Civil - Eng 2004 3

Vilnius Gediminas Technical University Lithuanian Academy of Sciences
Journal of Civil Engineering

and Management
2004, Vol X, No 3
Vilnius Technika 2004
ISSN 1392-3730
EDITORIAL BOARD
Editor-in-Chief
Prof Edmundas K. ZAVADSKAS, Lithuanian Academy of Sciences,
Vilnius Gediminas Technical University, Saultekio al. 11, LT-10223 Vilnius-40, Lithuania
Editors
Managing editor
Assoc Prof Darius BAINSKAS, Vilnius Gediminas Technical University,
Saultekio al. 11, LT-10223 Vilnius-40, Lithuania
Dr Rogerio BAIRRAO, Portuguese National Laboratory for
Civil Engineering, Av. Brasil, 101, 1700-066 Lisboa, Portugal
Prof Gyrgy L. BALZS, Budapest University of Technology
and Economics, Megyetem rkp.3, H-1111 Budapest, Hun-
gary
Assoc Prof Erik BEJDER, Aalborg University, .ibigerstraede
16, 9220 Aalborg, Denmark
Prof Adam BORKOWSKI, Institute of .undamental Techno-
logical Research, Switokrzyska 21, 00-049 Warsaw, Poland
Prof Micha BOLTRYK, Biaystok Technical University,
Wiejska 45A, 15-351 Biaystok, Poland
Prof Patrick J. DOWLING, .elow Royal Society, University
of Surrey, Guildford GU25XH, UK
Prof Aleksandr A. GUSAKOV, Moscow State University of
Civil Engineering, Dorogomilevskaja, 5/114, 121059 Moscow,
Russia
Prof Boris V. GUSEV, International and Russian Engineering
Academies, Tverskaja 11, 103905 Moscow, Russia
Assoc Prof Edward J. JASELSKIS, Iowa State University,
Ames, IA 50011, USA
Prof Oleg KAPLISKI, Poznan University of Technology,
Piotrovo 5, 60-965 Poznan, Poland
Prof Herbert A. MANG, Austrian Academy of Sciences,
Vienna University of Technology, Karlsplatz 13, A-1040
Vienna, Austria
Prof Antanas ALIKONIS, Vilnius Gediminas Technical Uni-
versity, Saultekio al. 11, LT-10223 Vilnius-40, Lithuania
Prof Juozas ATKOINAS, Vilnius Gediminas Technical
University, Saultekio al. 11, LT-10223 Vilnius-40, Lithuania
Prof Algirdas E. IAS, Vilnius Gediminas Technical Uni-
Assoc Prof Juozas DELTUVA, Kaunas University of Tech-
nology, Student g. 48, LT-3028 Kaunas, Lithuania
Prof Romualdas GINEVIIUS, Vilnius Gediminas Technical
Prof Arvydas JUODIS, Kaunas University of Technology,
Student g. 48, LT-3028 Kaunas, Lithuania
Prof Prancikus JUKEVIIUS, Vilnius Gediminas Techni-
cal University, Saultekio al. 11, LT-10223 Vilnius-40,
Lithuania
Prof Rimantas KAIANAUSKAS, Lithuanian Academy of Sci-
ences, Vilnius Gediminas Technical University, Saultekio al.
