cmde guidelines, the distribution of values of w. used in these calculations was defined by a
tmncated exponential distribution with a mean value of a= 0.08, a maximum of 0.16 and a miniinuin of O.

Using the distributions shown in Figure 8.2, the program @RISK was used, with Latin Hypercube
sampling to carry out 1,000 iterations o n the factor of safety. The resulting probability distribution was not
a smooth curve, indicating that an insufficient number ociterations had been performed for this coinbina-
tion of variables. A second analysis was carried out using 10,000 iterations and the resulting factor of safety
distribution is plotted in the lower right hand corner of Figure 8.2. Note that this distribution closely re-
semliles a norinal distribution.
From the statistical tables produced by the program @RISKit was determined that tlie prohability of
failure for this slope is approximately 7%. This value is given by the ration of the area under the distribution
cnrve for F<l (shown in black in Figure 8.2) divided by the total area under the distribution curve. This
means that, for the combination of slope geometry, shear strength, water pressure and earthquake accelera-
tion parameters assumed, 7 out of 100 similar slopes could be expected to fail at some time during the life of
the slope. Alternatively, a length of 70 m could be expected to fail in every 1000 m of slope.
This is a small but not negligible risl; of failure and it confirms the earlier conclusion, discussed i11
Chapter 7, that this slope was not adequately stable for a densely populatedregion such as Kowlooii. Inci-
dentally, a risk of this magnitude would probably be acceptable in an open pit mine, with limited access of
trained miners, and even o n a rural road. The decisions reached in Chapter 7 o n the long term stabilisation
measures for this slope are considered appropriate and the type of analysis described here could be used to
evaluate the effectiveness of these remedial measures.
 Analysis of rockfall hazards

9.1 Introduction
Roclcfalls are a major hazard in rock cuts for highways and railways in mountainous terrain. While
rodcfalls d o not pose the same leve1 of economic risk as large scale failures which can and d o close major
transportation routes for days at a time, the number of people killed by roclzfalls tends to be of the same
order as people Icilled by all other forms of rock slope instability. Badger and Lowell(1983) summarised the
experience of the Washington State Department of Highways. They stated that 'A significant number of
accidents and nearly a half dozen fatalities Iiave occurred because of rockfalls in the last 30 years ... [and] ...
45 percent of all unstable slope ~ r o b l e m sare rockfall related'. ~ u n g and
r Evans (1989) note that, in Canada,
there have heen 13 rockfall deaths in the past 87 years. Almost a11 of these deaths have been o n the mountain
highways of British Columbia.
I11 some circumstances, where no alternative access is available, it becomes necessary to carry out
construction activities on highway slopes while maintaining partia1 traffic flow. This increases the roclcfall
hazard many times and can only be considered acceptable if the road can be closed during the most hazard-
ous constructioil activities.

9.2 Mechanics of rockfalls

Roclcfalls are generally initiated by some climatic or biological event that causes a change in the forces
acting on a rodz. These events may include pore pressure increases due to rainfall infiltration, erosion of
surrounding material during heavy rain storms, freeze-thaw processes in cold climates, chemical degradation
or weathering of the rock, root growth or leverage by roots moving in high winds. In an active construction
environment, the potential for mechanical initiation of a rockfall will probably be one or two orders of
magnitude higher than these climatic and biological initiating events.
Once movement of a rockperched on the top of a slope has been initiated, the inost iinportant factor
controlling its fall trajectory is the geometry of the slope. In particular, dip slope faces, such as those created
by the sheet joints in granites, are itnportant because they impart a horizontal compoileilt to the path taken
by a rock after it bounces on the slope or 1011s off the slope. The inost datigerous ofthese surfaces actas 'ski-
jumps' and impart a higli horizoiital velocity to the falling rock, causing it to bounce a Ioilg way out froin the
toe of the slope.
Clean faces of hard unweathered rock are the most dangerous because they d o not retard the move-
ment of the falling or rolling roclz to any significant degree. 'On the other hand, surfaces covered in talus
material, scree or grave1 ahsorb a considerable amount of the energy of the falling rode and, in inany cases,
will stop it completely.
 181sof rockf all hazards

Figure 9.1: A rock siope on

a rnauntain highway.
Rockfalli are a major hazard
an iuch highwayi.
 Rock Engirieering Cor~ r s enotes t

Figure 9.2: Construction

an an active roadway,
iornetlmes neieisary when
there ii abiolutely
no alternative aciess.
increases the rockfall hazard
many times ovei that for
slopei without constructlan
or for iituations in which
lhe road i a n be iioied
during ionstrurtion.

This retarding capacity of the surface material is expressed mathematically hy a term called the iocfl-
cientofrestitution. The value of this coefficient depends upon the nature of the materiais that form the impact
surface. Clean surfaces of Iiard rod< have high coefficients of restitution while soil, gravel and completeiy
decoinposed granite have low coefficients ofrestitution. This is why gravel layers are placed on catch benches
in order to prevent further bouncing of falling rodes.
 er 09 Anal fall hazards

Other factors such as the size and shape of the rock boulders, the coefficientsoffriction of the rod<surfaces
and whether or not the rodc breaks into smaller pieces on impact are all of lesser significance than the slope
geometry and tbe coefficients of restitution descrihed above. Consequently, relative crude rockfall simulation
models, such as the program written hy Hoek (1986), are capable ofproducing reasonably accurate predictions of
roddall trajectories. Obviously more refined models will produce better results, provided that realistic input infor-
mation is available. Some of the more recent roddall models are those of Bozzolo et a1 (1988), Hungr and Evans
(1988), Spang aiid Rautensuauch (1988), Azzoni et a1 (1995) and the University of Toronto'.
Most of these roclcfall models include a Monte Carlo simulation technique to vary the paraineters
included in the analysis. This technique, named after the gamhling casinos of Monte Carlo, is similar to the
random process of throwing dice - one for each parameter being considered. A typical rockfall analysis is
reproduced in Figure 9.3.
Tbe analysis illusuated in Figure 9.3 was carried out using the program developed by Hung?. n i e
principal advantage of this program is that it includes a plasticity function which absorbs the impact energy of
boulders, depending upon their size. Tbis simulares the process in which large bo~ilderswill he damaged or will
indent the impact surface while small boulders will hounce off the impact surface with little energy loss.
In the analysis reproduced in Figure 9.3b, the road surface was assigued a coefficient of restitution
close to zero so that any hounce after the first impact was suppressed. The purpose of this study was to
determine the spread of first impacts so tbat an effective catch ditch and barrier fence could be desigi~ed.

9.3 Possible measurs which could be taken to reduce rockfall hazards

9.3.1 Identification of potential rockfall problems
It is either possible or practical to detect all potential rockfall hazards by any techniques currently in
use in rock etigineering.
In some cases, for example, when dealing with boulders on the top of slopes, the rodcfall hazards are
obvious. However, the most dangerous types ofrock failure occur when a blode is suddenly released froin an
apparently sound face hy relatively small deformations in the surrounding rode inass. This can occur when
the forces actiiig across discontinuity planes, which isolate a block from its neighboun, change as a result of
water pressures in the discontinuities or a reduction of the shear strength of these planes because of Iotig
term deterioration due to weathering. This release of 'keyblodcs' can sometimes precipitate roclcfalls of sig-
nificant size or, in extreme cases, large scale slope failures.
While it is not suggested that rock faces should not be carefully inspected for potential roclcfall prob-
leins, it should not be assumed that all rockfall hazards will be detected by such inspections.

9.3.2 Reduction of e n e r g y levels associated w i t h excavation

Traditional excavation methods for hard rodc slopes involve the use of explosives. Even when very
carefully planned controlled blasts are carried out, high intensity short duration forces act o n the rock inass.
Blocks and wedges w h i k are at risk can he dislodged by these forces. Hence, ali ohvious method for reduc-
ing roclcfall hazards is to eliminate excavation by blasting or by any other method, such as ripping, whicli
imposes concentrated, sliort duration forces or vibrations on the roclc mass.
Mechanical and hand excavation methods can used and, wliere massive roclc lias to be brolcen, chetni-
cal expanding rock breaking agents may be appropriate.
 .ering Coui

a)Typical irajectory for a 1000 kg boulder.

b) Trajectories for 1000 boulders weighing behveen 200 and 20,000

kg released within the range shown in a) above.

Figure 9.3:Typcal exarnple of a rockfall trajectory for a granite slope.

 iys~sof rockfall hazards

9.3.3 Physical restraint of rockfalls

If it is accepted that it is not possible to detect or to prevent all rockfalls, then methods for restraining
those rocl<falls,which do occur, must be considered. These methods are illustrated in Figure 9.4.
Berms are a very effective rneans of catching rockfalls and are frequently used on permanent slopes.
However, berms can only be excavated from the top downwards and they are of limited use in minimisitig
the ris]<of rodifalls during construction.
Rocksheds or avalanche shelters are widely used on steep slopes above narrow railways or roadways. An
effective shelter requires a steeply sloping roof covering a relatively narrow span. In the case of a wide multi-Iatie
highway, it may not be possible to design a rockshed stmcture with suffcient suength to withstand iarge roddalls.
Rock traps work well in catching rod<fallsprovided that there is sufficient roorn at the toe of the slope
to accommodate these rock traps. In the case of very narrow roadways at the toe of steep slopes, there may
not be sufficient room to accommodate rock traps. This restriction also applies to earth or rode fills and to
gabion walls or massive concrete walls.
Catch fences or barrier fences in common use are estimated to have an energy absorption capacity of
100 kNmi. This is equivalent to a 250 kg rock moving at about 20 metres per second. More robust barrier
fences, such as those used in the European Alps4, have an energy absorbing capacity of up to 2500 1<Nm
whicli rneans tbat tbey could stop a 6250 kg boulder moving at approximately 20 metres per second. Details
of a typical high capacity net are illustrated in Figure 9.5.
Another restraint system which merits further consideration is the use of mesh draped over the face.
This type of restraint is commonly used for permanent slopes and is illustrated in-Figure 9.6. The mesh is
draped over the rock face and attached at severa1 locations along the slope. Tbe purpose of the mesh is not to

,h
stop rocl<fallsbut to trap the falling rodi between the inesh and the rock face and so to reduce the horizontal
velocity component which causes the rock to bounce out onto the roadway below.

b. Rockshed

ure 9.4: Poiiible measures to reduce the damage due to rockfalls. (After Spang 1987).

'The kinetic energy ai a falling bady is given by 0.5 x mas8 x velacity2.

Wire mesh fence which incorporate5 cabes and energy absorbing slipping jointi i8 manufaitured by Geobrugg Proteitive Systerns. CHM590
nanihorn, Switzerland, Fax: 141 71466 81 50.
 r Evert Hoek

a: Anchor grouted into rock

with cables attached.

b: Geobrugg ring net sliown

restraining a boulder. These nets
can be designed witli energy
absorbiiig capaciiies of up to
2500 kNm wliicli is equivaleiii to
a 6 tome boulder rnoving at 20 rn
per second.

c: Geobrugg energy ahsorbing rinç.

Wlien subjected to iinpact loadinç
tlie ring deforms plastically and Figure 9.5: Detailr of
a rockfai net syitem
absorbs tlie energy of tlie bouldec manufactured by

. Probably the rnost effective perinanent rockfall protective system for most highways is the construc-
tion of a catch ditch at the toe o f the slope. The base of this ditch should be covered by a layer o f grave1 to
absorb the energy o f falling rocl<s and a sturdy barrier fence shouid be placed between the ditcb and the
 Chapter 1s of rockfal

roadaiay. The loc;itionof the barrier fence can be estimated by means of a roclcfall analysis such as that used
to cal'culate the tr;ajectories presented in Figure 9.3. The criterion for the ininimum distaiice benveen the toe
ofthe siope and the rode fence is that no rocks can be allowed to strike the fence before their Icinetic energy
has heen diminished by the first impact on the grave1 layer in the rode trap.
Asimple design chart for ditch design, based upon work hy Ritcliie (1963), is reproduced in Figure 9.7.

9.4 Rockfall Hazard Rating System

Highviay and railway construction in mountainous regions presents a special challenge to geolo-
gistsandgeotechnical engineers. This is because the extended length of these projects makes it difficult to
obtain sufficient information to permit stability assessments to be carried out for each of the slopes along
the route. This means that, except for sections which are identified as particularly critical, most highway
slopes tend to be designed o n the basis of rather rudimentaty geotechnical analyses. Those aiialyses which
are carried out are almost always concerned with the overall stability of the slopes against major slidiiig or
toppling failures which could jeopardise the operation of the highway or railway. It is very rare to find an
analysis of rockfall hazards except in heavily populated regions in highly developed countries such
as Switzerland.
In recognition of the seriousness of this problem and of the difficulty ofcarrying out detailed inves-
iigations and aiialyses o11the hundreds of kilometres of mountain highway in the western United States aiid
Canada, highway and railway departments have worlced o n classification schemes which can be carried out
by visual inspection and simple calculations. The purpose of these classifications is to identify slopes which
are particularly hazardous and which require urgent remedial work or further detailed study.
In terms ofrockfall hazard assessment, one of the most widely acceptedqs the Rocl<fallHazard Ratitig
System (RHRS) developed hy the Oregon State Highway Division (Pierson et al 1990). Table 9.1 gives a
summary of the scores for different categories included in the classification while Figure 9.8 shows a graph
~hvhich can be iised for more refined estimates of categoiy scores. The curve shown in Figure 9.8 is calculated
from the equationy = Ywhere, in tliis case, x = (Slope height- feet)/25. Similar curves for other category
scores can be c;ilculated from the following values of the exponent x.

SIope height x = slope height (feet) / 25

nverage vehicle risk x = % time / 25
Sight distance x = (120 - % Decision sight distance) / 20
Roadway width x = (52 - Koadway width (feet)) / 8
l o c k size x =Blocl< size (feet)
>lume x = Volume (cu.ft.) / 3

9.4.1 Slope Height

inis item represents the vertical height of the slope not the slope distance. Rocks o n high slopes have
more potential energy thau rocks on lower slopes, thus they present a greater hazard and receive a higher
ratiiig. Measurement is to the highest point from which rockfall is expected. If rocks are corning froin the
tiatural slope above the cut, use the cut height plus the additioiial slope height (vertical distaiice). A good
approximation ofvertical slope height can be obtained using the relationships shown below.

5Thii5yrtem has been adopted by the States of Oregon, Washington. Nex Mexico and Idaho and. in ilighty modified farm. by Caiifornia, Coiorado
andBritish Columbia.
 tock Engine

FB"CB ar waii

Figure 9.6: Rackfall cantrol rneasuies (After Faokes and Sweeney 1976)

Figures taken from FHWA Manuai 'Rock Siopes'

Novembei 1991. USDOT Chaplei 12 Page 19.

Figure 9.7: Roikfall dtch d e i g n chart based upon work by Ritchie (1963)
 151sof rockfall hazards

Table 9.1: Rockfall Hazard Ratina Svstem

I RATING CRITERIA AND SCORE

CATEGORY I POINTS 3 I POINTS 9 I POINTS 27 I POINTS 81

SLOPE HEIGHT 1 25 FT I . 50 FT 1 75 FT 1 100 FT

DITCH EFFECTIVENESS Good Moderate Limited No

catchment catchment catchment catchment

AVERAGE VEHICLE RISK 25% 75% 50% 100%

o f the time of the time of the time of the time

PERCENT OF DECISION Adequate site Moderate sight Limited site Very imited
SIGHT DISTANCE distance. 100% distance.80% distance. 6 0 % sight distance.
of low design of low design of low design 40% of low
value value value design value

ROADWAY WIDTH INCLUDING

PAVED SHOULDERS
/ 44feet
I 36 feet / 28 feet
I 20 feet

-
W
STRUCTURAL
CONDITION
Discontinuous
jointi. favorable
orientation
Discontinuous
joints. random
orientation
Discontinuous
joints, adverse
orientation
Continuous
joints, adverse
orientation
vi
Q
ROCK FRICTION Rough, irregular. Undulating Planar Clay infilling or
slickensided
I
STRUCTURAL Few differential . Occasional Many erosion Major erosion
w CONDITION erosion features erosion features features features
rn
Q
u DIFFERENCE IN Srnall Moderate Large Extreme
EROSION RAIES difference ' difference difference difference

BLOCK SIZE

QUANTITY OF 3 cubic 6 cubic 9 cubic 12 cubic

ROCKFALUEVENT vards

WATE AND PRESENCE L0w t o Moderate High High

WATER ON SLOPE moderate precipitation or precipitation precipitation
precipitation; short freezing or long and long
n o freezing periods or freezing periodz freezing
periods. n o intermittent or continua1 periods or
water on water o n water on continua1
slope S ~ O D ~ slope water on
slope and
lonq freezinq
I perrods -
ROCKFALL HISTORY Few falls Occasionai falls Many falls Constant falls
 eering Caiirse notes I:

I0 75 100

Sope Height - leet

Figure 9.8:Category wore graph for ilope height,

I DITCH
X
HIGHWAY

"
I H.1
TOTAL sLoPE H E m n =

where X = distance beween angle rneaiurernenls

H . ! = height of lhe instrument.

Figure 9.9: Measuiement of sope height. 1

 iII hazards

9.4.2 Ditch Effectiveness

The effectiveness of a ditch is measured by its ability to prevent falling rodc from reaching the road-
way. In estimating the ditch effectiveness, the rater should consider severa1 factors, such as: 1) slope height
and angle; 2) ditcli width, depth and shape; 3) anticipated block size and quantity of rockfall; 4) impact of
slope irregularities (la~inchiiigfeatures) on falling rocks. It's especially important for the rater to evaluate the
impact of slope irregularities because a launching feature can negate the benefits expected from a fallout area.
The rater should first evaluate whether any of the irregularities, natural or man-made, o n a slope will launch
falling rocla onto the paved roadway. Then based on the number and size of the launching features estimate
what portion of the falling rocks will be effécted. Valuable information on ditch perforrnance can be ob-
tained from maintenance personnel. Rating points should he assigned as follows:

3 points Good Catchment. All or nearly all of falling rocks are retained in the catch ditch
9 points Moderate Catchment. Falling rocks occasionally reach the roadway.
27 points Limited Catchmmt. Falling rocks frequently reach the roadway.
81 points No Catchment. No ditch or ditch is totally ineffective. ALI or nearly ali falling
rocks reach the roadway.
Reference should also be made to Figure 9.7 in evaluating ditch effectiveness.

9.4.3 Average Vehicle Risk (AVR)

This category measures the percentage of time that a vehicle will be present in the rockfall hazard
zone. The percentage is obtained by using a formula (shown below) based o n slope length, average daily
traffic (ADT), and the posted speed limit at the site. A rating of 100% means that o n average a car can be
expected to be within the hazard section 100% of the time. Care should be taken to measure only the length
of a slope where rodcfall is a problem. Over estimated lengths will strongly skew the forinula results. Where
high ADT's os longer slope lengths exist values greater than 1000io will result. When this occurs it ineans that
at any particular time more than one caris present within the measured section. The formula used is:

ADT ícarslhour) x Slope Lensth ímiles) x 100°10 = AVR

Posted Speed Limit (miles per hour)

9.4.4 Percent of Decision Sight Distance

The decision sight distance (DSD) is used to determine the length of roadway in feet a driver must
have to make a complex or instantaneous decision. The D S D is critical when obstacles on the road are
difficult to perceive, or when unexpected or unusual manoeuvres are required. Sight distance is the sbortest
.
distance along a roadway that an object of specified height is continuously visible to the driver.
Throughout a rockfall section the sight distance can change appreciably. Horizontal and vertical
highway curves along with obstmctions such as rock outcrops and roadside vegetation can severely limit a
driver's ability to notice a rock in the road. To determine where these impacts are most severe, first drive
through the rockfall section from both directions. Decide which direction has the shortest line of sight. Botli
horizontal and vertical sight distances should be evaluated. Normally an object will be rnost obscured wheti
it is located just beyond the sharpest part of a curve. Place a six-inch object in that position ori the fogiine os
on the edge of pavement if there is no fogline. The rater then walks along the fogline (edge ofpavement) in
the opposite direction of traffic flow, measuring the distance it takes for the object to disappear wheri your
eye lieight is 3.5 ft above the road surface. This is the measured sight distance. The decision sight distance can
be determined by the table below. The distances listed represent the low design value. The posted speed liinit
thro~ighthe rockfall section should be used.
 Rock Engineenng Cor

Posted Srieed Limit ímoh) Decision Siaht Distance íft)

These two values can he suhstituted into the formula below to calculate the 'Percent ofDecision Sight Distance.'

Actual Site Distance ( ) x 100% = 010

Decision Site Distance ( )

9.4.5 Roadway Width

This dirnension is measured perpendicular to the highway centreline from edge of pavement to edge
of pavement. This measurement represents the available manoeuvring room to avoid a rod<fall. This inea-
surement should be the minimum width when the roadway width is not consistent.

9.4.6 Geologic Character

The geologic conditions ofthe slope are evaluated with this category. Case 1 is for slopes where joints,
bedding planes, or other discontinuities, are the dominant structural feature of a rode slope. Case 2 is for
slopes where differential erosiou or oversteepened slopes is the dominant condition that coiltrols rockfall.
The rater should use whichever case best fits the slope when doing the evaluation. If both situations are
present, both are scored but only the worst case (highest score) is used in the rating.

Case 1
Strnctnral Condihon Adverse joint orientation, as it is used here, involves considering sucb things as
rock friction angle, joint filling, and hydrostatic head ifwater is present. Adverse joints are those that cause
hlod<, wedge or toppling failures. 'Continuous' refers to joints greater than 10 feet in length.

3 points Di~continuousJoints,Favourable OrientationJointed rock with no adversely oriented joints, bedding

planes, etc.
9 points Discontinuous]oints Random Orientation Rock slopes with randomly oriented joints creating a
three-dimensional pattern. This type of pattern is likely to have some scattered blocks with ad-
versely oriented joints but no dominant adverse joint pattern is present.
27 points Discontinnousoint~,Adverse Oriatation Rock slope exhibits a prominent joint pattern, bedding
plane, or other discontinuity, with an adverse orientation. These features have less than 10 feet of
continuous length.
81 points ContinuousJoints, Aduersc Orientation Rock slope exhibits a dominant joint pattern, bedding plane,
or other discontinuity, with an adverse orientation and a length of greater than 10 feet.

Rock Friction This parameter directly effects the potential for a block to move relative to another.
Friction along a joint, bedding plane or other discontinuity is governed by the macro and micro roughness
of a surface. Macro roughness is the degree of undulation of the joint. Micro roughness is the texture of the
surface of the joint. In areas where joints contain highly weathered or hydrothermally altered products,
where movement has occurred causing slickensides or fault gouge to form, where open joints dominate the
 is of rockfal

slope, I,r where joints are water filled, the rockfall potential is greater. Noting the failure angles from previous
rod<fdl1s on a slope can aid in estimating general rocli friction along discontinuities.

3 points Rocigh, Irregular The surface of the joints are rough and the joint planes are irregular enough to
cause interloclíing. This rnacro and rnicro roughness provides an optimal friction situation.
9 points Undulating Also macro and micro rongh hut without the interlocking ability.
27 points Planar Macro smooth and inicro rough joint surfaces. Surface contains no undulations. Friction
is derived stiictly from the roughness of the rock surface.
81 points Clay Infllingor Slidieniided Low friction materials, such as clay and weathered rock, separate the
rodí surfaces negating any micro or macro roughness of the joint planes. These infilling materials
have much lower friction angles than a rodí on rodí contact. Sliclíensided joiilts also have a very
low friction angle and helong in this category.

Case 2
Structural Condition This case is used for slopes where differential erosion or oversteepening is the domi-
~nditionthat leads to rockfall. Erosion features include oversteepened slopes, unsupported rod<uiiits or
d resistant rocks on a slope that rnay eventually lead to a roddall event. Roddall is caused by a loss of
t either locally or throughout the slope. Common slopes that are susceptible to this condition are:
laye!red units containing easily weathered rocli that erodes undermining inore durable rock; talus slopes;
liigh ly variable nnits such as conglornerates, mudflows, etc. that weather causing resistant rock aild hlocks to
fall, and rocklsoil slopes that weather allowing rocks to fall as the soil matrix material is eroded.

3 points FmDiffmntialErosion Features Minor differentia! erosioii features that are not distributed through-
out the slope.
9 points OccaionalErorionFeatldres Minor differential erosion features that are widely distrihuted through-
out the slope.
27 points Many Erosion Features Differential erosion features are large and numerous throughout the slope.
81 points M&r Erosion Features Severe cases such as dangerous erosion-created overhaiigs; or sigiiificantly
oversteepened soil/rock slopes or talus slopes.

Différence in Erosion Rates The Rate of Erosion on a Case 2 slope directly relates to the potential for a
roclífall event. As erosion progresses, unsupported or oversteepened slope conditions develop. The
impaci:of the commou physical and chemical erosion processes as well as the effects of man's actions shouid
be conisidered. The degree of hazard caused hy erosion and thus the score given this category should reflect
I1ow q uickly erosion is occnrring; the size of ro&, blocks,,or units heing exposed; the frequency of rockfdll
events; and the amount of material released during an event.

3 points SmallDzfference The difference in erosion rates is such that erosioii features develop over inany
years. Slopes that are near equilibrium with their environment are covered by this category.
9 poin ts Moderate Dférmce The difference in erosion rates is such that erosion features develop over a few years.
27 poi nts LargeDiffnence The difference in erosion rates is such that erosion features develop annually.
81 uoints ExtremeDzz!erence The difference in erosion rates is such that erosion features develop rapidly

9.4.7 Block Size or Quantity of Rockfall Per Event

This measurement should he representative ofwhichever type of roclífall event is inost likely to occur.
If individual bloclís are typical of the rockfall, the block size should he used for scoring. If a mass of hloclís
 Rock Engir

tends to be the dominant type of rodifall, the quantity per event should be used. This can be deterrnined
from the maintenance history or estimated froin obsewed conditions when no history is available. This
measurement will also be beneficia1 in determining remedial measures.

9.4.8 Climate and Presence of Water on Slope

Water a t ~ dfreeze/thaw cycles both contribute to the weathering and movement of rock materials. If
water is known to flow continually or intermittently from the slope it is rated accordingly. Areas receiving
less than 20 inches per year are 'low precipitation areas.' Areas receiving more than 50 inches per year are
considered 'high precipitation areas.' The impact of freezelthaw cycles can be interpreted from linowledge of
the freezing conditions and its effects at the site.
The rater should note that the 27-point category is for sites with long freezing periods or water prob-
lems such as high precipitation or continually flowing water. The 81-point category is reseived for sites that
have both long freezingperiods and one of the two extreme water conditions.

9.4.9 Rockfall History

This information is best obtained from the maintenance person responsible for the slope in question.
It directly represents the known rodífall activity at the site. There may be n o history available at newly
constmcted sites or where poor documentation practices have been followed and a turnover of personnel
has occurred. In these cases, the maintenance cost at a particular site may be the only information that
reflects the rockfall activity at that site. This information is an important check on the potential for future
rocl<falls. If the score you give a section does not compare with the roclífall history, a review should be
performed. As a better database of rockfall occurrences is developed, more accurate conclusions for the
rockfall potential can be made.

3 points Few Falls - Roclífalls have occurred several times according to historical information but it is not
a persistent problem. If rockfall only occurs a few times a year or less, or only during severe
storms this category should be used. This category is also used if n o rod<fall history data is
available.
9 points OccasionalFalls - Rockfall occurs regularly. Rod<fall can be expected several times per year and
during inost storms.
27 points Many Falls - Typically rockfall occurs frequently during a certain season, such as the winter or
spring wet period, o r the winter freeze-thaw, etc. This category is for sites where frequent roclifalls
occur during a certain season and is not a significant problem during the rest of the year This
category may also be used where severe rockfall events have occurred.
81 points Constant Falls - Rockfalls occur frequently throughout the year. This category is also for sites
where severe rockfall events are common.

In addition to scoring the above categories, the rating team should gather enough field information to
recommend which rockfall remedial measure is best suited to the rockfall problem. Both total fixes and
hazard reduction approaches should be considered. A preliminary cost estimate should be prepared.

9.5 Risk analysis of rockfalls on highways

The analysis ofthe risk ofdamage to vehicles or the death ofvehicle occupants as a result ofroddalls on
highways has not received vety extensive coverage in the geotechnical literature. Papers which deal directly with
the probability of a slope failure event and the resulting death, injury or damage have been published by Hunt
(1984), Fell (1994), Morgan (1991), Morgan et a1 (1992) and Vames (1984). Most of these papers deal with
 ill hazards

landslides rather than with rod<falls. An excellent study of risi< analysis applied to roddalls on highways is
contained in an MSc thesis by Christopher M. Bunce (1994), submitted to the Department of Civil Engineer-
ing at the University ofAlberta. This thesis reviews risk assessment methodology and then applies this inethod-
ology to a specific case in which a rocldall killed a passenger and injured the driver of a vehicle.

9.5.1 RHRS rating for Argillite Cut

Bunce carried out a study using the Rockfall Hazard Rating System for the Argillite Cut in which the
rockfall occurred. A suininary of his ratings for the section in which tlie roddail happened and for the entire cut is
presented in Tahle 9.2. The ratings which he ohtained were 394 for the rocldall section and 493 for the entire cut.
The RHRS system-does not iiiclude recommendations o n actions to be taken for different ratings.
This is hecause decisions on remedial action for a specific slope depend upon many factors such as tlie
budget allocation for highway work which cannot be talcen into account in the ratings. However, in personal
discussions with Mr Lawrence Pierson, the principal author of the RHRS, I was informed that in tbe State of
Oregon, slopes with a rating of less than 300 are assigned a very low priority while slopes with a rating iil
excess of 500 are identified for urgent remedial action.

9.5.2 Risk analysis for Argillite Cut

Bunce (1994) presented a number of approaches for the estimation of the annual probability of a
fatality occurring as a result of a rockfall in the Argillite Cut. Some of these approaches are relatively sophis-
ticated and I have to question whether tliis level ofsophistication is consistent with the quality of the input
informatioli which is availahle o n highway projects.
One approach which I consider to be compatible with the rockfail problem and with quality of input
information availahle is the event tree analysis. This technique is best explained by means of the practical
example ofthe analysis for the Argillite Cut, shown in Figure 9.10. I have inodified the event tree presented
by Bniice (1994) to make it simpler to follow.

! 9.10: The Argtlllte

t an Highway 99 in
i h Calurnbia, Canada
 Rock Engin

Table 9.2: RHRS ratingsfor Argillite Cut on Highway 9 9 in British Columbia (after Bunce 1994
Section where rockfall occurred Rating for entirt
Parameter Value Rating Value Rating
Slope height 36 1O0 35 100
Ditch effectiveness Limited 27 Limited 27
Average vehicle risk 7 1 225 100
Sight distance 42 73 42 73
Roadway width 9.5 17 9.5 17
Geological structure Very adverse 81 Adverse 60
Rock friction Planar 2.7 Planar 27
Block size 0.3 m 3 I rn 35
Cimate and water High precip. 27 High precip. 27
Rockfall history Many falls 40 Many falls 27

Total score 394 493

In the event tree analysis, a probability of occurrence is assigned to each event in a sequence which could
lead to arocldall fatality. For example, in Figure 9.11, it is assumed that it rains 33% of the time, that rockfalls
occur o11 5% of rainy days, that vehicles are impacted by 2% of these rod<falls, that 50% of these impacts are
significant, i.e. they would result in at least one fatality. Hence, the annual probability of fatality resulting from
a vehicle being hit by a rocbfall triggered hy rain is given hy (0.333 " 0.05 ' 0.02 " 0.5) = 1.67"104.
The event tree has been extended to consider the annual probability of occurrence ofone, two and
three or more fatalities in a single accident. These probabilities are shown in the final column of Figure 9.11.
Since there would be at least one fatality in any of these accidents, the total probability of occurrence ofa
single fatality is (8.33 + 5.56 + 2.78)"10-'= 1.7 " 10-', as calculated above. Tlie total probability of at least mo
fatalities is (5.56 + 2.78) " lO~'= 8.34 " 10.' while the probability of three or more fitalities remains a t 2.78 *
10"s shown in Figure 9.1 1.
Suppose that it is required to carry out construction work on the slopes of the Argillite cut and that,
because this is an important access road to an intemational ski resort area, it is required to maintain traffic
flow duritig this construction. It is assumed that the construction worklasts for 6 months (50°/0 of a year) and
that rodzfalls are initiated 20% of tlie working time, i.e. o n 36 days. All other factors in the event tree remain
the same as those assumed in Figure 9.11. T h e ~ e s u l t sof this analysis are presented'in Figure 9.12 which
sliows that there is an almost ten fold increase in the rislz of fatalities from rockfalls as a result of the ongoiug
construction activities. (Note thai this is a hypothetical example only and that n o such construction activi-
ties are planned o11 this higliway).

9.6 Comparison between assessed risk and acceptable risk

The estimated annual probabilities of fatalities from rodcfalls, discussed in the previous sections, have
little meaning unless they are compared with acceptable risk guidelines used o n other major civil engineering
construction projects.
O n e of the earliest attempts to develop an acceptable risk criterion was published by Whitman (1984).
This paper was very speculative and was published in order to provide a hasis for discussion on this important
topic. In the ten years since this paper was published a great deal ofwork has been done to refine the concepts
of acceptable rislc and there are now more reliable acceptability criteria than those suggested by Whitman.
 ysis of rockfall hazards

Amua1 huai
Initiaring Vehicle Poten'ial
lrnpact probabiliiy of probsbility 01
event Rackfuli beneath
,igificant occumnce occmence
(a~usl) failure
iam
33%-
no
95% 0317 - nii
I 5
no
% 98% 1.63'101 - nil

Annual probabiliry ora ~ i nfatoi>iy

~ l ~
h u a l probab~liryoftwofaraliticr
= ( 8 3 3 i 5.56
=(5.56+ 2.78) -
- 2.78)
10'
10' - 1.67
= 8.31
10'
10'
Annuai probability of hree or more fstalities =2.78' 10.' = 2.78 * 10'

Figure 9.11: Event tree analysis of rockfals in the Argillite Cut in Britirh Colurnbia (After Bunce 1994)

Annual XMuai
In,tiating Vehisle Pofential
hpact probability of probability af
eveni Rockiall bcncaih "Umberof
,ignifissnt Occmencs Occ~enC~
(mual) frilurc frtaliticr
consm~uon
50% --r "O
80% 0.40 - nii

figure 9.12: Evenr tree for a hypathetiial exarnple in which construction aitiutiei on the Argillite Cut are carried ou1 for a period of iix months while
the hghway i i kept open.

Figure 9.13, based on a graph published by Nielsen, Hartford and MacDonald (1994), suinmarises
pubiisbed and proposed guidelines for tolerable risk. The line marlced 'Proposed BC Hydro Societal RisB' is
particularly interesting since this defines an annual probability of occurrence of fatalities due to dain iàilures
as 0.001 lives per year or 1 fatality per 1000 years. A great deal of effort has gone into defining this line and
I consider it to be directly applicable to rock slopes on highways which, like dams, must be classed as major
civil engineering structures for which the rislis to the public must be reduced to acceptable levels.
 Rock Engint?erlng Cou

Figure 9.13: Compariion between risks of falal$t>es

due to ioikfalls with published and propoied arceptabie risk critera

Another point to be noted in Figure 9.13 is that marlíed 'Proposed BC Hydro Individual risk'. R i s
annual probability of fatalities of 10' (1 in 10,000) is based upon the concept that the rislí to an individual
from a dam failure should not exceed tlie individual 'natural death' risk m n by the safest population group
(10 to 14 year old children). Consensus is also developing that the annual probability of fatality OS 10'
defines the boundaiy between voluntaty (restricted access to site personnel) and involuntaty (general public
access) risk (Nielsen, Hartford and MacDonald 1994).
On Figure 9.13, I have plotted the estimated annual probabilities of fatalities froin rodífalls on the
Argillite Cut on BC Highway 99, with and without construction. These plots show tliat the estimated riskfor
these slopes, without constmction, is significantly lower than the 0.001 lives per year line. The estimated riskfor
the Argillite Cut slopes during active constmction is approximately ten times higher and is marginally higher
than the 0.001 lives per year criterion. Given tbe fact that courts tend to be unsympathetic to engineers who
knowingly put the public at risk, it would be unwise to proceed with construction while attempting to keep the
traffic flowing. A more pmdent course of action would be to close the highway during periods of active
construction on tbe slopes, even if this meant having to deal with the anger of fnistrated motorists.

9.7 Conclusions
The Rockfall Hazard Rating System and the Event Tree rislí assessments, discussed on the previous
pages, are veiy crude tools which can only be regarded as semi-quantitative. However, the trends indicated hy
these tools together with common sense engineering judgement, give a reasonable assessment of the relativt
hazards due to rockfalls from cut slopes adjacent to highways and railways.
, .
 In situ and induced stresses

10.1 Introduction
Rock at depth is subjected to stresses resulting from the weight ofthe overlying strata and fiom locked
in stresses oftectonic origin. When an opening is excavated in this roclí, the stress field is locally dismpted
and a new set of stresses are induced in the roclí surrouiidiiig the opening. A knowledge o f t h e magnitudes
and directions of tliese in situ and induced stresses is an essential component of underground excavation
:ign since, in many cases, the strength of the rock is exceeded and the resulting instability can have serious
isequences on the behaviour of tlie excavations.
This chapter deals with the question of in situ stresses and also with the stress changes that are induced
en tunnels or caverns are excavated in stressed rock. Problems, associated with failure of the roclí around
derground openings and with the desigti of support for these openings, will be dealt with in later chapters.
The presentation, which follows, is intended to cover only those topics which are essential for the
rea der to línow about when dealing with the analysis of stress induced instability and the design ofsupport
stabilise the rodí under these conditions.

10.2 In situ stresses

Consider ali element of rock at a depth of 1,000 m below the surface. Tlie weight of the vertical
lumn of rock restiiig on this element is the product of the depth and the unit weight of tlie overlying rodc
iss (typically about 2.7 tonnes/m' or 0.027 MN/m3). Hence the vertical stress oii the eleinent is 2,700
.-ts/m2 or 27 MPa. This stress is estimated from the simple relationship:
YZ (10.1)

e o,,is the vertical stress

y is the unit weight of the overlying rock and
z is the depth below surface.

Measurements ofvertical stress at various mining and civil engineering sites around the world coii-
firm that this relationship is valid although, as illustrated in Figure 10.1, there is a significant arnouiit of
scatti:r in the measurements.
The horizontal stresses acting on an element of rock at a depth z below the surface are inucli inore
:nlt to estimate than the vertical stresses. Normally, the ratio of the average horizontal stress to the
:a1 stress is denoted by the letter k such that:
 eering Coi
Rock Eng~n

Vertical stress, 0, (MPa)

O 20 40 60 80

Figure 10.1: Vertical itreis measurementr from minintg and civil engineering pio)ecti around the warld. (Afret Hoek and Brown 1978)

Terzaghi and Richart (1952) suggested that, for a gravitationally loaded rock mass i11 which no lateral
strain was permitted during formation of the overlying strata, the value of k is independent of depth and is
given by k = V / (1 -v), where V is the Poisson's ratio of the rock mass. This relationship was widely used iii
the early days ofrocl<inechanics but, as discussed below, it proved to be inaccurate and is seldoin used today.
Measureinents of horizontal stresses at civil and mining sites around the world show that the ratia k
tends to he high at shallow depth and that it decreases at depth (Brown and Hoek, 1978, Herget, 1988). In
order to uiiderstand the reason for these horizontal stress variations it is necessary to consider the prohlein
ou a much larger scale than that of a single site.
Sheorey (1994) developed an elasto-static therinal stress model of the earth. This model considers
curvature of the crust and variation of elastic constants, density and thermal expansion coefficients through
the crust and mantle. A detailed discussioii o n Sheorey's model is heyond the scope of this chapter, but he
did provide a simplified equation which can he used for estimating the horizontal to vertical stress ratio k.
This equation is:

where z (m) is the depth below surface and E,, (GPa) is the average deformation modulus of the upper part of the
earth's cmst measured in a horizontal direction. This direction ofineasurement is important particularly in layered
sedimentary roda, in which the deformation modulus may be significailtly different i11different directions.
A plot of this equation is given in Figure 10.2 for a range of deformation moduli. Tlie curves relatiiig
h with depth below surface z are similar to those puhlished by Brown and Hoek (1978), Herget (1988) and
 dtu and i n duced strew

others for measured in situ stresses. IHence equation 7.3 is considered to provide a reasonable basis for
estimating the value of h.
As pointed out by Sheorey, his work does not explain the occurrence ofineasured vertical stresses that
are higher than the calculated overburden pressure, the presence of very high horizontal stresses at some
locations or why the two horizontal stresses are seldom equal. These differences are probably due to local
topographic and geological features that cannot be taken into account in a large scale model such as that
proposed by Sheorey.
Where sensitivity studies have shown that the in situ stresses are likely to have a significant influente
on the hehaviour of underground openings, it is recommended that the in situ stresses should he measured.
Suggestions for setting up a stress measuring prograinme are discussed later in this chapter.

10.3 The World stress map

The World Stress Map project, completed in July 1992, involved over 30 scientists from 18 countries
and was carried out under the auspices of the International Lithosphere Project (Zoback, 1992). The aim of
the project was to compile a global database of contemporary tectonic stress data. Currently over 7,300 stress
orientation entries are included in a digital database. Of these approximately 4,400 observations are consid-
ered reliable tectonic stress indicators, recording horizontal stress orientations to within < I25',.

k = horizontal çtress / vertical stress

o 1 2 3 4

'e 10.2: Ratio of horizontal to vertical itresi for different deformatio" moduli based upan Sheorey's equation. (Aiter Shearey 1994)
 ?ering. Cou

The data iiicluded in tlie World Stress Map are derived inainly from geological o h s e ~ a t i o n on
s earth- ,
qualce focal inechanisms, volcanic alignments and fault slip interpretations. Less than 5% of the data is based
upoii liydraulic fracturing or overcoring measurements of the type commonly used in mining aiid civil
engineering projects.
Figure 10.3 is a version of the World Stress Map in which the orieiitations of maximuni horizontal
stress o~,,,,,~
are plotted o n a base of average topography. Major tectonic plate boundaries are shown as heavy
lines oii this inap. Figure 10.4 is a generalisrd version of the World Stress Map wliich shows mean stress
directions based o11 averages of clusters of data shown in Figure 10.3.
The stress syinhols in Figure 10.4 are defiiied as follows:
I
i
!
A single set of thiclc inward pointing arrows indicates o,,,,~.,~
.... orientations in a thrust faulting stress regime E
("h., > Ohiiiiii > 4,).
A single set of outward pointing arrows indicates crhnlinorientations in a iiormal faultiiig stress regime (q2
> Ohx,,2*> ohlnill).
Thiclc inward pointing arrows, indicating ohn,ax , together with thin outward pointing arrows, indicating
are located in strike-slip faulting stress regimes (ohn- > q, > o J.
ohii,,,,
In discus~iiigIiydraulic fracturing and overcoring stress measurements, Zoback (1992) has the follow-
ing comments: :: ,
'Detailed hydraulic fracturing testing in a number of boreholes heginning very close to surface (10-
20 m depth) has revealed marked changes in stress orientations and relative iiiagnitudes with depth in tlie
upper feur hundred metres, possibly related to effects of nearhy topography or a high degree of near
surface fracturing.
Included in the category of 'overcoring' stress ineasuremeiits are a variety of stress or strain relief
measureinent techniques. These techniques involve a three-dimeiisional measureinent of the straiti reliefin
n body of rock wlieii isolated from the surrounding rock volume; the three-dimensional stress tensor cai1
suhsequently be calculated with a Icnowledge of the complete compliance tensor of tlie rode There are two
primary drawhacks with this technique which restricts its usefulness as a tectonic stress iiidicator: measure-
ments inust be made near a free surface, and strain relief is determined over veiy srnall areas (a few square
millimetres to square centimetres). Furthermore, near surface measurements (by far the most cornmoii) have
been shown to be subject to effects of local topography, rode anisotropy, and natural fracturing (Engelder
and Sbar, 1984). I11 addition, many of these measurements have been made for specific engineering applica-
tions (e.g. dam site evaluation, mining work), places where topography, fracturing or nearby excavations
could strongly perturh the regional stress field.'
Ohviously, from a global or eveii a regional scale, the type of engineering stress measurements carried
out in a mine or on a civil engineering site are not regarded as very reliahle. Conversely, the World Stress
Map versions presented in Figures 10.3 and 10.4 can only be used to give first order estimates of the stress
directions which are lilcely to he encountered on a specific site. Since hoth stress directions and stress magiii-
hides are criticaily important in the desigii of underground excavations, it follows that a stress measuring
programme is essential in any major underground mining or civil engineering project.

10.4 Developing a stress measuring programme

Cotisider the example of a tunnel to be driven a depth of 1,000 m helow surface in a hard rock
environment. The depth of the tunnel is such tliat it is probable that in situ and induced stresses will be an
importaiit consideration in the design of the excavation. Typical steps tliat could be followed in the analysis
of this prohlem are:
 t u and l n d ~

Figure 10.3: World streri

map giving rnaximvm
horizontal stress orenta-
tions on a base of average
topogiaphy (~ndicatedby

Zobark from a paper by

Zaback (1992) ,

Figure 10.4: Generalised

stres map rhawing mean
drettions baied an average
cluiteii of data shown in
Figure 7.3. The meaning of
theiymboli ir deiiribed n
the text. Map piovided by
Dr M.L. Zoback from a
paper by Zoback (1 992).
 lock Engineering Cou~

a. During preliminary design, the information presented in equations 10.1, 10.2 and 10.3 can he used to
obtain a first rougli estimate of the vertical and average horizontal stress in the vicinity of the tunnel. For
a depth of 1,000 m, these equations give the vertical stress o= 27 MPa ,the ratio h = 1.3 (for E,=75 GPa]
and hence the average horizontal stress oh=35.1 MPa. A preliminary analysis of the stresses induced
around the proposed tutinel (as described later in this chapter) shows that these induced stresses are likelg
t o exceed the strength of the rock aiid that the question of stress measureinent inust be considered in
more detail. Note that for many openings in strong'rock at shallow depth, stress problems may not be
significatit and the analysis need not proceed any further.
b. For this particular case, stress probletns are considered to he important. A typical next step would be to
search the literature in an effort to determine whether the results ofin situ stress measurement prograinmes
are available for mines or civil engineering projects within a radius of say 50 km of the site. With lu&, a
few stress measurement results will be available for the region in which the tunnel is located and these
results can he used to refine the analysis discussed ahove.
c. Assutning that the results of the analysis o€ induced stresses in the rodi surrounditig the proposed tuniiel
indicate that significant zones ofrodc failure are likely to develop, and that support costs are likely to be high,
it is probahly justifiahle to set up a stress measurement project on the site. These measurements can be carried
out in deep horeholes from the surface, using hydraulic íiacturing techniques, or from underground access
using overcoring methods. The choice of the method and the numher of measurements to be carried oiit
depends upon the urgency ofthe problem, the availahility ofunderground access and the costs involved in the
project. Note that very few project organisations have access to the equipmeiit required to carry out a stress
measurement project and, rather than purchase this equipment, it may be worth briiiging in an organisation
which has the equipment and which specialises in such measurements.

Where regional tectonic features such as major faults are liliely to be encountered the in situ stressesin
the vicinity of the feature tnay be rotated with respect to the regional stress field. The stresses rnay be
significantly different in magnitude from the values estimated from the general trends described earlier.
Tliese differences can be very important in the design of the openings and in the selection of support and,
where it is suspected that this is likely to he the case, in situ stress measurements become an essential compo-
nent of the overali design process.

10.5 Analysis of induced stresses

When an underground opening is excavated into a stressed rock mass, the stresses in the vicinity of
the new opening are re-distributed. Consider the example of the stresses induced in the rocli surrounding a
horizontal circular tunnel as illustrated in Figure 10.5, showing a vertical slice normal to the tunnel a i s .
Before the tunnel is excavated, the in situ stresses q2, o,,, and o,, are uniformly distributed in the slice
of rod<under consideration. After removal of the rocli from within the tunnel, the stresses in the iintnediate
vicinity of the tunnel are changed and new stresses are induced. Three principal stresses o,, o, and o, acting
o11 a typical element of rock are shown in Figure 10.5.
The convention used in roclc mechanics is that compressive stresses are always positivc and the tliree
principal stresses are numhered such that o, is the largest and o, is the smallest (algebraically) of the three.
The three principal stresses are mutually perpendicular, but they may he inclined to the direction of the
applied in situ stress. This is evident in Figure 10.6, which shows the directions of the stresses in the rock
surrounding a horizontal tunnel suhjected to a horizontal in situ stress oh,equal to three times the vertical in
sitit stress on.The longer hars in this figure represent the directions of the maximum principal stress o,,while the
shorter bars give the directions of the minimum principal stress o3at each element considered. In this particular
case, o, is coaxial with the in situ stress o,,but the other principal stresses o, and o, are inclined to obtand q
 er10 I n s i tu and i n d u ired stresses

1
Figure 10.5: Illustration of principal stresies induced in an elernent
o1 rotk close to a horizontal tunnei subjected to a venical in situ
itreri4, a horizontal in iitu stress o,,,
in a plane normal to the tunnel
axls and a horizontal in situ streis o, paralel to the tunnel axii.
Figure 10.6: Prnripsi stress directoni in lhe iock iur-
rounding a hor~zontaltunnei iubjected to a horizontal
equa t o 3 4,.where o,, ii the vertical in
in situ strea o,,,
situ sties~.

Contours of the magnitudes of the maximum principal stress s, and the miniinum principal stress o,
are given in Figure 10.7. This figure shows that the redistribution of stresses is concentrated in the rod<close
to the tunnel and that, a t a distance of say three times the radius from the centre of the hole, the disturbance
to the in situ stress field is negligible.
An analytical solution for the stress distribution in a stressed elastic plate containing a circular hole
was published by Kirsch (1898) and this formed the basis.for many early studies of rock behaviour around
tlinnels and shafts.
Following along the path pioneered by Kirsch, researchers such as Love (1927), Muskhelishvili (1953)
and Saviu (1961) published solutions for excavations ofvarious shapes in elastic plates. A useful suminaiy of
these solutions and their application in rod< mechanics was published by Brown in an introduction to a
volume entitled AnaCyticaland ComputationalMethodsin Engineering Ro& Mechanics (1987).
Closed form solutions still possess great value for conceptual understand-ing of behaviour aiid for the
testing and calibration of numerical models. For design purposes, however, these models are restricted to
veiy simple geometries and material models. They are of limited practical value. Fortunately, with the devel-
opment of computers, many powerful programs which provide numerical solutions to these problems are
now readily available. A brief review of some of these numerical solutions is given below.

10.6 Numerical methods of stress analysis

Most underground excavations are irregular in shape and are frequently grouped close to other excava-
tions. These groups of excavations can form a set of complex three-dimensional shapes. In addition, because of
thepresence of geological features such as faultr and intrusions, the rockproperties are seldom uniforin within
the rod<volume of interest. Consequently, the closed form solutions described earlier are of limited value i11
calculating the stresses, displacements and failure of the rock mass surrounding undergrouiid excavations.
Fortunately a number of computer-based numerical methods have been developed over the past few decades
and these methods provide the means for obtaining approximate solutions to these problems.
 Rock Engirieering Coi

I Maximum p ~ c i p a stress
l G l /ov

principal stress G3 /

Numerical methods for the analysis of stress driven problems in rod< inechanics can be divided into
two classes:

Bonnday methods, in which only the boundary of the excavation is divided into elements and the interior
of the rod< inass is represented mathematically as an infinite continuuin.
Domain methods, in which the interior of the rock mass is divided into geometrically siinple elements each
with assumed properties. The collective hehaviour and interaction of these simplified elements model
the more complex overall hehaviour of the rock mass. Finitc element andfinite dzffiencc methods are
domain techniques which treat the rock mass as a continuum. The disknct elcment method is also a do-
main method which models each individual block of rock as a unique element.

These two classes of analysis can be comhined in the form of hybridmodels in order to maxiinise the
advatitages and minimise the disadvantages of each method.
It is possible to make some general obsewations ahout the two types of approaches discussed above.
In domain methods, a significant amount of effort is required to create the mesh that is used to divide the
rock mass into eleinents. In the case of complex models, such as those contaiuing multiple openings, mesh-
iug can become extremely difficult. The availability of highly optimised mesh-generators in many inodels
mal<es this task inuch simpler than was the case when the mesh had to be created manually. In contrast,
boundary methods require only that the excavation boundary be discretized and the surrounding rock mass
is treated as an infinite continuum. Since fewer elements are required in the boundary method, the demand
on computer memory and on the skill and experience of the user is reduced.
 u and indu:ed stresses

In the case of domain methods, the outer boundaries of the model must be placed sufficiently far
tom the excavations in order that errors, arising from the interaction between these outer boundaries
and the excavations, ale reduced to an acceptable minimum. O n the other hand, since boundary methods
treat the rock mass as an infinite continuum, the far field conditions need only be specified as stresses acting
on the entire rock mass and n o outer boundaries are reqiiired. The main strength ooilioundaiy inethods lies
in the siinplicity achieved by representing the rock mass as a continuum of infinite extent. It is this represen-
tation, however, that malies it difficult to incorporate variable material properties and the modelling of rocii-
support interaction. While techtiiques have been developed to allow some boundary element modelling of
variable rock properties, these types of problems are more conveniently modelled by domain inethods.
Before selecting the appropriate modelling technique for particular types of problems, it is necessary
to undentand the basic components of each technique.

10.6.1 Boundary Element Method

The boundary element method derives its name from the fact that only the boundaries of the prob-
lem geometry are divided into elements. In other words, only the excavation surfaces, the free surface for
shallow problems, joint surfaces where joints are considered explicitly and material interfaces for multi-material
problems are divided into eleinents. In fact, several types of boundary elemetlt inodels are collectively re-
ferred to as 'the boundary element method'. These models may be grouped as follows:
1. Indirect (Fictitious Stress) method, so named because the first step in the solution is to find a set of
fictitious stresses that satisfy prescribed boundary conditions. These stresses are then used in the calcula-
tion of actual stresses and displacements in the rocli mass.
2. Direct method, so named because the displacements are solved directly for the specified boundary conditiotls.
Displacement Discontinuity mrthod, so named because it represents the result of an eloilgated slit in ali
rlastic continuum being pulled apart.

The differences between the first two methods are not apparent to the program user. The direct
method has certain advantages in terms of program development, as wili be discussed later in the section on
Hybrid approaches.
The fact that a boundaiy element model extends 'to infinity' can also be a disadvantage. For example,
a heterogeneous rock mass coilsists of regions of finite, not infinite, extent. Special techniques inust be used
to handle these situations. Joints are modelled explicitly in the boundary element method using the displace-
rneiit discontinuity approach, but this can result in a considerable increase in computational effort. Numeri-
cal convergence is often found to be a problem for modils incorporating many joints. For these reasons,
problt:ms, requiring explicit consideration of several joints and/or sophisticated modelling ofjoint constitu-
tive biohaviour, are often better handled by one of the domain methods such as finite eleinents.
Awidely-used application of displaceinent discontinuity bo~indaryelements is in the modelling of tabular
ore bcbdies. Here, the entire ore seam is represented as a 'discontinuity' which is initially filled with ore. Minitlg is
simui;~tedby reduction of the ore stiffness to zero in those areas where mining has occurred, and the resulting
stress Iredistribution to the surrounding pillars may be exainined (Salamon, 1974, von Kimmelmann et ai., 1984).
Further details on boundary element methods can be found in the book Boundag clrinentmethnds in
rechnnics by Crouch and Starfield (1983).

10.6.2 Finite element and finite difference methods

In practice, the finite element method is usually indistinguishable from the finite difference inethod;
thus, they will be treated here as one and the same. For the boundary element method, it was seen that
conditions on a surface could be related to the state at a11 points throughout the reinaining rocli, even to
 Rock Engi urse nott

infinity. In comparison, the finite element method relates the conditions at a few points within the rock
(nodal points) to the state within a finite closed region formed by these points (the element). The physical
problem is inodelled numerically by dividing the entire problem region into elements.
The finite element method is well suited to solving problems involving heterogeneous or non-linear mate-
rial properties, since each element explicitly models the response of its contained material. Howrver, finite ele-
inents are not well suited to modelling infinite boundaries, such as occur in underground excavation problems.
One technique for handling infinite houndaries is to discretize beyond the zone of influence of the excavation
and to apply appropriate boundary conditions to the outer edges. Another approach has been to develop ele-
ments for which one edge extends to infinity i.e. so-called 'infinity' finite elements. In practice, efficient pre- and
post-processors allow the user to perform parametric analyses and assess the influence of approximated far-field
boundary conditions. The time requirèd for this process is negligihle compared to the total analysis time.
Joints can be represented explicitly using specific 'joint elements'. Different techniques have been
proposed for handling such elements, but no single technique has found universal favour. Joint interfaces
may be inodelled, using quite general cotistitutive relations, though possibly at increased coinputational
expense depending on the solution technique.
Once the model has been divided into elements, material properties have been assigned and loads have
been prescribed, some technique must be used to redistribute any unbalanced loads and thus determine tlie
solution to the new equilibrium state. Available solution tecbniques can be broadly divided into two classes -
implicit and explicit. Implicit techniques assemble systems of linear equations that are then solved using staudard
matrix reduction techniques. Any material noii-linearity is accounted for by modifying stiffness coefficients (secaiit
approach) and/or hy adjustingprescribed variables (initial stress ar initial strain approach). These changes are inade
in an iterative manner such that all constitutive and equilibrium equations are satisfied for the giveii load state.
The response of a iion-linear system generally depends upon the sequence of loading. Thus it is
uecessary that the load path modelled be representative of the actual load path experienced by the body.
This is achieved by breaking the total applied load into load increments, each increment being sufficiently
sinall, that solution convergence for the increment is achieved after only a few iterations. However, as the
system being modelled becomes increasingly non-linear and the load increment represents an ever smaller
portion of the total load, the incremental solution technique becomes similar to modelling tlie quasi-dy-
namic behaviour of the body, as it responds to gradual application of the total load.
In arder to overcoine this, a 'dynamic relaxation' solutioii technique was proposed (Otter et al., 1966)
and first applied to geomecliaiiics modelling by Cundall (I971). In this technique no matrices are formed.
Ratber, the solution proceeds explicitly - unbalanced forces, acting at a material integration point, result in
acceleratioii of the mass associated with the point; applying Newton's law of inotion expressed as a differ-
ence equation yields incremental displacenients; applying the appropriate constitutive relation produces the
new set offorces, atid so on marchiiig in time, for each material integration point in the model. Tliis solution
technique has tlie advantage that both geometric and material non-linearities are accommodated, with rela-
tively little additional computational effort as compared to a corresponding linear analysis, and computa-
tional expense increases only linearly with the number of elements used. A further practical advantage lies in
the fact that numerical divergence usually results in the model predicting obviously anomalous physical
behaviour. Thus, even relatively inexperienced users may recognise numerical divergence.
Most commercially available finite element packages use implicit (i.e. matrix) solution techniques.
For linear problems and problems of moderate non-linearity, implicit techniques tend to perform faster than
explicit solution techniques. However, as the degree of non-linearity of the system increases, imposed loads
must be applied in smaller iucrements which implies a greater number of matrix re-formations and reduc-
tions, and hence increased computational expense. Therefore, highly non-linear problems are best handled
by packages using an explicit solution technique.
 u and induc:ed stresses

10.6.3 Distinct Element Method

In ground conditions conventionally described as bloclcy (i.e. where the spacing of the joints is of the
same order of magnitude as the excavation dimensions), intersecting joints form wedges of rodc that tnay be
regarded as rigid bodies. That is, these individual pieces of roclc may be free to rotate and translate, and the
deformation, that takes place at block contacts, may be significantly greater than the deformation of the
intact rock, so that individual wedges may be considered rigid. For such conditions it is usually necessary to
model many joints explicitly. However, the behaviour of such systems is so liighly noii-linear, that even a
jointed finite element code, employing an explicit solution technique, niay perforrn relatively inefficieiitly.
An alternative modelling approacli is to develop data structures that represeiit the bloclcy nature of
the system being analysed. Each bloclcis considered a unique free body that may interact at contact locations
with surrounding blocks. Contacts may be represented by the overlaps of adjacent blocks, therehy avoiding
the necessity of unique joint elements. This has the added advantage that arbitrarily large relative displace-
ments at the contact may occur, a situation not generally tractable in finite element codes.
Due to the high degree of non-linearity of the systems being modelled, explicit solution techniques
are favoured for distinct element codes. As is the case for finite element codes employing explicit solution
techniques, this permits very general constitutive modelling ofjoint behaviour with little increase in compu-
tational effort and results in coinputation time being only linearly dependent on the number of elements
used. The use of explicit solution techniques places fewer demands on the sl<illsand experience than the use
of codes employing implicit solution techniques.
Although the distinct eleinent method has been used most extensively in academic environrnents to
date, it is finding its way into the offices of consultants, planners and designers. Further experience in tlie
application of this powerful modelling too1 to practical design situations and subsequent documentation of
these case histories is required, so that an understanding may be developed of where, when and how the
distinct element method is best applied.

10.6.4 Hybrid approaches

The objective ofa hybrid method is to combine the above methods in order to eliminate undesirable
characteristics while retaining as many advantages as possible. For example, in modelling an underground
excavation, inost non-linearity will occur close to the excavation boundary, while the rock inass at some
distance will behave in an elastic fashion. Thus, the near-field rock inass migbt be modelled, using a distinct
elernent or finite element rnethod, which is then linked at its outer limits to a boundary elemeiit model, so
thr far-field boundary conditions are modelled exactly. In such an approacli, the direct boundary ele-
merit technique is favoured as it results in increased programming and solution efficiency.
Lorig and Brady (1984) used a hybrid inodel consisting ofa discrete element rnodel for the near field
and a boundary element model for the far field in a to& mass surrounding a circular tunnel.

10.6.5 Two-dimensional a n d three-dimensional models

A two-dimensional model, such as that illustrated in,Figure 10.5, can be used for the aiialysis of stresses
displacements in the ro& surrounding a tunnel, shaft or borehole, where the length of the opening is mudi
er than its cross-sectional dimensions. The stresses and displacements in a plane, normal to the axis of the
ning, are not influenced by tbe ends of the opening, provided that these ends are far enough away.
On the other hand, a an underground powerhouse of crusher chamber has a much more equi-diinensiorial
shape and the effect of the end walls cannot be neglected. In this case, it is much more appropriate to carry out a
three-dimensional analysis of the stresses and displacements in the surrounding rock mass. Unforhlnateiy, this
switch from two to three dimensions is notas simple as it sounds and there are relatively few good three-diinen-
ia1 numerical models, which are suitable for routine stress analysis worlc in a typical rnining environment.
 Rock E n g i rieenng Coi

EXAMINE"' is a three-dimeiisional boundary element programs that provide a starting point for an
analysis of a problem in which the three-dimensional geometry of the openings is important. Such three-
dimensional analyses provide clear indications of stress concentrations and of the iilfluence of three-diinen-
sional geometry. In many cases, it is possible to simplify the problem to two-dimensions by considering the
stresses on critical sections identified in the three-dimensional modei.
More sophisticated three-dimensional finite element models such as VISAGE5re available, but are
not particularly easy to use at the present time. In addition, definition of the input pararneten and interpre-
tation of the results of these inodels would stretch the capabilities of all but the most experieliced rnodellers.
It is probably best to leave this type of modeiling in the hands of these specialists.
It is recommended that, where the problem being considered is obviousiy three-dimensional. a prelimi-
nary elastic analysis be carried out by means of one of the three-dimensional boundary eleineiit programs. The
results can then be used to decide whether further three-dimensional analyses are required or whether appropriate
two-dimensional sections can he modelled using a program such as PHASE2, described in the following section.

10.6.6 Stress analysis using the program PHASE'

In order to meet the requirements of modelling the post-failure hehaviour of rock masses and the
iiiteraction of these roclcs with support, a two-dimensional finite element model called PHASE" was devel-
oped by the Rode Engineering Group in the Department of Civil Engineering at the University ofToronto.
This program is very powerful hut user-friendly and it will generally meet the needs of most underground
excavation design projects. More sophisticated analyses such as those involving thremal stresses, fluid flow
or dynainic loading will require the use of more powerful software such as the program FLAC4.
The capability of the program PHASE' is best demonstrated by aprachcal example such as that presented below.

10.7 Practical example of two-dimensional stress analysis

The details included in this example, based upon ali actual case, are as follows:
A spillway tunnel for an einbankment dam is to be constructed in a poor quality sandstone. The
excavated diameter of the tunnel is about 13 m and the cover over the roof is 8 m. The tunnel is to have a 1.3
m thick un-reinforced concrete lining and, after placement of this lining, a 28 m high portiori of the roddill
dam will be constructed over the tunnel.
The questions to be addresses are:
1. What support is required on order to excavate the tunnel safely under the very shallow cover?
2. 1s the proposed top heading and bench excavation sequence, using drill and hlast methods, appropriate
for this tunnel?
3. How will the concrete lining respond to the loading imposed by the placement of 28 m of rocl<fillover
the tunnel?
In order to answer these questions a series of two-dimensional finite element analyses were carried
using the prograin PHASE'. The first of these analyses exainined the stability and support requirements for
the top heading excavation. The final analysis included the entire excavation and support sequence and the
placement of the roclcfill over the tunnel. The complete finite element model is illustrated in Figure 10.8. An
enlarged view of the excavation and the final support system is given in Figure 10.9.

Availablefrom Availablefiam Rocscience lnc. 31 8alsamAvenue.Toronto. Ontario. Canada M4E 385. Fax 1416 6980908. Phone 1416 6'388217.
Emai: software~rocscienie,iom,lnternet http://w.rocscience.com.
'Available from Vector Internatjonal Procesring Systems Ltd., Suitei 805 and 806, Surrey Houie, 34 Eden Street. Ki-iton on Tharnes, KTl 1ER.
England. Fax 44 81 541 4550. Phane 44 81 549 3444.
3Avalable from Rocscience Inc. 31 Baisam Avenue. Toronto. Ontario. Canada M4E 385, Fax 1 416 698 0908, Phone 1 416 6'38 8217,
Email: iofware@rociiience.ca~.inteinet http://w.rocsrienreeiomm
Ean
"vailable from IIASCA Consuiting Group lnc., Thresher Square 708 Southihird S t r w Suite310, Minneapolir, Mnnesota 5541 5. USA. Fau 1 612 371 4717
 u and ~ndu
ced stresses

The rod<mass is a poor quality sandstone that, being close to surface, is heavily jointed. The mechani-
cal properties' assumed for this rock mass are a cohesive strength c = 0.04 MPa, a friction angle @ = 40" and
a modulus of deformation E = 1334 MPa. No in situ stress measurements are availahle but, because of the
location of the tuntiel i11 the valley side, it has been assumed that the Iiorizontal stress normal to the tunnei
axis has been reduced by stress relief. The model is loaded by gravity and a ratio of horizontal to vertical
stress or 0.5 is assumed.

10.7.1 Analysis of top heading stability

A siinplified version of tlie model illustrated in Figures 10.8 aiid 10.9 was used to analyse the stability and
support requiremenci for the top heading. This model did excluded the concrete lining and the bench excavations.
Tlie first inodel was used to examine the conditions for a full-face excavation of the top lieading
without any support. This is always a useful starting point in any tunnel support design study since it gives
the designer a clear picture of the magnitude of the prohlems that have to be dealt witli.
The model was loaded in two stages. The first stage involved the model without any excavations and this
was created by assigning the material within the excavatioti boundary the properties of the surrounding rock
mass. This first stage is carried out in order to allow the model to consolidate under gravitational loading It is
required in order to create a reference against which subsequent displacements in the inodel can be measured.
The results of the analysis are illustrated in Figure 10.10, that shown the extent ofyield in the rod<inass
surrounding the top heading, and Figure 10.11 that shows the induced displacements aroiind tlie tunnel.
The large amount ofyield in the rock mass overlying the top heading suggests that this excavation will
be unstable without support. This view is supported by the displacemeiits shown in Figure 10.1 1.
Tlie reader may be surprised that the displacement in the roof ofthe tunnel is otily 26 tnin when the
extent of the yield zone suggests complete collapse of the rooi. It has to be reineinbered that PHASE' is a
small strain finite element model and that it cannot accommodate the very large strains associated with the
complete collapse of a tunnel. In examining Figure 10.11 it is more important to lool< at the shape of the
overall displacement profile than the magnitude of the displacements. A rocl< mass will not tolerate the
differeiitial displacements illustrated and progressive raveiling leading to ultiinate collapse would almost
certainly result from the excavation of an unsupported top heading.
A general nile of thumb used by experienced tunnellers is that ai1 underground excavation will not be self-
supporting unless the cover over the tunnel exceeds 1 112 times the span of the opening. This is a typical situatio11
that occurs when excavating tuntiel portals are there are severa1 options available for dealing with the problem.
One of these options is to use a shotcrete lining to stabilise the rock inass above the tunnel. A finite
element analysis of this option shows that a 50 mm thid<layer of fully hardened shotcrete (uniaxiai compres-
sive strength of 30 MPa) is sufficient to stabilise the tunnel. The problein is Iiow to get a fully hardened layer
i of shotcrete into an advancing tunnel heading. A second problem is whether the worl<erswould have suffi-
cient confidence in such a solution to work in the tunnel.
Oiie project on which this solution was used was the construction of an 8 in span diversion tunnel Tor
a dar. The rock mass was a veiy weakly cemented limestone that could be excavated by hand but which had
sufficient strength that it was inarginally self-supporting. The Scandinavian contractor on the project Iiad
used shotcrete for inany years and the very experienced tunnellers had complete confidence in working
under a cover of shotcrete. The tunnel was not on the critical path of the project aiid so constructioi~could
proceed at a sufticiently slow pace to allow the shotcrete to set before the next advance. A layer of uti-
reinforced shotcrete was the sole support used in this tunnel, with occasional steel sets einbedded in the
shotcrete where ground conditions were particulariy difficult.

Afull discuiiion on rnethods for estimating rock mass propertiej ii given in Chapter 11
 Rock Engirieering Coinse notes t

, Free suriaface

Figure 10.8: Fnite element mbdel ihowing mesh geometry and boundary conditions. The final suppon njitem used for this case i asa shawn and
w i l be dircursed in the text whch folows.

Figure 10.9: Detais of the finite element mesh , excavation requence, suppon iyaem and externa1 laad for the complete modei.

5 mm suriaçe subsidence
/

Figure 10.10: Yieid in the iack masi suiraundng the top heading exiavation wtth no ruppart initaled.
 tu and i n diced
~ stresse

Figure 10.11: Diiplacements indured in the rock masi by the excavation a i an unsupponed top heading.

111 tlie case of the top heading in sandstone under consideration here, the shotcrete solution was
rejected because, in spite of the finite element analysis, the designers did iiot have sufficient confidente in
the ability of the shotcrete layer t o support the iarge span of blodcy sandstone. In addition, the contractor on
this dam project did not have a great deal of experience in using shotcrete in tunnels and it wasunlikely that
the worliers would have been prepared to operate under a cover of shotcrete only.
Another alternative that is commonly used in excavating tunnel portals is to use steel sets to stabilise
the initial portion of the tunnel under low cover. This solution worlis well in the case of small tunnels b ~ i tin
,
this case, a 13 m span tunnel would require very heavy sets. An additional disadvantage in this case is that the
installation of sets would permit too inuch deformation in the rocli mass. This is because the steel sets are a
passive support system and they only cariy a load when the rodi mass has deformed onto the sets. Since this
tuniiel is in the foundation of a dam, excessive deformation is clearly not acceptable because of the addi-
tional lealrage paths which would be created through the rock mass.
The solution finally adopted was 'borrowed' from the mining industiy where un-tensioned fully grouted
doweis are frequently used to pre-support the rock inass above underground excavations. In this case, a
pattern 3 m x 3 m pattern of 15 m long 60 ton capacity cables were installed froin the ground surface before
excavation of the top headiiig was commenced. When these cables were exposed in the excavatioii, face
plates were attached and the excess cable length was cut off. In addition a 2 m x 2 in patterii of 6 in long
rnechanically anchored rockbolts were installed radially from the roof of the top heading.
The results of an analysis of this support system are iilustrated in Figures 10.12 and 10.13 which show
the extent of the yield zone and the deformations in the rock rnass above the top heading.
Coinparing Figures 10.10 and 10.12 shows that that the extet of the yield zone is only reduced by a
srnall arnount by the enstallation of the support system. This is not surprising sinre soine deformation of the
rock mass is required in order to mobilise the supporting loads in the untensioned cables. This deforination
occurs as a reullt of failure of the rock mass.
Figure 10.13 shows that the displacements in the roofof the top heading have been reduced substan-
tially as a result of the placement of the support. However, a small problem remains and that is the excessive
displacement of the rock between the rockbolt faceplates which are spaced o n a 2 m x 2 m grid. Unless this
displacement is controlled it can lead to progressive ravelling o f t h e rock mass.
Only a small surface pressure is required to control this ravelling and this could be achieved by ineans
of a layer of mesh or shotcrete of by the installation of light steel sets. I11 this case the latter soltion was
adopted because of the sense of security which these gave for the worliers in the tunnel. The appearance of
the supported top heading is illustrated in Figure 10.14.
 ieering Coi

10.7.2 Analysis of complete excavation

Having successfully excavated tlie top heading of this spillway tunnel, as discussed in the previous
section, the next question was how the excavation would behave during benching down and how the con-
crete lining would tolerate the additional loads imposed by the placement of 28 m of compacted rocl<fill.
The complete model, illustrated in Figure 10.9, was used for this investigation and it was found tliat
there were n o prohlems with the excavation ofthe benches. Figures 10.15 and 10.16 illustrate the yield zone
and the displaceinents of the rock mass surrounding the complete excavation supported hy meails of cables
and rod<bolts and, for the purposes of this study, a thin shotcrete lining.
The self-weight of the concrete lining indices displacements in the rock mass and these are increased
by the imposition of the surface load due to the placement of 28 m of compacted rocl<fillahove the tunnel.
The induced displacements are illustrated in Figures 10117 aind 10.18.
15 m long 60 ton capacity un
tensioned cabies grouted into
vertical holes before excavation of
the top heading commenced

6 m iong rnechanicaily
anchored and tensioned

Figure 10.12:Yield rane in the roik rnassiuirounding the tap heading iupporled by meand of pre-placed giouted iabled and meihancaly ancliaied
and tensioned rackbolts.

mm surface subsidence

3 mm displacment a!
rockbolt faceplates

7 mm displacement
behveen rockbolts

:. I 4
4 mm tloor heave
' c , " ' a .

Figure 10.13: Displacementi in the rack mais supponed by means of pre-placed grouted iablei and tensioned rorkbolts.
 t u and indirced stresse

Figure 11.14: Top heading

iuppaned byiablei and
rnck~boltsand ight steel seis.

Figure 10.16: Displacementi in fhe rotk mas i u r

Figure 10.15: Extent ofyieid in the rock maaiuriound-
n g the compete spillway tunnei excavation befare
placement of the rancrete lining.

Figures 10.17 and 10.18 show that significant displacements are induced as a result of the castiiig of
the concrete lining and the subsequent placement ofthe rockfill above the tunnel. No failure of the concrete
lining was shown by this analysis, in spite of the assumption of a very wealc concrete (10 MPa uniaxial
compressive strength). The only problem that could be anticipated from the placeinent of tlie rod<fill was
the possibility of bending of the entire length of the concrete lining aild the formation of tensilc cradcs
normal to the tunnel x i s . It was therefore recommended that the concrete lining be carefully inspected for
such crad<safter the completion of the rod<fill. Repair ofsuch cradcs by dental concrete and grouting would
not be a major problein hut, in any case, it proved not to be necessary.
 Rock Eng~n~
lering Cou

2 mm suríace subsidence
/

22 mm downward

18 mrn downward

.. . . . .

Figure 10.17: Displacernents induced as a result af the self-weight ofthe concrete iining.

Suríace load due to placement

/of 28 m of rockfill

mm suríace displacement

30 mm downward

8 I

Figure 10.18: Displacernents induced a i a reiuit of the plaiement of 28 rn of rockfiil above the cornpleted tunnel.

10.7.3 Conclusion
The analysis presented on the preceding pages is intended to deinonstrate how a numerical analysis
should he used as a too1 to aid designers. In all cases, practical issues take precedence and the results of the
analysis should only he used to guide the practical decisions and to clarifi matters of doubt or uncertainty.
Given the assumptions that have to he made in the construction of an analysis of this type, it would he veq
unwise for the designer to place too much credence in the results of the analysis and to allow a11 his or her
decisions to be driven by these results.
A discussion of the results with an experienced tunnel contractor will soon dispel any misconceptions
that the tunnel designer wiil have acquired as the results of suEh a theoretical analysis and such a discussion
is ali essential part of any practical design process.
 Rock mass properties

11.1 lntroduction
Reliable estimates of the strength and deformation characteristics of rodi masses are required for
almost any form of analysis used for the design of slopes, foundations and underground excavations. Hoeli
and Brown (1980a, 1980b) proposed a method for obtaining estimates of the strength ofjointed roclc inasses,
based upoii an assessment of the interlocking of rock blocks and the condition of the surfaces between these
blodis. This method was modified over the years in order to meet the needs of users who were applying it to
problems that were iiot considered when the original criterion was developed (Hoeli 1983, Hoeli and Brown
1988). The applicatioil of the method to veiy poor quality rodc masses required further clianges (Hoek,
Wood and Shah 1992) and, eventually, the development of a new classification called the Geological Streilgtli
Index (Hoek, Kaiser and Bawden 1995, Hoek 1995, Hoek and Brown 1997). A review of the development of
the criterion and of the equations proposed at various stages in tliis development is given iii Hoek and
Brown (1997).
This chapter presents the Hoeli-Brown criterioii in a form that has been found practical in the field
and that appears to provide the inost reliahle set ofresults for use as input for inethods ofanalysis in current
use in rocli engineering.

11.2 Generalised Hoek-Brown criterion

The Generalised Hoek-Brown failure criterion for jointed rodi masses is defined by:

whereú; and 0;are the inaximum and minimum effective stresses at failure,
m, is the value of the Hoek-Brown constant m for the rock mass,
s and a are constants which depend upon the rock mass characteristics, and
0,;is the uniaxial compressive strength of the intact rock pieces.
The Mohr envelope, relating normal and shear stresses, cai1 be determined by the method proposed
by Hoek and Brown (1980a). In this approach, equation 11.1 is used to generate a series of triaxial test values,
siinulating full scale field tests, and a statistical curve fitting process is used to derive ai1 equivalent Mohr
envelope defined by the equation:
 Rock Enginseerlng. Cou

where A and E are material constants

o, is tlie normal efièctive stress, and
O,,,, is the 'tensile' strength of the rock mass.
This 'tensile' strength, which reflects the interlodíing of the rock particles when they are not free to
dilate, is given hy:

In order to use the Hoek-Brown criterion for estimating the strength and deformability of jointed
rod<inasses, three 'properties' of the rock inass have to be estimated. These are
1. the uniaxial compressive strength oc,of the intact rode pieces,
2. the value of the Hoek-Brown constant mifor these intact roclí pieces, and
3. the value of the Geological Strengtli Index GSI for the rock mass.

11.3 Intact rock properties

For the intact rock pieces that make up the rock mass equation 11.1 simplifies to:
, \O.S

The relationsliip between the principal stresses at fail~irefor a given rod< is defined hy two constants, tlie
uniaxial coinpressive strength o(,and a constatit m, Wherever possible the values of tliese constants shouid
be deterinined by statistical analysis of the results of a set of triaxial tests on carefully prepared core samples.
Note that the range of iniiior principal stress (,J;) values over which these tests are carried out is
critical in deterrnining reliable values for the two constants. In deriving the original values of 4; and m;,
Hoelí and Brown (1980a) used a range of O <o; < 0.50~;and, in order to be consistent, it is essential that tlie
saine range be used in any laboratory triaxial tests on intact roclí specitnens. At least five data points should
be included in the aiialysis.
One type of triaxial cell that can be ~isedfor these tests is illustrated in Figure 11.1. This cell, described
by Hoek and Franklin (1968), does not require draining between tests and is convenieiit for the rapid testing
o r a large number of speciinens. More sophisticated cells are available for research purposes but the results
obtained froin the cell illustrated in Figure 11.1 are adequate for the rodc strengtli estiinates required for
estimatiiig q,and m,. This cell has the additional advantage that it cai1 be used i11 tlie field when testing
materials sucli as coals, shales and phyllites that are extremely difficult to preserve during transportation atld
normal specimen preparation for laboratory testing.
Laboratory tests should be carried out at moisture contents as close as possible to tliose which occur
in the iieid. Many rod<s show a significant strength decrease with increasing moisture content and tests on
sainples, which have been left to dry in a core shed for severa1 months, can give a inisleading impression of
the intact rock strengtli.
Once the five or more triaxial test results have been obtained, they can be analysed to determine the
uniaxial compfessive strength ocjand the Hoek-Brown constant mias described by Hoelí and Brown (1980a).
In this analysis, equation 11.1 is re-written in the form:
y = mor,x + tq; (11.5)
where x = O, andy = (o, - OJ
 Rock mas! a properties

hardened and ground steel

clearance gap for gauge wires

mild steel ceil body

rock specimen with ground ends and

vith a length to diameter ratio of 2

oil inlet - maximum pressure

-
strain gauges if required

ubber sealing sleeve

u
Figure 11.1: Cur-away view of a trianiai cell for teiting rock specimens.

For n specimens the uniaxial coinpressive strength a , the constant miand the coefficient of determi
nation r" are calculated from:
 Rock Eng~ni
lering Cou

A spreadsheet for the analysis of triaxial test data is given in Table 11.1. Note that high quality triaxial
test data will usually give a coefficient of determination ? ofgreater than 0.9.
When laboratory tests are not possible, Table 11.2 and Table 11.3 can be used to obtain estimates of
o<,and m,.
Sliort-term laboratory tests on very hard brittle rocks tend to overestimate the iil situ rock rnass sirength.
Laboratory tests and field studies o n excellent quality Lac du Bonnet granite, reported by Martin and Chan-
dler (1994), show that the in situ strength of this rock is only about 70% of that rneasured iii the laboratoly
This appears to be due to damage resulting from micro-cracking of the rock which initiates and develops
critical intensities at lower stress levels in tiie field than in laboratory tests carried out at higher loading rates
on srnaller specimens. Hence, when analysing the results of Iaboratory tests on these types of rocks to esti-
mate the values ofo';and m,, it is prudent to reduce the values ofthe major effective principal stress at failure
to 70% of the measured values.
Anisotropic and foliated rocks such as slates, schists and phyllites, the hehaviour of which is domi-
nated by closely spaced planes of weakness, cleavage or schistosity, present particular difficulties in the
determination of the uniaxial compressive strengths.
Salcedo (1983) bas reported the results of a set of directional uniaxial compressive tests o n a graphitic
phyllite froin Venezuela. These results are summarised in Figure 11.2. It will be noted that the uniaxial
coinpressive strength of this material varies by a factor of about 5, depending upon the directioil of loadiiig.
Evideilce of the behaviour of this graphitic phyllite in the field suggests that the rod< mass properties are
dependent upon the strength parallel to sdiistosity rather than that normal to it.

Table 11.1: Spreadsheet for the calculation of o.: and m:frorn triaxial test data
Triaxial test data
X Y XY xsq Ysq
sig3 sig I
O 38.3 1466.89 0.0 0.0 2151766
5 72.4 4542.76 22713.8 25.0 20636668
7.5 80.5 5329.00 39967.5 56.3 28398241
15 1 15.6 10912.0.36 151805.4 225.0 102421687
20 134.5 13064.49 261289.8 400.0 170680899

47.5 441.1 34523.50 475776.5 706.3 324289261

sumx surny sumxy sumxsq sumysq
Calculation results
Nurnber of tests n = 5
Uniaxial strengh sigci = 37.4
Hoek-Brown constant rni = 15.50
Hoek-Brown constant s = 1.O0
Coefficient of deterrnination r2 = 0.997
Cell formulae
y = (sigl-sig3)"2
siqci = SQRT(surny/n - (surnxy-surnx*surny/n)/(surnxsq-(surnx"2)/n)*sumx/n)
 . Rock mas: . properties

In deciding upon the value of ocjfor foliated roclís, a decision has to be made on whether to use the
st or the lowest uniaxial compressive strength ohtained from results such as those given iri Figure 11.1.
Mineral composition, grain size, grade of inetamorphism and tectonic history a11 play a role in deterinining
the characteristics of the rodí mass. The author cannot offer any precise guidance on the choice of o', hut
suggest that the maximum value should be used for hard, well interlocked rock masses such as good quality
slates. The lowesf uniaxial compressive strength should he used for tectonically dish~rhed,poor quality rode
masses such as the graphitic phyllite tested by Salcedo (1983).
Unlilíe otlier rocks, coa1 is organic in origin and therefore has unique coiistituents and properties.
Uiiless these properties are recognised and allowed for in characterising the coal, the results of any tests will
exhibit a large amount of scatter. Medhurst, Brown and Trueman (1995) have shown that, hy talíiiig into
account the 'brightness'which reflects the composition 2nd the cleating of the coal, it is possihle to differeii-
tiate benveen the mechanical characteristics of different coals.

11.4 Influente of sample size

The influence of sample size upon rock strength has heen widely discussed in geotechnical literature
aiid it is generally assumed that there is a significant reduction in strength witli increasing sainple size. Based
upon an analysis of puhlished data, Hoek and Brown (1980a) have suggested that tlie uniaxial compressive
strength o,of a rock specimen with a diameter of d mm is related to the uniaxial compressive strength o,,,,of
a 50 min diameter sample by the following relationship:

This relationship, together with the data upon which it was based, is illustrated in Figure 11.3

I Angle of schistosity to loading direction

Figure 11.2: nfluence of loading direction an the strength of graphitic phyllite tested by Salcedo (1983).
 Rock Engin eering COL

Cornp. Load Field eslimate

Grade* Terrn Strength Index of strength Exarnples
(MPa) (MPa)
R6 Extrernely > 250 >I0 Specirnen can only Fresh basalt, chert,
Stronq be chipped with a diabase, gneiss,
geological harnrner granite, quartzite

R5 Very 100 - 250 4 - 10 Specirnen requires Arnphibolite,

strong rnany blows of a sandstone, basalt,
geological hammer gabbro, gneiss,
t o fracture it granodiorite,
lirnestone, marble,
rhyolite, tuff

R4 ' Strong 50 - 100 2-4 Specirnen requires Lirnestone, rnarble,

more than one blow phyllite, sandstone,
of a geological harnrner schist, shale
t o fracture it

R3 Medium 25 - 50 1-2 Cannot be scraped Claystone, coa,

strong or peeled with a concrete, schist,
pocket knife, specimen shale. siltstone
can be fractured with
a single blow from a
geologicai harnrner

R2 Weak 5-25 ** Can be peeled with Chalk, rocksalt,

a pocket knife with ootash
difficulty, shallow
indentation made
by firrn blow with point
of a geological harnrner

R1 Very 1-5 ** Crurnbes under firrn blows Highly weathered or

weak with point of a geological altered rock
hamrner, can be peeled
by a pocket knife

R0 Extrernely 0.25 - 1 ** Indented by thumbnail Stiff fault gouge

weak
* Grade according t o Brown (1981)
* * Point load tests on rocks with a uniaxial cornpressive strength below 25 MPa are likely t o yield highly
 Rock mass I

Table 11.3: Valuei of the constant m, for intact rock, by rock group Note that values in parenthesis are
estimates
~ock I Class 1 Group Texture

type Coarse Medium Fine Very fine

1 1 Conalomerate Sandstone Siltstone Clavstone
Clastic (22) 19 9 4
Greywacke
(1 8)
Chalk

Clastic
C Organic

Carbonate

Chemical
Breccia
(20)
Sparitic
Limestone
(10)

Gypstone
Micritic
Limestone
8

Anhydrite
16 13
Non Foliated Marble Hornfels Quartzite
9 (19) 24
Õ Slightly foliated Migmatite Amphibolite Mylonitei
>
4 (30) 25 - 31 (6)

> Foliated* Gneiss Schists Phyllites Slate

33 4-8 (10) 9
Granite Rhyolite Obsidian
Light 33 (16) (1 9)
Granodiorite Dacite
(30) (17)
Diorite Andesite
(28) 19
Dark Gabbro Dolerite Basalt
27 (1 9) 17
Norite
22
Extrusive Agglomerate Breccia Tuff
pyroclastic type (20) (18) (1 5)
lese values are for intact rock specimens tested normal t o bedding or foliation. The value of m,will be
iificantly different if failure occurs along a weakness plane.
 Rock Eng~r

The author suggests that the reduction in strength is due to the greater opportunity for failure through
and around grains, the 'building blocks' of the intact rock, as more and more of these grains are included in
the test sample. Eventually, when a sufficiently large number of grains are included in the sample, the
strength reaches a constant value.
Medhurst and Brown (1996) have reported the results of laboratory triaxial tests on samples of 61,
101, 146 and 300 mm diameter samples of a highly cleated mid-brightness coa1 from the Moura mine in
Australia. The results of these tests are summarised in Table 11.4 and Figure 11.4.
The results obtained by Medhurst and Brown show a significant decrease in strength with increasilig
sample size. This is attributed to the effects of cleat spacing. For this coal, the persistent cleats are spaced at 0.3
to 1.0 m while non-persistent cleats withinvitrain bands aiid individual lithotypes define blodcs of 1 cm orless.
This cleating results in a 'critical' sample size of about 1 m above which the strengtli remains constant.
It is reasonable to extend this argument further and to suggest that, when dealing with large scale roclt
masses, the strength will reach a constant ?alue when the size of individual rodcpieces is sufficiently small in
relation to the overall size of the structure being considered. This suggestion is embodied in Figure 11.5
which shows the transition from an isotropic intact rockspecimen, through a highly anisotropic rock mass in
wliicli failure is controlled by one or m o discontinuities, to an isotropic heavily jointed rock mass.
The Hoek-Brown failure criterion, which assumes isotropic rock and roclí mass behaviour, should
only be applied to those rock masses in which there are a sufficient number of closely spaced discontinuities,
with similar surface characteristics, that isotropic behaviour involving failure on discontinuities can be as-
surned. When the structure being analysed is large and the bloclc size small in comparisou, the rock inass can
be treated as a Hoek-Brown material.
Where the bloclí size is ofthe same order as that of the structure being analysed ar when one oftlie
discontinuity sets is significantly weaker than the others, the Hoek-Brown criterion should not be used. In
these cases, the stability of the structure should be analysed by considering failure mechanisins involving the
sliding or rotation of blocl<sand wedges defined by intersecting stnictural features.

11.5 Geological stength Index

The strength of a jointed rock mass depends on the properties of the intact roclc pieces and also upon
the freedom of these pieces to slide and rotate under different stress conditions. This freedom is controlled
by the geometrical shape of the intact rock pieces as well as the condition of the surfaces separating the
pieces. Angular rodc pieces with clean, rough discontinuity surfaces will result in a much stronger rock inass
than one whicli contains rounded particles surrounded by weathered and altered material.
The Geological Strength Index (GSI), introduced by Hoek (1995) and Hoek, Kaiser and Bawden
(1995) provides a system for estimating the reduction in rock mass strength for different geological condi-
tions. This systein is presented in Table 11.5 and Table 11.6. Experience has shown that Table 11.5 is suffi-
cient for field observations since the letter code that identifies each rock mass category can be entered into a
field log. Later, these codes can be used to estimate the GSI value from Table 11.6.
Once the Geological Strength Index has been estimated, the parameters that describe the rock mass
strength characteristics, are calculated as follows:

GSI - 100
(11 10)

For GSI > 25, i.e. ro&.masses of good to reasonable quality, the original Hoek-Brown criterion is applicahle with
'I

r = -P[
GSI -100
]
 Rock rnass 1

and
a = 0.5 (11.12)
For GSI < 25,i.e. rock masses ofvery poor quality, the modified Hoek-Brown criterion applies with
s=0 (11.13)
and
GSI
a=0.65--
200

P.