11, LT-10223 Vilnius-40, Lithuania
Prof Gintaris KAKLAUSKAS, Vilnius Gediminas Technical
International Editorial Board
Prof Rene MAQUOI, University of Liege, Building B52/3,
Chemin des Chevreuils 1, B 4000 Liege, Belgium
Prof Yoshihiko OHAMA, Nihon University, Koriyama,
.ukushima-Ken, 963-8642, Japan
Prof .riedel PELDSCHUS, Leipzig University of Applied
Science, 132 Karl Liebknecht St, 04227 Leipzig, Germany
Prof Karlis ROCENS, Latvian Academy of Sciences, Riga
Technical University, zenes str. 16, Riga, LV-1048 Latvia
Prof Les RUDDOCK, University of Salford, Salford, Greater
Manchester M5 4WT, UK
Prof Miroslaw J. SKIBNIEWSKI, Purdue University, West
Lafayette, Indiana 47907-1294, USA
Prof Martin SKITMORE, Queensland University of Techno-
logy, Brisbane QLD 4001, Australia
Prof Zenon WASZCZYSZYN, Cracow University of Techno-
logy, Warszawska 24, 31-155 Krakow, Poland
Prof .rank WERNER, Bauhaus University, Marienstrasse 5,
99423, Weimar, Germany
Prof Nils-Erik WIBERG, Chalmers University of Technology,
SE - 412 96 Gteborg, Sweden
Prof Ji WITZANY, Czech Technical University, Prague,
Thkurova 7, CZ 166 29 Praha 6, Czech Republic
Local Editorial Board
Prof Stanislovas KALANTA, Vilnius Gediminas Technical
Prof Ipolitas Z. KAMAITIS, Lithuanian Academy of Sciences,
Vilnius Gediminas Technical University, Saultekio al. 11,
LT-10223 Vilnius-40, Lithuania
Prof Romualdas MAIULAITIS, Vilnius Gediminas Techni-
cal University, Saultekio al. 11, LT-10223 Vilnius-40,
Lithuania
Prof Gediminas J. MARIUKAITIS, Vilnius Gediminas Tech-
nical University, Saultekio al. 11, LT-10223 Vilnius-40,
Lithuania
Prof Josifas PARASONIS, Vilnius Gediminas Technical Uni-
Prof Vytautas STANKEVIIUS, Lithuanian Academy of
Sciences, Lithuanian Institute of Architecture and Building
Construction, Tunelio g. 60, LT-3035 Kaunas, Lithuania
Prof Vytautas J. STAUSKIS, Vilnius Gediminas Technical
University, Trak g. 1/26, LT-01132 Vilnius, Lithuania
d n a s l a i r e t a M g n i d l i u B
s e r u t c u r t S
, s c i s y h P d n a s c i n a h c e M l a r u t c u r t S
s e i g o l o n h c e T n o i t a m r o f n I
d n a y g o l o n h c e T n o i t c u r t s n o C
t n e m e g a n a M
f o r P . K s i n o r d u A S A R A D E V K ,
, y t i s r e v i n U l a c i n h c e T s a n i m i d e G s u i n l i V
, 1 1 . l a o i k e t l u a S
a i n a u h t i L , 0 4 - s u i n l i V 3 2 2 0 1 - T L
f o r P s a d l a u m o R S Y U A B ,
, 1 1 . l a o i k e t l u a S
f o r P K s a r t r A S A K S U A L K A ,
, 1 1 . l a o i k e t l u a S
169
169
2004, Vol X, No 3, 169175
JOURNAL OF CIVIL ENGINEERING AND MANAGEMENT
http:/www.vtu.lt/english/editions
ISSN 13923730
REVIEW O. .IRE HAZARD ANALYSIS O. IGNALINA NUCLEAR POWER PLANT
Aleksandr Alejev
1
, Meislovas Grikeviius
2
, Ritoldas ukys
3
, Ian .rith
4
, Antti Norta
5
1
Lithuanian State Nuclear Power Safety Inspectorate (VATESI), ermukni g. 3, LT-01106 Vilnius, Lithuania.
E-mail: alejev@vatesi.lt
2
Ministry of Interior

Affairs, R. Jankausko g. 2/28, LT 2050 Vilnius, Lithuania.
E-mail: m.griskevicius@vpgt.lt
3
Vilnius Gediminas Technical University, Saultekio al. 11, LT-2040 Vilnius,
Lithuania. E-mail: ritis@st.vtu.lt
4
NNC Limited, Booths Hall, Chelford Road, Knutsford, Cheshire, WA16 8QZ
England. E-mail: ian.frith@nnc.co.uk
5
Enprima Limited, POB 61 (Rajatorpantie 8), .IN-01601 Vantaa, .inland.
E-mail: antti.norta@enprima.com
Received 18 .eb 2004; accepted 28 June 2004
Abstract. .ire safety is important throughout the lifetime of a plant, from design to construction and commissioning,
throughout plant operation and to decommissioning. The article is the document that will present the review results of
the documents by VATESI during licensing Unit 1, 2 and provides information how Ignalina NPP to meet State require-
ments and international recommendations to achieve and maintain satisfactory fire safety. .ire hazard analysis of opera-
tion RBMK-1500 type reactor is done for first time in the world practice.
Keywords: safety, fire safety, nuclear power plant, fire hazard analysis (.HA), protection, licensing, analysis, review,
safety improvements.
1. Introduction
Ignalina Nuclear Power Plant is Lithuanias the only
nuclear power plant. It dominates electricity production
in Lithuania; in 2003, INPP contributed 80,6% of total
generated output. Consequently, the Lithuanian economy
is heavily dependent upon this plant. INPP was built as
part of the Soviet Unions North-West Unified Power
System. It consists of two units, commissioned in De-
cember 1983 and August 1987. Both units of INPP are
Soviet designed RBMK reactors, and are different from
those operating in Russia and Ukraine. They have a de-
sign power of 4800MW
th
1500MW
e
per unit, but
currently each is restricted to 4200MW
th
1300MW
e.