- O Maible
o Lirnestone
a Gmnite
n Basalt
D Basalt-andesiie lava
a Gabbro
Marbls
m Nonte
A Granite
r Quartz dionte

I
O 50 100 750 200 250 3W

Specimen diameter d mm

Figure 11.3: lnfluence of ipecirnen size an the strength of intact rock. After Haek and Brown (1980a).

Table 11.4: Peak strength of Moura DU coa1 in terms of the parameters contained in equation ( 1 1.1)
based upon a value of o<,= 32.7 MPa.
Dia.ímrn) m. s a
61 19.4 1 .O 0.5
101 13.3 0.555 0.5
146 10.0 0.236 0.5
300 5.7 0.184 0.6
rnass 2.6 0.052 0.65
 Rock Engirleenng Coi

1W

40

80

70

E
2 50

ri-

-=e
5
i0

30

20

10

O
0 1 2 3 4 5 6 7 8 9 1 0

Canfming pressure o; MPa

Figure 11.4: Peak irrength f o r Austialian Maura Figure 11.5: Idealised dagram ihawing the transitionfrorn intact toa heavly
coa1 After Medhurst and Brown (1996). jointed rock marr w ~ t hncreaiing sarnple siie.

The choice of GSI = 25 for the switch hetween the original and modified criteria is purely arbitrary. It could
be argued that a switch at GSI = 30 would not introduce a discontinuity in the value of a, but extensive trials
have shown that the exact location of this switch has negligible practical significance.
For better quality rod< inasses (GSI > 25), the value of GSI can he estiinated directly from the 1976
version of Bieniawski's Rock Mass Rating, with the Groundwater rating set to 10 (dry) and the Adjustment
forfoint Orientation set to O (very favourable) (Bieniawski 1976). For very poor quality roclí masses the value
of RMR is very difficult to estimate and the balance hetween the ratings n o longer gives a reliahle basis for
estimating roclí mass strength. Consequently, Bieniawski's RMR classification should not be used for esti-
mating the GSI values for poor quality ro& masses.
If the 1989 version of Bieniawsl<i's RMR classification (Bieniawski 1989) is used, then GSI = RMq,'
- 5 where RMR,,' has the Groundwater rating ser to 15 and the Adjustment forfoint Orientation set to zero.
One of the practical prohlems which arises when assessing the value of GSI in the field is related to
blast damage. As illustrated in Figure 11.6, there is a considerahle difference in the appearance of a rock face
which has heen excavated by controlled blasting and a face which has heen damaged by bulk blasting.
Wherever possihle, the undamaged face should be used to estimate the value of GSI since the overall aiin is
to determine the properties of the undisturhed ro& mass.
Where all the visible faces have been damaged hy blasting, some attempt should be made to coinpen-
sare for the lower values of GSI obtained from such faces.'In recently blasted faces, new discontinuity
surfaces will have been created by the blast and these will give a GSI value that may he as much as 10 points
 ock mass piroperties

lower than that for the undisturbed rock mass. In other words, severe blast damage can be allowed for by
inoving LI^ one row in Table 11.5 and Tahle 11.6.
Where blast dainaged faces have been exposed for a nuinher ofyears, it may also be necessary to step
as mucb as one column to the left in arder to allow for surface weathering which will have occurred during
this exposure. Hence, for example, a badly hlast dainaged weathered rodí surface which has the appearance
of a BLOCKYIDISTURBED and FAIR (BD/F in Table 11.5) rodí inass may actually be VERY BLOCKY
and GOOD (VB/G) i11its unweathered and undisturbed in situ state.
An additional practical question is whether borehole cores can be used to estimate the GSI value
behind the visible faces? For reasonable quality rock masses (GSI > 25) the best approach is to evaluate the
core in terins of Bieniawslci's RMR classificatioii and then, as described above, to estimate the GSI value
from RMR. For poor quality rock masses (GSI i 25), relatively few intact core pieces longer than 100 inin are
recovered and it becomes difticult to deterinine a reliable value for RMR. In these circumstances, the physi-
cal appearance of the material recovered in the core should be used as a basis for estiinating GSI.

11.6 Mohr-Coulomb parameters

Most geotechnical software is written in teims of the Mohr-Couloinb failure criterion in wliicli the
roclc mass strength is defined by tlie cohesive strength L and the angle of friction $'. The linear relationship
between the inajor and minar principal stresses, õ, and o,, for the Mohr-Coulomb criterioii is

where o',,, is the uniaxial compressive strength of the roclí mass and Ris the slope of the liiie relatiiig o, 2nd
o,. The values of $' and c' can Le calculated from

-sina')
C = (11.17)
2 cos l)
There is no direct correlation hetween equation 11.15 and the non-linear Hoelí-Brown criterion defined by
equation 11.1. Consequently, determination of the values ofc'and$' for a rodí inass that has been evaluated
as a Hoek-Brown material is a difficult problem.
The author helieves tliat tlie inost rigorous approach available, for the original Hoek-Brown criterion,
is that developed by Dr J.W. Bray aiid reported by Iloek (1983). For any point o n a surface of concern in an
analysis such as a slope stability calculation, the effective normal stress is calculated using an appropriate
stress analysis technique. The shear strength developed at that value of effective normal stress is then calcu-
lated froin the equations given in Hoek and Brown (1997). Tlie difficulty in applying tliis approadi iii
practice is that inost of the geotechnical software currently available provides for constant rather thaii effec-
tive normal stress dependeiit values of c'and $'.
Having evaluated a large number of possible approacbes to this problein, it has been concluded tliat
the most practical solution is to treat the problem as an analysis of a set of full-scale triaxiai strength tests.
The results of such tests are siinulated by using the Hoelí-Brown equation 11.1 to geiierate a series of triaxial
testvalues. Equation 11.15 is then fitted to these test results by linear regression analysis and the values o f i
and 4' are determined from equations 11.17 and 11.16. The steps required to determine the parameters A, B,
I and $' are given below. A spreadsheet for carrying out this analysis, with a listing of all the cell forinulae, is
given in Figure 11.7.
 neering Cc
Rock Eng~

Table 11.5: Character isation of rock masses o n the basis o f interlocking and joint alterationl.
r- v,
ROCK MASS CHARACTERISTICS FOR
STRENGTH ESTIMATES
", E ",
0
u> 5 E
Based upon the appearance of the rock, choose the
category that you think gives the best description of v .i v,

the 'averege' undisturbed in situ conditions. Note E

v,
-
3 8:
that exposed rock faces that have been creeted by
m
L
g3 0
23
"7 O v,c v,
blasting may give a misleading impression of the n c n u m v
quality of the underlying rock. Some adjustment for L e
.- E
a
õ L
"
blast darnage rnay be necessary and examination of n- c
5 5 5% 5 8,
diernond drill core or of faces created by pre-split or Z g L
a È 8' " C

srnooth blasting may be heipful in matdng these 2 g I; -

p-
adjustments. It is alsa important to recognize that t 3 m 2 ô
$ô
the Hoek-Brawn criterion shauld only be appiied to g r;a> 5.0, 0
.
rock rnasses where the size of individual blocks is v L h r,& r 2
srnail compared M t h the size of the excavation
under consideration.

STRUCTURE DECREASING SURFACE QUALITY .3

-
BLOCKY very weli interiocked
undisturbed rock rnass consisting
of cubical blocks formed by three BNG BIG BIF BIP BNP
orthogonai discontinuity sets
V)
W
O
&!
a
Y
, ; ,, VERY BLOCICY - interlocked, O
.,, . \; . partially disturbed rock mass with
multifaceted angular blocksfoimed u VBNG VBIG VBIF VBIP VBIVP
.
'\I
-...
. . . ;,.
'. , by four or more discontinuity sets O
Z
3
8
iI
(L
ill
C
BLOCKYIDISTURBED- folded z
formed by many interseding 2
V>
BDNG BDIG BDIF BDIP BDNP

9E
O
W
n
-
DISINTEGRATED poarly inter-
locked, heavily broken rock mass
v
with a mixture or angular and DNG DIG DIF DNP
DIP
rounded rock pieces

'Ir earlier verstons of tlis table the terms BLOCKYISEAMY and CRUSHED were uied. falowing the terminoogy uied by Tenaghi (1946) Hawever
these terms proved to be misleading and they have been replaced. in this table by BLOCKYIDISTURBED, which more accurately refleits the increaied
mobiity of a rock mas5 which has undergane somefoiding andlor faulting, and DISINTEGRATEDwhich encompanes a wider range of particleihapei
 Rock mas8

Table 11.6: i,' t. ?'! 3 >:

i.' . . 3 - 1 : - I-(l€, 68 :3scci 2.- ?r:- ..,; ! .? ,:F ;:I pl 71.:

GEOLOGICAL STRENGTH INDEX

Frorn the letier codes describing the structure

and surface conditions of the rock rnass (from
Table 41, pick-the appropriate box in this chart.
Estimate the average value of the Geological
Strength index (GSI) from the contours.
Do not atiempt to be too precise. Quoting a
range of GSI from 36 to 42 is more realistic
than stating that GSi = 38.
 eering COL

Figure 11.6: Cornpariion

between the results
achieved using controlled
blasting (an the eft) and
normal b u k blasting fora
surface excauation in gneiii.

The relatioiisliip between the normal and sliear stresses can be expressed i11 terins of the correspond-
ing principal effective stresses as suggested by Balmer (1952):

7 = (o; -o ; ) - \ i a
For the GSI> 25, wheii a = 0.5:

For GSI< 25, when r = 0:

The tensile strength of the rock mass is calculated from:

The equivalent Mohr envelope, defined by equation 11.2, may be written in the forin:
Y=logA+BX (11.23)
 Rock rnass properties

Using the value ofot,,,calculated from equation 11.22 and a range of values of .I a n d o , calculated froin
eqiiations 11.19 and 11.18 the values o f A and B are determined Ly linear regression where :

A= 10^(2Y/T- B ( 5 / 7 ) ) (1 1.26)
and Tis the total number of data pairs included in the regression analysis.
Tlie most critical step iti this process is the selection ofthe range ofo; values. As far as the author is awari.,
there are no theoretically correct inetliods for choosing this range and a trial and error method, based upon
practical compromise, has been used for selecting the range included i11 the spreadsheet presented in Figure 11.7.
For a Moiir envelope defined by equation 11.2, the friction angle$i for a specified normal stress o,,,
is given by:

The corresponding cohesive strength is given by:

and tiie corresponding uniaxial compressive strength of the rock inass is :

21, tos$,
0cm1 = (1 1.29)
1 - sin 9,
Note that the cohesive strength c; given by equation 11.29 is an upper bound value and that it is prudent to
reduce this to about 75% ofthe calculated value for practical applications.
The values of C and Q' obtained from this analysis are veiy sensitive to the range of values of the
rninorprinc/pal stress o; used to generate the simulated full-scale triaxial test results. O n the basis of trial and
error, it has been foiind that the most consistent results are obtained when 8 equally spaced values ofõ; are
used in the range O < o,' < 0.250~;.
Ali example ofthe resiilts, which are obtained from this analysis, is given in Figure 11.8. Plots of the values
ofthe ratio c Y q , and the friction angle $', for different combinations of GSI and miare given in Figure 11.9.
The spreadsheet includes a calculation for a tangent to the Mohr envelope defiiied by equatiori 11.2.
Anormal stress has to be specified in order to calculate this tangent and, in Figure 11.8, this stress has heen
chosen so that the friction angle $' is the same for both the tangent and the line defined by c'= 3.3 MPa and
$' = 30.1", determined by the linear regression analysis described earlier. The cohesion intercept for the
 Rock Engirieerlng Coi

tangent is c'=4.1 MPa which is approximately 25% higher than that obtained by linear regression analysis of
the simulated tria+al test data.
Fitting a tangent to the cuwed Mohr envelope gives an upper bound value for the cohesive intercept
c'. It is recomineiided that this value be reduced by about 25% in order to avoid over-estiination of the rod<
mass strength.
There is a particular class of problem for which extreme caution sliould be exercised when applyilig
the approach outlined above. In some roclc slope stability problems, the effective norinal stress on some
parts of the failure surface can be quite low, certainly less than 1 MPa. It will be noted that iti the exarnple
given in Figure 11.8, for values ofon of less than about 5 MPa, the straight line, constant L and @' method
overestiinates the available shear strength of the rock mass by increasingly significant arnouiits asõ, ap-
proaches zero. Uiider such circumstances, it would be pmdeiit to use values of 'c and $' based on a tangent
to the sliear strength curve in the range ofu, values applying in practice

11.7 Deformation modulus

Serafim aiid Pereira (1983) proposed a relationship between the in situ modulus of deforrnatiou aiid
Bieniawsla's RMR classification. This relationship is based upon badc analysis of dain foundation deforina-
tions and it has been fouiid to worliwell for better quality rocks. However, for maiiy of the poor quality rocl<s
it appears to predict deformation modulus values which are too high. Based upon practical obsematioiis and
back analysis of excavation behaviour in poor quality rock masses, the following modification to Serafiin and
Pereira's equation is proposed for 0,; < 100:

(F)
E,,, = &I0

Note tliat GSI lias been substituted for RMR in this equation and that the modulus E,,, is reduced progres-
sively as the value of falls below 100. This reduction is based upon the reasoniiig that the deformation of
better quality roclc masses is controlled by the discontinuities while, for poorer quality roclc tnasses, tlie
deformatioti of the intact rodc pieces contributes to the overall deformation process.
Based upon measured deformations, equation 11.30 appears to worl<reasonably well in those cases wliere
it has been applied. However, as more field evidence is gathered it may be necessary to modify this reiationship.

11.8 Post-failure behaviour

Wheii using numerical models to s ~ d the y progressive failure of rodc masses, estiinates of the post-pealc or
post-failure characteristics of the rodi mass are required. In some of these inodels, the Hoelc-Brown failure uiterion
is treated as a yield critenon and the analysis is carried out usingplasticity theory (e.g. Pan and Hudson 1988). No
definite mles for dealing with this problern can be given but, based upon experience in nurnerical analysis of avariety
of practical probleins, the post-failure characteristics illustrated iti Figure 11.10 are suggested as a startiiig poiiit.

11.8.1 Very good quality hard rock masses

For very good quality hard rodi masses, such as massive granites or quartzites, the analysis of spallirig
around highly stressedopenings (Hoelí, Kaiser and Bawden 1995) suggests that the rocli mass behaves i11an
elastic brittle manner as shown in Figure 11.10(a). When the strength of the rock mass is exceeded, a suddeii
strength drop occurs. This is associated with significant dilation of the brolcen roclipieces. Ifthis broken ro&
is confined, for example by rocli support, then it can be assumed to behave as a rock f i I I with a friction angle
of approximately $' = 38" and zero cohesive strength.
 Rock mas5 properties

Hoek-Brown and equivalent Mohr Coulomb failure criteria

input: I sigci = 85 MPa I mi= 10 I GSI= 45 I

Output:
I "
mb=
siritm
k=
=
1.40
-0.13
3.01
MPa I s = 0.0022
A= 0.50
phi= 30.12 degrees
I a=
3=
i
coh=
0.5
0.70
3.27 MPa
I
sigcm = 11.36 MPa E = 6913.7 MPa

Tangent: I siqnt= 15.97 MPa I phit= 30.12 deqrees I coht= 4.12 MPa I
Caiculation:
Sums
sig3 1E-10 3.04 6.07 9.1 12.14 15.18 18.21 21.25 85.00
sigl 4.00 22.48 33.27 42.30 50.40 57.91 64.98 71.74 347.08
dslds3 15.89 4.07 3.19 2.80 2.56 2.40 2.27 2.18 35.35
sign 0.24 6.87 12.56 17.85 22.90 27.76 32.50 37.13 157.80
ta" 0.94 7.74 11.59 14.62 17.20 19.48 21.54 23.44 116.55
x -2.36 -1.08 -0.83 -0.67 -0.57 -0.48 -0.42 -0.36 -6.77
V -1.95 -1.04 -0.87 -0.76 -0.89 -0.84 -0.60 -0.56 -7.11
xy 4.61 1.13 0.71 0.52 0.39 0.31 0.25 0.20 8.12
xsq 5.57 1.17 0.68 0.45 0.32 0.23 0.17 0.13 8.74
sig3sigl 0.00 68.23 202.01 385.23 612.01 878.92 1183.65 1524.51 4855
sig3sq 0.00 9.22 36.86 82.94 147.45 230.39 331.76 451.56 1290
taucalc 0.96 7.48 11.33 14.45 17.18 19.64 21.91 24.04
siqlsiq3fit
. . 11.36 20.51 29.66 38.81 47.96 57.11 66.26 75.42
signtaufit 3.41 7.26 10.56 13.63 16.55 19.38 22.12 24.81
tangent 4.25309 8.10321 11.4032 14.4729 17.3991 20.2235 22.9702 25.655

Cefl forrnulae:

a= IF(GS1~25.0.5,0.65-GSI1200)
siglm = O.s'sigci'(mb-sQRT(mW+4-s))
A= -
acalc = lO"(sumyl8 bcalc'surnxl8)
B= -
bcalc = (sumxv (sumx'sumv)/8)/(sumxsq - (sumxA2)18)

C= ' 59: ,100 tn00'10'(,GSI-101 .10 .SQRTis gc 100.~1000'10'. GS'-10,13 ,

ph I = ATAF. aca~c~oca~c' sipnt-sig'n, s'gc)' oca c-I, :i83 1'1,
cnni - acyic's qc'( s qii1.s qln> s q i 'o:nl:.sigrii'TAh(rihI'Pi 10:)
5.~= 3 Slan ai 1 C - 1 J 17 aloia zero crrois, ano ncrcmeri n 7 sicps o1 sgc 28 l i 0.25's ac
s.qi = S Q ~ ~ S I ~nlo.sig3
C.', S!QC~ . s v a
051~153- F CS ~ 2 5 li , rno'sgc 2' sgl-s:33 , I. a'iiio.3 ' s;3sqc ',>.I
sian =
tau =
-. "
sio3+(sial-sia3l/(t+dslds3)
(sign-sig3)'SQRT(dsIds3)
x = LOG((sign-sigtm)isigci)
y = LOG(tau1sigci)
xy = x'y . x sq = xA2 sig3sigl= sig8sigl sig3sq = sig3"2
taucalc = acaic*sigci'((sign-sigtm)/sigci)nbcalc
s3sifit = sigcm+k'sig3
sntaufit = coh+sign'TAN(phi'Pi()/l80)
tangent = coht+sign'TAN(phit*Pl()ll80)

Figure 11.7 Spieadiheet f o i raiiuation of Hoek~Brownand equivalent Mohr~Coulomb

pararneten.
 O 10 20 30 O 10 20 30 40
Minoiprinopalrlrerr MPa Normal rirerr o; MPa

Figure 11.8: Pot of reiults from simulated full scale triaxial tests on a rock m a s defned by a unaxial compreaive strength o, = 85 MPa. a Haek
Brown canstant m, = 10 anda Geological Strength index G i l = 45.

5s
m,
85
30
so
28
20
.I

-.
w
p '"
*
-
",
-
ai

80

-
c

Y ií
L

20 / / A -

,/ I
Geological Sfienglh n d e x G S I ,o 20 30 <o so 60 ,o 80 ao
. . G e o l o g c a l Strsnglh n d e x G S I

Figure 11.9: Plots of iaheiive Strength and f i i i t o n angier for different GS1 and m,valuei.
 Rock masc,properties

O 000 0001

strain
O 002

(a) Very good quality hard rock mass

O 003
",
e,
C
W
5

to L
OO000 0001

Slrain
Strain sofiening

O 002

(b) Average quality rock mass

O 003

Slrain

(C)Very poor quality soft rock mass

Figure 11.10: Suggested post failure characterstics for different quality rock masei.

Typical properties for this very good quality hard roclc mass may be as shown in Table 11.7. Note that,
in some nuinerical analyses, it may be necessary to assign a very sinal1 cohesive strength in ordei- to avoid
numerical instability.

11.8.2 Average quality rock mass

In the case of an average quality rock mass it is reasonable to assume that the post-fàiiure cliaracteris-
tics can be estimated by reducing the GSIvalue from the in situ value to a lowervalue which characterises the
broken rock mass.
The reduction of the roclc mass strength from the in situ to the broken state corresponds to the strain
 Rock Engin

softening behaviour illustrated in Figure 11.10(b). In this figure it has been assutned that post failure defor-
mation occurs a t a constant stress level, defined hy the compressive strength of the brolcen rode rnass. The
validity of this assumption is unknown.
Typical properties for this average quality rodc mass may be as follows:

Table 11.7: Typical properties for a very good quality hard rock mass
lntact rock strength o<, 150 MPa
Hoek-Brown constant 8'"
25
Geological Strength Index GSI 75
Friction angle 46"
Cohesive strength 1 3 MPa
Rock mass compressive strength 64.8 MPa
Rock mass tensile strength -0.9 MPa
Deformation modulus 42000 MPa
Poisson's ratio 0.2
Dilation angle q / 4 = 11.5"
Post-peak characteristics
Friction angle
Cohesive strenath
Deformation modulus E~m
-, 10000 MPa

Table 11.8: Tvi-iical ~roisertiesfor an averaoe rock mass.

lntact rock strength O, 8 0 MPa
Hoek-Brown constant "', 12
Geological Strength Index GSI 50
Friction angle $' 33"
Cohesive strength C' 3.5 MPa
Rock mass compressive strength 'im 1 3 MPa
Rock rnass tensile strength Otm -0.1 5
Deformation modulus Em 9000 MPa
Poisson's ratio v 0.25
Dilation angle a v18 = 4"
Post-peak characteristics
Broken rock mass strength Ofcm 8 MPa
Deformation modulus E?rn 5000 MPa

11.8.3 Very poor quality rock mass

Analysis of the progressive failure of very poor quality rodc masses surrounding tunnels suggests that
the post-failure characteristics of the rock are adequately represented by assuming that it behaves perfectly
plastically. This means that it continues to deform at a constant stress level and that n o volume change is
associated with this ongoing failure. This type of behaviour is illustrated in Figure 10.10(c).
Typical properties for this very poor quality rodc mass may be as follows:
 Rock mass properties

Table 11.9: Tvnical nronerties for a verv ooor oualitv rock mas8
Intact rock strength OC, 20 MPa
Hoek-Brown constant 8' "
8
Geological Strength Index GSI 30
Friction angle Q' 24"
Cohesive strength c' 0.55 MPa
Rock mass compressive strength Orm 1.7 MPa
Rock mass tensile strength -0.01 MPa
Deformation modulus 1400 MPa
Poisson's ratio 0.3
Dilation angle zero
Post-peak characteristics
Broken rock mass strength 1.7 MPa
Deformation rnodulus 1400 MPa

11.9 Reliability of rock mass strength estimates

The techniques described in the preceding sections ofthis chapter can be used to estimate tlie strength
and deformation characteristics of isotropic jointed rock masses. When applying this procedure to rock
engineering design problems, most users consider only the 'average' or mean properties. In fact, all of these
properties exhibit a distribution about the mean, even under the most ideal conditions, and these distribu-
tions can have a significant impact upon the design calculatioiis.
In the text that follows, a slope stability calculation and a tunnel support desigii calculatioii are
carried out in order to evaluate influente of these distributions. In each case the strength and deforination
characteristics of the rock mass are estimated by means of the Hoelc-Brown procedure, assuming tliat tlie
three input parameters are defined by normal distributions.

11.9.1 Input parameters

Figure 11.11 has been used to estimate the value of the value of GSI from field observations of
blockiness and discontinuity surface conditions. Included in this figure is a crosshatched circle representing
the 90% confidence limits of a GSI value of25 -t 5 (equivalent to.a standard deviation of approxiinately 2.5).
This represents the range of values that an experienced geologist would assign to a rock mass described as
BLOCKY/DISTURBED or DISINTEGRATED and POOR. Typically, roclu such as flysch, scliist and soine
phyilites may fall within this range of rod< mass descriptions.
In the author's experience, some geologists go to extraordinary lengths to try to determine an 'exact'
value of GSI (or RMR). Geology does not lend itself to such precision and it is simply not realistic to assign
a single value. A range ofvalues, such as that illustrated in Figure 11.1 1 is more appropriate. In fact, iii soine
complex geological environments, the range indicated by the crosshatched circle inay be too optimistic.
The two laboratory properties required for the appli&tion of the Hoelc-Brown criterion are the ui~iaxial
compressive strength of &e intact rock(o,J and the intact rodr material constant 5 Ideally these iwo paraineters
should be determined by triaxial tests on carefully prepared specimens as described by Hoek aiid Brown (1997).
It is assumed that all tliree input parameters can be represented by iiorinal distributions as illustrated
in Figure 11.12. The standard deviations assigned to these three distrihutions are based upon the author's
experience of geotechnical programs for major civil and mining projects where adequate funds are available
for high quality investigations. For preliminary field investigations or 'low budget' projects, it is prudent to
assume larger standard deviations for the input parameters.
 Rock Englneerfng Coi

GEOLOGICAL STRENGTH INDEX ",

Om
"7
a
O

3m
-"7
c
1 0
.?E
From the leiter codes describing the structure
and surface conditions of the rock mass (from
5
V>
v 3: $
Table 4), pick the appropriate box in this chait.
",

ai
O u &'f &'O
c -
gs 3
a>
Estimate the average value of the Geological .-c m
m 3 m
Strength Index (GSI) from the contours. Y) V> õ c Y)
Do not aitempt to be too precise. Quoting a v
range of GSI from 36 to 42 is more realistic
than staiing that GSI = 38.

STRUCTURE DECREASING SURFACE QUALITY e

/ V / A / A / /I//

BLOCKYIDISTURBED- folded
andlor faulted with angular blocks
formed by many intersecting
discontinuity sets

-
DISINTEGRATED poorly inter-
locked, heavily broken rock mais
with a mixture or angular and
rounded rock pieces

Figure 11.11: Ertimõte of Geological Strength Inden GSI baied on geological descriptions.
 Rock mas8 properties

11.9.2 Output parameters

The values o f t h e friction angle I), the cohesive strength c', the iiniaxial compressive strength of the
, ~ the deformation inodulus E,,, of the rock mass were calculated by the procedure de-
rock inass o ~aiid
scribed in previous sections of this chapter. The Excel add-on prograin @RISKZwas used for a Monte
Carlo analysis iii which 1000 calculations were carried out for randomly selected values of the iiiput
parameters. The results of tliese calculatioiis were analysed using the program BESTFIT' and it was fouiid
tliat all four output parameters could be adequately described hy the normal distributions illustrated iii
Figure 11.12.
In severa1 trials it was found that the output parameten 4, c'and n?,,, were always well represented by
iiormal distributions. O n the other hand, for GSIvalues ofmore than 40, the deformatioii inodulus E,,,was
better represented by a lognormal distribution.

11.9.3 Slope stability calculation

In order to assess the impact of the variation in output parameters, illustrated in Figure 11.12, a
calculation of the factor of safety for a homogeneous slope was carried out using Bishop's circular failure
analysis in the program SLIDE1.The geometry of the slope and the phreatic surface, the rock inass properties
and the critical failure surface for the 'average' properties are sliown in Figure 11.13.
Tlie distribution of the factor of safety was determined by Rosenbleuth's Point Estiinate inetliod
(Rosenhleuth 1976, Harr 1987) in which the two values are chosen at one standard deviation o n either side of
the mean for each variable. The factor of safety is calculated for every possible comhination of poiiit esti-
mates, producing 2" solutions, where m is the number of variables considered. The ineaii and srandard
deviation of the hctor of safety are then calculated froin these 2"' solutions.
This calculation of the mean and standard deviation is given in Table 11.10. Based upoii tlie fact that
the two variables included in this aiialysis are defined by normal distributioiis and considering the foriii of
the equatioiis used to calculate the factor ofsafety, it is reasonahle to assume that the factor of safety will be
adequately represented by a normal distribution. This distribution is illustrated in Figure 11.13.
The inean factor of safety for this slope is 1.3 that is a value frequently used in the desigii of slopes for
opeu pit mines. It is interesting that the probability offailure, given by the portion of the distrihution curve
for SF < 1, is very small. This suggests that, for a high quality geotechiiical investigation such as that assumed
in this study, a safety factor of 1.3 is adequate to ensure stability under the assumed conditions.

11.9.4 Tunnel stability calculations

Considera circular tunnel of radius r. in a stress field in which the horizontal and vertical stresses are
bothpo. If the stiesses are Iiigh enougli, a 'plastic' zone of damaged rode of radius r,, surrounds the tunnel. A
unifbrm support pressurep; is provided arouiid the perimeter of the tunnel. Tliis situation is illustrated iii
Figure 11.15.
Assuining that the rock inass fails with zero plastic volume change, the critical stress ievelp6,atwhich
failure initiates is giveii by :

'From Paiiade Corporation, 31 Deiker Road, Newfieid. New York 14867, USA.
'Available from Available from Rocscienie Ini. 31 Basam Avenue. Toronto, Ontario. Canada M4E 385, Fax 1 416 698 0908, Phone 1 416 698 8217
Email: ioftwarearocscence com, Internet http:ll~w.rociiiencecam.
160 Rock Engineering. Cours

INPUT 1 OUTPUT

Mean 10 MPa
Mean 22 85'
Stdev 2.5 MPa
Stdev 1.31'

O 5 10 15 20 18 20 22 24 26 28

Intact rock strength - MPa Fnction anga - degrees

.-2
- -.
i0 Mean 8 Mean O 23 MPa
Stdev 1 Stdev O 068 MPa
a
0.10

0.00

1
4 6 8 10 0.0 01 0.2 0.3 0.4 0.5

Haek-Brown constaot mi Cohesive strength MPa -

Mean 25
Stdev 2.5

15 20 25 30 35 00 O5 1.0 1.5

Geologlcai Strength Index GSi Rock mass strength - MPa

00100 r

-.,
.-
.-
n
Mean 373 MPa
Stdev 48 MPa
,m 0.ooso
e
a

200 250 300 350 400 450 500 550

Defonnation rnodulus - MPa

Figure 11.12: Asiumed normal distributons for nput parameteo and calculated distributionsfor output parameteri.
 Rock rnass p r o p e r t i e s

z $ r " ~ y $ z z
Factor of Safety

Figure 11.13: Sope and phreatic surfaie geornetry, rock masi proper Figure 11.14: Normal distribution of the factor of iafety
tiei and citical faiiure surface for a hornogeneoui ilope. of lhe rlope defined Agure 11.13.

Figure 11.15: Deveopment of a plaitic rone around a circular rvnnei in a hydrostatic stresi fieid.

Table 11.10: Calculations for Rosenbleuth's Point Estimate rnethod using + one standard deviation.

Case Friction Angle Cohesion ' Safety Factor (SF-F,)'

- 1n
Mean Safety Factor = SF = - CSF; = 1.31
n ,=i

2
1
Standard deviation = S' = - c(G
n

n - l ,=i
- SF;) = 0.11
where

Where the support pressurep, is less than the critical pressurep,, the radius r? of the plastic zone aiid
tlie inward deforination of the tunnel wall a,? are given by:

L[2
7,
( (1 O',)

lL
(1+ 4 ( ( k -l)P, + o o n )
(k-1)

In order to study the influente of the variation in the input parameters, a Moiite Carlo aiialysis was per-
forined using tlie prograni @RISKin an Excel spreadsheet that had been prograinined to periòrin the analy-
sis defined above. It was assuined that a 5 m diameter tunnel ( r = 2.5 m) was subjected to uiliform in situ
stress o f y o = 2.5 MPa. The rock mass properties were defined by the norinal distributions for @, c, q,,, aiid E
defined in Figure 11.12.
This analysis was carried out for a tunnel with n o support. A second analysis was perforined for a
tunnel with a support pressure offi = 0.3 MPa which is approximately that which can be achieved with a
closed ring o f 5 0 inm thick shotcrete with a uniaxial compressive strength of 14 MPa (after 1 day of curiiig).
This would represent the early support that would be achieved by the imrnediate application ofshotcrete
behind the advancing face. A third analysis was performed for a support pressure p; = 0.8 MPa. This is
approximately tlie support which can be achieved in this size of tunnel by a 75 m m thick shotcrete lining
with a uniaxial coinpressive strength of 35 MPa (cured for 28 days). The results of these analyses are suininarised
graphically in Figures 11.16 and 11.17.
Figures 11.16 and 11.17 show that the size of the plastic zone and the tunnel deformation can be
represented hy lognorinal distributions. As would be expected, the mean values for tlie size of the plastic zone
aiid tlie magnitude of the sidewall displacernents are reduced significantly by the installation of support.
What is surprising is the drainatic reduction in the standard deviations with increasing s~ipportpres-
sure. This is because of the strong dependence ofthe size of the plastic zone upon the difference betweeii the
critical pressurep<,aiid the support pressurep,. A detailed discussion on this dependeilce is beyond the scope
of this technical note and is the subject of ongoing researdi by the author.
From the results of the analysis described above it is evident that tlie installation o f a relatively siinple
support system is very effective in controlling the behaviour of this tunnel. Without support there is ai1
approximate 50% probability of severe instability and possibie collapse of the tuimel. A plastic zone diain-
eter of 15 in and a tunnel closure of 50 mm in a 5 m diameter tunnel would certainly cause visible signs of
distress. The fact that a relatively thin shotcrete lining can control the size ofthe plastic zone aild the closure
of the tunnel provides confirmation of the effectiveness of support.
A word ofwarning is required at this point. The example described above is for a 5 in diaineter tunnel
at a depth of approximately 100 m below surface. For larger tunnels at greater depths, the plastic zone and
the displacernents can be significantly larger. The demands on the support system may be such that it inay br
very difficult to support a large tunnel in poor ground at considerable depth below surface.
 Rock mass I

suppori pressure = 0.8 MPa suppoTt pressure = 0.8 MPa

lognornai distribution iognormai distribution
mean = 1.31, stdev = 0.11 mean = 0.0016. stdev = 0.0005
4.0 r
suppori pressure = 0.3 MPa suppori pressure = 0.3 MPa
lognormal distribution lognormal distribution
mean = 1.94. stdev = 0.30 mean = 0.004, stdev = 0.0019
.-+Ii.
- Ii
0 no support .-
0
m 2.0 lognormai distribution 0 no suppon
0 mean 2.88. stdev = 0.75 lognormal distribution
n 2 mean = 0.0102. stdev = 0.0079
1.0
a
250

0.0 o
. 1 2 3 4 5
0.W 0.01 0.02. 0.03 0.04
Plastic z o n e radius / tunnel radius Tunnel d e f o n a t i o n / tunnel radius
I I
2.5 5.0 75 100 12.5
O 25 50 75 1O0
Plastic zone radius -m Tunnel defomation - rnm

Figure 11.16: Lognorrnal distrbutions repreienting the rangeof Figure 11.17: Lognorrnal ditribul~onrrepreienting the range of
pastc ione i a d for
~ different iupporting preiiurei. tunnei deformationi for different support preiiurei.

11.9.5 Conclusions
The uncertainty associated with estimating the properties of in situ roclí rnasses lias a sigiiificant
impact or the desigii of slopes and excavations in roclc. Tlie examples that have been explored in tliis sectioii
show that, even when using the 'best' estimates currently available, the range of calculated factors oisafety or
tuiiiiel behaviour are uncomfortably large. These ranges become alarmingly large when poor site iiivestiga-
tion techniques and inadequate laboratory procedures are used.
Given the inherent difficulty of assigning reliable numerical values to rode inass characteristics, it is
unlikely tliat 'accurate' methods for estimating rock mass properties will be developed in the foresçeablç
future. Consequently, the user of the Hoelc-Brown procedure or of any other equivaleiit procedure for
estimating rodc mass properties should not assume that the calculations produce unique reliable iiuinbers.
Tlie simple techiiiques described in this section can be used to explore the possihle range of values aiid the
impact of tliese variations on engineeriiig design.

11.10 Practical examples of rock mass property estimates

Tlie following examples are presented i11 order to illustrate the range of rock rnass properties that cai1
be encountered in the field and to give the reader some insight ofhow the estirnation of rock mass properties
was tackled in a number of actual projects.

11.10.1 Massive w e a k rock

Karzulovic and Diaz (1994) have described the results of a prograin of triaxial tests o11 a ceinented
breccia luiown as Braden Breccia from the E1 Teniente mine in Chile. In order to desigii undergroiind
openings in this roclc, attempts were made to classify the rod< mass in accordance with Bieniawski's RMR
system. However, as iilustrated i11 Figure 11.18, this rock rnass has very few discontinuities and so assigiiiilg
 Rock Engtneertng CciuBe notes

realistic iiuinbers to terms depending upon joint spacing and condition proved to be veiy difficult. Finalli
it was decided to treat the rock inass as a wealc but homogeneous 'almost intact' rode and to determine its
properties by means of triaxial tests o n large diameter specimens.
A series of triaxial tests was carried out o n 100 mm diaineter core samples, illustrated in Figure
11.19. The results of these tests were analysed by means of the regressioil analysis presented in Section
11.3. Back analysis of the behaviour of underground openings in this roclc iiidicate that the in-situ
GSI valiie is approximately 75. From the spreadsheet presented in Figure 11.7 the following paraineters
were obtained:

Intact rock strength O'; 51 MPa Friction angle a' 42"

Hoek-Brown constant m, 16.3 Cohesive strength 4.32 MPa
Geological Strength Index GSI 75 Deformation inodulus E8,a 30000 MPa
Hoelc-Brown constant ' s 0.062

Figure 11.18: Braden

Breccia at €1 Teniente Mine
n Chile. Ths rack i i a
cernented brecca with
praiticaly na jaintr. fuiar
d e a t with n a manner
similar to weakconcrefe
and tests were iarried out
on 100 mrn diameter
ipeiimeni iluitrated n
Figure 11.19
 Rock mass properties

Figure 11.19: 100 rnm diameter by 200 mm long ipeii- Figure 11.20: Excelent quality caiewtir veiy f e w d i -
mensof Biaden Brerca from the E1 Teniente mne n Chile. continuitiesfiom the massve gneiisaf the Ro Gnnde
piojert n Aigentna

Figure 11.21: Top heading

ofthe12mspan. 1 8 m
high taliaiétunnel f o i
the Ria Grande Pumped
Storage Pro~ect.
 eering COL

11.10.2 Massive strong rock masses

Tlie Rio Grande Puinped Storage Project in Argentina includes a large undergroiind powerhouse
aiid surge control complex and a 6 Icm loiig tailrace tunnel. The rock inass surrouiiding tliese excava-
tioris is a inassive gneiss with very few joints. A typical core froin this rod< inass is illustrated i11 Figure
11.20. The appearaiice of the rock at the surface is illustrated in Figiire 11.6, whicli shows a cuttiiig for
the dain spillway.
The rock mass can be described as BLOCKYNERY G O O D and the GSI value, froin Table 11.6, is
75. Typical characteristics for the rock inass are as follows:

Intact roclc strength o?) 110 MPa Friction angle a' 43"
Hoel<-Brown constant m; 17.7 Cohesive strength C' 9.4 MPa
Geological Streiigth Iiidex GSI 75 Roclc mass compressive strength o,,,, 43 MPa
Hoelc-Brown constant m,, 7.25 Roclí mass tensile strengtli o,,j, -0.94 MPa
Hoelí-Brown coiistaiit s 0.062 Deforination rnodiilus E,j, 42000 MPa
Constaiit a 0.5

Figure 11.21 illustrates the 8 m high 12 m spaii top heading for the tailrace tuiinel. Tlie final tuiinel
lieight of 18 m was achieved by blasting two 5 m benches. The top heading was excavated by full-face drill
and blast atid, because of the excellent quality of the rod< mass and the tight control o11 blastiiig quality,
rnost of the top heading did not require any support.
Details of this project are to be found in Moretto et al (1993). Haininett and Hoelí (1981) have
described the desigii of the support systern for tlie 25 in span underground powerhouse in which a f i w
structurally coiitrolled wedges were identified and stabilised during excavation.

11.10.3 Average quality rock mass

The partially excavated powerliouse cavern in the Nathpa Jliakri Hydroelectric project i11 Hiinachel
Pradesh, India is illustrated in Fig. 14. Tlie rock is a jointed quartz mica schist, wliicli has been extensively
evaliiated by the Geological Suivey of India as described by Jalote et al [23]. An average GSI value of 65 was
chosen to estimate the rod< inass properties wliich were used for the cavern support design. Additiorial
support, instalied on the instnictions of the Engineers, was placed iii weaker rodc zones.
Tlie assumed rodi mass properties are as follows:

Intact roclc strength 0, 30 MPa Friction angle @' 40"

Hoelc-Brown constant m, 15.6 Coliesive strength C 2.0 MPa
Geological Strengtli Iiidex GSI 65 Roclc mass compressive strength o,,,, 8.2 MPa
Hoelc-Brown constant "i 4.5 Roclc rnass tensile strength otod -0.14 MFa
Hoek-Browii constaiit i 0.02 Deformation modulus E,,, 13000 MPa
Constant a 0.5

Two and three diinensional stress analyses of the nine stages used to excavate the cavern were carried
out to determine the extent of potential rodc mass failure and to provide guidance in tlie design of the
support systein. An isometric view of one of the three dimensional models is given iii Figure 11.23.
The support for the powerhouse cavem coiisists ofrockbolts and inesh reinforced shotcrete. Alternat-
ing 6 and 8 in long 32 inm diameter bolts o n 1 x 1 m and 1.5 x 1.5 m centres are used in the arch. Alternatiiip
9 and 7.5 in long 32 inin diameter bolts are used in the upper and lower sidewalls with alternating 9 and 11
in long 32 mm rocldiiolts in the ceiitre of tlie sidewalls, all a t a grid spacing of 1.5 m. Shotcrete coiisists oftwo
 Rock mas.

50 inin thick layers of piain shotcrete with an interbedded'layer ofweldmesh. The support provided by the
shotcrete was not iiicluded in the support design analysis, which relies upon the roclcboits to provide ali tlie
support required.
In thr headrace tunnel, some zones of sheared quartz mica scliist have been encountered arid thesi.
have resulted in large displacernents as illustrated in Figure 11.24. This is a cominon problein in liard roclc
tunnelling where the excavatioi~sequence and support systein have been desigiied for 'average' rode inass
conditioiis. Unless very iapid clianges in the length of blast rounds and the iiistalled support are inade when
an a b r ~ ~ chailge
pt to poor rock conditions occurs, for example when a fault is encountered, probleins with
controlliiig tunnel deforinatioii can arise.
Tlie only effective way lcnown to the authors for anticipating this type of problein is to keçp a probe
liole ahead o i the advancing cace at all tiines. Typically, a long probe hole is percussion drilled during a
inaitltenai~ceshift and the pei~etrationrate, return water flow and chippings are constantly inonitored duririg
drilling. Wherç significant probleins are indicated by this percussioi~drilling, one or m o diainoild-drilled
Iioles inay be required to investigate these problems in more detail. In some special cases, the use of a pilot
tunnel may be more effective i11 that it perinits the ground properties to be defiiied more accuratrly tliari is
possible with probe hole drilling. In addition, pilot tunnels allow pre-drainage and pre-tkinfbrcement ai. the
roclc ahead of the development ofthe f~illexcavation profile.

Figure 11.23: liametric view of the 3DEC" model of tlie iundergiound

powerhouie cavern and traniformer galieiy of tlie Natlipa ilrakri Hyrlro-
eiectrir Project. analyied by Dr 8 Daigupta'.

dAvaiable from ITASCA Consultng Graup i n i . Thieiher Square Eait.

708 South Third Street, Suite 310. Mnneapolii. Minnerota 55415. USA.
Fan 1 6 1 2 371 4717
'For,merly at the lnstitute of Roik Mecliano (Koiar), Kolar Gald feldi.
Karnataka, now with of Advanced Teclinoogy and Engineering Sewcer,
Delhi. ndia
 Rock Englneering COL

Figure 11.24: Large dspacementi

in the too heading of the headraie
tunnel of the Nathpa Jhakri
Hydroeectric Praject. These
diplacemeno are the result
of deterorating iack masi q u a t y
when tunneling thiough a
fault ione.

11.10.4 Poor quality rock mass at shallow depth

Kawadas et al(1996) Iiave described some of the geotechnical issues associated with the constructioii
of 18 I<inof tuiinels and the 21 underground stations of the Atliens Metro. Thrse excavatioils are all shallow
with typical depths to tuiinel crown of between 15 and 20 in. The principal problein is one of surfice
subsideiice rather tliaii fjilurc of the rode mass surrounding the openings.
The rode mass is locally knowii as Athenian schist which is a term erroiieously used to descrihe a
sequence of Upper Cretaceous flysch-type sediinents iiicluding thinly bedded clayey and calcareous saiid-
stones, siltstoiles (greywackes), slates, shales and liinestones. During the Eoceiie, the Athenian schist forina-
tioils were subjected to inteilse foldiiig and thrusting. Later extensive faulting caused extensional fracturiilg
and widespread weathering and alteration of the deposits.
The GSI values, estirnated from Bieniawski's 1976 RMR classiiication, modified as recommended by
Hoeb, Kaiser 2nd Bawden (11995) ranges from about 15 to about 45. The higher values correspond to tlie
intercalated layers of saudstones and limestones, which can be described as BLOCKY/DISTURBED aiid
POOR (Table 11.6). The completely decomposed schist can be described as DISINTEGRATED and VERY
 t properties

POOR and has GSI values ranging from 15 to 20. Rod< tnass properties for the completely decornposed
schist, using a GSI value of 20, are as foilows:

Intact rock strength 0 5-10 MPa Constant a 0.55

Hoelc-Brown constant m, 9.6 Friction angle $' 22.4
Geological Strength Index GSI 20 Cohesive strength C' 0.09-0.18 MPa
Hoel<-Brownconstant
Hoelc-Brown constant
m,
3
0.55
O
RocB mas's streiigth
Deformation modulus
o,, 0.27-0.53 MPa
E,,, 398-562 MPa

The Academia, Syntagina, Omonia and Olyinpion stations were constructed using tlie New Austrian
Tuiinelling Method twin side drift and central pillar inethod as illustrated in Figure 11.25. The more coiiveii-
tional top headiiig and bench inethod, illustrated in Figure 11.26, was used for the excavation of the Ainl-ielol<ipi
station. Tliese stations are all 16.5 rn wide and 12.7 in high. The appearance of the rod; rnass in oiie of tlie
Olympion station side drift excavations is illustrated in Figures 11.27 and 11.28.
Nuinerical analyses ofthe two excavation methods showed that the twin side drift method resulted in
slightly less rock mass failure in tlie crown of the excavatioii. However, the final surface displacements
induced by the two excavation inethods were practically identical.
Maximum vertical displacements ofthe surface above tlie ceiitre-line of the Oinoiiiastation arnounted
to 51 mm. O f this, 28 inm occurred during the excavation of the side drifts, 14 inin during the removal ofthe
central pillar and a fiirther 9 mm occurred as a time dependent settlement after completion of tlie excava-
tion. According to Kawadas et a1 (1996), this time dependent settlement is due to the dissipation of excess
pore water pressures whicli were built up during excavation. In tlie case of theOinonia station, the excava-
tion of recesses towards the eastern end of the station, after coinpletion of the station excavation, added a
furtlier 10 to 12 mm ofvertical surface displacement at this end of the statioii.

11.10.5 P o o r quality rock mass under high stress

The Yacambú Quibor tunnel in Venezuela is considered to be one oftlie rnost difficult tuniiels in
the world. This 26 I<m loiig water supply tuniiel through the Andes is being excavated iii saiidstones aiid
phyllites at depths of up to 1200 m below surface. The graphitic phyllite is a very poor quality rock aiid
gives rise to serious squeezing problems which, without adequate support, result in complete closure o[
the tunnel. A full-face tunnel-boring machine was completely destroyed in 1979 wheii trapped by squeez-
ing ground conditions.
At its worst, the graphitic phyllite has an unconfined compressive strength of abput 15 MPa ancl ~ l i e
estirnated GSI value is about 24. Typical rock mass properties are as follows:

Intact rocl; strength O,, 15 MPa Constant a 0.53

Hoel<-Brown constant m; 10 Frictioii aiigle @' 24"
Geological Strength Iiidex GSI 24 Cohesive strength C' 0.34 MPa
Hoelc-Brown constant m, 0.66 Rock mass streiigth o,88A 1 MPa
Moel<-Browncoiistaiit s O Deformation inodulus E,,, R70 MPa

Various support methods have been used on this tunnel and oiily one will be considered here. Tliis was
a trial section of tunnel, a t a depth of about 600 m, constructed in 1989. n i e support of the 5.5 m span tuniiel
was by ineans o f a complete ring of 5 in long, 32 mm diameter untensioned grouted dowels with a 200 rnm
tliicl<shell of reinforced sliotcrete. This support system proved to be very effective hut was later abandoried i11
favour of yielding steel sets (steel sets with sliding joints) because of construction schedule coiisiderat'ions.
Figure 11.25: Twn side drift and central pilar e x ~
Cavatlon method. Temporary suppart coni~stsof
double wlre meih reinforced 250 - 300 mm thck
shotirete ihells witii embedded lattice girdea or
HEB 1 6 0 i t e e iets at 0.75 - 1 m ipacing.
Figure 11.26: Top head~ngand bench method of excavatio~i.Tem-
poraiy ~upportconsisti of a 200 mm thick ihotrrete s h e with 4
and 6 m long untensianed grouted rockbaltiat 1 . 0 1.5 rn spacing.

Figure 11.27: Side drift ~n

the Athens Metro Olympion
station exravaton that wai
excavated by the method
ilustrated in Figure 1 1 2 5 .
The staton h a i a cover
deptii of approxmately 10
m over the crown.
 Chapter 11 Rock mas!, properties

Figure 11.28:
Appearance of the very
paoi quality Athenian
Sihist at the face af the
iide heading illustrated
~n Figure 11.27.

Examples ofthe results of a typical numerical stress'analysis ofthis trial section, carried out using tlie
prograin PHASE" are given iu Figures 11.28 and 11.29. Figure 11.28 shows the exteilt of iailure, witli aiid
without support, while Figure 11.29 shows the displaceineiits in tlie rock mass surrouriding the tunnel. Note
tliat the criteria used to judge the effectiveness of the support design are that the zone of failure surroundiiig
the tuiinel should lie within the envelope of the rockbolt support, the rocl~boltsshould not he stressed to
failure aiid the displaceinents sliould be ofreasonahle inagnimde and should be uniforrnly clistriblited arouiid
the tunnel. Ali ofthese objectives were achieved by the support systern describecl earlier.

11.10.6 Slope stability considerations

When dealing with slope stahility problems iii rock inasses, great care has to be talcen i11attetnpting to
appiy the Hoelc-Brown Failure criterioli, particularly for sinal1 steep slopes. As illustrated in Figure 11.30,
even rocl< masses that appear to be good candidates for the application of tlie criterioii cai1 suffer sliallow
stnicturally controlled failures uiider tlie very iow stress conditions which exist i11 such slopes.
As a general rule, wlien desigtiing slopes in r o k , tlie iuitial approach sliould always be to seardi for
potei~tialfailures controlled by adverse structural conditions. Tliese may take tlie forin o f p l a ~'iri , fAI'1 ures o11
outward dipping features, wedge failures on intersecting features, toppling failures o11inward dippiiig fiiilures
or coinplex failure inodes involving all of these processes. Only when the potential for structurally con-
trolled fail~ireshas been eliminated should consideration he given to treating the rock inass as a l i isotropic
material as required by the Hoek-Brown failure criterion.
 '\'\ ,
, h' Failure zone
,L---- with suppm . ... . ' . ' . . proiile wilh
Failuie zane with n o suppon

Figure 11.29: Rerultiaf a numerica analyiisof thefaiiuieaf

the roik masi iurrounding the Yacarnbu-Quibor tunne when
excavated in graphitic phyilte at a depth of about 600 rn be-
o w iuifaie.

Figure 11.31: Structuraiy

controiied faiiure in the face
af a steep bench n a heavily
jonted roik m a s .
 Rock mas8 properties

Figure 11.31 illustrates a case in which the base of a slope failure is defined hy an outwarcl dippiilg
fault which does not daylight at the toe of the slope. Circular failure througli the poor quality rode inass
overlying the fault allows failure of the toe of the slope. Analysis of this problein was carried out by assigning
the rod< rnass at the toe properties that had heen deterrnined by applicatioil of the Hoelc-Brown criteriori. A
search for the critical failure surface was carried out utilising the program XSTABL7 which allows coinplex
failure surfaces to be analysed and which includes facilities for the input of non-linear failure characteristics
as defined by equation 11.2.

Tension crack

Faiiure through weak rock

rnass

ailure on fault dipping a1 25"

Failure through weak rock rnass

u
100 m

Figure 11.32: Complen ilopefa!lure controlled by a n outward dipping baial fault and circular failure through the poor quality roik m a s oveily~ngtlie
toe of the slope.

iilable from Interarrive Sofiware Derigni. lnc., 953 N. Cleveland Street. Moicaw, Idaho, USA 83843. Fax + 1 208 885 6608
 Tunnels in weak rock

12.1 lntroduction
Tunnelling iii weak rock presents soine special challenges to the geotechnical engineer since
inisjudgeinents in the desigii of support systems can lead to very costly failures. In order to understand the
issues involved in tlie process of designing support for this type of tunnel it is necessary to exainiiie soine
very basic coiicepts of how a rock inass surrounding a tunnel deforins and how the support systems acts to
control this deforination. Once these basic concepts have been explored, exainples ofpractical support
desigiis for different conditions will be considered.

12.2 Deformation around an advancing tunnel

Figure 12.1 shows the results o f a three-dirnensional finite eleineiit analysis of the deformation of the
rod<inass surrouiiding a circular tiinnel advancing through a weak rocl<inass subjected to equal stresses iii ai1
directions. The plot shows displacement vectors iii the rock inass as well as the shapc of the deforined tunnel
profile. Figure 12.2 gives a graphical summary or the most importaiit features of this aiialysis.
Deforrnation o i t h e rock mass starts about one half a tuiinel diameter ahead of the advancing face 2nd
reaches its inaxiinuin value about one and oiie half diameters behind the face.At the face position about one
tliird of tlie total radial closure of tlie tunnel has already occurred and tlie tunnel face deforins itiwards as
illustrated in Figures 12.1 and 12.2. Whether or not these deformations induce stability probleins iii tlie
tunnel depeiids upon the ratio of roa<mass strength to the in situ stress leve], as will be deinotistrated iil the
following pages.
Note that it is assurned that deforination process described occurs iinrnediately upon excavatioli of
tlie fice. This is a reasonable approximation for inost tunnels in rock. Tlie effects of tiine dependent defor-
inations upoii the perforinance of the tuiinel and tlie design of the support systeiu will be not be discussed
in this cliapter.

12.3 Tunnel deformation analysis

111 order to explore the concepts of rock support interaction in a forin whidi can readily he under-
stood, a very siinple analytical model will be utilised. This model involves a circular tunnel subjected to a
hydrostatic stress field in which the horizonral and vertical stresses are equal.
In this analysis it is assurned that the surrounding rock heavily jointed mass behaves as a l i elastic-
perfectly plastic material in which failnre involving slip along intenecting discontinuities is assumed to occur
witli zero plastic voluine change (Duiicaii Fama, 1993). Support is modelled as an equivalent intern'il pres-
sure aiid, although this is an idealised inodel, it provides useful insights on how support operates.
 unnels i n vumeak rock

'Deformed
profile

Figure 12.1: Vertical secton throuyh a three~dmeniianalf n t e element rnodel of the faiiure and deformation of the rark mass sunoundng lhe face
af an advaniing circular tunnel. The plot showi dispaiernent vectors as w e l a i the ihape of the deformed tunnel profile.

inward deformation of -
tunnel face

Direction of
tunnel advance [Radial displacement staris about one half a
tunnel diameter ahead o1 the advancing lace

figure 12.2: Pattern of defarmation in the rock m a s surrounding an advancing tunnel

 12.3.1 Definition o f failure criterion
It is assuined that the onset ofplastic failure, for different values of the effective confining stress 0 3 ,
is defiried hy the Mohr-Couloinb criterion and expressed as:

+ ko3
01 = O<># (12.1)
is defined by:
The uniaxial compressive strength of the rock mass o',,,

and the slope h of tlie o;versus 0;line as:

wliere o;is the axial stress at which failure occurs

o; is the confining stress
c' is the cohesive strength and
+' is tlie angle of friction of the rock mass
In order to estimate the cohesive strength L' and the friction angle 4'for an actual rod< inass, tlie
Hoek-Brown criterion (Hoek and Brown 1997) can he utilised. Having estimated the parameters for failure
criterion, values for c'and c)' can be calculated as described in Chapter 11.

12.3.2 Analysis of tunnel behaviour

Assume that a circular tunnel of radius r0 is subjected to hydrostatic stressesp" and a uniform iiiternal
support pressurep, as illustrated in Figure 12.3. Failure ofthe rock mass surrounding the tunnel occurs wheii tlie
internal pressure provided by the tunnel lining is less than a critical support pressurep<,,which is defined by:

If the internal support pressurep, is greater than the critica1 support pressurep<,,n o failure occurs, the
heliaviour of the rodc mass surrounding the tunnel is elastic and the inward radial elastic displacement of the
tunnel wall is given by:

where E,,,is the Young's modulus or deformation modulus and

V is the Poissoii's ratio.
When the internal support pressurepjis less than the critical support pressurep,,, failure occurs and the radi~is
5 of the plastic zone around the tunnel is given by:
 unnels In w'eak rock

- -

Figure 12.3: Plaitii zone surrounding a circular tunnel

For plastic failure, the total inward radial displacement of the walls of the tunnel is:

A spreadsheet for the determinatiori of the strength and deforination characteristics oftiie rocic inass
and the behaviour of the rock mass surrounding thr tunnel is given in Figure 12.4.

2.4 Dimensionless plots of tunnel deformation

A useful means of s t u d ~ i n ggeneral behavioural trends is to create diinensionless plots froin the
results of parametric studies. Two such diinensionless plots are iresented in Figures 12.5 aiid 12.6. Thesr
plots were constmcted froin the results o f a Monte Carlo analysis in which the input parameters for rock
mass strengtli and tunnel deformation were varied at random in 2000 iterations'. It is rernarl<abIe that, in
spite of tbe very wide range of conditions included in these analyses, the results follow a very similar trend
and that it is possible to fit curves which give a very good indication of the average treiid.
Figure 12.5 gives a plot of the ratio to plastic zone radius to tunnel radius versus tlie ratio ofrod<inass
strength to in situ stress. This plot shows that the plastic zone size increases very rapidly once tlie rodi inass
streiigth falls below 20% of the rodi mass strength. Practical experience suggests that, once this rapid growth
stage is reached it becomes very difficult to control the stahility of the tunnel.

I Uring the program a R S K in conjunrton with a Miiioioft Excel ipread~heetfor estimating roik mass strength and tunnel behavour (equationi 4 ta
7). Uniform distributioni were iampled for the foilowing input paiameters, the two figures in bracketi define the minimum and rnaximum values
uied: intact rock itiength o,, (1.30 MPa). Hoek~Brownionitant m, (5.12). Geoogical Strength index G S (10.35). n iitu i t r e a (2. 20 MPa), Tunnel
radiui (2, 8 m)
 Rock Engineering Cc

Input: sigci= 10 MPa mi= 10 GSI = 25

mu = 0.30 ro = 3.0 m po = 2.0 Mpa
pi = 0.0 MPa pilpo = 0.00

Output: mb = 0.69 s = 0.0000 a = 0.525

k= 2.44 phi = 24.72 degrees coh = 0.22 MPa
sigcm = 0.69 MPa E = 749.9 MPa pcr = 0.96 MPa
rp = 6.43 m ui = 0.0306 m ui= 30.5957 mm

Isigcmlpo 0.3468 I rplro = 2.14 I uilro = 0.0102

Calculation:
Sums
sig3 1E-10 0.36 0.71 1.1 1.43 1.79 2.14 2.50 10.00
sigl 0.00 1.78 2.77 3.61 4.38 5.1 1 5.80 6.46 29.92
sig3sigl 0.00 0.64 1.98 3.87 6.26 9.12 12.43 16.16 50
sig3sq 0.00 0.13 0.51 1.15 2.04 3.19 4.59 6.25 18

Cell formulae:
mb = mi'EXP((GS1-100)/28)
s = lF(GS1>25,EXP((GSI-100)/9),0)
a = IF(GS1>25,0.5,0.65-GSII200)
sig3 = Start at 1E-10 (to avoid zero errors) and increment in 7 steps of sigcil28 to 0.25'sigci
sigl = sig3+sigci*(((mb*sig3)/sigci)+ç)"a
-
k = (sumsig3sigl (sumsig3*sumsigl)/8)I(çumsig3sq-(sumsig3A2)/8)
phi = ASIN((k-l)l(k+l))'l8OiPI()
coh = (sigcm'(l-SIN(phi'PI()I180)))/(2*COS(phi*PI()l180))
sigcm = sumsigll8 - k'sumsig318
E = lF(sigci>100,1000*1OA((GSI-10)/40),SQRT(sigci/100)'1000*1O"((GSI-10)/40))
pcr = (2'po-sigcm)/(k+l)
rp = IF(pi<pcr,ro'(2'(po'(k-l)+sigcm)l((l+k)'((k-1 )*pi+sigcm)))A(lI(k-l)),ro)
ui = IF(rp>ro,ro'((l+mu)lE)"(2*(l-mu)*(po-pcr)~((rplro)~2)-(l-2*mu)*(po-pi)),ro~(l+mu)*(po-pi)lE)

Figure 12.4: Spreadsheet for the calculation of rock mass chaiacteriitici and the behaviour af the rock mais iurrounding a circular tunnel in a
hydroitatic i t r e a field.

Figiire 12.6 is a plot of the ratio of tunnel deformation to tunnel radius against the ratio of rodc mass
strength to in situ stress. Once the rock inass strength falls below 20% ofthe in situ stress Ievel, deforinatioiis
iilcrease substantially and, unless these deformations are controlled, collapse of the tunnel is likely to occur.
Figures 12.5 and 12.6 are for the condition of zero support pressure @;=O). Similar analyses were niii
for a range of support pressure versus in situ stress ratios @jpJ and a statistical curve fitting process was iised
to determine the best fit curves for the generated data for eachp,lpo value. These curves are given in Figures
12.7 and 12.8.
 eak rock

00 O.? 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.8 1.0 0.0 O l 0.2 03 0.4 0.5 0.8 0.7 0.8 0.9 I 0

ROCX mass slrength l i n ritu stress R o c k m a 5 9 strenglh i i n situ stiess

Figure 12.5: Relat~onihpbetween size of plaitic 2one and ratio of Figure 12.6: Tunnel deformation versui ratio of rock mais itrengtli
iork mais itrength to in i i t u streir. to in situ itreii.

The series of curves shown in Figures 12.7 and 12.8 are defined by the equations:

where rp = Plastic zone radius

u, = Tunnel sidewall deformation
r"= Original tunnel radius in metres
p,= Interna1 support pressure
po = In situ stress = depth below surface x unit weight of rock mass
o,,,,= Rocic mass strength = 2C cos @'/(1-sin$1
ALI alternative plot of the data used to constmct Figure 12.8 is given i11 Figure 12.9. For readers who have
studied roclc support interaction analyses this plot will be familiar and it gives a good indication of t11e
influente ofsupport pressures o n tunnel deformation.

12.5 Estimates of support capacity

Hoelc and Brown (l980a) and Brady and Brown (1985) have published equations which cai1 be used
to calculate the capacity of iriechanically anchored rodcbolts, shotcrete or concrete linings or steel sets for a
circular tunnel. No useful purpose would be sewed by reproducing these equations here but they have been
used to estimate the values plotted in Figure 12.10. This plot gives maximuin support pressures @,,,) and
maximum elastic displacements (u,,,,)for different support systems installed in circular tunnels of different
 Rock Engir

diameters. Note that, in all cases, the support is assumed to act over the entire surface o f t h e tunnel walls. In
other words, tlie shotcrete and concrete linings are closed rings; the steel sets are coinplete circles; and the
mechanically aiichored rocl<bolts are iiistalled in a regular pattern that coinpletely surrounds the tunnel.
Because this inodel assumes perfect syrninetry under hydrostatic loading of circular hlilneis, no bending
inoments are induced in the support. 111reality, there will always be some asyminetric loading, particularly for steel
sets and shotcrete placed on rough rodi surfices. Hence, induced bending will result in support capacities that are
lower than those given iii Figure 12.10. Furthermore, the effect ofnot closing the support ring, as is frequently tlie
case, leads to a drastic reduction in the capacity and stiffness of steel sets aiid concrete or shotcrete liniiigs.

12.6 Practical example

In order to illustrate the application of the concepts presented in this chapter, the following practical
example is considered.
A 4 m span drainage tunnel is to be driven in the rod<mass behind tlie slope of an open pit mine. The
tunnel is at a depth ofapproxiinately 150 m below surface and the general rocb is a granodiorite of fair
quality. A zone of heavily altered porphyry associated with a fault has to be crossed by the tunnel aiid the
properties of this zone, which has been exposed in the open pit, are I<nowii to be very poor. Mine manage-
ment has requested an iiiitial estiinate of the behaviour of the tunnel and of the probahle support require-
inents. The tunnel is to link up with an old mine drainage tunnel that was constructed several decades ago.

12.6.1 Estimate of rock mass properties

Figures 12.6 and 12.7 show that a crude estimate of the behaviour of the tunnel can be made if the
ratio of rocl< mass strength to in situ stress is available. For the purpose of this analysis the in situ stress is
estimated from the depth below surface and the unit weight of the rodi. For a deptli of 150 m and a uiiit
weight of 0.027 MN/in3, tlie vertical in situ stress is approximately 4 MPa. The fault material is considered
incapable of sustaiiiingliigh differential stress levels and it is assumed that the horizontal and vertical stresses
are equal withiii tlie fault zone.
It has heen found that the ratio of tlie uniaxial compressive strengths in the field and the lahoratoiy
(q,,,ía,Jcai1 he estiinated froin the following equation:

where GSIis the Geological Strength Index proposed by Hoek and Brown (1997).
The relationship expressed in equation 12.10 is illustrated graphically in Figure 12.11.
111 the case of the granodiorite, the laboratory uniaxial compressive strength is approxirnately 100
MPa. However for the fault material, specimens can easily be brolcen by hand as shown in Figure 12.12. The
laboratory uniaxial compressive strength of this material is estimated at approxiinately 10 MPa.
Based upon observations in the open pit mine slopes, the granodiorite is estiinated to have a GSI
value of approximately 55. The fault zone has heen assigned GSI = 15. The rock inass descriptions that forrn
the basis of these estiinates are illustrated i11 T'g
.i ure 12.13.
For the granodiorite, substitution of GSI = 55 and q,= 100 MPa into equation 12.10 gives ai1
approxiinate value for the uniaxial coinpressive strength of the rod<mass as 18 MPa. For an i11 situ stress of
4 MPa, this gives a ratio of roclc inass strength to in situ stress in excess of4. Figures 12.5 and 112.6 show that
the size of the plastic zone and also the induced deformations will be negligibly small for is ratio. This
conclusion is confirmed by tlie appearance of the old drainage tunnel that has stood for several decades
without any form of support.
 ~ e a krock

--
0.0 O1 02 0.3 01 05 0.6 0.7 0.8
Rock mas$ strength l i n situ stress

Figure 12.7: Ratio of plaiticzone ta tunne radiui uersus the rato of Figure 12.8: Ratioof tunnei defarrnation to tunnel raduiveriui tiie ratia
rock mais itrength to n situ stress for dfferent support preisures. of rock masi strength ta in situ streiifor different support preiiurei

Figure 12.9: Reationshp between supparr preriure and tunnel de- Figure 12.10: Estimateiofiupport iapactyfartunnelsof dfferent sizei.
foirnaton for different rattoi of rock rnass strength ta in i i t u stresi.
 Rock Englneerfng COLirse notes b

Gealogical Strength Index - GSI

Figure 12.11: Approxmate relationihip between in iitu and laboratory uniaxia compreriive itrengthi and the Geological Strength Index

Figure 12.12: Heavily

altered porphyry can easily
be broken by hand.
 Tunnels in iveak rock

Figure 12.13: Table for eitimating GSI "alue (Hoek and Brown 1997) showing ranges af vaues for gianodioire and f a u t ione.
 Rock Engirieering Coc

B a e d upon this evaluation, n o permanent support should be required for the tunnel in the iãir
quality granodiorite. Temporary support in the forrn of spot bolts and shotcrete inay be required for safety
where the rodi inass is heavily jointed.
Iii tlie case of the altered porphyry and fault material, substitution ofGSI = 15 and a,,= 10 MPa into
equation 12.10 gives a rock mass strength of approxirnately 0.4. Tliis, in turn, gives a ratio of rock inass
strength to in situ stress of-0.1.
Froiii Figure 12.5, the radiiis of plastic zone for a 2 in radius tunnel in this material is approxiinately
9.5 in without support. The tunnel deformation in approximately 0.4 in, giving a closure of 0.8 in. Tliese
conditioiis are clearly unacceptable and substantial support is required in order to prevent convergence and
possible collapse of tliis section. Siiice this is a drainage tunnel, the final size is iiot a inajor issue aiid a
significant ainount of closure can be tolerated. However, experieiice suggests tliat the ratio of tuniiel defor-
inatioii to tunnel radius sliould be Iiept below about 0.02 in order to avoid serious instability problems.
Figure 12.9 indicates that a ratio of support pressure to in situ stress of approxiinately 0.35 is required to
restrain tlie deformatioil to this leve1 for a iock mass witli a ratio of rocli inass strength to ir1 situ stress of 0.1.
This translates into a required support pressure of 1.4 MPa.
Because of tlie very poor quality of the rocli mass and the presence of significant amount of clay, tlie
use of roclibolts or cables is not appropriate because of the difficulty of achieving adequate anchorage.
Consequeiitly, support has to be in the forin of shotcrete or coiicrete lining or closely spaced steel sets as
suggested by Figure 12.10. Ohviously, placeinent of a full concrete lining during tiinnel driving is iiot prac-
tical and Iience the reinaining choice for support is the use of steel sets.
The problein of using heavy steel sets in a smali tunnel is that bending of the sets is difficult. A
practical rule of thumb is that an H or I section can only be bent to a radius of about 14 times the deptli of
the sectioii. Tliis problem is illustrated in Figure 12.14 that shows a heavy H section set beiiig bent. In spite
of the presence of teinporary stiffeners, there is significant buckling of the inside flange of the set and a lot of
additioiial work is required before the set can be sent underg~ound.
The practical solution adopted in the actual case upoii which this example is based was to use sliding
joint top hat sectioii sets. Tliese sets, as delivered to site, are shown in Figure 12.15 whidi illustrates how tlie
sectioiis fit into each other. Tlie asseinbly of these sets to forma slidiiig joint is illustrated iii Figure 12.16 and
the iiistallatioii of the sets in the tunnel is illustrated in Figure 12.17.
The sets are installed immediately behind the advanciiig case which, iii a rocli inass such as that
coiisidered here, is usually excavated by hand . The clamps holding the joints are tiglitened to control tlie
fiictioiial force in the joints which slide progressively as the face is advaiiced and the rocli load is applied to
the sets.
The use ofslidingjoints in steel sets allows vety inuch lighter section sets to be used than would be the
case for sets with rigid joints. These sets provide iminediate protection for the worliers behind the face but
they permit sigiiificailt deforination of the tuiiiiel to take place as the face is advaiiced. In inost cases, a
positive stop is welded onto the sets so that, after a pre-determined ainount of deformatioii has occurred, the
joint loc1i.s and the set becoines rigid. A trial and error process has to be used to find the amount of deforina-
tion that cai1 be perinitted before the set locl<s.Too little deformation will result in obvious bucliling of the
set while too inuch deforination will result in loosening of the surrouiiding rock inass.
111 the case of tlie tuiinel illustrated in Figure 12.17, lagging behind the sets coiisists of woodeii
poles of about 100 inin diaineter. A variety ofinaterials can be used for laggiiig but wood, iii the forin of
planks or poles, is still the most cominon. In addition to the lagging, a timber inat has beeii propped
against the face to improve the stahility of the face. This is an important practical precaution since insta-
bility of the tunnel face can result in progressive ravelling ahead of the steel sets and, in some cases,
collapse of the tunnel. \
 hapter 12. Tunnels i n iweak rock

. .
r cgt.r~ 12
.., - 1.1
r r,,, , , I
,
. .
.
c
- ,,
.
,
.
I ,i c-
I
,-.r-- .-
~ ~ J
r:, .,
.,,v-<.-
1I) 15
P I
. i i ' ,. 8 T

I I , , .
~~ ~~
. I
~~~
I= I, - I , .' ' , ' . J ,. l
itraghten the flanges after tlieie itiffeneri hõve been rernoved.

Figure 12.16: Asiernbiy of a frictionjanr n a top hiit iection i t e e ser

 i.erlng Cou rse notes b!f Evert Hoel

Figure 12.17: Installatian oi

i i d i n g jaint tap hat iertion
i t e e s e t ~immediately
hehnd the face of a tunnei
being advanied tlirough
veiy poor quaiity rock.

As ai1 alternative to supporting the face as illustrated in Figure 12.17, forepoles or spiles can be used
to create an uinbrella of reinforced rock ahead of the advancing face. Figures 12.18 and 12.19 illustrate the
general principies of tlie technique. 111 tlie example illustrated, forepoling is heing used to advance a 10 in
span, 5 in Iiigh tunnel top heading through a closely jointed rod< mass.
Forepoles are usually solid steel reinforcing bars that are grouted into holes drilled ahead of the face.
In some cases, steel pipes are used and the grout is injected through the pipes. In oder for the forepoles to
work effectively tlie rock mass should behave in a fiictional inanner so that arches or bridges can forin
beiween i~idividualforepoles. The techilique is not very effective in fault gouge material containing a sigii~ifcant
proportion of clay.
 Ihapter 12 T u n n e l s i n weak rock

optional 6breglasr dawels

grouted inta face to improve
slability

Figure 12.18: Use of forepoles to advance a 10 rn ipan 5 m high tunnel Figure 12.19: Typica 6 rn long forepole instalation iequence for a 10
tap headng through cosely lotnted rock. m ipan 5 m high top heading.

Figure 12.20: Spiing in veiy

p001 quality ilay~richfault
ione material

i
! Where the rockinass is suitable for tlie application offorepoles, consideration can be giveli to stabilising
the face by means of fibreglass dowels grouted into the face as illustrated ir1 Figure 12.19.
Spiles generally consist of steel bars that are driven iilto the rode inass hetween steel sets as illustrated
in Figure 12.20 which sliows a situation wliere the spiles were driveil into the very poor quality fault material
by sledge hammer. This is a veiy crude technique and the anthor prefers tlié use of forepoles wherever
possible. However, in veiy poor quality clay-ri& fault gouge, forepoles may may work and spiling is one of
the few alternatives remaining.
 Large Powerhouse caverns in weak rock

13.1 lntroduction
In the context of this discussion, weak rodc is rock that will fail when subjected to the stress levels
induced by the excavation of large underground caverns at depths of 100 to 300 m helow surface. Sedimen-
tary rocks such as hedded sandstones, shales, siltstones and mudstones are amoiigst the roclis which fall into
this category. Tunnels and caverns associated with underground hydroelectric projects are soinetiines exca-
vated in rodi masses of this type.
The design concepts discussed are based upon experiente drawn from a nuinber of hydioelectric
projects. The clioice of the size 2nd shape of power and transformer caverns, the locatioii of these caveriis
relative to eacli other and to the ground surface, the influence ofjoints and beddiiig planes o n the stability
of the excavaticins 2nd the choice of the m'ost appropriate support systems are issues which are cominon to
all of these schemes. In applying these principies to a uew scheme, the reader should be aware that each
scherne will Iiave its own set ofrockrnass properties, in situ stress conditions and design constraints imposed
by inechanical, electrical and hydraulic considerations. Consequently, the general design concepts outliued
here have to be inodified to suit each scheme.

13.2 Rock mass strength

Iii most strongigneous 2nd metainorphic rocks such as dolerites, basalts, granites, gneisses andquartz-
ites, the stability of large caveriis at depths of less than 500 in below surface depends almost entirely upoli
structurally controlled wedges and bloclis that are released by the creation ofthe excavations. In these condi-
tions, the excavation profile can be controlled with good blasting procedures, and the rock mass atid any
support placed in it are subjected to relatively small displaceinents. Tlie strength of the rock itselfplays a
ininor role i11 tlie hehaviour of the rode inass which is controlled by intersectiiig joints, scliistosity, bedding
planes, shear zones and faults. Away from the major structures, the roclc may be capahle ofstanding unsup-
ported during excavation for considerable periods of time, and the excavation and support of tuniiel inter-
sectioiis with the cavern poses no particular problein. The evaluation of the stability of these excavations is
carried out by means of limit equilibrium analyses. The factor of safety is calculated by comparing the shear
strength of the discontinuities, which bound the potentially unstable hloclcs and wedges, with the driving
forces due to the gravitational weight of these blodis and wedges as discussed in Chapter 5.
In the case of wealier sedimentaty rocls, the strength of the rock material is geilerally Lower and the
roclc iiiass is frequently more heavily jointed and more deeply weathered.The control of overbreak during
excavation hy drilling and blasting will be more difficult, and the rodc inass and tlie supports placed iii it inay
be subjected to deforinations of up to 50 or 100 inm at the surface of the excavation. The sequence of
excavation and support of tunnel iiitenections with the cavern will have to he controlled aiid will require
careful eugineering design. For these caverns, support will usually be required immediately after excavatioii.
In designing these supports, failure of the rodi material and sliding and rotation of individual hlodcs of rock
 )use caverns in weak rc

within the rock inass have to be coiisidered in addition to failure along structural features such as bedding
planes, shear zones'and faults.
The Hoelz and Brown failure criterion, described in Chapter 11, provides a basis for estiinating the
strength of rodz inassrs of the type uuder consideration here. For all the exarnples discussed, it lias been
assumed that the rodz mass is a fair to poor quality siltstones and that its propertics ;ire defined by:

Uniaxial compressive strength of intact roclz o' = 100 MPa

Constantm; for intact siitstone .m;= 10
Geological Strength Index GSI=48
Rock mass constaiit m,= 1.56
Rock mass constant s = 0.003
Deformatioti inodulus E = 8900 MPa
Poisson's ratio v=0.3
Friction angle @ = 31"
Cohesive strength c'= 4 MPa

Figure 13.1 gives plots of the relationship between the maximum and ininimum principal stresses and tlie
shear and normal stresses at failure defined by the Hoek-Brown failure criterion.

13.3 In situ stress conditions

A study of the results of in situ stress measurements from around the world suggests that the horizon-
tal stress is generally significantly greater than the vertical stress at depths below surface of less thaii 1000 in
(Brown and Hoelz (1978), Sheory (1994)). The vertical stress is normally assumed to be equal to tlie product
of the unit weight of the rock mass and the depth below surface aiid measured in situ stresses are usually in
reasonable agreeineiit with this assuinption.The ratio of average horizontal stress to vertical stress can be as
higli as 3 and values of 1.5 or 2 are frequently assumed for preliminary analyses.
It is always advisable to ineasure the in situ stresses in the viciiiity of major uilderground caverns as
early in the project feasibilitystudy as possible. Dnring early site investigations, when n o underground access
is available, the most coinmonly used method for measuring in situ stresses is hydrofracturing (Haiinson
(1978)). The hydraulic pressure required to produce fresh crad« and subsequently close aiid re-opeii thein is
used to estiinate in situ stress levels. Once underground iccess is available, overcoring techniques can be
used, as discussed in Chapter 10.

13.3.1 Stresses around underground caverns near the toes of slopes

Siiice hydroelectric projects are frequently located in mouiitainous areas, the influence of surfice
topography upon the in situ stress field has to be taken into account in deciding upon the exact location of
the underground powerhouse.
Figure 13.2 sliows the maximum and minimum principal stresses in a gravitationally loaded slope
with a far field horizontal to vertical in situ stress ratio o f 3 : 1.The in situ stresses, particularly the iniiiimum
principal stress s,, are significantly altered i11 the vicinity of the slope face as compared with the f i r field
stresses. These local changes in the in situ stress field influence tlie stresses induced in the rock inass sur-
rounding an underground cavern located near the slope toe'.

'There analysei were carried out using the elastic baundary elernent prograrn EXAMINE" deudoped in the Department of Civil Enginernng at lhe
Unversity of Toronto. Tlhe piogram is available from Rociiience lnc., 31 Balsam Avenue, Toronto, Ontario, Canada M4E 385. Fax 1 416 693 0908,
Phane 1 416 698 8217, Ernail: ioftware~rocscience.corn,Internet http:llwwwracsciencecom.
 O 5 10 15 20 25 30
Effecüve normal stress a, MPa

O 5
Minimum principal
stress õ3 MPo

Figure 13.1: Pot af reatiomhipi between maximum and minimum principal itreiiei and normal and shear itreses for f a u r e of a f a r ta poar qually
5titoneThe properties of the iitstone are defined on the pievious page.

Figure 13.3 illustrates tlie results of a boundary elemeiit analysis in which an uiiderground power-
liouse cavern has been located at different distances froin the toe of the slope analysed in Figure 13.2.
Contours showing zones in which the tensile and shear strength of the rock mass have been exceeded are
plotted in this figure.Failure trajectories in these overstressed zones indicate the direction in which fdilure of
the roclc would propagate, assuining the rock inass to be hom,ogeneous.
In all cases, vertical tensioil cradcs would be generated at the crest of this particular slope and a minor
ainount of shear failure would occur near the toe. Such failures are common in slopes in weab rock inasses
and, once forined, are generally not a cause for concern since they are local in nature and result in stress relief
and the re-establishment of equilibrium. Tension cracks, running parallel to the crests of slopes in sedimeii-
tary rock masses are common and have been obsemed to remain stable for many years. Usually, they liave
had n o significant influence o n the overall behaviour of the slope.
The zones of overstress in the rock surrounding the cavem are significantly different in the three cases
illustrated. For the case in which the power cavern is closest to the slope toe, the extension of the overstressed
zone to the slope face could result in the generation of local slope instability and could also result in the
formation of a highly permeable zone between the slope face and the dowiistream wall of the cavern. In
addition, the extent and the asyinmetrical shape of the zone of overstressed rod< suggests that substantial
support in the forin of long grouted cables would be required to stabilise the rock mass surrounding the
cavern. The autlior considers this cavern to be too close to the toe of the slope and would recommeiid
moving it hrther into the rod< mass.
 use caverns In weak ro

7 Maxirnurn principal stress

MPa

o 1 O0 200
I I
Scale - rn

Figure 13.2: Distribution af maximum and minimum principal stresses in a gravitationaily ioaded rlope with far field in i t u rtreaei defined by a ratio
of horizontal to vertical streii of 3:l
 Rock Engi

, Tension cracks at top of slope

Potentiol tensile foilure

shear failure

Figure 13.3: Zonei of overitreis and faiure trajectoriei for a siltstone raik masr surrounding an underground cauern at different ditances from the
toe of the rlope.
 se caverns IIn weak roc

The zonès of overstress in the case of the cavern located furtliest from the slope toe are mucli sinaller
thaii for tlie other two cases and the rock inass surrounding tliis cavern could probably be stabilised witli a
relatively modest airay of grouted cables.
111considering the three options illustrated in Figure 13.3, it must he remembered that geotechnical
factors are not the only items which have to be considered in decidiiig upon the cavern location. Another
itnportant Factor is the length of tlie tailrace tunnels. Hydraulic engineers usually want to keep these as short as
possible in order to avoid the need for a downstream surge shaft to compensate for pressure fluctuations. Once
the tailrace tunnel length approadies 100 m tbe need-for a downstream surge shaft has to be considered.

13.3.2 Determination of steel lining length for pressure tunnels

Although not related to cavern design, the determination of the length of steel linings for pressure
tunnels is an in situ stress related prohlein that deserves special consideration.The very higli cost ofpressure
tunnel steel linings can sometimes impose significant constraints upon the feasibility of hydroelectric projects.
In inost wealc rock inasses, concrete linings are required in all hydraulic conduits in order to provide
protection against erosion of weak seatns and to improve the hydraulic characteristics of the tuniiels. Assum-
ing that the tunnels have been correctly supported hy means of grouted rockholts and cables diiring coil-
struction, the concrete linings can he of minimum thidmess and lightly reinforced.It is assumed that these
concrete linings will crack under operating pressure and that they will be slightly leaky. Provided that the
craclcs do not propagate a significant distance into the rock tnass and cause hydraulic jacking, the slight
ainount of leakage is not a prohlein. However, conditions which give rise to hydraulic jadcing have to be
avoided and this is usually achieved by the provision of an internal steel lining in the tunnel.
The criterion for deciding when a tunnel should be steel lined is when the minimum principal stress
in the roclc mass falls below the rnaximum dynamic water pressure in the tunnel. This is a function of the
maximum static head ofwater in the tunnel, the operation ofthe gates and the characteristics of the turbines.
An allowance of 20% over the maximuin static head is usually considered adequate for a pressure tunnel
associated with the operation of a Pelton wheel since this does not induce large pressure fluctiiations. In the
case of a Fraiicis turbine, larger pressure fluctuations can be induced and an allowance of 30% ahove the
maximum static head is normally used.
Figure 13.4 illustrates the case of a proposed pressure tunnel arrangement for a surface powerhouse at
the toe of a s1ope.A headrace tunnel at elevation 680 m feeds water into a vertical shaft and then into a
horizontal pressure tunnel at elevation 550 m.The maximum reservoir pool leve1 is 780 m and this results in
an internal static pressure in the upper headrace tunnel of(780 - 680) x 0.01 = 1.0 MPa, where the unit weight
ofwater is 0.01 MN/tn3.Assuming that an allowance óf 30% over this static pressure is required for a Francis
tiirhine system, a steel lining will he required for any section of the upper headrace tunnel where the inini-
mum principal stress in the rock inass is less than 1.3 x 1.0 = 1.3 MPa. As shown in Figure 13.4, these
conditions require about 80 m ofsteel lining where the headrace tunnel passes beiieath the valley upstreain
of the surge tanl<.An option which should be considered in this case would be to increase the grade of thr
headrace tunnel so that the cover depth in the viciniiy of the valley is increased.A relatively modest increase
of 15 to 20 m iii cover depth would probably eliminate the need for the steel lining resulting in significant
cost savings and the removal of a construction impediment.
For the lower elevation tunnel, the internal static pressure in the tutinel is given by (780 - 550) x 0.01
= 2.3 MPa. Allowing an additional 30°/n for dynamic pressure, the minimutn principal stress at which the
steel lining should commence is 3 MPa.As shown in Figure 13.4, the lower pressure tunnel should be liiied
froin x co-ordinate 500 m to the surface powerhouse.
The ininimum principal stress contours plotted in Figure 13.4 were deteriniiied by ineans o f a bound-
a r y element analysis assuming a gravitational loading of the slope and a far field horizontal to vertical in situ
 Rock Engi

stress ratio of 1.5:l. This analysis assumes that the rock mass is homogeneous and isotropic and that the
minimum principal stress lies in the plane of the drawing. It is essential that an accurate topograpliic inap of
the surrounding area be checl<edto ensure that there are no valleys or iow points in a direction normal to the
plane of the drawing which could give rise to stress relief in that direction.If in doubt, similar stress analyses
to that illustrated in Figure 13.4 should be carried out for other sections perpendicular to the tunnel a i s to
dieclí whether the minimum principal stresses are low enough to cause probleins.
Since these stress analyses assume ideal rock conditions and do not take into account possible leakage
paths along faults, shear zones or other geological discontinuities, I strongly recommend that the findiiigs of
the type ofstress analysis illustrated in Figure 13.4 be confirmed by hydraulic acceptance tests.These involve
drilling boreholes from the surface to the points at which the steel linings are to be terminated, packing off
the lower 1 to 3 m of the borehole and then suhjecting the pacl<ed-off sections to hydraulic pressures ili-
creased increinentally up to the inaximum dynamic pressure anticipated in the tunnel. The pump pressures
and flow rates should be carefully monitored to determine whether any excessive joint opening or hydraulic
fiacturing occurs during the pressure tests. The test pressure should he maintained for at least an liour to
estahlish that there is n o significant lealíage through the rock mass from the points at which the steel linings
are to be terminated. Only when these tests have confirmed the theoretical calculations can the steel lining
lengths he established with confidente. I ain aware of severa1 pressure t~tnnelfailures where lining Iength
calculations were carried out hut n o pressure acceptance tests were performed. Had such tests been done,
they would have revealed anomalies or deviations from the conditions assumed in the calculatioils and these
deficiencies could have been allowed for in deciding upon the steel lining lengths.

13.4 Pillar size between excavations

In some cases, more than one cavern is required in an underground hydroelectric project. For ex-
ample, the transformers may be placed in a smaller cavern parallel to the power cavern. This has the advan-
tage of reducing the size of the inain cavern and of isolating the transformers in case of fire. When this
arrangement is used, there is frequently a demaiid from the electrical engineers to place the two caverns as
close together as possihle in order to reduce the length and hence the cost of the husbars that liiil<the
generators to the transformers. However, placing tlie two caverns close together may give rise to unfavourable
stress conditions in the pillar between the caverns.

Figure 13.4: Contovis of minimvm prncipal stress(MPa) in a gravitationally loaded slope with a far fieid in situ stress field defned by a ratio of
horizontal ta vertical stress of 1.5 : 1 . Lengths of iteei iining required in the prenure tunnel a o rhown.
 s e caverns in weak roc

Figure 13.5 illustrates the results of a series of analyses in which the distances between two parallel
caverns were varied.These analyses assume that the rock mass is a fair to poor quality siltstone, as defined
earlier in this chapter, and that the ratio of horizontal to vertical in situ stress is 1.5:l. The caverns are located
at a deptli of 280 in below surface.
The contours plotted in Figure 13.5 define the ratios of the roclí mass strength to the inaximuin
principal stress induced in the rodí mass surrounding the caverns.Zones defined by contours with strength/
stress ratios of less than 1 are zones of potential rockmass failure.Because of the complex process of stress re-
distributioii associated with progressive failure of the rodc mass surrounding the excavations, these zones of
overstressed rodc d o not necessarily coincide with the actual extent ofrock fracture. The zones of overstress
do, however, give a reasonable basis for comparison and experience suggests that the designer should at-
tempt to Iíeep these zones as sinal1 as possible with a priority o n ininiinising the extent ofpotential zones of
tensile fracture'.
In the case of the upper plot in ~ i g u r e13.5,
' the distance between the two caverns is approxiinately
equal to one half of the height of the larger of the two caverns and the zoiie ofoverstress extends across the
entire pillar. Note that the central portion of the overstressed zone has strengtli/stress values of less than zero
and will be prone to tensile failure. This pillar wonld almost certainly suffer severe damage aiid would require
very suhstantial support in the form of tensioned grouted'cables spanning the width of the pillar.
The lower plot in Figure 13.5, witli the pillar width approximately equal to 1.5 times the lieiglit of the
larger of the two caverns, shows that tlie zones of overstress are of liinited extent and that the core of the
pillar has strength/stress ratios in excess of4.This means that the stress fields surrounding the two caverns are
alrnost independent of one another.The extent of the overstressed zones suggests that a relatively inodest
ainount of support would be required to stabilise the rock inass surrounding the caverns.
The plot in the iniddle of Figure 13.5 represents a situation that could he considered a reasonable
comproinise for many underground hydroelectric projects.The distance between the two caverns is approxi-
mately equal to the height of the larger of the two caverns and this is generally acceptahle iu terms of busbar
1ength.The stress fields surrounding the two caverns obviously interact to a certain extent but the zones of
overstress are not so large that major changes in the support pattern would be required.Tlie zone of potential
tensile failure between the two caverns has been eliminated in this layout.
For caverns in weak rock masses such as those considered in this chapter, I recominend tliat pillar
widths should not be less than the height of the larger of the two caverns, and that, wherever possible, they
should be slightly greater. In very poor quality rock inasses, in which the overstressed zones are larger, i t inay
he advisable to increase the pillar width to 1.5 times the height of the larger cavern.In all cases, a comparative
study similar to that illustrated in Figure 13.5 should be carried out in order to confirm these decisions.

13.5 Problems in using a concrete arch in weak rock

Many underground caverns have been constructed with roof support provided by a cast-in-place
concrete arch. As illustrated in Figure 13.6, the cavern arch is excavated to its full width and inclined haunches
are provided to carry the reaction of the concrete arch.The reinforced concrete arch is cast in place when the
floor of the excavation is leve1 with the bottom of the inclined haunches.The lower part of the cavern is then
excavated, usually by benching downwards.

'Thii analyris w a i originally ianied aut using the program EXAMINEzDthati5 only iutable for elastiianalysei. Since that time, new programi iucli a i
PHASE' have become availableand theie can be used for a full progresrive faiiure anaiyits. It is recommended that analyiei of the type deioibed heie
rhould be canied aut with a program such as PHASE'. whch isavailable from Racscience Inc. 31 Balsam Avenue, Toronto. Ontaiio. Canadõ M4E 385,
Fan 1 416 698 0908, Phone 1 416 698 8217, Email: saftware@rocscience.com,nternet h t t p : l / w . r o i i r i e n ~ e . ~ ~ m .
 urse nbes I

Figure 13.5: Cantours of itrengrhhtreii ratioi and faiiure trajectories in the rock m a s wrrounding two adjacent caverns in a iiltsrone rock mais w t h '
diiferent spaiing between the caverns.
 use cavernr, in weak ro

The concrete ardi provides support for any rode that may become loosened in the cavern roof The
problem with this arch is that, if it is very rigid as compared'with the surrounding rock mass, the deforinations
induced as a result of the excavation of the lower part oftlie cavern can cause excessive bendiiig in the coiicrete
arch. In addition, large amounts of teinporary support in the form of rodc reinforceineiit may be required to
allow excavation to proceed to the stage where the concrete arch can be constructed. This reinforcement will
have to be placed in the same manner as would normally he employed if it formed the permanent support of
the rock ar&. This would involve excavating a small top heading in the central crown, installing the reinforce-
ment, widening the heading out into the hauncbes, installing more reinforceinent, and so on. An exainple
where pre- treatment and pre-support of a very poor quality rock mass was carried out to avoid temporary
support problems during arch excavation is presented in a paper by Cheng and Liu (1990).
Figure 13.6 is a plot of the displacements induced hy excavation of the lower part of thc cavern.This
plot was obtained by subtracting the displacements induced by the creatioii of the top heading fioin the
displacements induced by the excavation of the entire cavern. These displacement vectors show that the
upper part of the sidewalls displace inwards ahout 60 mm while the centre of the arch displaces upwards
about 10 inin. Figure 13.7 is a diagraminatic representation of the displaceinents imposed o n the concrete
arch as a result of the displaceinents in the rock mass. Depending upon the magnitude of tlie displaceinents
iil the rock inass and the curvature and thidaiess oSthk concrete arch, tlie stresses in the coiicrete aiid in the
reinforcing steel can exceed the safe workiiig loads in these materials. This cai1 give rise to critical conditions
during construction since the repair of a damaged concrete roof arch is an extreinely difficult and expeiisive
process. Cases exist where severe cradcing of the concrete arcli led to the installation OSadditional steel ardi
support in local areas of the power cavern roof
In general, I consider that the use of concrete roof arches should be avoided when designing large
underground powerhouse caverns in wealc rodc masses. Experience has shown that the use of a more flexible
support system such as that provided hy grouted cables and a surface layer of shotcrete provides a more
satisfactory solution. Where local problems occur, these can usually he dealt with hy the installation of
additional cables or the application of an additional layer OSshotcrete. As long as access to the roof is
maintaiiied, remedial worics cai1 he carried out without disrupting the other construction ac~ivitiesin the
cavern.Two scheines, with which I have been involved, have utilised platforms o11 top of-a teinporaiy crane
to ensure that such access is avaiiable.

13.6 Crane beams

Underground powerhouse and transformer caverns require cranes of significant capacity to inove
heavy pieces of equipment around during installation and maintenance operatioils. Since these cranes are
designed by structural engiiieers, it has heen the practice that the cranes are supported o11 beains supported
by columns. Iu many cases these stmctures are designed to he completely independent of the surrounding
rock inass, just as they would he in a surface powerhouse structure. In my opinion, this is an inappropriate
design approach since it does not utilise the enormous carrying capacity of the roclc mass surrouiidiilg the
cavern. Whenever possible the cranes beams should be suspended directly from the cavern walls as described
in the examples given below.
Tlie concerns whicb are norinally expressed by designers, who are iiot iãiniiiar with tlie installation OS
suspended crane bea'ms, are:

1. What capacity and length of cables or rockholts are required to support the beains and what assurance
can be given about the security of this support system?
2. How can displacements in the roclc mass surrounding the cavern he accominodated in the case OScrane
beams attached to the walls?
 _ _ _ . . . . 1 1 1 1 < < . . . _ _ . _ .

_ _ _ .. -. ,
. . . . . I . . I . . . . . - - - - - _ C

I . . . . . - - _ _ . _ I

. . _ _ - . . . . a . . . . - - - C . I _
_ _ . _ - - e I *
- _ - _ - - - I , . . . - - - - - C C . #

-- -------- . Concrete arch - . - # H . -

* C & . - . -

--
--e".-

-- -------.
- .-----.-
------
-----...
-------- -- c-ccc-c-
----c-

-----.
c----- CC

-- -------
-------C

--
Lower port e-------

-----
-- -- -----__ --
of Ci,"e,-" e------
e+- ..-c-

- - - -__ C- --c-- CC

---------_
C--CC--

- - - - - - - --_. ------
- # - - - - _ _ _ C C C

e--.------
c-----
-

-------___
- - e - - - _ _ 9

- - - - - - - - . \ < I , , , , - - - - -
-- --- -- -- --- -. ,- , .\ ., .I ,, , ,, ., .<< ,L ,_ . r - - -

- - - - - - - . . \ , , , , < , , , - - -
- - - - - . . . , , <. ,
- - - - - . - . . . , a < < ,
, , . , , . -
, , r . - -

Figure 13.6: Plot af net dsplacernents indured by excavation af the lower pan of a large cavei" in which the roof iupport i i piovided by rneaniofa
cast~in~placeconirete arch.

Figure 13.7: Dispacements imposed an a cait-ir-place cancrete a n h as a result of exiavatron of the iower part af the cavern
 ieenng: CoiJrse notes t

Figure 13.8: Example of a

iast n plaie crane b e m in
the Diakensbery Pumped
storage proieit in Sauth
Afiica. The beam i i
supported againit a vertical
face of interbedded
sanditone. sltitone and
mudstone by meani of

Figure 13.9: Example of iuspended irane beami n the
powerhouie cavern of the Singkaiak Hydroeiectric Project
in Indanesia.
Figure 13.10: Example of anihored i r a n e beami in the
undeiyiaund powerhoure cavem for theThinavroi Hydro~
electric pro)ect in Greeie. ln this caie the beami ihown
were used for a ionitrucfion oane and ionirete colurnnr
were added later to cany the full crane loadr.
 use cavern: , in weak ro

The principal reason for the choice of this system on many projects is that it a1low.s for eariy installa-
tion ofthe cranes. Frequently, a light construction gantry, which runs o n the craiie rails, ir installed iininedi-
ately after the construction of the crane heams. This provides access for the rnonitoring ofinstruments in the
cavern roof and for repair worli to the roof support system if required. The inain craile cai1 bc asseiiihled
early in tbe sd~eduleatid it is then available to assist in excavatioii of the lower benches and iii tlie concrete
work in the base of the cavern. These scheduling advantages offer considerable benefits wheil compared with
a norinal column supported crane.

13.7 Choice of cavern shapes

In strong roclc masses, for which rocli inass failure is not a problein, the coilventioiial shape choseii iõr
an undeground powerliouse cavern is similar to that illustrated in Figure 13.11. The arched roof provides
stability in the rock above the cavern roof and also provides convenient headroom for an overhead craiie.
The sidewalls are siinple to excavate by vertical drill-and-blast benchiilg aiid provide clean ~~iicoinplicated
walls for craiie colurnn location and the accominodation of services.
The prohlein with this cavern shape when used in wealiroclc inasses, particularly with high horizoiital
iii situ stresses, is that tal1 straight sidewalls are deflected iiiwards (see Figure 13.6) and tensile failure is
induced as shown in Figure 13.11. Zones of failure are illustrated in this figure and the maxitnuin sidewall
movement is 38.5 inm. The stabilisation of the rock mass surrouiiding this caveri1 will require significant
reinforcement in the form of gro~itedcables or rocl<bolts.
An alternative cavem shape is illustrated in Figure 13.12. This elliptical shape has been used on
schemes such as Waldeck I1 in Germany (Lottes, (1972)) and Singliarak in Indonesia (see Figure 13.9). As
shown in Figure 13.12, the dept ofthe failure zones in the sidewalls has been reduces as coinpared to that in
tlie conventional cavern. Tliis results in a more stable overall cavern and a reduced support requireineiit.
While this cavern shape is better from a geotechnical point of view, it has some practical disadvan-
tages. The cavern shape is such that the construction has to be more carefully executed tliati the cotiveii-
tional straight-walled cavern aiid items such as the cranes and services liave to be desigiled to fit into the
cavern shape. These differeiices cai1 create significant probleins where the slili of tiie labour force is liinited.
In the wealcrodischeines in whidi I have heen involved, the conventional caveri1 shape has been chosen
in preference to tlie elliptical shape hecause the overall advantages of tlie elliptical caverii have iiot heen
considered of critical iinportance when compared with the siinplicity of tlie conventiotial caverii sliape. l'resent
support design techniques, discussed later in this chapter, are relatively unsophisticated and the stress clianges
resulting froin a chaiige in cavern shape are probably too small to have very much iinpact upon the support
design. Consequently, it is doubtf~ilwhether the overall costs and time involved in the construction of converi-
tiooally shaped caverns would be higher than those that would be incurred in constructing ellipticnl caverns.
I recoininend that each scheine should be investigated o n its own merits, taliing into account cori-
stniction problems as well as geotechnical factors.1n some cases, the use of an elliptical caverii sli;ipe may be
justified but, in general, the conventioiial cavern sliape illustrated in Figure 13.11 will be found suirable for
all but the very weakest rock masses.
Before leaving this topic, attention is drawn to the unfavourahle stress and poteiitial iàilure condi-
tions created by the stepped base of the cavern. A step ofsoine sort is generally required to liousi. tlie draft
tubes and the lower parts of the turbines. It could be argued that this step should be created by cast-iii-piace
concrete after a cavern of optiinum shape has been excavated. In practice, tliis somewhat theoretical ap-
proach is found to be both unnecessary and uneconomical since failure in the base of the caverii is relatively
easy to control. The instability of this hench is due to stress relief resulting froin the creation of ali ui~sup-
ported vertical face and minimal support iti the forin of untensioned grouted steel rods (dowels), installed
from the cavern floor before excavation of the lower benches, will counteract this instability.
 ouse caverr1s in weak rock

r g ~ r 13
t 11. : . -.C ,
~. . . _. Figure 13.12: Elliptical cavern i l i o w n g zoneiof faiure for
. ; I .. .- - .
... ,
.:.r i.. .i- t i .< I.. the same rock mais strengtli and in iitu strea conditons
.,, , .-i , ,,r:.. . 1: a i for lhe conventionai iavern ilustrated n Figure 13 1 1
times the vertical streri. The deformed excavation The manimum sidewal displaiement ir 3 3 mm.
ihape is shown and maximum sidewall displaie-
ment i8 38.5 mm. Analysed uiing PHASE2.

Figure 13.13: nfluence of two sets afjointi. indined at 45.. on the displacementson therock mais iurraundiny a convenlional cavern The rock m õ i i
propeitles and the in silu stressesarethe same as those aisumed in the previoui sectian. Thejointi have a cohes~veitrength of 0 2 MPa and a fricton
angle a i 30".
 Rock Engin eering. Cou ne notes b3

A good starting point for any support design is a study of the literature to deterinine what others liave
dane iil similar circuinstances. These precedents have also been usefully summarised in the roclz mass classi-
fication schemes of Bieniawski (1989) and Barton, Lien and Lunde (1974). Figure 13.14, adapted from Barton
(1989), gives the type of support which has been successfully used in underground excavatioiis in weak rock
masses. In this figure,fibercrete is an abbreviation for steel fibre reinforced shotcrete and bolts refer to either
grouted bolts or cahles whicli inay or inay not be tensioned, depending upon the sequence of installation.
Note that the design of cast concrete liiiings, shown in the upper left hand segtneiit of Figure 13.14,
must take into account the relative deformability of the concrete and the surrouiiditig rock inass. As dis-
cussed earlier, tlie choice o f a n inappropriate shape and thickness for a concrete liiiiilg cai1 result in serious
overstressiiig of the concrete.This problem is more acute for acavern roof than for a tunnel lining. Tlie roof
arch is vuliierable to deformations resulting from benching after its construction and there is n o opportunity
to close the arch by constructing a concrete invert, as is cominoiily done for a tunnel lining.
I unreservedly recoininend the use of one of the existing classification scheines to obtain a first ap-
proxiination to tlie Ievel of support necessary for a large cavern in a wealc roclc mass. Mowever, caution is
advised against the uncritical acceptance of the recommendations coming from them. The collection of the
data necessary to calculate RMR a r Qvalues for a rock mass usually provides enough inforination for the
design eiigineei to formulate a clear impression of the particular modes of failure that his desigil must resist.
Careful consideration of the poteiitial behaviour patterns of the rock mass, particularly the iiifluence of
geological structures, inay lead to significant inodification of the support recominendations obtained from
the applicatioil of classification systems.

13.9.1 Estimating support pressures

One approach i11 estitnatiiig support pressure is to consider the extent of the zones of overstressed and
blast damaged rod< above the caveri1 roof and to assuine that this rodc acts as a dead weight which has to be
supported. Consideriiig tlie exainple illustrated in Figure 13.1 1 for a rock mass with an RMRvalue of 48 and
Q= 1.5, the zone of overstress extends approximately 3 m into the rodc inass above the crown of tliis 20 m
span caverii.
In addition to the stress induced fracturing that may occur, the fkacturing and loosening of the rock tnass
due to blasting should also be considered.From experience it is suggested that blast damage may extend 1.5 to 3
m into the rock adjacent to the roof dependingupon how much care has been talcen to control the blastirlg.
Assuining tliat 3 in of rodc have been damaged by either stress induced or blast induced fracturing, a
dead weight of broken roclc of up to 8 tonneslmz of exposed surface of the cavern roof has to be supported.
Where this support is to he provided by means of rodzbolts or cables, a factor of safety of 1.5 to 2 is usually
allowed fo account for installation probleins and to provide some reserve support capacity. Hence, the total
capacity of the installed bolts or cables should be of the arder of 12 to 16 toi1iies/m~(0.12 to 0.16 MPa).
Figure 13.15, adapted frorn apaper by Barto11(1989), shows that this estimate ofa support pressure of 12
to 16 tonneslrnqor a rodz inass with Q=1.5 is in line with recommendatioiis hased upon previous experience.
Some simplified closed form theoretical solutions have beeii developed which permit the study oftlie
iilteraction of different types of support with the zone of failed rocl<surrounding underground excavations,
and these solutioiis liave heen summarised by Brown et al(1983).All of these solutions coiisider the develop-
ment of a 'plastic' failure zone in a homogeneous rock mass surrounding a circular tunnel in a hydrostatic
stress field aiid these assumptions impose severe limitations upon the application of these solutions to
practical rock support desigii. Nevertheless, these simplified models have proved to be very useful in the
development of an understanding of the basic concepts of rocl< support interaction and I urge interested
readers to hecome familiar with these models and c0ncepts.A discussion o n this topic together with a listing
of the simple calculatioii steps required can be found in Hoek and Brown (1988).
 ster 13 Lar luse caverns in weak rt

.c
c
Z Z
c m

e
W

@
a ,O
k
2
"7

I
mi I, 1 t m, 111
B4RR)N'S NNNOUNC W INOM Q WRTON'S N N N U J N C OUALm INOa Q
esmeromo
,
mo
81ENkWSKl'S ROCK UAS RAllNG RUR

Figure 13.14: Summary af precedem experiente on suppart of weak Figure 13.15: Relatiooship between suppori presíure and rock mais
racks in terms of'rock m a s ilaiiificationi. Adapted fram Barton (1989). quality. Adapted from Bartan (1989).

Today, the availability ofpowerful nuinerical analysis tools such as FLAC and PHASE' inalces it
possible to study support optioiis in great detail. Sucli analyses are no longer constrained by thc capability of
tlie programs but ratlier by the quality of the input information. Tlie reader should avoid the teinptation to
believe the results of a single analysis, however convincing the output niay appear. It is always necessary to
carry out parametric studies to cover the range ofpossible input paraineters. It is only in this way that a
sound understanding of the behaviour of the rod<mass surrounding the excavation can be established.

13.9.2 Design of rockbolt and cable support

The excavation of a large caverti in weak rock wiil usually require the installation of systeinatic rock
support or reinforcement during excavation as an integral part of the excavation cycle.Even wlien the use of
cast concrete arches and/or sidewalls is envisaged for final supi~ort,consideration inust be given to the
requireinents of teinporary stability during excavatioii, which will often require extensive support if safe and
controlled excavatioil is t o be achieved. l n many cases, the only significant difference between supports nsed
for temporary and permaneiit purposes relates to corrosion protection requirements. Corrosion protectioii is
an important desi,gn consideration in environments where corrosive water is encountered during preliminary
investigation or excavation. In modern practice, corrosiol~protection measures usually require at ieast oile
pliysical barrier (single corrosion protection) to protect the individual rockbolts ar rock ancllors against
corrosion (British Standards, 1987). This physical barrier often talces the form of a corrugated plastic slieatli
which completely encapsulates the steel bar or cable from which the anchor is formed.
Rock inass reinforcement for large caverns usually involves the installation of rodcbolts forined froin
deformed steel bar or cables made froin pre-stressing strand. Rock bolts are generally cheaper and quiclcer to
install but cables can provide bigher capacity and may be easier to iiistall when long reinforcement eieinents
are required. Coinbinations ofrockbolts and cables are cominonly nsed to combine the best features of both
systems. The rockbolts are installed close to the face for immediate support while the cables are installed
subsequently as tlie primary reinforcement system.
 Rock Engineering COL

The possibility has to he considered that, during excavation, cahles will pidc up excess load over tlieir
installation load to the extent that they become seriously overstressed. In this circumstance, predictioti of the
filial cable load is an important input into decisions on the appropriate installed load, the tirning of the instal-
lation and tensioning and the choice between fully grouted ar adjustable cables. In the author's experience,
fully grouted calles are more convenient for the contractor and provide stiffer support, particularly i11response
to deforinations occurring at a large angle to the cables. However, the use of adjustable (i.e. te-tensionbable)
cables may be appropriate i11 circiiinstances where very large deformations are anticipated. These cables are
rnade by placing a plastic sheath over a sigtiificant portion of tlie lengtli of tlie cable. n i i s sheath brealcs the
bond between the cable and the grout and allows the cable to deform independently of the snrrouiiding roclc.
It is important to recognise that there are two types of rodibolt or cable support cominonly used in
underground excavatioiis. In good quality rodí masses in which the stability is controlled by intersectiiig
joints, beddiiig planes, shear zones and faults, the support has to he desigtied to reitiforce specific biocks atid
wedges which may fali or slide into the excavation. This type of support, frequently referred to as 'spot
bolting', iiivolves the installation of a few bolts or cables at clearly defitied locations, with their length,
orientation and capacity chosen to provide adequate support for the wedge or hloclí under consideratioii.
While 'spot bolting' may be required for isolated blods or wedges in wealirodc inasses, it is the second type
of support, frequently called 'pattern bolting', which is more releGant to this chapter. 'Pattern bolting' involves the
installation of rockbolts or cables in a regular pattern that is designed to reinforce the entire rockinass in inucli tlie
same way as reinforcing steel acts in reinforced concrete. Typically, 5 m long 20 tonne capacity rodibolts installed
in a 2 m x 2 m grid pattern could be used over the entire roof and wall area of a large underground cavern. niese
bolts would provide a support pressure of 2.5 tonnes/m2 if loaded to 50% of their capacity.
The first question to be decided is the length of the rodibolts or cahles. Analyses of tlie extent of zones of
oventressed rodi, such as those presented in Figures 13.11 and 12.12, are useful in determining the approximate
extent ofrodirequiring support.Generally, the bolts or cables should extend 2 or 3 in beyond the limit of the zone
of overstressed material. As previously stated, great care must be taken in using this approadi to select reinforce-
ment lengths since relatively modest dianges in rock mass properties or in situ stresses can result in significant
changes in the zones of overstress. Consequently, parametric studies in which these iiiput data are varied over the
inaxiinum credible range ar: esseiitial if the reinforcement lengths are to he based upon sudi studies.
An alternative approach is to use previous experience. Figures 13.16 aiid 13.17 give the lengths ofroof
and sidewall rockbolts and cables in some typical large powerhouse caverns in weak rocli masses. Plotted on
the same graphs are einpirical relationships suggested by Barton (1989) for bolts and cables. For under-
ground powerhouse excavations these relationships are simplified to:

Roof rodcbolts +
L = 2 0.15 x SPAN in
cables L=0.4xSPANm
Walls rodibolts +
L = 2 0.15 x HEIGHT m
cahles L = 0.35 x HEIGHT m

The choice of rocliholt and cable spacing is based upon the following considerations:
a) In order to ensure that the bolts or cables interact with each other to form a zone ofuniformly reinforced
roclc (Lang, 1961), the spacing S of the bolts or cables should be less than one half of the length L, i.e. S
< L/2.
ort pressure P and a worliing load in the bolt or cable T,the spacing for a square grid is given hy
S = a?= T / P .
h) For
Typical holt and cable spacing range from 1 to 3 m with 1.5 m being a cornrnon spacing for bolts and 2 m
being widely used for cables. Where additional support capacity is required to support local areas of weaker
 Chapter 13 LE use caverns i n weak ro,

rock, holts or cahles placed at the centre of each grid square will sometimes suffice. Alternatively, when
cahles are used, additional strands can he placed in each bole to increase the capacity of the cahles.

13.9.3 Use of shotcrete linings

In the past two decades, sliotcrete has developed into a versatile support systein that is ideally suited
to the requireinents of cavern support in a deformahle rock mass.
Two significant benefits ofshotcrete are that it can be applied quidcly to freslily exposed rod<during
excavation and that it develops strength steadily after application. Green shotcrete is resilient to dainage
from nearby hlasting and associated stress redistribution and gains strength at the same time as load is being
transferred onto it. High early strength can be achieved by the addition of accelerators without a s'erious loss
in long-term strength. Dosage of the accelerator has to be controlled carefiilly to ensure good intermixing
and avoid local overdosing.
Wheii used in combination with systematic rock reinforcement, shotcrete can provide immediate
temporary support during excavation or form the permanent lining for the roof and sidewalls during the
design life of the cavern. Irrespective of the form of temporary support, shotcrete application for permanent
support purposes can often he delayed until all adjacent excavation has been completed. This allows the
permanent lining to he sprayed in a better worlcing environment than that existing close to the working face
and this allows hetter control over thicluiess and quality. Some designers prefer to utilise additive free shotcrete
for final lining application on the grounds that high early strength is not needed for the final lining aiid the
addition of additives may he detrimental to the long-terin perforinance of the shotcrete.
Research and development into shotcrete mixes, additives and equipinent have progressed to the
stage that shotcrete quality now rests almost entirely with the choice of compatihle equipinent and the
equipment operators. Wet mix application requires careful attention to the supply of mix and air to ensure
that the material leaves the nozzle in a continuous unintermpted stream that can he applied by the operator
ornozzleman in such a fashion as to maximise compaction and ininimise rebound.With the dry inix method
the supply of water also has to be controlled hy the nozzleman.

Wildeck II .
C i b l a : L = 0.4 x rpan

-
Imaichi ( u b l s ) .
Hongtin ( u b h )
@ Mingtan ( o b b -'fauk zom)
,' ' Min~hu
/
Wingtan ( u b k - bedded *andatene) " Nevada tert rite
ò
Mingtrn (bcdded randstonc)@

.
/---
/---
' Randpo íbohr)
Drakenrberg (boltr)
/

Ponage Mountarn (boltn)

Poatina (boltr)
+ 0.15 x rprn + 0.15 x height
I0
Rackbaltr : L = 2

4 . . . . 3s, . . . . , . . . . ,,, . . . ,x,..... ..,

iU
, ,
w
Rockboltr : L = 2

Roof Spin -m
Figure 13.16: Rockboit and cable lengths for roaf iuppolt in some Figure 13.17: Roikboit and cable lengrhifar iidewail iupport n some
large caverni in weak raik. lacge cauerni in weak rock
 Rock Engir iy Evert Hol

For the engiiieer who is not an expert in shotcrete technology, there is a bewildering choice of equip-
ment and additives, starting with the basic selection of the wet mix a r the dry mix method. No hard aiid fdst
rules apply to this selection, other than to say that, depending on local circumstai~ces,either system may be
suitable to form the teinporary a r permanent lining of a large cavern. In general, higher production rates are
possible with the wet mix processes, hut this is not necessarily a major factor in considering the system to be
used for cavern lining purposes. Remotely controlled robot applicator systems are widely used, but hand
lield nozzles will also provide a satisfactory product, if used properly. Liquid plasticizers and water reducing
agents are commonly used as ai1 aid to application and dust suppressants have r e c e n t l ~been developed to
improve the working environment for tlie nozzleinan.It is common policy to liinit the total amount of all
additives in a mix to a figure of the order of 5 a r 6% by weight of cement.
A significant advance in shotcrete technology has been brought about by the introduction of inicro
silica into the mix at u p to 10% by weight of cement.This results in a significant reduction in rebound, and an
increase in the thickness that can be built up in a single application. It is also beneficia1 for application outo
damp a r wet surfaces.The addition of micro silica produces a deiiser product with an increase in early strength,
and does not appear to have a detrimental effect on long-term strength. Problems with the use of steel fibre
reinforced shotcrete have been greatly reduced by the use of micro silica. Rebound of both shotcrete and the
fibres is significantly less than it used to be.Balling of the steel fibres and excessive equipment wear have also
beeil largely overcome so that, over the last tive years, micro silica fibre reinforced shotcrete technology has
become aviable and frequently preferable option to the accepted use of steel mesh embedded in plain shotcrete.
For steel fibre shotcrete, the iiumber ofsbotcrete applications can soinetimes be reduced when com-
pared with the more complex installation of layers ofplain andlor mesh reinforced shotcrete. Although the
initial bending strength of the two products is similar, performance is improved because the post crack load
bearing capacity is significantly better in the case ofsteel fibre shotcrete.Typica1 current practice involves the
use of fibres in the range of 20 to 40 mm long and approximately 0.5 mm diameter. Current research is
examining the use of longer tibres (to improve post crack strength furtlier) and inaterials other than steel.
Decisions o n lining thickness are usually based on a combination of empirical and practical consider-
ations rather than concern about stress levels in the lining. Where concern about stress does exist, delayed
application of the final layer or the application of an additional layer are available options. The thickness
built up in a single application is typically of the order of 40 to 80 mm and total thickness of the order 100
to 200 mm. With layers less than 40 mm thick concern will sometiines exist that an effectively continuous
layer will not be achieved if application to a very irregular surface is required.
When designing perinanent shotcrete linings, it is important to specify quality control or acceptance
tests as a design requirement.The use ofsteel fibre shotcrete does not readily allow rigorous checking during the
execution of the work in the manner that rockbolt work or reinforced concrete constmction does, since it is
dependent on the skill of the operator.Thus, it is essential to do routine acceptance testing by coring througli
the completed lining in order to &e& the density and strength ofthe sprayed product, the adhesion to the rock
surface, the inter-layer adhesion where two a r more layers have been applied, and the total thickness achieved.
Inter-layer adhesion can be a particular problem if a long time period elapses between the application of
temporary support and permanent support shotcrete, since it is difficult to remove the grime that accumulates
on the surface of the initial layer if diesel powered equipment is used in association with blasting, mudung and
support activities. Where large deforinations are expected after the completion of the final lining, some design-
ers may stillprefer a fully engineered solution, with mesh layers incorporated into a shotcrete lining and
positive connection of the mesh to rockbolts and cables, over the use of simple fibre reinforced linings.
Uncertainties of this type in relation to shotcrete linings give rise to a tendency to overspeciiy the
product in terms of the strength properties to be achieved at 3 days or 7 days. While it is generally tme that
high early strength is a desirable feature ofshotcrete, nothing is gained hy forcing the contractor to produce
 Chapter 13 Lars use caverns in weak orm

higher strengths than are needed. For exainple, a permanent lining sprayed onto a rode surface, which already
had a reasonahle level of support for temporary purposes, may not need to be accelerated at all. I11 this
circumstaiice, the 7 and 28 day strengths generally accepted in structural concrete usage inay be as good a
criterion as the 3 and 7 day strengths which have come to be associated with shotcrete usage. It lias already
been stated that concern ahout the long- term effects of additives on strength has caused some designers to
opt for an additive free sliotcrete for final liiiiiig purposes.

13.9.4 Support installation sequences

Should roclíbolts aiid cables be tensioned and at what stage of the excavation sequence should a
shotcrete layer be applied?These are questions which arise during discussions o n the design of support for
underground excavations in weak rock masses. They are dealt with liere hy ineans of a practical exainple
hased upon the support instailation sequence for the power cavern of the Miilgtan Puinped Storage Project
in Taiwan. Tl~isproject is described in detail by Cheng and Liu (1990) 2nd further details can be found in a
paper by Moy and Hoek (1989). Figure 13.18 gives a summary of the principal support installation stages for
this 25 in span 46 m high cavern in fair to poor quality bedded sandstone.
Duriiig a preliininary contract an existing exploration/drainage gallery and two longitudinal working
galleries were utilised to instali grouted cables in the pattern iliustrated in Figure 13.18a.These 50 tonne
capacity cables were double corrosion protected and installed dowiiwards from the exploration/draiiiage
gallery, located 10 m above the crown of the arch, and upwards from the two worlcing galleries. A light
straighteiiing load of 5 toiiiies was applied to each cahle before grouting and hence the cables were effectively
untensioned but straight and fully grouted iiito the roclí inass. Since these cahles were installed hefore aiiy
excavation of the cavern had taken place, n o significant displaceinents had occurred in the rock mass at tlie
time of cable grouting.
Excavation of the cavern roof, illustrated in Figure 13.18b, induced significant displaceineiits (Moy
and Hoelc (1989)) and these tensioned the grouted cables. Had the cables heen tensioned hefore grouting,
tlie additional tension induced by the displaceinents in the rock inass could have resulted in overstressing of
the cahles.The purpose behind the installation of these uiitensioned grouted cables was to reinforce the roclc
mass in much the same way as the placement of reinforcing bars in concrete acts to strengtheil the concrete.
The process was intended to improve the overall quality of the rodí inass so that the main excavation
contract could proceed with fewei roclí stahility problems than would have been the case had tlie pre-
reinforceinent not beeii in place.
The cavem roof was excavated by ineans of a central 6 m x 6 in heading which was suhsequently
slashed out to the full cavern width as illustrated in Figure 13.18b. Upon exposure of the final roofsurface at
each stage of tliis excavation process, a 50 mm layer of steel fibre reinforced inicro-silica shotcrete was
applied within 5 to 10 in of the face.The purpose of this shotcrete layer was to provide support for the sinal1
blocks and wedges which would othenvise have been free to fall from behveen tlie reinforcing cables. In
addition, the sliotcrete provided iminediate sealing against moisture changes which could cause slaking iii
some o i t h e siltstone rock units exposed on the surface.
A relatively thin shotcrete layer was used at this stage in order to allow for displacements which would
be induced by adjacent excavation of the upper part of the cavern.Even if mitior cracking of the shotcrete
had been induced hy these displacements, the presence of the steel fihre reinforcemeiit provided a high post-
crack deforination capacity for the shotcrete and hence maintained its support capacity.
As soo11 as tlie ends of the pre-placed reinforcing cables were exposed, faceplates were installed on them
and a tensioil of 20% of the ultimate capacity of the cahle was applied to ensure positive anchorage. Note that
this tension acts over a very short length of eadi cable near its exposed end since the reinainder of the cable is
fully grouted into the rock mass. Experience in tlie inining industry has shown that the installation of faccplates
 Rock Eng by Evert H1

oiipre-placed untensioned groute cables is very beneficia1 in providing support for the near surface Llast
damaged material which othenvise tends to fall away from the ends of the cable. In inost areas in the Miiigtaii
cavern, 5 m long 25 mm diameter mechanically anchored rodcholts were placed at the centre of each square in
the 2 m x 2 in grid of cables.These bolts were tensioned to 70% of their yield load before grouting since they
would not be subjected to sigiiificant displacements during the excavation of the lower part of the
cavern.Excavation of the lower part of the cavem was carried out by meais of 2.5 m high vertical benches.
Sidewall support was provided by a 3 m x 3 m pattern of tensioned, grouted, double corrosion protected 75,
112 or 131 toiine cables, installed a t a downward angle of 15" to ensure crossing of bedding planes wliich strilce
across the cavem axis (Moy and Hoek (1989)).These cables were tensioned at between 38 and 45% of their yield
load, depeiiding upon their Ievel in the cavem sidewall and their location in relationship to the position of the
bench floor at tlie time of installation.The tension was reduced for those cables which were installed close to
the bench floor in the lower walis of the cavern. Mechanically anchored roclcbolts, 6 in long and 25 inin in
diameter, were installed between the cables as illustrated in Figure 13.18~.These bolts were tensioned to 70010of
their yield load before grouting.
At an early stage of benching, wlien excavation had progressed to a stage beyond which further defor-
mations induced in the roofwere relatively small, an additional thiclcness of 100 inm of steel fibre reinforced
inicro-silica sliotcrete was applied to the roofand upper sidewalls. The total thiclcness of 150 inm of shotcrete
represents the final lining for the roof and upper sidewalls of this cavern. Full details of the shotcrete specifi-
cations and mix design have been given by Moy, Hsieh and Li (1990).
The lower cavern sidewalls were reinforced with cables in the same way as the upper sidewalls
shown in Figure 1 3 . 1 8 ~Only
. 50 min of steel fibre reinforced micro-silica shotcrete was used o n the lower
sidewalls siiice most of these surfaces were subsequently covered by concrete as the turbine foundatioiis
were cast in place.

13.10 Excavation method

The stability of a large uiiderground excavation is verymuch dependent upon the integrity of tlie roclc
immediately surrounding it. In particular, the tendency for roof falls is directly related to the iiiterlocking of
the immediate roof strata. For weak rock masses with clay filled joints and localised soft or altered zones, tlie
capacity of the rock mass to contribute to its own support temporarily during excavation can be totally
destroyed by careless excavation or poor sequencing of excavation and support activities.Thus, the clioice of
excavation method assuines a degree of importance which has not always been catered for by specificatioiis
and construction procedures.
The most frequently adopted method for cavern excavation is drilling and blasting. Strictly spealcing,
blasting control is iiot a cavern design requirement hut it exerts the biggest single influence on the outcome
of the excavation process, and should be considered accordingly. In a wealc roclc mass, the sequenciiig of
excavation and support will typically follow the top heading, slash and 2.5 to 5 m beiich procedure described
in the previous section.
The control of rodcbolting (e.g. maintainingdesign spacing) and shotcrete operations (e.g. achiev-
ing uniform design thickness) is far more easily effected when good control is maintained over excavated
profiles, so blasting specitications are normally written in terins of maxiinum permissible overbreak or the
presence of "half barrels" (charge hole drill marks) o n the profile produced hy the blast. Even in weak
rocks, some half barrels can be expected with a well-balanced blast, especially if pre-splitting or sinooth
hlasting is utilised, but care should be talcen to ensure that the specifications d o not demand results which
are iinpossible to achieve i11practice.After the initial heading, extensive free faces exist and provide a void
into which the broken rock can move. In this circumstance, the hlast energy absorbed into the rodc inass
should be minimal.
 use caverns in weak ros

Installation of double co7rosion protected cables from

uploration/drainage gallery located 10 m above center
o f rwf arch and from tw longitudinal waking galleriez.
The 50 tonne capacity cables were installed on a 2 rn
x 2 m grid pattern and a small straightening load o f 5
tonnes was appli:d b e f a c grouting.

Excavation o f cavern roof by center heading and slash-

ing with application o f fint 50 m m thick layer of steel
fiber reinforced microillica shotcrete. Faceplatcr were
added t o projecting endr o f cables and tensioned t o

centers o f 2 rn x 2 rn grid o f cable reinforccment.

dependirig upon their location relative to the bench.

Intermediate 6 m long 25 rnm diameter tensioned and

Complete ucavation of the cavern with 150 m m total

thickntss o f steel tiber reinforced rnicrc-silica shotcrcte
on the r w f and upper ridewalls and 50 m m thickness
on the lower ridewalls. Accerr t o r w f for inspection
and minar remedial bolting maintaincd from temporary

Figure 13.18: Suppart nstailaton iequenie for the power cavern of the Mingtan Purnped Storage Project in Taiwan.
Rock Engirieering Coiirse notes L

Figure 13.19: Caber and

ihotcrete were uied to
i roof a i the
~ u p p o t the
powei iavern in the
Mingtan Pumped Storage
Pmject n Taiwan.

Figure 13.20: Installation of

rables in the iidewal of lhe
power cavern in the
Mngtan Pumped Storòge
Prolect in Taiwan
 )use cavern

For a well-balanced blast, each successive delay should produce even breakage and leave the appropri-
ate burden to be removed by the next delay. Whether a pre-split or a sinooth blast is employed, the locatioii
and chargiiig of the easier holes drilled closest to the perimeter holes is critical to tlie outcoine of the blast,
since overcharging or poor alignment of these holes will result in damage beyond the final perimeter wliich
cannot be rectified once it has occurred. Hole alignment.of the perimeter holes themselves is ofobvious
importance, and can be maintained by the use of parallel drill hole facilities o n modern jumbos. Alterna-
tively, drilling inspectors can help the operator inaintain the required hole alignment during the early stages
ofdrilling. For vertical drilling, templates inade up fiom flat plates with steel tubes welded onto them to act
as a guide for the drill, can be used when the hole is collared.
Where good drilling control and apparently well-balanced charges d o not produce good blasting
results, it is sometimes useful for the inspection teams to request a pieceineal hlast. This requires eacli delay
of holes to be fired individually, so that the profile created by each delay can be inspected to ensure that
breakage has occurred in the way it should. If such a process confirins that the holes and charges are well-
balanced but the production blasts still d o not produce the desired results, then there may be a problem with
the initiation system, e.g. excessive scatter on the delay detonators producing out of sequence firing.
The ultiinate in damage control is machine excavation, and for cavern excavation, this usually implies
the use of road headers. The possibility of using road headers for general cavern excavation should be
considered wherever the intact rod<streiigth is less than 60 MPa, and the viability of the method, as opposed
to drilling and blasting, will be dependent on a comparison ofthe costs and required excavatioii rates rather
than the ability of the road Iieader to cut the rod<.The absolute limiting rod<strength for cuttiiig with a road
header has been put at 125 to 130 MPa, and this can only be achieved with great difficulty over short
distances with pick destnictioii being the limiting factor (Pearce, 1988). The lack of disturbaiice to the rode
and the possible reductions in support required are major advantages iii the use of road lieaders.
The use of road headers becomes even more attractive when there is a need to control vibrations
induced by blasting. This can be the case when cavern excavation is required adjacent to an existing under-
ground installation, or when cavems are to be excavated relatively close to tlie surface. Langefors and Khilstrorn
(1973) and others have published blast damage criteria for kuilding and surface structures.Alinost all oftliese
relate blast dainage to peak particle velocity resulting from the dynamic stresses induced by the explosioii.
Wliere these generally applicable guidelines impose unreasonable restriction o n the blasting requireinents of
a project, monitoring at the site allows site specific limits of the cliarge weight to be determined. These lirnits
are defined by the charge weight in the cavern which will produce unsatisfactorily high particle velocities at
the surface or the adjacent underground structures. An example o f this type of monitoring is described in a
paper o n the Tai Koo cavern in Hong Kong (Sliarp et a1 (1986)).

13.11 Cavern instrumentation

The installation ofrocl<mass monitoring systeins around a large cavern in a weak rod<mass is consid-
ered essential in arder to eiisure that control is maintained over stability conditions during and iininediately
followiiig excavation.The stress redistributions that accompaiiy excavation in a weak rode mass can produce
large deforinations, whicli will in turn modify the loads carried by the rocli reinforcement systein and the
stress carried by the shotcrete lining. Given the uncertainties of support design, tlie design eiigiiieer will
require confirmation that his assumptions, o n the level of deformation and load and stress clianges that will
occur, are not invalidated by the actual response to excavation.
O f the three effects listed above, the most reliable data usually come froin displacement moiiitor-
ing, since this can be conducted on a scale comparable with the size of the excavation and the voluine of
rock affected by stress redistribution. By comparison, measurements of stress change in the rocli or the
liiiiiig cai1 only be conducted at isolated points which may not be representative o f t h e average condition.
 Rock Engineering. Coi

Load inonitoring in the rock reinforcement is possihle for unbonded holts and anchors, but the results are
of questionable applicability to fully honded reinforcement for which highly localised strains and load
changes may occur where the bolt or cahle crosses a specific joint.
Displacement monitoring'may be relative or absolute.An example ofthe former is the installation of
a multipoint extensometer in the sidewall of a cavern, with the deepest anchor inside the zone ofrocl<where
movement may he expected. Movements beyond the deepest anchor will not be registered by the extensom-
eter. An example of absolute movement monitoring is the measurement of the horizontal convergeuce of
the two sidewalls of the cavern hy meails of a tape extensometer stretched betweeti the walls. However deep
the movement, it will all he registered by tbe tape extensometer. Where relative displacement is monitored,
the opportunity exists to extrapolate to the ahsolute displacement hy calihrating a numerical model agaiilst
the relative inovements monitored, and using the model's predictions outside the monitored zone. Nuineri-
cal models also allow the estimation of any inovements occurring before the installation of the itlstruineilts,
which are usually iiistalled from inside the excavation. In some circuinstances, it is possible to instali exten-
someters, hefore the coininenceinent of cavern excavation, from adjacent exploration or drainage galleries,
and tlie advisabiiity of doing this should always be assessed. In deciding the layout of exteiisoineters, it is
usually advailtageous to be ahle to distinguish hetween movements inside and outside the reinforced zone,
since the foriner will affect the loads carried hy the reinforcement and the latter will not. During the later
stages of excavation of a cavern in weak rodc, large movements may continue at depth in the upper sidewall,
but this may he of n o coilcern to tlie designer if the reinforced zone has stabilised.
Stress changes in the rodc can he calculated from monitored displacemei~tsby the assumption of a
value for the rodc mass modulus. The use of analyses of this kind during the early stages of excavation
sometimes indicates the need for additioilal supports or modifications of the design requireinents during a
later stage of excavation. This will be backed up hy load change data ohtained fiam load cells fitted to
isolated elemeiits of the reinforcement system.
Large stress changes i11 shotcrete linings are usually fairly apparent froin the occurrence of crad<s in
the lining. For this reasoil, vislial observation maintains its status as an important data gathering inethod. It
is also still the inost effective way of assessing the groundwater coilditions in tlie rock mass'surrounding ai1
opening. Where groundwater discharges into a cavern, piezometer installations are advisable to check tliat
excessive pressures cannot huild up in the roof or hehind the sidewalls.

13.12 Summary and conclusions

The design of large powerhouse caverns in weak roclc masses differs from that of caverns in stronger
rocl<s in that failure of the roclc inass surrounding the excavations and large deformations of the roof and
walls will have to be accommodated in the design. This requi&s an uilderstanding of the behaviour of weak
roclc inasses and of the interaction of the support with these rock inasses during excavation and subsequent
operation of the caverns.
Estimatiilg the strength and deformation characteristics of wealc rock masses is an uncertain process
and large variations in properties can be anticipated, particularly in hedded sedimentary rocks. This means
that precise analysis of the stresses and deformations induced by the excavation of the caverns is not pos-
sible, and the designer has to rely o n parametric studies in which the in situ stresses and material'properties
are varied over their inaxiinum credihle range in order to estahlish general hehavioural trends. Exainples of
such parainetric studies, using a two-dimensional elastic boundary element analysis, have been presented in
this chapter. More retined studies, using non-linear progressive failure analyses, are only justified when suffi-
cient data have heen gathered from the monitoring ofactual excavation hehaviour to provide realistic input
data forsuch analyses.An example Ofthis more refined type ofanalysis is presented in Cheng and Liu (1990).
Issues such as the location of the caverns relative to the toes of slopes aild the determination ofthe
 >use caverris in weak ri

lengths ofsteel linings in pressure tunneis, while not central to the question of cavern design, have important
practical and financia1 implications and have heen considered hriefly in this chapter.
The priiicipal issues which have been addressed are those o f t h e failure and deformatioris induced in
tlie rock inass surrounding large caverns and how these are dealt with in tlie choice of the excavation shape
and the type of reinforcement used. Concrete arches, traditionally used to provide support for the rode inass
ahove large powerhouse caveriis, can suffer from excessive bending as a result of the large deformations
wliich occur in these weak rocks. Consequently, the author recommends that concrete arches should not he
used or that, if they are used, very careful attention be given to inatching the deforination characteristics of
the arch to the displacements which occur in tlie rock mass. A preferred ineans of support involves tlie
iiistallation ofgrouted cables aiid roclcholts in the rock mass and tlie application of a surface layer of shotcrete
to stahilise the iiear surface blast-damaged rocl<.This system is very flexible, as compared with the concrete
arch, and can move with the rock mass to accominodate the large displaceinents associated with caveril
excavatioil. Corrosion protection of the cables is essential since these provide the priinary permanent sup-
port for the rock mass and inust have a worlcing life in excess of that o f t h e cavem itseli:
The sequeiice of installation ofcables, roclcbolts and shotcrete is an important issue which has been
illustrated by ineans ofa practical example.Tne questions ofwhether rod<holtsand cables should be terisioned
hefore grouting and when different shotcrete layers should be applied are all related to the developmeiit of
the deformation pattern iii the rode surrounding the excavation. Consequeiltly, the sequeiice of s u p ~ ~ o r t
insrallation must be carefully matched to tlie sequence of excavation in order to provide adequate support
without the rislc of overstressing the support elements.
All o f the care whicli has heeii taken in estimating the in situ stresses, the roclc inass strength and
deformatioii characteristics and in carrying out the support design can he wasted if excessive dainage is
inflicted oii the rock hy careless blasting.Techniques for controlling this blast dainage are available and have
proved to he very effective when Owners and Engiiieers work with the Contractor to ensure that these
techniques are used during critica1 stages of a project.
 Rockbolts and cables

14.1 fntroduction
Roclcbolts and dowels have heen used for many years for the support ofunderground excavations and
a wide variety of holt and dowel types have been developed to meet different iieeds which arise i11 inining
and civil engineering.
Rockbolts generally coiisist of plain steel rods with a mechanical anchor at one eiid and a face plate
and nnt at tlie other. They are always tensioned after installation. For short terrn applications the bolts are
generally left nngrouted. For more permanent applications or in rodi- in which corrosive groundwater is
present, the space between the bolt and the rock can be filled with cemeiit or resin grout.
Dowels or anchor bars generally consist of deformed steel ban which are grouted into the roclc.
Tensioning is not possible and the load in the dowels is generated by movements in the rodc mass. In order
to be effective, dowels have to be installed before significant movement in tlie rodc inass has talcen place.
Figure 14.1 illustrates a number of typical rockbolt and dowel applicatioiis that can be used to control
different types of failure that occur in rode masses around underground openings.
The move towards larger underground excavations in both mining and civil engineering has resulted
in the gradual development of cable reinforcement technology to take on the support duties which exceed
the capacity of traditional rocldiiolts and dowels. Some of the hardware issues that are critical in the success-
ful application of cahles in underground excavations are reviewed in this chapter.

14.2 Rockbolts
14.2.1 Mechanically anchored rockbolts
Expansion shell rockbolt anchors come iii a wide variety of styles but the basic principle of operation
is the same in all of these anchors. As shown in Figure 14.2, tlie components of a typical expansion shell
anchor are a tapered cone with an interna1 thread and a pair of wedges held in place hy a bail. The cone is
screwed oiito the threaded eiid of the bolt and the entire assembly is inserted into the hole that has been
drilled to receive the rod<bolt. The length o f t h e hole should be at least 100 mm longer than the bolt
otherwise the hail will be dislodged by being forced against the end of the hole. Once the assembly is in
place, a sharp pull o n the end of the bolt will seat the anchor. Tightening the bolt will force tlie cone further
into the wedge thereby increasing the anchor force.
These expansion shell anchors workwell in hard rock hut they are not very effective in closely jointed
rocks and in soft rodcs, hecanse of deformation and failure of the rock in contact with the wedge grips. In
such rocks, the use of resin cartridge anchors, descrihed later in this chapter, are recommended.
 Chapter 14 Rockbo Its and cabl~

I I Low stress levels

I High stress levels
I
Y
O
ea3
.->
V)
V)

Massive rack subiected to low in situ stress Massive rock sublected to high in situ çtress
ievels. No permanent suppon. Light levels. Pallern ockbolts or dowels with
Suppolt may be required for canstiuction mesh or shatcrete to inhibit fracturing and to
safety. keep broken rock in place.

Y
O
e
u
m
4-
c Massive rock with reiatively few
.-O Massive rack with relativeiy few
discontinuities subiected to low in situ discontinuities subjected to high in situ
stress conditions. 'Spor bolts located to stress conditions. Heavy bolts or doweis,
prevent laiiure of individual blocks and 'nciined to cioss rock structure, with mesh or
wedges. Balts muçt be tensioned. steel fibre reinforced shotcrete an raof and
sidewails.

Figure 14.1: Typical roikboit and dowei appiicationi ta iantrol different typei of iork mais failure during tunnel driving
 rse notes bl

bail

grout injection tube

faceplate drilled for tubes

Figure 14.2: Cornponenti of a rneihanicaly anchared rockboit with provisian for groutng

At the other end of the rockbolt from the anchor, a fixed head or threaded end and nut systern cai1 be
used. In either case, soine forin of faceplate is required to distribute the load from the bolt onto the rock face.
In addition, a tapered washer or corlicai seat is needed to compensate for the fact that the rode face is very
seldoin at riglit angles to the bolt. A wide variety of faceplates and tapered or domed washers are available
from roclcbolt suppliers.
In general, threads on roclcbolts sliould be as coarse as possible and should be rolled rather than cut.
A fine thread is easily dainaged aiid will cause installation problems in a typical underground environment.
A cut thread weakens the bolt and it is not unusual to see bolts with cut threads that have failed at the first
thread at the baclc of the nut. Unfortunately, rolled thread bolts are more expensive to manufacture and the
added cost tends to limit their applicatioil to situations where high strength bolts are required.
Tensioning of rockbolts is important to ensure that all of the components are in contact and that a
positive force is applied to the rode. In the case of light 'safety' bolts, the amount of tension applied is not
critical and tightening the nut with a conventional wrench ar with a pneumatic torque wrench is adequate.
Where tlie bolts are required to carry a significant load, it is generally recommended that a tension of
approximately 70% ofthe capacity ofthe bolt be installed iilitially. This provides a'lmown load with a reserve
in case of additional load being induced by displacements in the rock mass.
One of the primary causes of rockbolt failure is msting or corrosion and this can be counteracted by
filling the gap between the bolt and the drillhole wall with grout. While this is not required in temporary
support applications, grouting should be considered where the ground-water is lilcely to induce corrosion or
where the bolts are required to perform a 'permanent' support function.
The traditional inethod ofgrouting uphole rodcbolts is to use a short grout tube to feed the grout into
the hole and a sinaller diameter breather tube, extending to the end of the hole, to bleed the air iiom the
hole. The breather tube is generally taped to the bolt shank and this tends to cause problems because this
tube and its attachrnents can be damaged during transportation ar insertion iiito the hole. In addition, the
faceplate has to be drilled to accommodate the two tubes, as illustrated in Figure 14.2. Sealing the system for
grout injection can be a problem.
Many of these difficulties are overcome by using a hollow core bolt. While more expensive than
 Chapte 14 Rockbolts and cablí

conventional bolts, these hollow b o l ~ srnake the groutingprocess much more reliable and should be consid-
ered wherever perinanent roclíbolt installations are required. Tlie grout should be injected thsough a short
grout tube inserted into the collar of the Iiole and the central hole in the bolt should be used as a breather
tube. Wlieii installing these bolts iii downholes, the grout should be fed through the bolt to the end of the
hole and the short tube used as a breather tube.
Since the priinary purpose of grouting mechanically anchored bolts is to prevent corrosion and to
loclc the inechanical anchor in place, the strength requirement for the grout is notas imporrant as it is in tlie
case ofgrouted dowels os cables (to be discussed later). The grout should be readily puinpable without being
too fluid and a typical waterlcement ratio of 0.4 to 0.5 is a good starting point for a grout mix for this
application. It is inost important to ensure that the annular space between the bolt and the drillhole wall is
completely filled with grout. Pumping should be continued until there is a clear indication that the air has
stopped bleeding tlirough the breather tube or that grout is seen to retum through this tuhe.

14.2.2 Resin anchored rockbolts

Mechanically anchored roclíbolts have a tendency to work loose when subjected to vibratiotis due to
nearby blasting or when anchored in weak rock. Consequeiitly, for applications wliere it is essentiai that the
support load be mainrained, the use of resin áncbors should be considered.
A typical resin product is inade up of two component cartridges containing a resin and a catalyst in
separate coinpartinents, as shown in Figure 14.3. The cartridges are pushed to the end of the drillliole aliead
of the bolt rod that is then spun into the resin cartridges by the driil. Tiie plastic sheath of tlie cartridges 'is
broken and the resin and catalyst inixed by this spiniiing action. Setting of tlie resin occurs within a few
minutes (depending upon the specifications of the resin mix) and a very strong anchor is created.
This type of anchor will work in most rodís, including the wealí shales and mudstones in which
expansion shell anchors are not suitable. For 'permanent' applications, consideration should be given to the
use of fully resin-grouted roci<bolts,illiistrated in Figure 14.4. In these applications, a i~urnberof slow-setting
resin cartridges are inserted iiito tlie drillhole behind the fast-setting anchor cartridges.
Spinning the bolt rod through all of tliese cartridges initiates the chemical reaction iii all of the resins
but, because the slow-setting 'grout' cartridges are tiined to set in u p to 30 millutes, the bolt can be tensioned
within two or three minutes ofinstallation (after tlie fast anchor resin has set). Tliis tension is tliei~locked iii
by the later-setting grout cartridges and tlie resulting installation is a fully tensioned, fully grouted roci<bolt.
The high unit cost ofresin cartridges is offset by the speed ofinstallation. The process described above
results in a completely tensioned and grouted rockbolt installation in oiie operatioii, something that cannot
be matched by any other system currently on the mariíet. However, tliere are potential problems with resins.
Most resin/catalyst systems have a limited slielf life which, depending upon stosage teinperatures and
conditions, may be as short as six months. Purchase of the resin cartridges should be limited to the quantities
to be used within tlie shelf life. Care sliould be taken to store the boxes uiider conditions that conforin to tlie
manufacturer's recommendations. In critical applications,'it is good practice to test the activity of the resin
by sacrificing one cartridge from each box, before the contents are used underground. This cai1 be done by
brealíing the compartment separating the resin and catalyst by hand and, after inixing the coinporieiits,
measuring the set time to check whetiier this is within the manufacturer's specificatioiis.
Brealíing the plastic sheath of the cartridges and mixing the resins effectively can also present practical
probleins. Cutting the end of the bolt rod at an angle to forin a sharp tapered point will help in tliis process, but
the user should also be prepared to do some experimentatiou to achieve the best sesults. Note that tlie length of
time or the number of rotations for spinning the resins is limited. Once the setting process has been initiated,
the sticture of the resin can be damaged and the overall installation wealíened by additional spinning. Most
manufacturers supply instructions on the number of rotations or the length of time for spinning.
Rock Engineering. Coiirse notes I:

Figure 14.3: Typical t w o ~

iornponent resin cartridge
uied for anchoring and
grouting ro~kbolts.

Figure 14.4: Typcal iet-up

for creating a reiin anchored
and arouted r o i k b o t R e i n

char cartridgei and ipnning

the bolt rod through them all
to mix the resin and catalyst.
The bolt i i teniioned after
the fait-ietting anrhor reiin
has set and the ilow-setting
resin iets Iater to giout the
rod in lace.
 4 Rockbol ts and cabl~

In some wealz argillaceous rodes, the drillhole surfaces becoine clay-coated during drilling. This causes
slipping of th'e resin cartridges duri~ngrotation, resulting in incomplete mixing and an unsatisfi~ctorybond.
In highly fractured rock masses, the resin may seep into the surrounding rock before setting, leavitlg voids iii
the resin column surrounding the rocl<bolt. In both of these cases, the use of cement grouting rather than
resin grouting may provide a more effective solution.
There is some uncertainty about the long-term corrosion protection offered by resin grouts and also
about the reaction of some of these resins with aggressive groundwater. For temporary applications, these
concerns are probably not an issue because of the limited design life for most rockbolt installatio~is.How-
ever, where very long service life is required, current wisdom suggests that cement grouted bolts inay provide
better long term protection.

14.3 Dowels
14.3.1 Grouted dowels
Wlien conditions are such that installation of support can be carried out very close to an advancing
face, or in anticipation ofstress changes that will occur at a later excavation stage, dowels cai1 be used in place
of rodzbolts. The essential difference between these systems is that tensioned rocl<bolts apply a positive force
to the roclz, while dowels depend upon inovement in the rock to activate the reinforcing actioti. Mining
drawpoints, which are rnined before the overlying stopes are blasted, are good exainples of excavations wliere
untensioned grouted dowels will work well.
The simplest form of dowel in use today is the cemen't grouted dowel as illustrated in Figure 14.5. A thick
grout (typically a 0.3 to 0.35 waterlcement ratio grout) is pumped into the hole by inserting the grout tube to the
end of tlie hole and slowly withdrawing the tube as the grout is pumped in. Provided that a sufficiently viscous
grout is used, it will not niti out of the hole. The dowel is pushed into the hole about halfway and theii given a
slight bend before pushing it hlly into the hole. This bend will seme to keep the dowel firmly lodged in the hole
while the grout sets. Once the grout has set, a face plate and nut c& be fitted onto the end of the dowel and pulled
up tight. Placing this face place is important since, if the dowel is called on to react to displacemeilts in the rodz
mass, the rock close to the borehole collar will tend to pull away from the dowel unless restrained by a faceplate.
In mining drawpoints and ore-passes, the flow of brolcen rod< can cause serious abrasion and iinpact
probleins. Tbe projecting ends of grouted rebars can obstruct the flow of the roclz. Alternatively, the rebar can
be bent, brolcen or ripped out of the rock mass. In sucli cases, grouted flexihle cable, illustrated iil Figure 14.6,
can be used in place ofthe more rigid rebar. This will allow great flexibility with impact and abrasion resistance.
Older type grouted dowels such as the Scandinavian 'perfobolt' or dowels, wliere the grout is injected
after the rod has been inserted, tend not to be used. The installation is more complex and time consuining
and the end product does not perform any better than the simple grouted dowel described above.

14.3.2 Friction dowels or 'Split Set' stabilisers

Split Set stabilisers were originally developed by Scott (1976, 1983) and are manufactured and distrib-
uted by Ingersoll-Rand. The system, illustrated in Figure 14.7, consists of a slotted high streiigtli steel tube
and a face plate. It is installed by pushing it into a sligbtly undersized hole and the radial spring iõrce
getierated, by the compressiou of the C shaped tube, provides the frictional ancliorage along the entire
length of the hole. A list oftypical Split Set stabiliser dimensions and capacities is given in Table 14.1.
Because the system is quick and simple to install, it has gained rapid acceptance by miners throughout
the world. The device is particularly useful in mild rockburst environments, because it will slip rather than
rupture and, when used with mesh, will retain the broken rock generated by a mild burst. Provided that the
demand imposed on Split Sets stabilisers does not exceed their capacity, the system worlzs well and can be
considered for many mining applications. They are seldom used iii civil engineering applications.
Rock E n g i n

Figure 14.5: Grouted dawei

usng a deforrned bar inserted
n t o a grout-flled h a e

,,
required or were impact and
abrasion can cause probems
w ~ t hn g d support.

Figure 14.7: Split Set

stabilirer Ingerial-Rand
photogiaph.
 .Rockbolts and cables

Table 14.1: Solit Set soecifications (After Solit Set Division, lnaersol-Rand Comisanv).
Split Set stabiliser model 55-33 55-39 55-46
Recommended nominal bit size 31 t o 3 3 mm 3 5 t o 38 mm 41 t o 4 5 mm
Breaking capacity, average 10.9 tonnes 12.7 tonnes 16.3 tonnes
minimum 7 . 3 tonnes 9.1 tonnes 13.6 tonnes
Recommended initial anchorage (tonnes) 2.7 t o 5 . 4 2.7 t o 5.4 4.5 t o 8 . 2
Tube lengths 0.9 to 2.4 m 0.9 t o 3 . 0 m 0 . 9 t o 3.6 m
Nominal outer diameter of tube 3 3 mm 3 9 mm 46 mm
Domed plate sizes 150x1 50 mm 150x1 5 0 mm 150x1 50 mm
125x125 mm 125x125 mm
Galvanised system available yes Yes yes
Stainless steel model available no Yes no

Corrosion remains ooe of the prime problems with Split Set stabilisers since protectioii of the outer
surface of the dowel is not feasible. Galvanising the tube helps to reduce corrosion, but is probably not a
preventative ineasure which can be relied upon for long term applications in aggressive environments.

14.3.3 'Swellex' dowel

Developed and marlceted by Atlas Copco, the 'Swellex' system is illustrated in Figure 14.8. Tlie dowel,
which may be up to 12 in long, consists of a 42 mm diameter tube which is folded during manufacture to create
a 25 to 28 mm diameter unir which can be inserted into a 32 to 39 mm diameter hole. No pushing force is
required during insertion and the dowel is activated by injection of high pressure water (approximately 30 MPa
or 4,300 psi) which inflates the folded tube into intimate contact with the walls of the borehole.
During 1993 the original Swellex dowel was replaced by the EXL Swellex which is manufactured fioin
a higii strength but ductile steel. This steel allows significant displaceinent without loss of capacity. Stillborg
(1994), carried out a series of tests in wliich bolts and dowels were installed across a simulated 'joint' and
subjected to tensile loading. 111the EXL Swellex dowel tests, opening of the joint concentrates loading onto
the portion ofthe dowel crossiug the joint, causing a reduction in diameter and a progressive 'de-bondinp' of
the dowel away from the joint. The ductile characteristics of the steel allows the de-bonded section to deform
under constant load until, eventually. failure occurs when the total displacement reaciies about 140 mrn a t a
constant load of approximately 11 tonnes. These tests are described in greater detail later in this Chapter.
Corrosion o i Swellex dowels is a matter of concern since the outer surface of the tube is in direct
contact with the roclc. Atlas Copco have worked with coating manufacturers to overcome this problem aild
claiin to have developed effective corrosion resistant coatings.
Speed of installation is the principal advantage of the Swellex system as compared with conventional
rockbolts and cerneut grouted dowels. In fact, the total installation cost ofSwellex dowels or Spilt Set stabilisers
tends to be less than that of alternative reinforcement systems, when installation time is talcen into account.
Both systems are ideal for use with automated rockbolters.

14.4 Load-deformation characteristics

Stillborg (1994) carried out a uumber of tests o n rockbolts and dowels installed across a simulated
'joint', using two blocks of high strength reinforced concrete. This type oftest gives a more accurate represen-
tation of conditions encountered underground than does a standard 'pull-out' test.
 Rock Engir

Figure 14.8: Atlas Capco 'Swelex' dowel

 Chapter 111 Rockbolt!i and cables

The rod<boltsand dowels tested were installed in percussion drilled holes using the installation techniques
used in a normal underground inining operation. The installed support systems were then tested by pulling the
two bloclcz of concrete apart at a fixed rate and measuring the displacement across the simulated 'joint:
The results of Stillborg's tests are summarised in Figure 14.9 which gives load deformation curves
for all the bolts and dowels tested. The configuration used in each test and the results obtained are
summarised below:
1. Expansion shell anchored rockbolt
Steel rod diaineter: 17.28 min
Ultimate tensile strength of bolt shank: approximately 12.7 tonnes
Expansion shell anchor: Bail type three wedge anchor
At the pre-load of 2.25 tonnes, n o deformation of the face plate.
At a load of 4 tonnes, the face plate has deformed 9.5 mm and is coinpletely flat, the bolt shank has
defornied an additional 3.5 m m giving a total deformation of 13 mm at 4 tonnes load.
Failure initiates at a load of 8 tonnes and a deformation of 25 mm with progressive failure of the expan-
sion shell anchor in which the cone is pulled through the wedge.
Maximum load is 9 tonnes at a deformation of 35 mm.

2. Cement grouted steel rebar

Steel bar diameter: 20 mm
Ultimate tensile strength of steel rebar: 18 tonnes
Faceplate: flat plate
Borehole diameter: 32 inm
Cement grout: 0.35 watertceinent ratio grout cured for 11 days
A t a load of 15 tonnes and an elastic deformation of about 1.5 mm, a sudden load drop is characteristic
of hot rolled rebar steel.
Maximum load is 18 tonnes at a deformation of 30 mm.

3. Resin grouted steel rebar

Steel rebar diameter: 20 mm
Ultimate tensile strength ofsteel rebar: 18 tonnes
Faceplate: flat plate
Borehole diameter: 32 mm
Resin grout: Five 580 mm long, 27 mm diameter polyester resin cartridges. Curing time 60 ininutes.
Mixed hy rotating rebar through cartridges in the borehole
A t a load of 15 tonnes and an elastic deformation of about 1.5 mm, a sudden load drop is characteristic
of hot rolled rebar steel.
Maximum load is 18 tonnes at a deformation of 20 mm
The resin is stronger than the cement grout aud local fracturing and bond failure in and near the joint is
limited as compared with the cement grouted rebar, leading to a reduced ultimate displaceinent at rebar
failure.

4. Resin grouted fibreglass rod

Fibreglass rod diameter: 22 mm
Ultimate tensile strength of fibreglass rod: 35 tonnes
Facepiate: special design by H. Weidmann AG. Switzerland (see margin drawing - after Stillborg)
Borehole diameter: 32 mm
 Resin grout: Five 580 min long, 27 mm diameter polyester resin cartridges. Curiug time 60 minutes.
Mixed by rotating fibreglass rod through cartridges in the borehole
At approximately 1.5 tonnes load, failure of the fibreglasslresin interface initiates and starts progressing
along the rod. As bond failure progresses, the fiherglass rod deforms over a progressively longer 'free' length.
General bond failure occurs at a load of approximately 26 tonnes and a deformation of 25 min.
The ultimate capacity of this assembly is determined by the bond strength between the resin and the
fibreglass rod and by the relatively low frictional resistance of the fibreglass.

5. Split Set stabiliser, type SS 39

Tube diameter: 39 mm
Ultimate tensile strength of steel tube: 11 tonnes
Faceplate: special design by manufacturer (see Figure 12.8)
Borehole diameter: 37 m m
Dowel starts to slide at approximately 5 tonnes and inaintaiiis this load for the duration of the test which,
in this case, was to a total displacement of 150 mm

6. EXL Swellex dowel

Tube diameter: 26 mm before expansion
Ultimate tensile strength of steel tube: 11.5 tonnes (before expansion)
Type of face plate: Domed plate (see rnargin drawing - after Stillborg)
Borehole diaineter: 37 min
Pump pressure for expansion of dowel: 30 MPa
At 5 tonnes load tlie dowel starts to deform locally at the joint and, at the same time, 'bond' iàilure
occurs at the joint and progresses outward from the joint as the load is increased. General 'bond' failure
occurs at 11.5 tonnes a t a deformation of approximately 10 mm. The dowel starts to slide at tbis load and
maintains the load for the duration of the test which, in this case, was to 150 mm.

14.5 Cables
A coinprehensive review of cable support in underground mining has been given in a boolí by
Hutchinson and Diederichs (1996). This boolí is highly recommended for anyone who is concerned with tlie
selection and installation of cable support for either inining or civil engineering applications.
Some of the main cable types used by mining have been summarised by Windsor (1992) and are
illustrated in Figure 14.10.

14.5.1 Bond strength

The forces and displacements associated with a stressed cable grouted into a borehole in rod< are
illustrated in Figure 14.11.
As the cable pulls out of tbe grout, the resultant interfeience of the spiral steel wires with their associ-
ated grout imprints or flutes causes radial displacement or dilation oftheinterface between the grout and the
cable. The radial dilation induces a confining pressure that is proportional to the combined stiffiiless of the
grout and the rock surrounding the borehole. Tbe shear stress, which resists sliding of the cable, is a product
of the confining pressure and the coefficient of friction between the steel wires and tbe grout. Shear strength,
therefore, increases with higher grout strength, increases in the grout and the rock stiffness and increases in
the confining stresses in the rodc after installation of the cable. Conversely, decrease in shear strength can be
expected if any of these fdctors decrease or if the grout crushes.
Theoretical inodels of the behaviour of this rock/grout/cable systein have been developed by Yazici
 4 Rockbolts and cable

and Kaiser (1992), Kaiser et a! (1992), Hyett et al, (1992). The second of these models bas been incorporated
into the program PHASE2.

14.5.2 Grouts a n d g r o u t i n g
The question of grout quality has always been a matter of concern in reinforcement systems for
underground construction. One of the critical factors in this matter has been the evolution of grout pumps
capable ofpumping grouts with a low enough watedcement ratio (by weight) to achieve adequate strengtlis.
Fortunately, this problem has now been overcoine arid there is a range of grout pumps o11 the inarket which
will pump very viscous grouts and will operate reliably under typical underground conditions.
The results of a comprehensive testing prograinme on Portland ceinent grouts have been sumrnarised
by Hyett et al (1992) and Figures 14.12, and 14.13 are based upon this summary. Figure 14.12 shows the
decrease in hoth 28 day uniaxial compressive strength and deformation modulus with increasing water/
cement ratio. Figure 14.13 gives Mohr failure envelopes for three watedcement ratios.
These results show that the properties of grouts with watedcement ratios of 0.35 to 0.4 are sigriifi-
cantly better than those with ratios in excess of 0.5. However, Hyett et a1 found that the scatter in test results
increased marlcedly for watedcement ratios less than 0.35. The implication is that the ideal watedcement
ratio for use with cahle reinforcement lies in the range of 0.35 to 0.4.
The cliaracteristics of grouts with different watedcement ratios are descrihed as follows (after Hyett et
al, 1992):

wlc ratio Characteristics at end of grout hose Characteristics when handled

c 0.30 Dry, stiff sausage structure. Sausage fractures when bent. Grout
too dry t o stick to hand. Can be rolled
into balls.
0.30 Moist sausage structure. Sausage is fully flexible. Grout w i l
'Melts' slightly with time stick t o hand. Easily rolled into wet,
soft balls.
0.35 Wet sausage structure. Grout sticks readily t o hand.
Structure 'melts' away with time. Hangs from hand when upturiied.
0.4 Sausage structure lost irnmediately. Grout readily sticks t o hand but can
Flows viscously under its own weight be shaken free.
t o form pancake.
0.5 Grout flows readily and splashes Grout will drip from hand - no
on i m ~ a cwith
t around. shakina reauired.

14.5.3 Cable installation

The left hand drawing in Figure 14.14 shows the traditional method of grouting a cable iii an uphole.
Tliis rnethod will be called tbe 'breather tube inethod: The grout, usually having a water/cernent ratio > 0.4,
is injected into the bottoin of the hole through a large diameter tube, typically 19 min diameter. Thc air is
bled through a smaller diameter tuhe which extends to the end of the hole and which is taped onto tlie cahle.
Both tubes and the cable are sealed into the bottorn of the hole by means o f a plug of cotton waste or of
quiclc setting mortar. As shown, the direction of grout trave1 is upwards in the hole and this tends to favour
a grout coluinn which is devoid of air gaps since any slump in the grout tends to fill these gaps.
 Rock Engi iurse notes

TYPE LONGITUDINAL SECTION CROSS SECTION

Multi-wire tendon
(Clifford. 1974)
o 0

Birdcaged multi-
wire tendon I
(Jirovec. 1978)
Antinode Node

Single strand
(Hunt & Askew,
1977) / 1

i984) Doubie-acting twin anchor Single anchor 3 component 2 component

Swaged anchor on
strand
(Schmuck. 1979)
Sauare Circular

nign wpacty Stee !-oe

snear aowe
,IAainews e1 a
1S86i Concrete

I
I
Biracagea sirana O O
:hLtcrtns
...
ei a =;c
1YYU)
Antinode Node

.?:0
Bulbed strand O o @
(Garford. 1990) O

Antinode Node

Ferwled strand @
(Windsor. 1990)
Antinode Node

Figure 14.10: Surnrnaiy of the deveopment af cable relnforcing systerns for underground mining (Windior (1992)).
 ts and cabli

Confining pressure

6. 16 D e f o d o n moduius - 80
Radial displacement O
4
-60 f,
.-

Shear resistance
e Uniaia1 compressive Y
a
Confining pressure
3

Radial displacernent

V- Tensile force

Figure 14.11: Forcei and dirplaiernenti aisoclated Figure 14.12: Relationshipbetween rhewaterlcernentratia and the aveiage uniaxal
with astresied iabie grouted intoa bareholein rack. compressive i!iength an deforrnation rnodulus for grouts testes at 28 dayi.

100,

Water cemmt ratios of

cem& gmut mixes

Figure 14.13: Mohr faiure enueiopes for the peak itrength of grours with different watericement ratios. teited ar 28 days.

wlc ratio g MPa constant m constants Friction angle @" Cohesion c MPa
0.32 78 3.05 1 24 25
0.41 54 2.14 1 20 19
0.52 38 1.67 1. 17 14
 Rock Engin~2erins. Course notes b y Evert Hoe

Apart from the difficulty ofsealing the collar of the hole, the main problem with this system is that it
is difficult to detect when the hole is full of grout. Typically, tlie hole is judged to he full when air ceases to
flow froin the bleed tube. This may occur preinaturely if air is vented into an open joint along the hole. In
addition, a void the size of the bleed tube is likely to be left in the grout column. Therefore, it is preferable
to stop groutitig the borehole oiily when grout returns along the bleed tube. However, a viscous grout will
not flow down a 9 mm bleed tube and so a larger tube is required.
An alternative inethod, called the 'grout tube method' is illustrated in the right hand drawing in
Figure 14.14. In this case a large diameter grout injection tube extends to the end of the hole and is taped
onto the cahle. The cable and tuhe are held in place in tlie Iiole by a wooden wedge inserted into the hole
collar. Note that care has to be talcen to avoid compressing the grout tube between the wedge aud the cable.
Grout is injected to the top of the hole and is pumped down the hole until it appears at the hole collar. If a
watery grout appears iirst at the collar of the hole, grout pumping is continued untii a consistently thick
rrout is ohserved.

alinost certaiiily result in air voids iti the grout column as a result of slumping of the grout. n i e principal advan-
tage of this inethod is that it is fairly obvious when the hole is h l l of grout and this, together with the sinaller
number of coinponents required, malces the method attractive when compared witli the traditioiial method for
groutitig plaiii straiid cables. In addition, the thidcer grout used i11 this inethod is not lilcely to flow into fractures
in tlie rodc, preferring iiistead the path of least flow resistance towards the borehole collar.
The procedure used for grouting downholes is similar to the grout tube method, described above,
without the wooden wedge iii the boreliole collar. The grout tube may be taped to the cable or retracted
slowly from the bottom of the hole as grouting progresses. It is important to ensure that the withdrawal rate
does not exceed the rate of filling the hole so the air voids are not introduced. This is achieved by applying,
by hand, a sliglit downward force to resist the upward force applied to the tube by the rising grout coluinn.
Grout of-anyconsistency is suitable for this method but the best range for plain strand cables is between 0.3
and 0.4 watedcement ratio.
Modified cables, such as hirdcage, ferruled or bulbed strand, should be grouted using a 0.4 waterl
cement ratio inix to ensure that the grout is fluid enough to fill the cage stmcture of these cables. Therefore,
the breather tube inethod must be used for these types ofcables, since the grout flow characteristics required
hy the grout tube method is liinited to grouts in the range of 0.3 to 0.35 water/cement ratio.
One of the most critical components in a cable installation is the grout column. Every possible care
must be talcen to ensure that the coluinn contains as few air voids as possible. In the breather tube inethod, a
large diaineter hreather tube will allow the return of grout as well as air. Wheii using the grout tube method in
uplioles, a 0.3 to 0.35 watedceinent ration grout will ensure that pumping is required to cause the grout column
to flow, and this will avoid slumping of the grout in tlie borehole. A grout witli a water/ceinent ratio of less than
0.3 sbould be avoided, since it will tend to form encapsulated air voids as it flows around the cable.
A liollow cable, illustrated in Figure 14.15, has been introduced by Atlas Copco and this could reduce
some of tlie grouting prohlerns discussed above.

14.5.4Cables for slope reinforcement

Most of the applications described in this chapter have been related to underground excavations.
However, under certain circuinstances, it may also be necessary to reinforce slopes and foundations aiid
cables have proved to he very effective in such applications.
Figure 14.16 illustrates a unit set up for drilling 140 in loiig 50 mm diameter holes for the installation
of cables, illustrated in Figure 14.17, in a slope.
 4 Rockbolts and cable

Figure 14.14: Alternative

methods for groutng cabes
in upholes.

Figure 14.15: Hollow cahle

by Atlas Copco.
Rock E n y i neenny Coiirse notes b

Figure 14.16: Driilng

rnachne foi the installation
of 40 rn long reinforcing
iables in 150 rnm diarnetei
holei in a dam excavation

being prepared for

installation n a dam
evcavation.
 14. Rockbolts and cabl~

These cables were installed to stabilise the slopes of a dam foundation in gneiss. Sheet jointing parallel
to the surface of the steep slopes would have resulted in large scale slope instability if the excavation, wliich
undercut these sheet joints, had not been reinforced.
The cables illustrated liave an ultimate capaciry of 312 tons and a working load of 200 tons. The
cables were fully grouted after tensioning. The cost of materials and installation for these cables was approxi-
mately US$ 500 per metre.
 Shotcrete support

15.1 lntroduction
The use of shotcrete for the support of underground excavations was pioneered by the civil engineer-
ing industry. Reviews of the development of shotcrete technology have been presented by Rose (1985),
Morgaii (1992) and Franzén (1992). Rabcewicz (1969) was largely responsible for the introduction of the use
of shotcrete for tunnel support in the 1930s, and for the development of the New Austrian Tunnelling
Method for excavating in weak ground.
In recent years the mining industry has become a major user of sliotcrete for underground support. It
can be expected to make its own contributions to this field as it has in other areas of undergroui~dsupport.
The siinultaneous working of multiple headings, difficulty of access and unusual loading conditions are
soine of the probleins which are peculiar to underground inining and which require new and innovative
applications of shotcrete technology.
An important area of shotcrete application in uiiderground inining is in the support of 'perinanent'
openings such as ramps, haulages, shaft stations and crusher chambers. Rehabilitation of conventional rod<bolt
and inesh support cai1 be very dismptive and expensive. Increasing numbers of these excavations are being
shotcreted immediateiy after excavation. The incorporation of steel fibre reinforcement into the shotcrete is an
irnportant factor in this escalating use, since it minimises the labour intensive process of mesh installation.
Recent trials and obsemations suggest that shotcrete can provide effective support in mild rockburst
conditions (McCreath and Kaiser, 1992, Langille and Burtney, 1992). While the results froin these studies are
still too liinited to perinit definite conclusions to be drawn, the indications are encouraging enough that
more serious attention will probably be paid to this application in the future.

15.2 Shotrete technology

Shotcrete is the generic name for cement, sand and tine aggregate concretes which are applied pneu-
rnatically and cornpacted dynamically under high velocity.

15.2.1 Dry mix shotcrete

As illustrated in Figure 15.1, the dry shotcrete components, whidi may he slightly pre-dampei~edto
reduce dust, are fed into a hopper with continuous agitation. ,Compressed air is introduced through a rotat-
ing barrel or feed bowl to coiivey the inaterials in a continuous stream through tlie delivery hose. Water is
added to the inix at the nozzle. Gunite, a proprietary name for dry-sprayed inortar used in the early 1900's,
has fallen into disuse in favour of the more general term shotcrete.
 - 15 Shotcrete suppori

shotcrete rnix
.->

cornpressed air
water iniection J

Figure 15.1:Simpified rketch a i a typical dry mxshotcrete system. After Mahar et ai (1975).

tube to normal shape

rubber nolzle tip

wet rnix shotcrete

'purnping tube

Figure 15.2: One typical type of wet mix shotcrete machine. After Mahai et al (1975).

15.2.2 Wet mix shotcrete

In this case the shotcrete components and the water are mixed (usually in a truck mouiited inixer)
before delivery into a positive displacement pumping unit, which then delivers the mix liydraulicaily to the
nozzle where air is added to project the material onto the rock surface.
The final product of either the dry or wet shotcrete process is very similar. The dry mix system tends
to be more widely used in rnining, because of inaccessibility for large transit mix truclis and because it
generally uses smaller and more compact equipment. This can be moved around relativeiy easily in an
underground mine environment. The wet mix system is ideal for high production applications in mining
and civil engineering, where a deep shaft or long tunnel is being driven and where access allows the applica-
tion equipment and delivery truclis to operate on a more or less continuous basis. Decisions to use the dry or
wet rnix shotcrete process are usually made on a site-by-site basis.
 Rock Engineering Co[irse notes b

15.2.3 Steel fibre reinforced micro silica shotcrete

O f t h e many developments in shotcrete technology in recent years, two of the most significant were
the introduction of silica fume, used as a cementitious admixture, and steel tihre reinforcement.
Silica fbme or micro silica is a by-product of the ferro silicon metal industiy and is an extremely fine
pozzolan. Pozzolans are ceinentitious materials which react with the calcium hydroxide produced during
cement hydration. Silica fuine, added in quantities of 8 to 13% hy weight of cement, can allow shotcrete to
achieve compressive strengths which are double or triple the value ofplain shotcrete mixes. The result is an
extremely strong, iinpermeahle and durable shotcrete. Other benefits include reduced rebound, improved
flexural strength, improved hond with the rock mass and the ability to place layers of u p to 200 mm thidc in
a single pass hecause of the shotcrete's 'stickiness'. However, when using wet mix shotcrete, this stickiness
decreases the worl<ability of the material and superplaticizers are required to restore tliis workability.
Steel fibre reinforced shotcrete was introduced in the 1970s and has since gained world-wide acceptaiice
as a replacement for traditiotial wire inesh rejnforced plain shotcrete. The main role that reinforceinent plays i11
shotcrere is to impart ductility to an othenvise brittle material. As pointed out earlier, rock support is only called
upon to carry sigtlificant loads once the roclc surrounding an underground excavation deforms. This ineans that
unevenly distributed non-elastic deformations of significant magnitude may overload and lead to failure of tlie
support system, unless that system has sufficient ductility to accommodate these deformatioils.
Typical steel fihre reinforced, silica fume shotcrete rnix designs are summarised in Table 15.1. These inixes
can he used as a starting point when embarking on a shotcreteprogramme, but it may be necessary to seek expert
assistance to 'fine tune' the mix desigi~sto suit site specific requirements. For many dry rnix applications it may he
advantageous to purchase pre-mixed shotcrete in bags of up to 1,500 kg capacity, as illustrated in Figure 15.3.
Figure 15.4 shows the steel fihre types which are currently available on the north American inarl<et.In
addition to their use in shotcrete, these fibres are also widely used in concrete floor slabs for buildings, in
airport runways and in similar concrete applications.
Wood et al(1993) have reported the results of a comp~ehensivecomparative study in which ai1 of the
fibres shown in Figure 15.4 were used to reiilforce shotcrete samples, which were then subjected to a range of
tests. Plain and fibre reinforced silica fume shotcrete samples were prepared by shooting onto vertical panels;
using hoth wet and dry mix processes. The fibre reinforced samples all contained the same steel fibre dosage
of 60 I<g/m3(see Table 15.1). A11 the sainples were cured under controlled relative humidity conditions aild
all were tested seven days after shooting.

Table 15.1: Ty~icalsteel fibre reinforced silica fume shotcrete mix desians (After Wood. 1992).
Components Dry rnix Wet mix
kg.lm3 % dry ' kg.im3 % wet
materials materiais
Cement 420 19.0 420 18.1
Silica fume addittve 50 2.2 40 1.7
Blended aggregate 1,670 75.5 1,600 68.9
Steel fibres 60 2.7 60 2.6
Accelerator 13 0.6 13 0.6
Superplasticizer 6 litres 0.3
Water reducer 2 litres O. 1
Air entratning admixture if required
Water controlled at nozzle 180 7.7
Total 2.213 100 2.321 1 O0
 Figure 15.3: Bagged pre-
mixed dry shotcrete
cornponents being delivered
nro a hopper feeding a
screw conveyor fitted with
a pre-darnpener, which
dischargei into the happer
of a ihotcrete rnaihine

These tests showed that the addition of steel fihres to silica &me shotcrete enhances both the com-
pressive and flexural strength ofthe hardened shotcrete by up to 20%. A significant increasc in ductility was
also obtained in all the tests on fibre reinforced samples, coinpared with plain sarnples. Wbile different fihres
gave different degrees of improvement, all of the fibres tested were fouud to exceed the levels ofperformance
commonly speciiied in north America (i.e. 7-day coinpressive strength of 30 MPa for dry inix, 25 MPa for
wet mix aud 7-day flexural strength of 4 MPa).
Kompen (1989) carried out bending tests on slabs of unreinforced shotcrete and shotcrete reiiiforced
with 'Dramix" steel fibres, shown in Figure 15.5. The shotcrete had an uiiconfined compressiue strengtli,
determined from tests on cuhes, of 50 MPa. The results of these tests are reproduced iii Figure 15.6. The pealc
strength of these slahs iiicreased hy approximately 85% and 185% for 1.0 2nd 1.5 volume Oo/ of fibres,
respectively. The ductility of the fihre reinforced slabs iiicreased by approxiinately 20 and 30 times for the 1.0
and 1.5 volume % of fibres, respectively.

' Manufactured by N V Bekaert 5 A . 8-8550 Zwevegern. Beiglum

 ieering COI

Figure 15.4. iteel fibre typei available an the north Arneriian rnarket. After Wood et a (1993) (Note: ali dirnensions are in mm)

Figure 15.5: 'Dramix' steel fibres used In slab bending tests by Kornpen (1989). Thefibres are glued together in bundlei w t h a water soluble glue
ta facilitate handling and hornogeneous distribution of the hbres in the shotiieie.

I I 1 I I
o 10 20 30 . 40 50 60
Deflection, d (mm)

Figure 15.6: Load deflertion curves for unreinforied and steel fibre reinfoned shotcrete slabs tested in bending. After Kornpen (1989).
 5 Shotcrete support

15.2.4 Mesh reinforced shotcrete

While steel fibre reinforced shotcrete has been widely accepted in both civil and mining engineering,
mesh reinforced shotcrete is still widely used and is preferred in some applicatioiis. In very poor quality,
loose rock inasses, where adhesion of tlie shotcrete to the rodisurface is poor, the mesh provides a significaiit
amount of reinforcement, even without shotcrete. Therefore, when stabilising slopes in very poor quality
rock masses a r when building bulliheads for underground fill, weldmesh is frequently used to stabilise the
surface or to provide reinforcernent. In such cases, plain shotcrete is applied later to provide additional
support and to protect the mesh against corrosioii.
Kirsten (1992, 1993) carried out a comprehensive set of laboratory hending tests o n hoth inesh and
fibre reinforced shotcrete slabs. The load versus deflection curves, that he obtained, were siinilar to those
ieported by Koinpen, reproduced in Figure 15.6. He found that tlie load carrying capacity of the inesli aiid
fihre reiiiforced shotcrete samples were not significantly different, hut that the mesh reinforced sainples were
superior in bending witli both point loads and uniformly distributed loads. H e concluded that this was due
to the more favourable locatioii of the mesh reiiiforcement in the slabs suhjected to hending.
Kinten also concluded that the quality control, required to ohtain a consistent dosage aiid uniforin distri-
hution of fibres in shotcrete, is more easily achieved in civil engineering thati in mining applications. This is a
reflection of the multiple worlcing headings and the dificulties of access that are conimon problems associated
with inany iniiies. Under these circumstances, more reliahle reinforcement will he obtaiiied witli inesh reiiiforced
rather than fibre reinforced shotcrete. However, in large mines, in wliich many of the 'perinanent' openiiigs are
siinilar to those on large civil engineering sites, these problems of quality control should not arise.
Chaiiilinli mesh, used iii many underground mining excavations to support loose rock, is not usually
suitable for shotcrete reinforceinent. This is hecause penetration of the shotcrete is iiiliihited by tlie twisted
joints as illustrated in Figure 15.7. This allows air cavities to form hehiiid the mesh aiid these inay allow water
to enter and cause corrosion of the mesh.
O n the other hand, weldinesh, tiglitly pinned against the rock face as illuatrated in Figure 15.8, is
generally ideal for shotcrete applicatioiis. Typically the weldmesh should he made from 4 min dianieter wire
welded into a 100 mm x 100 inm grid. Tliis type of mesh is strong eiiougli for inost undergroutid applica-
tions aiid tlie sheets are light enough to he handled by one man.

15.3 Shotcrete applications

The quality of the final shotcrete product is closely related to the application procedures used. These proce-
dures include: surface preparation, nozzling technique, lightiiig, ventilatioii, communications, and crew training.
Shotcrete should not be applied diiectly to a dry, dusty or frozeii rodi surface. n i e work area is usually
sprayed with ai1 air-water jet to remove loose ro& and dust from the surfice to be shot. The damp rodc will create
a good surface oii which to bond the initial layer ofshotcrete paste. The nozzleman coinmonly starts low on the
wall and moves the nozzle iii small circles working his way up towards the baclc, a r roof Care inust be taken to
avoid appiying fresh tnaterials on top of rebound a r oversprayed shotcrete. It is essential that the air supply is
consistent and has sufficient capacity to ensure the delivery ofa steady streain ofhigh velocity shotcrete to the rodi
iace. Shooting distantes are ideally ahout 1 to 1.5 metres. Holding the nozzle further from the rodi face will result
i11 a lower velocity flow ofmaterials which leads to poor compaction and a iiigher proportioti of rebound.
A well-trained operator can produce excellent quality shotcrete manually, when tlie work area is well-
lit and well-ventilated, and when the crew members are in good coinmunication witli eacli other using
prescribed hand signals a r voice activated FM radio headsets. However, this is a very tiring and uiicoinfort-
able job, especially for overhead shooting, and compact robotic systeins are increasingl~being used to
peiinit the operator to control the nozzle remotely. Typical rohotic spray booins, used for shotcrete applica-
tion in uiiderground excavations, are illustrated in Figures 15.9, 15.10 and 15.11.
Rock Engi n e e r i n g C<j u r s e notes iek

Figure 15.7: Chainink

meih. w h l e very itrong and
iiexibe. is not ideal ior
shatiiete appication
beiause t i i difficult
for the shotirete to
,ienetrate the mesh

Figure 15.8: Weided wire

rneih, firmly attached
to the roik suriace. piovidei
enielient reinforcement
for ihotcrete.
 Chapter 15 Shotcre te support

Figure 15.9: A truck

rnounted ihotcrete rohof
heing uied in a arge CIVII
engineering tunnel
Note that the distance
between the norzle an<l
the rock surface i
approximatey one metre

Figure 15.10: Carnpacl

 Rock E n g i rieering: Coiirse notes

Figure 15.11: Shotrrete aperator uiing a remotey contiolled Figure 15.12: Piaitic pipes uied to provde dranage for a iliatirete
u n t ra apply ihotcrete to a iock face in a large civil engineer- layer appled to a rack masi w t h water-bearing joints.
ing excauation.

When shotcrete is applied to rod<masseswith well-defined water-hearing joints, it is iinportant to provide

drainage through tlie shotcrete layer in arder to relieve high water pressures. Draiii holes, fitted with plastic pipes
as illustrated in Figure 15.12, are commonly used for this purpose. Where the water inflow is not restncted to a few
specific features, a porous fibre mat can he attached to the rod<surface before the shotcrete layer is applied. When
practical to doso, the water from these drains should be collected and directed into a drainage ditch or suinp.

15.4 Design of shotcrete support

The design of shotcrete support for underground excavations is a veiy imprecise process. However,
one observation, which is coininonly made by practical engineen with years ofexperience in using sliotcrete
underground, is that it almost always performs bettei than anticipated. There are many examples (veiy few of
which are documented) where shotcrete has been used as a last act of despeiation in ai1 effort to stabilise the
failing rod< around a tunnel and, to inost people's surprise, it has worI<ed.
The complex interaction between the failing rock mass around an underground opening, and a layer
of shotcrete ofvarying thidcness with properties which change as it hardens, defies inost attempts at theoreti-
cal analysis. It is only in recent years, with the development of powerful nuinerical tools, that it has heen
possihle to contemplate realistic analyses, which will explore the possihle support-interaction behaviour of
shotcrete. A clear understanding of shotcrete behaviour wili require many more years of experience in the
use of and in the interpretation of the results obtained from these programs. It is also important to recognise
 te support

that shotcrete is very seldorn used alone and its use iti combination with rocldiolts, cablebolts, lattice girders
or steel sets further complicates the problem of analysing its contrihution to support.
Current shotcrete support 'desigti' methodology relies very heavily upon niles of thurnb and prece-
dent experience. Wickham et al (1972) related the thickness of a shotcrete tuilnel lining to their Rock Struc-
ture Rating (RSR). Bieniawski (1989) gave recornmendations on shotcrete thicl<nesses (in conj~~iiction with
rockbolts or steel sets) for different Rode Mass Ratings (RMR) for a 10 m span opening. Gritnstad and Barton
(1993) have published an updated relating different support systems, including shotcrete and fibre reiti-
forced shotcrete, to the Tunnellitig Quality Index a
Vandewalle (1990) collected various rules of thumb
from a variety of sources and included them in his monograph.
Tdbie 15.2 is a compilation of current shotcrete practice by the present author, combining ali of these
empirical rules and adding i11iny own practical experience. The reader is warned, that this table cai1 only he
used as an approximate guide when deciding upon the type and thickness of shotcrete to be applied in a
specific application. Modifications will alrnost certainly be required to deal with local variations in rock
conditions and shotcrete quality.

Table 15.2: Summaiyof recommended shotcrete a~plicationsin underqround rnininq. for different rock mass conditons.
Rock mass
descriution I Rock mass
behaviour
I ~upport
reauirements
I Shotcrete application

Massive No spalliny, None None.

metarnorphic or slabbing or failure.
igneous rock.
Low stress conditions.
Massive Surfaces of some Sealing surface Apply 25 rnrn thckness of
sedimentary rock. shalei, siltstones. or to prevent slakiny plain shotcrete to permanent
LOW stress claystones rnay slake surfaces as soon as possible
conditions. as a result of rnoisture after excavation. Remir shotcrete
content change. damage due to blastiny.
Massive rock with Fault gouge may Provision of support Remove weak material to a
single wide fault or be weak and erodible and surface sealing in depth equal to width of f a u t
shear zone. and mav cause vicinitv of weak fault or shear zone and arout rebar
stabi~it~~roblems in of shiar zone into adjacent sound rock.
adjacent jointed rock. Weldrnesh can be used if
required to provide temporary
rockfall support Fill void with
plain shotcrete. Extend steel
fibre reinforced shotcrete laterally
for at least width of gouge zone.
Massive Surface slahbing, Retention of broken Apply 50 mm shotcrete over
metarnorphic spalling and rock and control of weldmesh anchored behnd
or igneous rock. possible rockburst rock mass dilation. bolt faceplates, or apply 50 mm
High stresi damage. of steel fibre reinforced shotcrete
conditions. on rock and install rockbolts with
faceplates; then apply second
25 mm shotcrete layer.
Extend shotcrete application
down sidewalls where reauired.
 neering Co

Tabie 15.2: (cont'd) Surnrnary of recommended shotcrete applications in underground rnining, for different rock
rnass conditions
Rock rnass Rock mas8 Support Shotcrete application
description behaviour requirernents
Massive Surface slabbing, Retention of broken Apply 75 rnm layer of fibre
sedimentary rock. spalling and possible rock and control of reinforced shotcrete directly
High stress squeezing in shales squeezing. o n clean rock.
conditions. and soft rocks. Rockbolts or dowels are alio
needed for additional support.
Metamorphic or Potential for wedges Provision of support Apply 50 rnm ofsteel fibre
igneous rock with or blocks to fali in addition t o that reinforced shotcrete t o rock
a few widely or slide due to available frorn surfaces on which joint traces
spaced joints. gravity loading. . rockbolts or'cables. are exposed.
Low stress
conditions.
Sedimentary rock Potential for wedges Provision of support Apply 50 mrn of steel fibre
with a few widely or blocks to fali in addition t o that reinforced shotcrete on rock
spaced bedding or slide due to available frorn surface o n which discontinuity
planes and joints. gravity loading. rockbolts or cables. traces are exposed. with
Low stress Bedding plane Sealing of weak particular attention to
conditions. exposures rnay bedding plane bedding plane traces.
deteriorate in time. exposures.
Jointed Cornbined structural Retention of broken Apply 75 rnrn plain shotcrete
rnetamorphic or and stress controlled rock and control over weldmesh anchored
igneous rock. failures around of rock mas8 dilation. behind bolt faceplates or
High stress opening boundary. apply 75 m m of steel fibre
conditions. reinforced shotcrete on rock,
install rockbolts with faceplates
and then apply second 25 rnm
shotcrete layer
Thicker shotcrete layers may
be required at high stress
concentrations.
Bedded and Slabbing, spalling Control of rock mas8 Apply 75 rnrn of steel fibre
jointed weak and possibly failure and squeezing. reinforced shotcrete t o clean
sedirnentary rock. squeezing. rock suríaces as soon as
High stress possible. install rockbolts,
conditions. with faceplates. through
shotcrete. apply second
75 rnm shotcrete layer.
Highly jointed Ravelling of srnall Prevention of Apply 50 rnm of steel fibre
metamorphic or wedges and blocks progressive ravelling. reinforced shotcrete on
igneous rock. defined by clean rock surface in roof
LOWstress intersecting joints. o f excavation.
conditions. Rockbolts or dowels rnay be
needed for additional support
for large blocks.
 15 Shotcrete support

Table 15.2: (cont'd) Summary of recomrnended shotcrete applications in underground mining, for different rock
mas8 conditions.
Rock mass Rock rnass Support Shotcrete application
description behaviour requirements
Highly jointed Bed separation Control of bed Rockbolts or dowels required
and bedded in wide s ~ a n se~arationand t o control bed seaaration.
sedimentary rock excavatiins and raielling. Apply 75 mrn of iibre
Low stress and ravelling of reinforced shotcrete t o
conditions. bedding traces in bedding plane traces
inclined faces. before boltina.
Heavily jointed Squeezing and Control of rock rnass Apply 100 rnm of steel fbre
igneous or 'plastic' flow of failure and dilation. reinforced shotcrete as soon
metarnorphic rock, rock rnass around as possible and install
conglomerates or rockbolts, with face-plates,
cernented rockfill. through shotcrete Apply
High stress additional 50 rnrn of shotcrete
conditions. if required Extend support
down sidewalls if necessary,
Heavily jointed Squeezing and Control o f rock mass Apply 50 rnm of steel fibre
sedirnentary rock 'plastic' flow of failure and dilation. reinforced shotcrete as soon
with clay coated rock rnass around as possible, install lattce
surfaces. opening. Clay rich girders or light steel sets, with
High stress rocks may swell. invert struts where required.
conditions. then more steel fibre reinforced
shotcrete t o cover sets or
girders. Forepoling or spiling
may be required t o stabilise face
ahead of excavation. Gaps may
be left in final shotcrete to
allow for movernent resulting
from squeezing or sweiling.
Gap should be closed once
opening is stable.
Mild rockburst Spalling, slabbing Retention of broken Apply 50 t o 100 rnm
conditions in and mild rockbursts rock and control of of shotcrete over mesh
massive rock failure propagation. or cable acing which is firmly
subjected t o high attached 1 to the rock surface
stress conditions. by rneans of yielding rockbolts
or cablebolts.
 Blasting damage in rock

,16.1 Introduction
The development of roclí mechanics as a practical engineering too1 iii both underground and surface
mining has followed a rathererratic path over the past few decades. Only the most naively optimistic amoiigst
us would claim that the end of tlie road lias been reached and that the subject has inatured ihto a fully
developed applied science. O n the other hand, there have been some real advances which only the most
cynical would discount.
O n e of the results of the erratic evolutionary path has beeii the emergente of different rates of ad-
vance of different branches of the subject of rock mechanics. Leading the field are subjects sucli as the
inechanics'ofslope instability, the monitoring of movernent in surface and undergrouiid excavations and tlie
analysis of iiiduced stresses aiound underground excavations. Trailing the field are subjects such as the
rational design of tuniiel support, the movernent of groundwater through jointed rock inasses aiid the inea-
sureinent of iii situ stresses. Bringing up the rear are those areas of application where rock inechanics has to
interact with otiier disciplines and one of these areas involves the influeiice of the excavation process upoii
the stability of rock excavations.

16.2 Historical perspective

By far the inost cominon technique of rock excavation is that of drilling and blasting. Froin the
earliest days of blasting with bladí powder, there have been steady developments in explosives, detonatiiig
aiid delaying techniques and in our uiiderstanding of the mechanics of rodí brealiage by explosives.
It is not the development in blasting technology that is of iiiterest in this discussion. It is the applica-
tioii of this technology to the creation of excavations in roJí and the influence of the excavation techniques
upon the stability of the reinaining roclí.
As is fiequently tlie case in engineering, subjects that develop as separate disciplines teiid to develop
in isolation. Hence, a handful of highly slíilled and dedicated researchers, frequently worlíing in association
witli explosives manufacturers, have developed techiiiques for producing optimum fragmentation and
ininimising damage in blasts. At the otlier eiid of the spectrum are miiiers who have learned their blasting
skills by traditional apprenticeship inethods, and who are either not familiar with the specialist blasting
control techniques or are iiot convinced that the results obtained from the use of these tediniques justify the
elfort aiid expeiise. At fault in this system are owners and managers who are more concerned with cost than
with safety and design or planning engineers who see both sides but are not prepared to get involved hecduse
they view blasting as a biadc art with the added threat of severe legal penalties for errors.
The need to change the present system is not widely recognised because the impact of blasting dain-
age upon the stability of structures in rodí is not widely recognised. It is the author's aim, in the reinainder
of this chapter, to explore this subject and to identify the causes of blast damage and to suggest possible
iinproveinents in the systein.
 Chapter 1 t> Blasting clamage in ri

A discussion on the influente of excavation processes upon the stability ofroclc structures would not
he complete without a discussion on machine excavation. The ultimate in excavation techniques, whicli
leave the rock as undisturbed as possible, is the full-face tunnelling machine. Partia1 face machiiies os
roadheaders, when used correctly, will also inflict very little damage on the rock. The characteristics of
tunnelling machines will not be discussed here but comparisoiis will be drawn between the ainouiit of
damage caused by these machines and hy blasting.

16.3 Blasting damage

It appears to me, a casual reader of theoretical papers on blasting, that the precise nature of the inechanism
of rock fraginentation as a result of detonation of an explosive charge is not íülly understood. However, from a
practical point ofview, it seeins reasonable to accept that hoth the dynamic stresses induced by tlie detonation and
the expandiiig gases produced by the explosion play important roles in the fiagmeiitation process.
Duvall aiid Fogelsoii (19621, Langefors and Khilstroin (1973) and others, have published blast dainage
criteria for buildings and other surface strnctures. Alrnost all of these criteria relate blast dainage to pealí particle
velocity resulting froin the dynainic stresses induced by the explosioii. While it is generally recogiiised tliat gas
pressure assists in tlie rodí fragineiitation process, there has been little attempt to quaiitify this damage.
Work on the strength ofjointed rodí masses suggests that this streiigth is influenced by the degree of
interloclíing between individual roclí bloclcs separated by discontinuities such as hedding plaiies and joints.
For all practical purposes, the tensile strength of these discontinuities can be taken as zero, aiid a small
arnount of opening a r shear displaceinent will result in a drainatic drop in the interlocking of the individual
blocks. It is easy to visualise how the high pressure gases expanding outwards from ali explosion will jet iiito
these discoiitinuities and cause a breakdown of tliis important block interloclíiiig. Obviously, the ainount of
damage or strength reduction will vary with distance from the explosive charge, and also with the in situ
stresses which have to be overcoine by the high pressure gases behre looseiiing of the roclc can talce place.
Coiisequently, the extent of the gas pressiire induced damage can be expected to decrease with deptli below
surfice, and surface structures sucli as slopes will he very susceptible to gas pressure induced blast dainage.
Ai1 additional cause of hlast inducted dainage is that of fracturiiig induced by release of load (Hagan
(1982)). This inechanisin is best expiained by the analogy of dropping a heavy steel plate onto a pile of
mbher inats. Tliese rubber inats are compressed until the moineiituin of the falling steel plate lias been
exhausted. Tlie highly compressed rubber mats then accelerate tlie plate in the opposite direction and, iri
ejecting it vertically upwards, separate from each other. Such separation between adjacent layers explaiiis tlie
'tension fractures' frequently ohsemed iii open pit and strip mine operations where poor blasting practices
encourage pit wall iiistability. McIntyre and Hagan (1976) report vertical cradcs parallel to and up to 55 in
beliind iiewly created open pit mine faces where large inulti-row blasts have been used.
Whether or iiot one agrees witli the postulated mechanisin of release ofload fracturing, tlie iact that
craclcs can be induced at very coiisiderable distance from the point of detonation of an explosive must be a
cause for serious concern. Obviously, these fractures, whatever their cause, will Iiave a inajor disruptive effect
upon the integrity of the roclc inass and this, in turii, will cause a reductioii in overall stability.
Hoelí (1975) has argued that blasting will not induce deep seated instability in large o p e i ~pit mine
slopes. This is because the failure surface can be severa1 hundred metres below the surhce iii a very large
slope, aiid also because this hilure surface will generally not be aligned in the same direction as blast induced
fractures. Hence, unless a slope is already vety close to the poiiit of failure, and the blast is simply the last
straw that brealcs the cainel's badí, blasting will not generally induce inajor deep-seated iiistability.
O n the other hand, near siirface damage to the rock mass can seriously reduce the stability of the
individual benches mhich inalce up the slope and which carry the haul roads. Consequently, in a badly
blasted slope, the overall slope iiiay be reasoriably stable, but the fi~ceinay reseinble a rubhle pilc.
 ieering: Coi

In the case of a tunnel os otlier large underground excavation, the problem is rather different. In this
case, the stability of tlie underground stnicture is very inucli dependem upon the integrity of the roclc
immediately surrounding the excavation. Iii particular, the tendency for roof falls is directly related to the
interlocking of the immediate roof strata. Since blast damage can easily extend several inetres into the rock
which has been poorly blasted, the halo of loosened ro& can give rise to serious instability probleins in the
rocl<surroundiiig the underground openings.

16.4 Damage control

The ultimate in damage control is machine excavation. Anyone who lias visited an underground
metal mine and looked u p a bored raise will have been impressed by the lack of disturbance to the rock and
the stability of the excavation. Even when the stresses in the rod< surrounding the raise are higli enough to
induce fracturing in the walls, the damage is usually limited to less than half a inetre in depth, and the overall
stability of tlie raise is seldom jeol~ardised.,
Full-face and roadheader type hinneiiiiig inacliines are becoining more and more common, particu-
larly for civil engineering tunnelling. These machines have bein developed to the point wliere advaiice rates
and overall costs are geiierally coinparable or better than the best drill and blast excavation methods. The
laclc of disturbance to the roclc and the decrease in the amount of support required are major advantages in
the use of tunnelling machines.
For surface excavations, there are a few cases in which inachine excavation can be used to great advan-
tage. In the Bougainville opeii pit copper mine in Papua New Guinea, trials were carried out on dozer cutting
of the final pit wall faces. The final vertical blastholes were placed about 19 m froin the ultiinate hench crest
position. The remaining rocl<was then ripped using a D-10 dozer, and the final 55 degree face was trirnmed with
the dozer blade. The rock is a veiy heavily jointed andesite, and the results of the dozer cutting were reinarkable
when compared with the bench fices created by the normal open pit blasting techniques.
The inachine excavation techniques described above are not widely applicable in underground min-
iiig situations, and consideration must therefore be given to what can be done about controlling dainage in
normal drill and blast operations.
A coininon misconception is that the only step required to control blasting dainage is to iiitroduce
pre-splitting os sinooth blasting techniques. These blasting methods, which involve the siinultaneous deto-
nation of a row of closely spaced, lightly charged holes, are designed to create a clean separation surface
hetween the rock to be blasted and the rockwhich is to remain. When correctly perforined, these blasts can
produce very cleaii faces with a ininimum of overbreal< and disturbance. However, controlling blasting
damage starts long before the introduction ofpre-splitting or smooth blasting.
As pointed out earlier, a poorly designed blast can induce craclcs several metres behind the last row o i
blastholes. Clearly, if such dainage has already been iiiflicted oii the rock, it is far too late to atteiiipt to
remedy the situation by using sinooth blasting to trim the last few inetres ofexcavation. O n the other haiid,
if the entire blast has been correctly designed and executed, smooth blasting can be very beneficia1 in trim-
ining the final excavation face.
Figure 16.1 illustrates a comparison between the results achieved by a normal blast and a face created by
[xesplit blastiiig iii a jointed gneiss. It is evident that, in spite of tlie fairly large geological stmctures visible i11
the face, a good clean face has been achieved by the pre-split. It is also not difficult to imagine that the pre-split
face is inore stable than the section which has been blasted without special attentiomto the final wall condition.
Tlie correct design of-ablast starts with the veiy first hole to be detonated. In the case of a tunnel blast,
tlie first requirement is to create a void into which rock broken by the blast can expand. This is generally
acliieved by a wedge or burn cut which is designed to create a cleaii void and to eject the rock origiiially
contaiiied in this void clear of the tunnel face.
 Chapter 16 arnage in r<

Figure 16.1: Cornpariion

berrveen the reiulti
achieved by pre~splt
biaitng (on the left) and
normal b u k blasting for a
iurface excavation in gneiii

Iii today's drill and blast tunnelling in which multi-booin drilling inachines are used, tlie inosl conve-
nient method for creating the initial void is the burn cut. This involves drilling a patterii ofcarefully spaced
parallel Iioles which are then chargcd with powerful expiosive and detonated sequentially using inillisccond
delays. A detailed discussion on the design of burn cuts is given by I-Iagaii (1980).
Once a void has heen created for the full length of the intended blast depth or 'pull', tlie iiext step is
to brealc the rock progressively iiito this void. This is generally achieved hy sequeiitially detoiiatitig carefi~lly
spaced parallel holes, using one-half second delays. The purpose of using such long delays is to ensure that
the rock broken by each successive blasthole has sufficient time to detach from the surrounding roclc aiid to
be ejected into the tunnel, leaving the necessary void into which the next blast will break.
A final step is to use a smooth blast iil which lightly charged perimeter Iioles are detonated simulta-
iieously in order to peel off the rernaining half to one metre of roclc, leaving a cleaii excavation surtice.
The details of such a tunnel blast are given in Figure 16.2. The developrneiit of the burn cut is
illustrated in Figure 16.3 and the sequence of detonatioil aiid fracture of thc remaiiider ofthe blast is sliowri
in F i g ~ ~ 16.4.
r e The results acliieved are illustrated iii a photograpli reproduced in Figure 16.5. In this particu-
lar project, a significant reduction in tlie amount of support insralled in the tunnel was achieved as 3 result ot
the implementation of the blasting design shown in Figure 16.2.
A final point o n blasting in underground excavatioi~sis that it is seldom practical to use prc-split
blastiiig, except in tlie case o f a beiicliing operatioii. In a pre-split blast, the closely spaced parallel Iioles
(similar to those numbered 9, 10 and 11 iii Figure 16.2) are detonated before the maiii blast iiistead ofafter,
as iil the case of a sinooth blast. Since a pre-split blast carried out under these circuinstances lias to take place
in almost coinpletely iindisturbed rock which inay also be subjected to relatively high induced stresses, tlie
chances of creating a clean brealc line are not very good. The craclcs, which sliould rui1 cleaiily from oiie Iiolc
to the next, will frequently veer offin the direction ofsome pre-existing weakness such as foliatioii. For these
reasons, sinooth blasting is preferred to pie-split blasting for tiinnelling ~~ieratioiis.
 urse notes t

In the case of rock slopes such as those in open pit mines, the tendency today is to use large diameter
hlastlioles on a relatively large spacing. These holes are generalky detonated using inillisecond delays which
are designed to give row by row blasting. Unfortunately, scatter in the delay times of the inost commonly
used open pit blasting systems can sometimes cause the blastholes to fire out of sequence, aiid this can
produce poor fragmentation as well as severe damage to the roclc which is to remain to form stable slopes.
Dowiihole delay systems which can reduce the problems associated with the detonation of charges i11
large diaineter hlastholes are available, but open pit blasting engineers are reluctant to use thein because of
the added c o ~ l i c a t i o n sof laying out the blasting pattern, and also because of a fear o i cut-offs due to
failure of the ground caused by the earlier firiiig blastholes. There is clearly a iieed for further development
of tlie technology atid the practical application of bench blasting detonatioli delaying, particularly for tlie
large blasts whicli are required in open pit miniiig operations.

Holes no Dia m m Explosives Total w t . kg Detonators

Burn 14 45 Gelamex 80. 18 stickslhole 57 Millisec
Lifters 9 45 Gelamex 80. 16 stickslhole 33 Half-sec
Perimeter 26 45 Gurit, 7 sticksthole and 26 Half-sec
Gelamex 80, 1 sticklhole
Others 44 45 Gelamex 80, 13 stickslhole 130 Half-sec
Relief 3 75 No charge

Total 96 246

Figure 16.2: Blaithole pattern and charge detas used by Balfaur Beatty - Nuttali on the Virtoria iiydroelectricprojert in Sri Lanka. Roman numerais refei
to the detonaton sequenie of miilsecond delayi in the burn cut. whie Arabic nurnerals iefer to the half-second deayi ~nthe rernainder of the blait
 Layout of holes Muliacond deiay VI

Millisecond delay O Millisecond delay IIX

Millisecond delay 11 Millisecond d i a y X

Milliaecond delay IV Hd-second delay 1

Figure 16.3: Development of a burn cut usiny miliiseiand delayi.

 Rock Engin~
eering. Cou / Evert Hoel

Haü-second delay 3 H d t - w n d delay 7

HaU-pecond delay 4 HaU-second delay 8

H&-second delay 5 H&-second delay 9

Ha-aecond delay 6 Liers & smooth blast

Figure 16.4: Use of haf-second delayi in the main blait and rrnooth blaiting of the perimeter of a tunnel.
 Chapter 16 Blasting d amage in rcick 2

Figure 16.5: Resulti

achieved using well
deiigned and carefully
iontroiled blaiting in a 19
foat diametertunne in
gneisi in the V~itoria
hydioelectric project n
Sri Lanka. Phatograph
ieproduced with
permisiion from the Britirli
Overseai Deveapment
Administratio" and from
Baliour Beatty - Nuttall.

16.5 Blasting design and control

While there is room for improvement in the actual techniques used in blasting, many of the existing
techniques, if correctly applied, could be used to reduce blasting damage in both surfdce aiid uiidergrouiid
rodz excavation. As pointed out earlier, poor coininunications and reluctance to becoine involved on the
part of most engineers, means that good blasting practices are generally not used o n rnining aiid civil engi-
neering projects.
What can be done to improve the situation? In tlie writer's opinion, the inost critical need is for a
inajor iinprovement in cominui~ications.Currently available, written information on control of blasting
damage is either grossly inadequate, as in the case ofblasting handbooks published by explosives inanufac-
turers, or it is hidden in teclinical journals or texts which are not read by practical blasting eiigineers. Ideally,
what is required is a clear, concise book, which sets out the principies of blasting design and control i11
unambiguous, non- mathematical language. Failing this, a series of articles, in similarly plain language, pub-
lished iii trade jouriials, would help a great deal.
In addition to the gradual improvement in the undentanding of the causes and control of blast
damage which will be achieved by the improvement in communicatioiis, there is aiso a need for more urgent
actioii oii tlie part of engineers involved in rock excavation projects. Such engineen, who sliould a t least be
aware ofthe damage being inflicted by poor blasting, should take a much stronger line with owners, inanag-
ers, contractors and blasting foreinen. While these engineers may not feel theinselves to be competerit to
redesign the blasts, they may be able to persuade the other parties to seek the advice of a blasting specialist.
Explosives inaiiufacturers can usually supply such specialist semices, or can recominend individuais who will
assist in improving the blast design. Incidentally, in addition to reducing the blasting damage, a well de-
signed blast is generally more efficient and may provide improved fragmeiitation and hetter inuclz-pile con-
ditions at the saine cost.
 Rock Engin

16.6 Conclusion
Needless dainage is being caused to both tunnels and surface excavation by poor blasting. This dam-
age results in a decrease in stability which, in turn, adds to the costs of a project hy the requireinent of greater
volumes of excavation or increased rod< support.
Tools and techniques are availahle to ininimise this dainage, but tliese are not being applied vely widely
in either the mining or civil engineering industries because o f a ladí of awaretless of the benefits to he gained,
and a fear of tlie costs involved in applying controlled blasting techniques. There is an urgent iieed for iin-
proved communications between the blastiiig specialists who are competent to desigi~optiinuin blasting sys-
teins and the owners, managers aiid blasting foremen who are responsible for the execution of these designs.
Research organisations involved in work on blasting should also recognise the current ladí of effective
cominunications and, in addition to their work in improving blasting tediniques, tliey should be more
willit~gto participate in field-oriented prograrns in co-operatioii with industiy. Not only will organisatioiis
gaiti invaluable practical Iaowledge but, byworlíing side-by-side with other engineers, they will d o a great
deal to improve tlie general awareness of what can be acbieved by good blasting practices.
 References

1. Amos, A. J., A. Granero Hemandez, and R. J. Rocca. 1981. Prohleinas de meteorizacion de1 geneis eii Ia
Presa Principal de1 complejo hidroeléctrico Rio Grande I. Proc. VIII Cong Geol. Arg. Actas 2, 123-135.
2. Anon. 1977. Description of rode masses for engineering purposes. Geological Society Engiiieerinp
Group Working Party Report. QJ. Engng Geol. 10, 355-388.
3. Azzoni, A,, La Barbera, G. and Zaninetti, A. 1995. Analysis and prediction of roclcfalls using a inath-
ematical inodel. InfernationalJournal ofRock Mechanics and Mining Science and Geomechanics Abstrncts.
Vol. 32., No. 7. pp. 709-724.
4. Badger, T. C. and Lowell, S. 1992. Rocl<fallControl Washington State. In Rock,füllPredzct~onand Controi
and Landslide Case Histories, Fansportation Ruearch Rccord, National Research Council, Wasliiiigtoii,
No 1342, pp 14-19.
5. Bajzelj, U., Lilcar, J., Zigman, E., Subelj, A. and Spek, S. 1992. Geotechnical analyses of the mining
method using long cable holts. In Rocksnpportin miningandundergronndconstrnction,proc. int. gmp. on
roLhsxpport, Sudbury, (eds. P. K. Kaiser and D. R. McCreath), 393-402. Rotterdam: Bali<ema.
6. Balcer, D. G. 1991. Wahleach power tunnel monitoring. Proc. 3rdInt. Symp. on FieldMea's~remcntsin
Geomechanics, Oslo, Norway, in press.
7. Balmer, G. 1952. A general analytical soIution for Mohr's envelope. Am. Soc. Zst. Mat. 52, 1260-1271.
8. Bandis, S. C. 1980. Experimentalstudiesofscaleeffectsonshearshength,anddcormation ofrock~oints.Pli. D.
thesis, Uiiiversity of Leeds.
9. Bandis, S. C. 1990. Meclianical properties of rode joints. In Proc. Int. Soe. Rodi Mcch. gmp. on rorlz~oints,
Loen, Nonvay, (eds N. Barton and 0. Stephansson), 125-140. Rotterdam: Ballema.
10. Barton, N. 1976. The shear streiigth of rod<and rock joints. Int.J. RockMech. Min. Sci & Gcomech.Abstr.
13, 1-24.
11. Barton, N. 1989. Cavern design for Hong Kong rocks. Proc. Rork Cavern Seminar- HongKong(eds A. W.
Malone and P. G. D. Whiteside), pp. 179-202. London: Institutioii of Mining aiid Metallnrgy.
12. Barton, N., By, S. L., Chryssanthakis, L., Tunbridge, L., Kristiansen, J., L ~ s e t ,F., Bliasin, R. K.,
Westerdahl, H. and Vik, G. 1992. Comparison ofprediction aiid performaiice for a 62 m spaii sports
hall in jointed gneiss. Proc 4th. int. rock merhanics nndrock engineeringconj, Torino. Paper 17.
13. Barton, N., Laset, F., Lien, R. and Lunde, J. 1980. Application ofthe Qsystein in design decisions. In
Snbsufacespace, (ed. M. Bergman) 2, 553-561. New York: Pergainoii.
14. Barton, N. R. 1973. Review of a ilew shear strengtli criterion for rock joints. Engng Geol. 7,287-332.
15. Barton, N. R. 1974. A reaie~uof the shear strength ofjllied discontinuities in rock. Norwegian Geotech. Inst.
Puhl. No. 105. Oslo: Nomegian Geotech. Inst.
16. Barton, N. R. 1976. The shear strength of rock and rock joints. Int J. Mech. Min. Sei & Geomech. Abstr.
13(10), 1-24.
17. Barton, N. R. 1987. Predicting the behauiour of nnderground openings in rodi. Manuel Rocha Memona1 Lec-
tnre, Lisbon. Oslo: Nonvegian Geotech. Inst.
 Rock Engirieering Coitrse notes t~yEvert Hoc

18. Barton, N. R. and Bandis, S. C. 1782. Effects ofblock size on the the shear behaviour ofjointed rodc.
23rd U.S. symp. on rock mcchanics, Berkeley, 737-760.
17. Barton, N. R. and Bandis, S. C. 1770. Review ofpredictive capabilites ofJRC-JCS model in engineer-
ing practice. In Rock joints, proc. int. symp. on rockjoints, Loeti, Norway, (eds N. Bartoti and 0.
Stephansson), 603-610. Rotterdam: Ballcema.
20. Barton, N. R. and Choubey, V. 1777. The shear strength of roclc joints in theory aiid practice. Rock
Mech. lO(1-2), 1-54.
1 . Barton, N. R., Lien, R. aiid Lunde, J. 1774. Engineering classification of rodc masses for the design of
tunnel support. Rock Mech. 6(4), 187-239.
22. Bétournay, M. C . 1987. A design philosophy for surface crown pillars in hard roclc mines. Bull Cana-
dian Inst. Min. Metall. 80(703), 45-61.
23. Bieniawski Z. T 1789. EngineeringRock Mass Clars$cations. Wiley, New Yorlc. 251 pages.
24. Bieniawski, Z. T 1967. Mechanism of brittle fracture of roclc, parts I, I1 and 111. Int. J. Rock Mech. Min.
Sei. & Geomech. Abstr. 4(4), 395-430.
25. Bieniawslci, 2.T. 1773. Engineering classification of jointed roclc inasses. Trans S. A f i Inst Civ. Engrs
15, 335-344.
26. Bieniawslci, 2.'L 1774. Estiinating the strengtli of rock materials. J. South Afncan Inst. Min. Metail.
74(8), 312-320.
27. Bieniawslci, Z. 'L 1774. Geomechanics classification ofroclc masses aiid its application in tu~inelling.In
Aduancer in Rock Mechanics 2, part A: pp.27-32. Washington, D. C.: National Academy of Sciences.
28. Bieniawslci, Z. T. 1976. Rock mass classification in roclc engineering. In Exploration for rock engineering,
proc. of~hesymp.,(ed. Z.T. Bieniawski) 1, 77-106. Cape Town: Balkeina.
29. Bieniawslei, Z. T 1978. Determining roclc mass deformability - experiences from case histories. Int. J:
Rock Mech. Min. Sei. & Geomech. Abstr. 15, 237-247.
30. Bieniawslci, 2.T. 1779. The geomechanics classification in rodc engineering applications. Proc. 4th.
cong/.,Int. Soc Rock Mech., Montreux 2, 41-48.
31. BieniawsBi, 2. T. 1789. Enginrerinyrock mais classficationr. New Yorlc: Wiley.
32. Bieniawski. Z. T., 1767. Mechanism ofbrittle fracture ofroclc, parts I, I1 and 111. IniJ RockMech. Min.
Sei. & Geomech. Abstr 4(4), 375-30.
33. Bouchier, F., Dib, E. and O'Flaherty, M. 1792. Practical improvements to installatioii ofcable bolts:
progress at Campbell Mine. In Rock support in minfnxandunde&oundconstruction,proc. int. symp. on rock
support, Sudbury, (eds P. K. Kaiser and D. R. McCreath), 311-318. Rotterdam: Ballcetna.
34. Bowcoclc, J. B., Boyd, J. M., Hoelc, E. and Sharp, J. C. 1776. Dralcensberg puinped storage scheme,
rock engineeriiig aspects. In Explorationfor Rock Engineering (ed. Z.T. Bieniawslci) 2, pp. 121-139.
Rotterdam: Ballcema.
35. Bozzolo, D., Pamini, R. and Hutter, K. 1788. Rockfall analysis - a matheinatical model and its test
witli field data. Proc. 5'~1nemational~ym~orinm on Landslides, Lusanne. July 1788, Vol. 1, pp. 555-560.
36. 'Brady, B. H. G. and Brown, E. T 1785. Rock mechanicsfor underground mining. London: Allen and
Unwin.
37. Brand, E. W. 1788. Special Lecture: Landslide rislc assesstnent in Hong Kong. Proc. 5'"nternational
Symposium on Landslides, Lusanne. July 1988, Vol. 2, pp. 1057-1074.
38. Brawner, C. O . and Hoek, E. 1777. Design, Constmction and Maintenance ofroclc slopes on highway
projects. Proc. VIIIlh International RoadFederatfon World Meeting, Tok~o.October, 1977,
37. British Standard Code of Practice, 1787. BS 8081, GroundAnchorage, 87-115.
40. Brown E. T, BrayJ. W., Ladanyi B. and HoekE. 1983. Characteristic line calculations for rodc tunnels.
L, Geotcch. Engng Div., Am. Soe. Ciu. E n p 109, 15-37.
 Chapter 17 Referenc

41. Brown, A. 1982. The influente aiid control of groundwater in large slopes. In Srnbilily in Su$ie Mzn-
ing (ed. C. O. Brawner), pp. 19-41. New Yorlí: Society of Miniiig Engineers, AIME.
42. Brown, E. T.1970. Strength ofinodels ofrodí with intermittent joints. J. SoilMech. FoundnDiv,ASCE
96, SM6,1935-1949.
43. Brown, E. T. 1987. Introduction. Ana~ticalandcon~putational mcthods in cnginceringr.ock mechanicr, (ed.
E. 'L Brown), 1-31. Loiidoii: Allen and Unwin.
44. Browii, E. T. and Bray, J. W. 1982. Roclí-support interaction calculatioiis for pressure sliafts niid tun-
nels. In RoúzMechanics: CavemsandPressureShafis(ed. W.Wittlíe) 2, pp. 555-565. Rotterdain: Balkema
45. Brown, E. T.and Ferguson, G. A. 1979. Progressive hánging wall caving at Gatli's inine, Rhodesia.
Tmns. Instn Min. Metall. (Section A : Min. indushy) 88, A92-105.
46. Brown, E. T and Hoelc, E. 1978. Trends in relationships between ineasured rode in situ stresses and
depth. Int.J. Rock Mech. Min. Sei. & Gconicch. Abstr. 15, 211-215.
47. Browii, E. T.,Bray, J.W., Ladanyi, B. and Hoek, E. 1983. Characteristic liiie calculatioris for roclí
tunnels. J. Gcotech. EngngDiv., ASCE 109, 15-39.
48. Bucky, P. B. 1931. Use of inodels for the study of inining problems. Atn. Inst. Min. Meta11 Engrs,
Technical Publication 425.
49. Bunce, C. M. 1994. Risk Analysis for Rock Fali on Highu'ays. MSc thcsis submittcd to thc Dcpartmcnt of
Civil Enginccring, Universitj ofiilbcrta, Canada. 129 pages.
50. Bywater, S. and Fuiler, P. G. 1984. Cable support for lead open stope hanging walls at Mount Isa
Mines Limited. Iii Rock bolting: theory and appiication in rnining and undcground constructiofi, (ed. O .
Stephanssoii), 539-556. Rotterdam: Balkema.
51. Carter, T. G. 1992. A iiew approach to surface crown pillar design. Proc. 16'". Canadian RockMe~hanics
Symporir~m,Sndbur;7i, 75-83.
52. Carter, T G. 1992. Prediction and uncertainties in geological engineering and rock mass characteriza-
tion assessineiits. Proc. 4th. int rock n~cdianicsand rock enffrieering con,f, Torino. Paper 1
53. Clian, Y C., Clian C. F. atid Au S. W. C., 1986. Design of a boulder feiice in I-Iong Kotig. Cor$ OnRod
Engineeringand Excavatii>nin an Urban Environment Hong Koilg: Institutioii of Mining & Metallurgy.
54. Cheng, Y. 1987. New development in seain treatment of Feitsui arch dain fouiidation. Proc. 6'"Cong
ISRM, Montreal, 319-326.
55. Clieng, Y and Liu, S. C. 1990. Power caverns of the Miiigtaii Puinped Storage Project, Taiwaii. Iii
Comprehensi~~e Rock Engineering (ed. J . A. Hudsoii) 5, I1 1-132.
56. Clegg, I. D. and Hanson, D. S. 1992. Ore pass design and support at Falcotibridge Liinited. I11 Roch
szlpport in mining and underground C~IIitructi~n, proc. int. rymp. on ro1:I~support, Sudbury, (eds P. K. Kaiser
and D. R. McCreath), 219-225. Rotterdam: Ballcema.
57. Clifford, R. L. 1974. Long rockbolt support at New Broken Hill Consolidated Liinited. Prnc. Aus. Inst.
Min. Metall., No. 251, 21-26.
58. Clough, R. W. 1960. The finite eleinent method in plane stress analyses. Proc 2nd. ASCE Conferenceon
Electronic Computation, Pittrbnrgh, 345-378.
59. Coates, D. 1966. Roclt Mechanics Princ$lcs. Ottawa: Dept. Mines and Technical Surveys.
60. Cool<,N. G. W. (1965) Tlie failure ofroclc. Int.J. Rock Meiii. Min. Sei. Gcomerli. Absh 2, 389-403.
61. Cording, E. J. and Deere, D. U. 1972. Rock tuiinel supports and field ineasureinents. Pror. North
Amcrican rapid cxca71. tunneling conz, Chicago, (eds. K. S. Lane and L. A. Garfield) 1, 601-622. New
Yorlí: Soe. Min. Engrs, Am. Inst. Min. Metall. Petrolm Engrs.
62. Cording, E. J., Hendron, A. J. and Deere, D . U. 1971. Roclí engineering for uiiderground caveriis.
Proc. Symp. on UndcrgroundRock Chambers, 1971, pp. 567-600. New York: Ainerican Society of Civil
Engitieers.
 Rock Eng~n
eering Cor

63. Couloinb, C. A. 1776. Essai sur une application des regles de maximis et miniinis a quelques probleines
de statique, relatifs a l'architecture. Memoires deMathmatique & de Physique 7, 343- 82.
64. Croucli, S. L. and Starfield, A. M. 1983. Bonnday elementmethods in solidmechanics. London: Alleil and
Unwin.
65. Cuminings, R. A., Kendorski, F. S. and Bieniawski, 2. T. 1982. Cuuingrock mass classfieation and
support estimafion. U. S. Buieau of Mines Contract Report #J0100103. Chicago: Engineers Inter-
national Inc.
66. Cundall, P. A. 1971. A computer model for simulating progressive large scale movements in blocky
rock systems. Iii. Rock Fracture,Proc. Symp. ISRM, Nany 1, Paper 2-8.
67. Daeinan, J. J. K. 1977. Problems in tunnel support mechanics. UndergroundSpace 1, 163-172.
68. Davis, W. L. 1977. Initiation of cableboltiiig at West Coast Mines, Rosebury. Proc. Aust Insi Min.
Mctall. conf, Tasinania, 215-225.
69. Deere, D. U. 1989. ROLAquali& daignation (RQD) aficr 20years. U . S. Arrny Corps Engrs Contract
Report GL-89-1. Vicksburg, MS: Waterways Experimental Station.
70. Deere, D. U. and Deere, D. W. 1988. The rock quality designation (RQD) index iii practice. Iii Rock'
classfication iystemsfor en&neeringpurposes, (ed. L. Kirkaldie), ASTM Special Publication 984, 91-101.
Philadelphia: Am. Soc. Test. Mat.
71. Deere, D. U. and Miller, R. P. 1966. En~neerin~~lassificntionandindexpropertiesofrock.
- Tecliiiical Report
No. AFNL-TR-65-116. Albuquerque, NM: A& orce Weapons Laboratory.
72. Deere, D. U., Heiidron, A. J., Patton, F. D. and Cording, E. J. 1967. Design of surface and near surface
coiistructioii in rock. In Failure and breakage of r o k proc. 8th U.S. vmp. rock mech., (ed. C . Fairhurst),
237-302. New York: Soc. Min. Engrs, Am. Inst. Min. Metall. Petrolm Engrs.
73. Diederichs, M. S., Pieterse, E., Nosé, J. and Kaiser, P. K. 1993. A model for evaluating cablc bond
strength: an update. Proc. Eurodi '93, Lisbon. (in press).
74. Dorsten, V., Frederick, F. H. and Preston, H. K. 1984. Epoxy coated seveii-wire strand for prestressed
concrete. Prestressed Conmete Inst.J. 29(4), 1-11.
75. Domk, P. 1991. Analysis ofthe laboratoy strength data using the onginal and modfied Hoek-Bro~onjiilurc
criteria. MASc thesis, Dept. Civil Engineering, University ofToronto.
76. Duncaii Fama, M. E. 1993. Nuinerical inodelliiig ofyield zoiies iii weak rod<s. In Comprehensi~~e roiir
engineering, (ed. J.A. Hudson) 2, 49-75. Oxford: Pergamon.
77. Duvall, W. I. and Fogelson, D. E. 1962. Review of criteria for estiinating damage to residences froin
blastiiig vibrations. U S. Bur. MinesRep. Inuest. 5986. 19 pages.
78. Endenbee, L. A. and Hofto, E. O . 1963. Civil engineering design and studies in rod< mechanics for
Poatina underground power station, Tasmania. J. Instn. Engrs Australia 35, 187- 209.
79. Engelder, T. and Sbar, M. L. 1984. Near-surface in situ stress: introduction.j. Geoyhys. Res. 89, 9321-
9322. Princeton, NJ: Princeton University Press.
80. Ewy, R. T. and Cook, N. G. W. 1990. Deforination and fracture arouiid cylindrical openings in rock.
Parts I & 11. Int.J. Rock Mech. Min. Sei. Geomech. Absk 27, 387-427.
81. Fabjanczyk, M. W. 1982. Review of ground support practice in Australian underground metalliferous
mines. Proc. Ans. Inst. Min. Metall. conj, Melbourne, 337-349. Melbourne: Aust. Inst. Min. Metall.
82. Fairliurst, C. and Cook. N. G. b '.,1966. The phenomenon of rock splitting parallel to a free surfice
uiidcr coinpressive stress. Proc. 1st. Cong. ISRM, Lisbon 1, 687-692.
83. Fariner, I. W., and Slielton, P. D. 1980. Review of underground rock reinforcement systems. Faizs.
'
Instn Min. Metall. (Sect A: Min. indushyl89, A68-83.
84. Feat-Smith, I. 1982. Survey ofrock tunnelling machines available for miningprojects. Trans. Insin Min.
Metull. (Sect. A: Min. industy) 91, ,423-31.
 Chapter 17 Reference:

85. Federal Highways Administration. 1993. Rockfall Hazard Rating System, Participants Manual for NHI
Course No. 130220. Puhlication No. FHWA SA-93-057.
86. Fell, R. 1994. Landslide risk assessment and acceptahle rislc. Canadian GeotechnlcalJoumal. Vol. 31. pp.
261-272.
87. Fenner, R. 1938. Untersuchungen zur Erkenntnis des Gehirgsdruckes. Glukauf 74, 681-695, 705-
715.
88. Foolíes, P. G. and Sweeney, M. 1976. Stahilisation and control of local roclífalls and degrading of
slopes. QuarterlyJ. Enginecrrng Geology. Vol. 9, pp 37-55.
89. Franklin, J. A. and Hoek, E. 1970. Developments in triaxial testing equipment. RockMech. 2, 223-228.
Berlin: Springer-Verlag.
90. Franzén, T. 1992. Shotcrete for underground support - a state of the art report with focus on steel fibre
reinforcement. In Rocksupportinminingandundergroundconshu~tion,proc. int. symp. rocksupport, Sudhury,
(eds P. K. Kaiser and D. R. McCreath), 91-104. Rotterdam: Ballcema.
9 1. Preeze, A. R. and Cherry, J. A. 1979. Groundwater. Englewood Cliffs, NJ: Prentice-Hall 604 pages
92. Fuller, P. G. 1981. Pre-reinforcement of cut and fiI1 stopes. In Application ofrock mechanio to cutand,flI
mining, (eds O. Stephansson and M. J. Jones), 55-63. London: Instn Min. Metall.
93. Fuller, P.G. 1983. Tlie potential for support of long hole open stopes witli grouted cables. Proc. 5th.
Int. Cong.fir Rodz Mechanics, Melbome2, D39-D44. Rotterdam: Ballcema.
94. Fuller, P. G. 1984. Cable support in mining - a keynote lecture. In Rock bolting: theoy andapplication in
miningand undergoundconshuction, (ed. O . Stephansson), 51 1-522. Rotterdam: Ballcema.
95. Gane, P. G., Hales, A. L. and Oliver, H. A. 1946. A seismic investigatioii of Witwatersrand earth
trernors. Buli. Seism. Soc. Am. 36, 49-80.
96. Garhrd Pty Ltd. 1990. A n improved, economicalmethodfir rock stabilisation. 4p. Perth.
97. Greenwald, H.,.'i Howarth, H. C. and Hartinan, I. 1939. Experiments on the strerigth ofsrnall pillars
of coa1 in the Attshurg hed. U.S. Bureau qfMines íêch. Rep. No. 605.
98. Griffith, A. A. 1921. The phenoinenon ofrupture and tlow in solids. Phil. Trans. Roy. Soc., London
A221, 163.198.
99. Griffith, A. A. 1924. Theory of rupture. Proc. Istcongr. appliedmechanics, Delft, 55-63. Delft: Technische
Bockhandel en Drulckerij.
100. Griggs, D. T. 1936. Deforination ofrod<s under high confining pressures. J. Geol. 44, 541-577.
101. Grimstad, E. and Barton, N. 1993. Updating the QSystem for NMT. Proc. iut. symp. on sprayedconcrete
- modern use of wet mix sprayed conmetefor underground support, Fagernes, (eds Kompen, Opsahl and
Berg). Oslo: Nomegian Concrete Assn.
102. Hagan, T. N. 1980. Understanding the burn cut - a key to greater advance rates. Trans. Insm. Min.
Metall. (Sect. A: Min. Industy), 89, A30-36.
103. Hagan, T. N. 1982. Controlling hlast-induced cracking around large caverns. Proc. ISRM Symp., Rock
Mechanics Related to Caverns and Pressure Shajs, Aachen, West Gemany.
104. Haimson B. C. 1978. The hydrofracturing stress measuring method and recent field results. Int./ Rock
Mcch. Min. Sci. & Geomech.Abstr. 15, 167-178.
105. Hammett, R. D. and Hoelí, E. 1981. Design of large underground caverns for hydroelectric projects,
with reference to structurally controlled failure inechanisms. Proc. American Soe. CivilEngri. Int. ConJ
on Recent Developments in Geotechnical Engineerignfir Hydro Projem. pp. 192-206. New Yorlc: ASCE.
106. Harr, M. E. 1987. Reliabi1i~-basedd e s i in civil engineering. New Yorlc: McGraw-Hill.
107. Hatzor, Y and Goodman, R. E. 1992. Application of hlock theory and the critical key bloclí concept
in tunneling; two case histories. In Proc. Int. Soe. Rock Mech. conJonflactnredandjointedrock masses, Lake
Tahoe, California, 632-639.
 Rock Engiineering C(

108. Hatzor. Y. and Goodman. R. E. 1993. Determination of the 'design blocl<' for tunnel supports in
highly jointed rock. In ComprehensiveRock EnginePnng, Pnnciples, Practice andl'rojects. (ed. J. A. Houson)
2,263-292. Oxford: Pergamon.
109. Herget, G. 1988. Stressesin rock. Rotterdam: Balkema.
110. Heyman, J. 1972. Coulomb's Memoir on Statics. Cambridge: at the University Press.
111. Hoek E and Brown E. T. 1988. The Hoek-Brown failure criterion - a 1988 update. l i o c . 15th Canadian
Rock Mech. Symp. (ed. J . H. Curran), pp. 3 1-38. Toronto: Civil Engineering Dept., University ofToronto
112. Hoek E. and Brown E. T. 1980. UndergroundExcauationsin Rock. London: Institution of Mining and
Metallurgy 527 pages.
113. Hoelc, E. 1965. Rockflacture understatic stress conditions. PhD Thesis, Univ. Cape Town.
114. Hoek, E. 1968. Brittle failure of rock. In Rock Mechanics in Enginem'ngPrachce. (eds K. G. Stagg and 0.
C. Zienkiewicz), pp. 99.124. London: Wiley.
115. Hoek, E. 1975. Influence of drilling and blasting on the stability of slopes in open pit mines and
quarries. Proc. Atlas Copco Bench Drilling Days Symp., Stockholm, Sveden.
116. Hoelí, E. 1982. Geotechnical considerations in tunnel design and contract preparation. Fans. Instn
Min. Metall. (Sect. A: Min. industry) 91, A101-109.
117. Hoel<,E. 1983. Strength of jointed rock masses, 23rd. Rankine Lecture. Géotechnique 33(3), 187-223.
118. Hoek, E. 1986. Rockfall: a computer program for predictingrod<fall trajectories. Unpublished interna1
notes, Golder Associates, Vancouver.
119. Hoek, E. 1989. A limit equilibrium analysis of surface crown piilar stability. In Suface crovnpillnr
eualuationfor active and abandoned metal mines, (ed. M. C. Betourney), 3-13, Ottawa: Dept. Energy,
Mines & Resources Canada.
120. Hoek, E. 1990. Estimating Mohr-Coulomb friction and cohesion values from the Hoek-Brown failure
criterion. Intnl]. Rock Mech. & Mining Sci & GeomechanicsAbstracts. 12(3), 227-229.
121. Hoek, E. 1994. Strength of rock and rock masses, ISRMNewsJournal, 2(2), 4-16,
122. Hoek, E. and Bray, J. W. 1981. Rock Slope Engneering. 3rd edn. London: Institution of Mining and
Metallurgy 402 pages.
123. Hoek, E. and Brown, E. T. 1980b. Empirical strength criterion for rock masses. J. Geotech. Engng Din.,
ASCE 106(GT9), 1013-1035.
124. Hoek, E. and Brown, E. T. 1988. The Hoek-Brown failure criterion - a 1988 update. In Rockengineering
fir underground excauations, proc. 15'" Canadian rock me&. symp., (ed. J . C. Curran), 31-38. Toronto:
Dept. Civ. Engineering, University of Toronto.
125. Hoek, E. and Brown, E. T. 1997. Practical estimates or rock mass strength. Intnl.J. RockMech. & Mining
Sci. & Geomechanics Abstracts. 34(8), 1165-1186.
126. Hoek, E. and Moy, D. 1993. Design of large powerhouse caverns in weak rocl<. In Comprehensiverork
engineerin8 (ed. J. A. Hudson) 5, 85-110. Oxford: Pergamon.
127. Hoelí, E., and Brown, E. T. 1980a. Undergoundexcavations in rock. London: Instn Min. Metall.
128. Hoek, E., Kaiser, P. K and Bawden. W. F. 1995. Supportofundergomndexcauationsin hardrock. Rotterdam:
Balkema.
129. Hoek, E., Wood, D. and Shah, S. 1992. A modified Hoek-Brown criterion for jointed rock masses.
Proc. rock characterization, symp. Int. Soc. Rock Mech.: Eurock '92, (ed. J. A. Hudson), 209-214. London:
Brit. Geol. Soc.
130. Holland, C. T. and Gaddy. F. L. 1957. Some aspects ofpermanent support ofoverburden on coa1 beds.
Proc. W Virginia CoalMining Inst. 43-66.
131. Holmberg, R. and Persson, P-A. 1980. Design of a tunnel perimeter blasthole pattern to prevent rock
damage. Fans. Insm Min. Metall. (Sect. A: Min. industy) 89, A37-40.
 Reference!

132. Hungr, 0 . and Evans, S. G. 1989. Engineering aspects ofroclifall hazard in Canada. Geological Survey
of Canada, Open File 2061, 102 pages.
133. Hnnt, R. E. 1984. Slope failure risk mapping for highways: Methodology and case history. In Rockfall
prediction and Conhol and Landslide Case Histones. Transportatiou Research Record, National Research
Council, Washington, No. 1343. pp. 42-51.
134. Hunt, R. E. B. and Askew, J. E. 1977. Installation and design guidelines for cable dowel ground sup-
port at ZC/NBHC. Proc. Underground Operators Conferente, Broken Hill, 113-22.
135. Hutchins, W. R., Bywater, S., Thompson, A. G. and Windson, C. R. 1990. A versatile gronted cable
dowel reinforcing system for rock. Proc. Aus. Inst. Min. Metal[. 1, 25-29.
136. Hutchinson, D. J. and Diederichs, M. S. 1996. Cableboltingin undergroundmines. Vancouver: Bitech
137. Hyett, A. J., Bawden, W. F. and Coulson, A. L. 1992. Physical and mechanical properties of normal
Portland cement pertaining to fully grouted cable bolts. In ~ o c ksupport in mining and underground
conshuction, proc. int. symp. rock szpport, Sudbury, (eds. P. K. Kaiser and D. R. McCreath), 341-348,
Rotterdam: Balkema.
138. Hyett, A. J., Bawden, W. F. and Reichert, R. D. 1992. The effect ofrockmass confinemeilt on the bond
strength of fully grouted cable bolts. Int.J. Rock Mech. Min. Sei. & Geomech. Abstr. 29(5), 503-524.
139. Hyett, A. J., Bawden, W. F., Powers, R. and Rocque, P. 1993. The nutcase cable. In Innovativc mine
desipfor thc2lst~century,(eds W.F. Bawden and J. F. Archibald), 409-419. Rotterdam: Balkema.
140. Ide, J. M. 1936. Comparison ofstatically and dynamically determined Young's modulus of rode. Proc.
Nat. Acad. Sei 22,81- 92.
141. Iman, R. L., Davenport, J. M. and Zeigler, D. K. 1980. Latin Hypercubesamplin~(Aproyam
.. . -. . uscr'sm~de).
- .
Technical Report S A N D ~ ~ - 1 4 7~3 .l b i ~ u e r ~New
u e ,Mexico: Sandia Laboratories.
142. Iinrie, A. S. 1983. Taming the Downie Slide. Canadian Geographic 103.
143. Imrie, A. S., D. P. Moore and E. G. Enegren. 1992. Performance and maintenance of tlie drainage
system at Downie Slide. Proc. 6th Int. Symp. on Landslides, Christchurch, New Zeaiand, in press.
144. International Society for Rock Mechanics Commission on Standardisation of Laboratory and Field
Tests. 1978. Suggested methods for the quantitative description of discontinuities in rodi masses. Int.
1.Rock Mech. Min. Sci & Geomech. Abstl: 15,319-368.
145. International Society for Rock Mechanics. 1981. Rock characterisation, testing and monitoring - ISRM
suggested mcthods. Oxford: Pergainon.
146. Jaeger, C . 1972. Rock Mechanics andEngineaing. Cambridge: at the University Press 417 pages.
147. Jaeger, J. C. 1971. Friction of rocks and stability ofrock slopis. The 1l t h Rankine Lecture. G4otechniqnc
21(2), 97-134.
148. Jaeger, J. C. and Cook, N. G. W. 1969. FundamentalsofRockMtchanics. London: Chapmail and Hall.
149. Jirovec. P. 1978. Wechselwirliung zwischen anker und gebirge. Rock Mech. Suppl. 7, 139-155.
150. John, K. W. 1968. Graphical stability analyses of slopes in jointed rocli. Proc. SoilMech. Fndn Div.,
ASCE, SM2, paper no. 5865.
151. Kaiser, P. K., Hoek, E. and Bawden, W. F. 1990. A new initiative in Canadian rock mechanics research.
Proc. 31st USrockmech. symp., Denver, 11-14.
152. Kaiser, P. K, Yazici, S. and Nosé, J. 1992. Effect of stress change on the bond strength of fully grouted
cables. Int. J. Rock Me&. Min. Sei. Geomech. Abstl: 29(3), 293-306.
153. Kemeny, J. M. and Cooli, N. G. W. 1987. Crack models for the failure of rock under compression. In
Proc. 2nd int. conf:on constitutive lawsfor engineeringmaterials, theory and applications, (eds C. S. Desai, E.
Krempl, P. D. Kiousis and T. Kundu) 1, 879-887. Tucson, AZ: Elsevier.
154. Kendorski, F., Cummings, R., Bieniawski, Z. T. and Skinner, E. 1983. Rocli mass classification for blodi
caving mine drift support. Proc. 5th congx Int. Soe. Rock Mech., Melbourne, B51-B63. Rotterdam: Balkema.
155. Kiiig, L. V. 1912. O n the limiting strength of rocks under conditions of stress existing in the earth's
interior. j. Geol. 20, 119- 138.
156. Kirsch, G., 1898. Die theorie der elastizitat und die bedurfnisse der festigkeitslehre. Eit. Deit. Ing.
42(28), 797-807.
157. Kirsten, H. A. D. 1992. Comparative efficiency and ultimate strength of mesh- and fibre-reinforced
shotcrete as determined from full-scale bending tests./. S. Afr ~ n i tMin.
. Metall. Nov., 303-322.
158. iúrsten, H. A. D. 1993. Equivalence ofmesh- and fibre-reinforced shotcrete at large deflections. Can.
Geotech.J. 30, 418-440.
159. Kompeil, R. 1989. Wet process steel tihre reinforced shotcrete for rodc support and fire protection,
Norwegiail practice and experience. In Proc. undergroundcity conz, Munich, (ed. D. Morfeldt), 228-237.
160. Ladanyi, E. and Arcliamhault. G. 1970. Siinulation of shear behaviour of a jointed roclc mass. 111Rock
mcchanics - T h o y andPractice, Proc. 1Ith Symp. on Rock Mechanics, Berkeley, 1969, pp.105-25. New York:
Society of Mining Engineers, AIME.
161. Lajtai, E. 2. 1982. Thcfiacturc ofLac du Bonnetgranite. Contract Report. Pinawa, Ontario: Atoinic
Energy of Canada.
162. Lajtai, E. 2. aiid Lajtai, V. N. 1975. The collapse of cavities. Int.1.Rock Mech. Min. Sci Geomech. Absh
12, 81-86.
163. Lang, T. A. 1961. Theory and practice of rodcbolting. Pans Amcr Inst Min. Engrs 220,333-348.
164. Langefon, U. and Khilstrom, B. 1973. The modern technique of rock hlating. 2nd edn. New Yorlc: Wiley.
405 pages.
165. Langille, C. C. and Burtney, M. W. 1992. Effectiveness of shotcrete and mesh support in low energy
rodchurst conditions at INCO's Creighton mine. In Rock support in miningandundergroundconstruction,
proc. int. symp. rocksupport, Sudbuiy, (eds. P. K. Kaiser and D. R. McCreath), 633-638. Rotterdam: Balkema.
166. Lappalainen, P., Pulkkinen, J. and Kuparinen, J. 1984. Use of steel strands in cable bolting 2nd rodc
bolting. In Rock bolting: thcoy andapplication in miningand nndcrground construction,(ed. O. Stephanssoil),
557-562. Rotterdam: Balkema.
167. Lau. J. S. 0 . and Gorski. B. 1991. Thepostfailure hehauiour ofLac du Bonnctgreygranite. CANMET
Divisional Report MRL 91-079(TR). Ottawa: Dept. Energy Mines and Resources, Canada.
168. Laubscher, D. H. 1977. Geomechanics classification of jointed rock masses - mining applications.
Trdns. Instn. Min. Metall. 86, AI-8.
169. Laubscher, D. H. 1984. Design aspects and effectiveness of support systems in different miiling condi-
tions. Fans Instn. Min. Metall 93, A70 - A82.
170. Laubscher, D. H. and Taylor, H. W. 1976. The irnportance of geoinechanics classification of jointed
rock masses in iniiling operations. In Explorationfor rock engiueering, (ed. Z.T. Bieniawslzi) 1, 119-128.
Cape Town: Balkema.
171. Laubscher, D. M. and Page, C. H. 1990. The design of rodz support in high stress or wealz rodz
environinents. Proc. 92nd Cun. Inst. Min. Mctall. A G M , Ottawa, Paper # 91.
172. Lauffer, H. 1958. Gebirgsklassifizierung fur den Stollenbau. Geol. Bauwesen 24(1), 46-51.
173. Leeman, E. R. and Hayes, D. J. 1966. A technique for deterinining the complete state of stress in rock
using a single borehole. Proc 1st Cong. Int. Soc. Rock Mech, Lisbon 2, 17-24.
174. Lewis, M. R. and D. P. Moore. 1989. Construction of the Downiè Slide and Dutchman's Ridge drain-
age adits. Canadian Tunneiiing (ed. 2. Eisenstein), 163-172. Vancouver: Bi-Tech.
175. Lin, D and Fairhurst, C. 1988. Static analysis of the stability of three-dimensional hlocky systems
around excavations i; rodc. Int./. RockMech. Min. Sci & Gcomcch. Abstr. 25(3), 139-147.
176. Liu, S. C., Y Cheng and C. T. Cliang. 1988. Design ofthe Mingtan cavem. Proc.symp. ISRM. onRock
Mech. and Power Plants, Madrid, 199-208.
 Chapter 1 7 Referenci

177. Londe. P. 1965. Une inethode d'analvse a trois dimensions de Ia stabilite #une rive rocheuse. Annales
des Ponts et Chausseer 135(1), 37-60.
178. Londe, P., Vigier, G. and Vormeringer, R. 1969. The stability ofrodc slopes, a three-dimensional study.
J. SoilMech. ~oundnsDiv., ASCE 9 5 ( ~ MI), 235-262.
179. Londe, P., Vigier, G. aiid Vormeriiiger, R. 1970. Stahility of slopes - graphical methods. J. SoilMech.
Fndns Div., A S C E 96( SM 4), 1411.1434.
180. Lorig, L. J. and Brady, B. H. G. 1984. A hybrid computational scheme for excavation aiid support
design in jointed rode media. In Designandperjòmzanceofnndergoundexcavationr, (eds E. T. Brown and
J. A. Hudson), 105-112. London: Brit. Geotedi. Soe.
181. Lmset, F. 1992. Support needs compared at the Svartisen Road Tunnel. Tunnels and Enneliing, June.
182. Lottes, G. 1972. The development of European pumped-storage plants. WaterPower24, 22-33.
183. Love, A. E. H. 1927. A treatise on the mathematical theory ofelasticity. New York: Dover.
184. Lutton, R. J., Banks, D. C. and Strohm, W. E. 1979. Slides in the Gaillard Cut, Panama Canal Zone. In
Rochlides andAvalancher (ed. B. Voight) 2, pp. 151-224. New Yorlc: Elsevier.
185. Mahar, J. W., Parlcer, H. W. and Wuellner, W. W. 1975. Shotvetepracticein undergoundconstrnction. US
Dept. Transportation Report FRA-OR&D 75-90. Springfield, VA: Nat. Tech. Info. Semice.
186. Marachi, N. D., Chan, C. K. and Seed, H. B. 1972. Evaluation ofproperties ofrocl<fill inaterials. J. Soil
Mechs. Fdns. Div. ASCE 98(SM4), 95-114.
187. Marsliall, D. 1963. Hangingwall control at Willroy. Can. Min. Metall. Bnll. 56, 327-331.
188. Martin, C. D. 1990. Characteriziilg iii situ stress doinains at the AECL Underground Researcli Labo-
ratory. Can. Geotedl.J. 27, 631-646.
189. Martin, C. D. 1993. The strength ofmasrive Lac dn Bonnetgranite around nnderground openings. PIi. D.
thesis, Winnipeg, Manitoha: Dept. Civil Engineering, University of Manitoba.
190. Martin, C. D. and Siininons, G. R. 1992. The Underground Research Laboratory, an opportunity for
basic rock mechanics. ISRM Ne7us Joumall(l), 5-12.
191. Masur, C. I. and Kaufman, R. I. 1962. Dewatering. In Fouudation Engineering (ed. G.A. Leonards), pp.
241-350. New York: McGraw- Hill.
192. Mathews, K. E. and Edwards, D. B. 1969. Rock mechanics practice at Mount Isa Mines Limited,
Australia. Proc. 9th Commonwealth min. metall. congr, Paper 32. London: Instn Min. Metall.
193. Mathews, K. E., Hoek, E., Wyllie, D. C. aild Stewart, S. B. V. 1981. Prcdiction ofstable excauationsjor
miningatdepth belolu 1000metres in hardrock. CANMET Report DSS Serial No. OSQ80-00081, DSS
File No. 17SQ23440-0-9020. Ottawa: Dept. Energy, Mines and Resources.
194. Mathews, S. M., Tillman, V. H. and Worotniclci, G. 1983. A modified cablebolt system for support of
uiiderground openings. Proc. Aust. Inst. Min. Metall. annualconf, Brolcen Hill. 243-255.
195. Matthews, S. M., Thompson, A. G., Windsor, C . R. and O'Bryan, P. R. 1986. A novel reinforcing
system for large rock caveriis iil bloclcy rock masses. In Large rock caverns, (ed. K. H. O . Saari) 2, 1541-
1552. Oxford: Pergamon.
196. McCreath, D. R. and Kaiser, P. K 1992. Evaluation of current support
.. practices iii burst-prone xround aild
.
preliminary guidelines for Canadian hardrock mines. In Rock support in minzngand undmground construction,
proc. int. symp. rockrnpport, Sudbury, (eds P. K Kaiser and D. R McCreath), 611-619. Rotterdam: Ballcema.
197. McIntyre, J. S. and Hagan, T.N. 1976. Tlie design of overburden blasts to promote highwall stability
at a large strip mine. Proc. 11th Canadian Rock Mech. Symp., Ihncouver.
198. McMahon, B. K. 1971. A statistical method for the design of rod< slopes. Proc. 1st AustralIII-Nem
Zealand Confon Geomcchanics, Melhoirrue 1, 314-321.
199. McMahon, B. K. 1975. Probability of failure and expected voluine offailure in high rock slopes. Proc.
2nd Aust.-Nem Zealand Conf on Geomech., Brisbane. '
 Rock Engi~

200. Merritt, A. H. 1972. Geologic prediction for underground excavations. Proc. North American. rapid
excav. tunnelingconj, Chicago, (eds K. S. Lane and L. A. Garfield) 1, 115-132. New York: Soc. Min.
Engrs, Am. Inst. Min. Metall. Petrolm Engrs.
201. Moore, D. P., A. S. Imrie and D. G. Baker. 1991. Rockslide riskreduction using monitoring. Proc. Can.
Dum Safe@Assoc. AnnualMeeting, Whistler,British Columbia, in press.
202. Moretto, 0 . 1982. Mecánica de rocas en el complejo hidroeléectrico Rio Grande No. 1. Proc Primer.
Cong Sudamericano de Mecánica de Rocas, Bogotá, Colombia.
203. Morgan, D. R. 1993. Advances in shotcrete technology for support of underground openings in
Canada.In Shotcretejirnndergrouudsupportyproc. engineeringjiundation conj, Uppsala, (eds J. C. Sharp
and T Franzen), 358-382. New York: Am. Soc. Civ. Engrs.
204. Morgan, G. C. 1991. Qualification of risks from slope hazards. In Landslide Hazards in the Canadian
Cordillera. Geological Association of Canada, Special Publication.
205. Morgan, G. C., Rawlings, G. E. and Sohkowicz, J. C. 1992. Evaluation of total risk to cominunities
from large debris flows. GeotechnicalandNaturaiH~1~ards, Vancouver Geotechnical Society and Cana-
dian Geotechnical Society, Vancouver, BC, Canada, May 6-9, 1992, pp. 225-236.
206. Morgenstern, N. R. 1991. Limitations of stabilityanalysis in geo-technical practice. Geotecnia 61: 5-19.
207. Morrison, R. G. K. 1942. Report on the rockburst situation in Ontario mines. Trans. Can. Inst. Min.
Metali. 45.
208. Morrison, R. G. K. 1976. Aphilosophy ofground control: a úridge between theoy andpractice. rev. edn.
Montreal: Departinent of Mining and Metallurgical Engineering, McGill University 182 pages.
209. Morriss, P. and Stoter, H. J. 1983. Open-cut slope design using prohabilistic methods. Proc 5th. Cong.
ISRM., Melbourne 1, C107-C113. Rotterdam: Balkema.
210. Moy D., Hsieh C. S. and Li, H. C. 1990. The introduction of steel tiber shotcrete to underground
cavern support in Taiwan. Shotcretefor Undergrouud Support, Proc. of the Engineering Foundation Conj on
Shotcrete V , Uppsala, Sweden. in press.
211. Moy, D. and Hoek, E. 1989. Progress with the excavation and support of the Mingtan power cavern
roo6 Proc. Rock Cavern Seminar- HongKong (eds A. W. Malone and P. G. D. Whiteside), pp. 235-245.
London: Institution Mining and Metallurgy.
212. Muller, J. 1979.Josef Stini. Contributions to rngineering geology and slope movement investigations.
In Rocklides and Avalanches (ed. B. Voight), Part 2, pp. 95-109. New Y o k Elsevier.
213. Mnskhelishvili. N. I. 1953. Some basicproblems of the mathematicaltheoy ofeldstici@.4th edn, traiislated
by J. R. M. Radok. Gronigen: Noordhoff.
214. Nguyen, V. U. and Chowdhury, R. N. 1985. Simulation for risk analysis. Gcotechnique 35(1), 47-58.
215. Nickson, S. D. 1992. Cable supportguidelinesfor underground hard roúz mine operations. MASc. thesis,
Dept. Mining and Mineral Processing, University of British Columbia.
216. Nielsen, N.M., Hartford, D. N. D. and MacDonald. 1994. Selection of tolerahle risk criteria for dam
safes. decision making. Proc. 1994 Canadian Dum Saf.5 Conferente, Winn~peg,Manitoba. Vancouver:
BiTech Publishers, pp 355-369.
217. Obert, L and Duvall, W. I. 1967. Rock Mechanics andthe Design of Structures in Rock. New York: Wiley 65
pages.
218. Ortlepp, D. W. 1992. The design ofthe containment of rockhurst damage in tunnels - an engineering
approach. In Rock support in mining and undergound constrnction, proc. int. symp. on rock support, Sudbury,
(eds P. K. Kaiser and D. R. McCreath), 593-609. Rotterdam: Balkema.
219. Ortlepp, D. W. and Gay, N. C. 1984. Performance of an experimental tunnel subjected to stresses
ranging from 50 MPa to 230 MPa. Proc. gmp. ISRM on Design andi'erformance of UndergroztndExcava-
tions, Cambridgd. pp.337-346. London: British Geotechnical Society.
220. Ortlepp, W. D.1993. Invited lecture: The design ofsupport for the containment of rockburst dainage
in tunnels - an engineering approach. In Rock support in mining and undergound construction, proc. int.
symp. ou rock support, Sudbuty,(eds P. K. Kaiser and D. R McCreath), 593-609. Rotterdam: Ballíema.
221. Ortlepp, W. D., 1993. A philosophical consideration of support requirements in civil and mining
tunnels. In TUNCON '93, Proc. symp. Aspects ofcontrol in Tunnellzng,Johannesburg. 49-54.
222. Otter, J. R. H., Cassell, A. C. atid Hobbs, R .E. 1966. Dynamic relaxation. Proc. Instn Civ. Engrs 35,633-
665.
223. Pacher, F., Rabcewicz, L. and Golser, J. 1974. Zum der seitigen Stand der Gebirgsl<lassifizieruiigi11
Stollen-und Tunnelbau. Proc. XXII Geomech. colloq., Salzburg, 51-58.
224. Palmstrom, A. 1982. The volumetric joint count - a useful and simple measure of the degree of rock
jointing. h c . 4th congr. Int. Assn Engng Geol., Delbi 5,221-228.
225. Patton, F. D. 1966. Multiple modes of shear failure in rock. Proc. 1st congr. Int. Soc Rock Mech., Lisbon
1, 509-513.
226. Pearce G. E. 1988. Report on the Proc. 5th Int. Tunnelling Symp., Tunnelling 88, Fans Instn Min.
Metall. (Sect. A: Min. indust?y)97, A149-159.
227. Pelli, F., Kaiser, P. K. & Morgenstern, N. R. 1991. An interpretation of ground movements recorded
during construction of the Donkin-Morien tunnel:Can. Geotech.J. 28(2), 239-254.
228. Pierson, L. A,, Davis, S. A. and Van Vidcle, R. 1990. Rodífall Hazard Rating System Implementatioti
Manual. Federal Highway Administration (FHWA) Report FHWA-OR-EG-90-01. FHWA, U. S. De-
partrnent of Transportation.
229. Pine. R. J. 1992. Risk analysis design applications in mining geomechanics. Trans. Inst Min. Metull.
(Sect.A) 101, 149-158.
230. Pomin, Y 1988. EmpiricalopenstopedeslgB in Canada. Ph.D. thesis, Dept. Mining and Mineral Process-
ing, University of British Columbia.
231. Potvin, Y. and Milne, D. 1992. Empirical cable bolt support design. In Rock Support in mining and
undergroundconstruction,proc. int. symp. on rocksupport, Sudbury, (eds P. K. Kaiser and D. R. McCreath),
269-275. Rotterdam: Balkema.
232. Potvin, Y, Hudyma, M. R. and Miller, H. D. S. 1989. Design guidelines for open stope support. Bull.
Can. Min. Metall. 82(926), 53-62.
233.
-~~ Priest. S. D. aiid E. T. Brown. 1983. Probabilistic stabilitv analvsis of variable rock slooes. Trdns. Inst.
Min. Metall.(Sect.A) 92: 1-12,
234. Rabcewicz, L. 1969. Stabilit~of tunnels under rock load. L%terPouier21(6-8) 225-229,266-273,297-304.
235. Read, J. R. L. and Lye, G. N. 1983. Pit slope design methods, Bougainville Copper Liinited open cut.
h c . 5th Cong. ISRM., Melbourne pp. C93-C98. Rotterdam: Balkema.
236. Riemer, W., Pantzartzis, P., Krapp, L. and Scourtis. C., Investigation and monitoring of landslides at
the Polyphyton project in Greece. &c. Inml. Symposium on Landslides, Trondheim. 1996.
237. Ritchie, A. M., 1963. The evaluation of rockfall and its control. Highway Record Vol 17.
238. Ritter, W. 1879. Die Statik der Tunne[gcwolhe.Berlin: Springer.
239. Robbins, R. J. 1976. Mechanized tunnelling - progress and expectations: 12th Sir Julius Werhiier Me-
morial Lecture. In Tunnelling '76 (ed. M. J. Jones), xi-xx. London: Institution Mining and Metallurgy.
240. Rase, D. 1985. Steel fibre reinforced shotcrete for tunnel liiiings: the state of the art. Proc. North
American rapid excau. tunnelingconf: 1, 392-412. New Yorlí: Soc. Min. Engrs, Ain. Inst. Min. Metall.
Petroltn Engrs.
241. Rosenbleuth, E. 1981. Two-point estimates in probabi1ities.J Appl. Math. ModeUing5, October, 329-335.
242. Salamon, M. D. G. 1974. Rock mechanics of underground excavations. In Advances zn rock mechanics,
Proc. 3rd Cong. I S R M , Denuer lB, pp.951-1009. Washington, DC: National Academy of Sciences.
 ieertng COI

243. Salamon, M. D. G. and Munro, A. H. 1967. A study of the strength of coa1 pillars. J. S. A?. Inst. Min.
Metall. 65, 55- 67.
244. Salainon. M. D. G. and Oravecz. K. I. 1976. Rock mechanics in coalmining.Johannesburg: Chamber of
Mines of South Africa. 119 pages.
245. Sarma, S. K. 1979. Stability analysis ofemhankments and s1opes.J Geotech.Eng. Div., ASCE. 105(GT12),
1511- 1524.
246. Saviii, G. N. 1961. Stress concentrations aroundholes. London: Pergamon.
247. Schmuck, C. H. 1979. Cable bolting at the Homestake gold mine. Mining Engincering, Deceinber,
1677-1681.
248. Scott, J. J. 1976. Friction rock stabilizers - a new rodc reinforcement method. In Monograph on rock
mechanicsapplicationsinmining,(eds W.S. Brown, S. J. Green and W. A. Hustrulid), 242-249. New York:
Soc. Min. Engrs, Ain. Inst. Min. Metall. Petrolm Engrs.
249. Scott, J. J. 1983. Friction rock stabilizer iinpact upon anchos design and ground control practices. In Rock
bolting.theoy andapplication in undergroundconstruction,(ed. O. Stephansson), 407-418. Rotterdain: Balkema.
250. Serafim, J. L. and Pereira, J. P. 1983. Consideration of the geoinechanical classification ofBieniawslci.
Proc. int. sytnp. on engieeringgeologv and underground construction, Lisbon l(II), 33-44.
251. Shah, S. 1992. A sturly of thc behaviour ofjointed rock masses. Ph. D. thesis, Dept. Civil Engineering,
University of Toronto.
252. Sharp, J. C., Smith, M. C. F., Thoms, I. M. and Turner, V. D. 1986. Tai Koo Cavern, Hong Kong -
performance of a large metro excavation in a partially weathered rode mass. In LargeRock Caverns (ed.
K. H. O. Saari). 1, pp. 403-423. Oxford: Pergamon.
253. Sheory, P. R. 1994. A theory for in sitn stresses in isotropic and transversely isotropic rode. IniJ Rock
Mech. Min. Sci & Geomech.Abstr. 31(1), 23-34.
254. Shi G. H. and Goodman R. E. 1989. The key blocks of unrolled joint traces in developed maps of
tunnel walls. 1nt.J NumcricalandAnaiyticalMethods in Geomechanio 13, 131-158.
255. Shi G. H. and Goodman, R. E. 1981. A new concept for support of underground and surface excava-
tions in discontinuous rocks based on a keystone principie. In Rock mechanicsfiom Reseurch to Applcd-
tions. Proc. ZZnd. U.S. Symp.fòrRock Mechanics, Camhridge. pp. 290- 296. Cainbridge, Mass.: Massachu-
setts Institute for Technology.
256. Silveira, A. F. 1990. Some consideratioiis on the durability of dains. WatcrPouerandDam Construc-
tion, 19-28.
257. Soas, I. G. K. 1979. Uplift pressures in hydraulic structures. Water PoucrandDam Construction. 31(5)
21-24.
258. Spang, R. M. and Rautenstrauch, R. W. 1988. Empirical and mathematical approadies to rockfall
prediction and their practical applications. Proc. 5'"InternationalSymposium on Landslides, Lusanne. Vol.
2. 1237-1243.
259. Startzinan, R. A. and Wattenbarger, R. A. 1985. An improved computation procedure for risk analysis
problems with unusual probability functions. Proc. symp. Soc. Petrolm Enga hydrocarbon economics and
nialuation, Dallas.
260. Stillborg, B. 1994. Pro,+issional users handbookfòrrock bolting, 2nd edn. Clausthal- Zellerfeld: Trans Tecli
Publications.
261. Svanholm, B. O., Perssoii, P-A. and Larsson. B. 19n. Smooth blasting for relialille underground openings.
111Storagein cxcauatedrockcavems, Rochrtore 77:Storqe(ed. M. Berginan) 3, . . 573-579. Oxford: Per~ainoii.
pp.
262. Talobre, J. 1957. La mecanique des raches. Paris: Dunod.
263. Tatchell, G. E. i'991. Automatic data acquisition systems for monitoring dams and landslides. hoc. 3rd
Int. symp. on FieldMeasurements in Geomechanics, Oslo, Norway; in press.
264. Terzaghi, K. 1725. Erdbaumechanik auf Bodenphysikalischer Grundlage. Vienna: Franz Deuticke.
265. Terzaghi, K. 1736. Presidential Address. Proc. IstInt Conq!forSoilMechanics andFoundations Engneering,
Cumbridge, Masr. 1, 22-3.
266. Terzaghi, K. 1945. Stress conditions for the failure of saturated concrete and rock. Proc. Am. Soe %I.
~ u t e45,
i 777-801.
267. Terzaghi, K. 1746. Roclc defects and loads on tunnel supports. In Rock tunnelingzeith steelsnpports, (eds R.
V. Proctor aiid T. L. White) 1, 17-97. Youngstown, OH: Coininercial Shearing and Stainping Company.
268. Terzaghi, K. and Richart, F. E. 1752. Stresses in rodí ahout cavities. Geotechnique 3, 57-90.
267. Terzaghi, R. and Voight, B. 1979. Karl Terzaghi on rocl<slides: the perspective of a half-century. In
Rochrlides undAvalanches (ed. B. Voight), Part 2, pp. 111-131. New York: Elsevier.
270. Thorn, L. J. and Muller, D. S. 1964. Prestressed roofsupport in underground engine chambers at Free
State Geduld Mines Ltd. Eans. Assn Mine Mngrs S. A f , 411-428.
271. Tyler, D. B., Trueman, R. T. and Pine, R. J. 1771. Rodíbolt support design usiiig a prohaliiilistic method
of líey bloclí analysis. In Rock mechanics as a multidiscz$linary science, (ed. J . C. Roegiers), 1037.1047.
Rotterdam: Balkema.
272. Tyler. D. B., Trueman. R. and Pine. R. J. 1991. Rockbolt support design using a probahilistic metliod
of líey hlock analysis. Proc. 32nd U S . Symp. Rock Mechanics, Normun, Oklahoma, 1037-47.
273. Vandewalle, M. 1993. Dramix: Tunnellingthe world. 3rd edn. Zwevegem, Belgiuin: N.V. Belcacrt S.A.
274. Vaninarcke, E. H. 1980. Probalistic analysis of earth slopes. Engineering Geology 16: 29-50.
275. Varnes, D. J. 1784. Landslide hazard zonation: a reviewofprinciples and practice. NaturulHazards 3.
UNESCO, Paris, 63 pages.
276. Vogele, M., Fairhurst, C. and Cundall, P. A. 1978. Analysis of tunnel support loads using a large
displaceinent, distinct hlock model. In Storage in excavated rock cuverns (ed. M. Bergman) 2, pp. 247-
252. Oxford: Pergamon.
277. von Karman. Th. 1911. Festiglíeitsversucheuntei allseitigem Drudí. ZNtd VerDeuticherIng.55, 1749-1757.
278. von Kiininelmann, M. R., Hyde, B. and Madgwidí, R. J. 1784. The use of coinputer applications'at
BCL Liinited in planning pillar extraction and the design of inine layouts. In Design andpe$armunce of
underpundexcavutionr, (eds E. T. Browii and J. A. Hudsoii), 53-64. London: Brit. Geotecli. Soc.
277. VSL Systems Ltd. 1982. Slab post tensioning. 12p. Switzerlaud.
280. Warburton, P. M. 1781. Vector stability analysis of ali arbitrary polyhedral blodí with any nuinber of
free faces. Int. J. Rock Mech. Min. S i & Geomech.Abstr. 18, 415-427.
281. Whitinaii. R. V. 1784. Evaluating calculated risb in geotechnical engineering. J, Geotech. Enng, ASCE
110(2), 145-186.
282. Wickhain, G. E., Tiedemann, H. R. and Slíinner, E. H. 1972. Support deterinination hased on geologic
predictions. In Proc. North Americun rupidexcav. tunneling conf, Chicago, (eds K. S. Lane and L. A.
Gartield).,, 43-64. New York: Soc. Min. Eiiers.
" , Ain. Inst. Miii. Metall. Petrolm Eiiers.
~ ~~~

"~~
283. Windsor, C . R. 1990. Ferruledstrand. Unpublished inemorandum. Perth: CSIRO.
284. Windsor, C . R. 1792. Cable holting for underground and surface excavations. In Rocksupportin minin~
and underpund conshuction, proc.~int.symp. on rock support, Sudbury, (eds P. K. ai ser and D. R
McCreath), 347-376. Rotterdam: Balkema.
285. Wittke, W. W. 1765. Method to analyse the stability ofrock slopes with and without additional load-
ing. (in Gerinan) Felsmechunik undIngericurgeologi, Supp. 11,30, 52-77. English translation iii Imperial
College Rock Mechanics Research Report, no. 6, July 1771.
286. Wood, D. F. 1772. Specification and application of fibre reinforced shotcrete. In Rocksupportin miniq
and undergound construction, proc. int. symp. on rock support, Sudbury, (eds. P. K. Kaiser and D. R.
McCreath), 147-156. Rotterdam: Balkema.
 Rock Engin!Zering. Cou

,287. Wood, D. F., Banthia, N. and Trottier, J-F. 1993. A comparative study of different steel fibres in
shotcrete. In Shotnetefor under~oundsupportVI,Niagara Falls, 57-66. New York: Am. Soc. Civ. Engrs.
288. Worotnicki, G. and Walton, R. J. 1976. Triaxial 'hollow inclusion' gauges for determination of rode
stresses in situ. Proc symp. ISRM on Inuestigation of Shcsr in Rock, Sjidnq. Supplement 1-8. Sydney:
Institution of Engineers, Australia.
289. Yazici, S. and Kaiser, P. K. 1992. Bond strength of grouted cable boits. IntJ. Rock Mcch. Min. Sci. &
Geomech.Abrtr. 29(3), 279-292.
290. Zheng, Z., Kemeny, J. and Cook, N. G. W. 1989. Analysis of borehole breakouts.j. Geophys. Res.
94(B6), 7171-7182.
291. Zobadc, M. L. 1992. First- and second-orderpatterns ofstressin the lithosphere: the World Stress Map
Project.j. Gcophys. Rcr. 97(B8), 11761-11782.
 Acknowledgements
Some of the material contained in Chapters 3, 4; 5, 10, 11, 14 and 15 from a hool< by Hoelc, E,
Kaiser, P. K. and Bawden W. F. entitled Supportof Underg/oundExcavationsin HardRock that was published
hy A. A. Ballceina of Rotterdam in 1995. Permission from Mr A. A. Ballcema to reproduce this material is
gratefully acknowledged.
An order for this hoolc is reproduced overleaf. This order form should he sent to one of the following
addresses:

A. A. Balkema Puhlishers, P.O. Box 1675,3000 BRRotterdam, Netherlands

(Fax+31.10 4135947)
For customers in USA and Canada
A. A. Balkema Publishers, Old Post Road, Brookfield, VT 05036-9704
(Fax 802 276 3837)

A. A. Balkema Puhlishers can also he reached hy email at hallceina@ball<ema.nl and on the Internet a t
http://www.halkema.nl.

Some of the material contained in Chapter 11 on Rock Mass Properties is from a páper and a techni-
cal note puhlished in the InternationalJournal of Rock Mehanics and Mining Sciences and perinission from
Elsevier Science to re~roducethis material is gratefully acknowledged. The references are:

Hoelc, E. and Brown, E. T. Practical estimates of rock mass strength. Int.J. Rock Mech. Min. Sei. Vol. 34,
No. 8, pp 1165-1186, 1995.

Hoelc, E. Reliability of Hoek-Brown estimates of rock mass properties and their impact on design.
Int. J. Rock Mech. Min. Sei. Vol. 35, No. 1, pp 63-68, 1998.

The InternationalJoumal ofRock Mechanics and Mining Sciences can be accessed on the Internet at
http://www.elsevier.nl/inca/puhlications/store/2/5/6ndex.htt.
 A. A. Balkema Publishers

Order Form:
Hoelc, E., P. K. Kaiser & W. F. Bawden:
Support of underpund excavations in hard rock
1995,28 cm, 232 pages

I-Iard cover edition (90 5410 186 5) Dutch fl.95 / US $45.00 / UK £32
Student edition (paper baclc) (90 5410 187 3) Dutch fl.45 / US $10.50 / UK £16

A comprehensive volume dealing with the design ofrodcbolts, dowels, cables bolts a ~ i dshotcrete for
underground excavations in hard rock. Many practical examples are given and extensive use is inade of user-
friendly software developed spec*fically for this application (available separate). Topics covered include roclc
inass classification systems, shear strength of discontinuities, analysis of structurally controlled failnres, in
situ and included stresses, estimating rode mass strength,' support design for overstressed rock as well as
discussions on different types of underground support.

Please fill out the following to arder the book:

Penonal Inhrmatioil (Please print)

Name:
Address:
City:
Zipcode:
Country:
Telephone:
Telefax:
e-inail:

Type of payement
pro-forma invoice (postage will be charged extra)
Travellers cheque
Personal cheque (drawil o? a bank in USA or,UK)
Credit card (Deliveries are made only to the address of the account or credit card holder)

Card Type and Number: ........................ /, ..........................................

Expity Date: ......................... .
.........................................................

Signature: ................... .
...............................................................
Banlcers: ING Bank, Schiekade, Rotterdam. Account No.: 69.52.62.904
 A publicaqtlo deste livm, dc autoria do
Hoek, irA certamente aprimorar o nlvcl de M
atual do projeto e construflo de tCmcis. Trata
apenas de uma fonte de inform
qualidade, como tambkm um
refletindo a qualificaçtlo ímpar do
a obras subterrâneas
Recentemente, o projeto e
subterrâneas passaram a enfatiza
segurança, custos e meio ambiente.
reduçzo de custos das abras, a atual
relaçtlo ao meio ambiente, e novos mCto
que permitem otimizar o p j e t o e torna-
requerem uma maior utilizaçio
de engenharia, no projeto e con
subterrâneas.
Mtlo obstante os novos
pelos avanços recentes na infor
municaqóes, o aprendizado de novas ttc
engenharia ainda e feito de forma clássica.
de novaç tecnicas se dB atr
semindrios, congressos, e principalm
ttcnicas de qualidade. Cientes de q
natureza e pelo ser humano ser00
da Engenharia Civil no futum, os
livm esperam que o objetiw de divuigaro seu
tecnológih seja alcantado de forma rlrpida e

VETEC Engenharib
VENCE Engenharia e Emp~end
1
.
 Rock Engineering
The Application of Modern Techniques to Underground Design
The text presents a series o f case histories, each dcsigned to bring out specific aspects
o f the application o f modern rock engineering techniques to tunnel design.
While many fundamental issues are discussed, they are presented from a pratica1 point o f view
for use by geologists and engineerj who are concerned with real tunnclling problems.
e case histories, drawn from Dr Hoek's very wide international experienre, covered the rangf from
extremely poor quality rocks to hard jointed rock masses and included veiy shallow tunnels
as well as very deep tunnels. Site investigation and rock m a s classification techniques
are reviewed and the estimation of rock mass properties is discuswd. Special attention is given
esign of the excavations associated with hydroelectric projects and underground metro developments