Immediately after the Chernobyl accident, the first inter-
national programmes of safety evaluation of the RBMK
concept were undertaken, and a significant number of
bilateral initiatives was started too. A number of techni-
cal areas were identified for short-term safety improve-
ments, as well as for further in-depth safety analyses
[17]. .ire is one of the most hazardous events that may
affect the safety of a nuclear power plant. This is be-
cause the complexity of the effects of a fire. When
designing a nuclear power plant this fact is taken into
consideration by different means. .or example, rooms
housing important equipment is designed as separate fire
compartments, cable routings are separated by distance
and safety systems are physically separated.
During the lifetime of a nuclear power plant im-
provements are implemented to meet modern fire pro-
tection requirements. A tool to systematically look for
weaknesses regarding fire safety is a .ire Hazard Analy-
sis (.HA) [3, 5].
The Visaginas Township and the Ignalina NPP .ire
Protection Department are responsible for operational fire
fighting and rescue service at INPP. This department is
a structural sub-unit within the .ire Protection Depart-
ment of the Ministry of Internal Affairs, and consists of
three fire teams located outside the plant. As a contracted
service to INPP, this department also provides a fire in-
spection team located at the plant. The main functions
of the team are:
supervision of activities bearing the fire risk;
fire prevention and provision of technical and ad-
ministrative measures for fire fighting at the Ignalina
NPP;
supervision of the state of automatic fire fighting
systems;
supervision of the execution of remedial measures
aimed at fire protection enhancement and analysis
of the cases of fire.
The fire inspection function at INPP is arranged as
a 24-hour job performed by a team of inspectors who
conduct their activities based on relevant instructions and
.ire Safety Rules for INPP. In the event of a fire at the
INPP, the plant shift supervisor is in charge of operation
170
170 A. Alejev, et al / JOURNAL O. CIVIL ENGINEERING AND MANAGEMENT 2004, Vol X, No 3, 169175
until the officer of the .irst Brigade arrives at the plant.
Ignalina NPP must also report to the fire protection team
on cases of fire safety violations.
The fire relevant aspects in defence in depth con-
cept are also included in accordance with international
practices. This includes INPP measures to prevent the
occurrence of fire, a system analysis of fire protection
installations credited as well as the general and credited
manual fire fighting capability. .or the justification and
specification of the design measures for fire protection
of a NPP, and a definition of its adequacy against the
established requirements, the operating organisation
should carry out an overall fire hazard analysis.
Principles of fire protection were implemented in
the design and justified in the documents the technical
safety justification of reactor facilities (TOB RU) and
plant (TOB AS).
In comparison with the design documentation de-
veloped in the late 70s and early 80s, much fire protec-
tion work has been carried out at units, and more is in
progress.
Programme and measurements to improve fire safety
of INPP have been realised since the moment of Units
commissioning. Examples of such fire safety improve-
ments can be Program B-5 of European bank of Re-
construction and Development, SIP-2 Program.
2. .HA Unit-1 and Unit-2
Because no fire hazard analysis, corresponding to
Western approaches, had been carried out, VNIPIET
(Saint-Petersburg), Lithuanian Energy Institute, SiP/
SwedPower (Sweden), AEA-Technology (UK) and INPP
experts carried out a .HA for Units during 19982003.
Among measures to guarantee nuclear safety for suc-
cessful licensing of INPP Unit 1 it was necessary to per-
form an analysis of the existing fire protection. .HA for
INPP Unit 1 has been realised in two phases. In .HA
Phase 1 performed by VNIPIET, analysis of selected very
important main rooms with Safety Systems (SS) elements
was performed. After Phase 1 of INPP .HA measures
have been taken to eliminate weaknesses that were dis-
covered and shown. In this second phase of the .HA a
deeper analysis of the consequences of a fire has been
performed [811].
Primary goal of the .ire Hazard Analysis for Unit
1 of Ignalina NPP is to analyse and draw conclusions on
the existing fire protection measures from the point of
view of INPP nuclear safety. The primary goal can also
be described as to demonstrate documentary that exist-
ing fire protection measures guarantee INPP Unit 1
nuclear safety.
Phase 1 of the .ire Hazard Analysis for INPP Unit
1 (.HA1/1) [9] was completed in 1998. The report was
accepted by VATESI but with remarks. The long-term
operation license issued by VATESI for Unit 1 includes
the condition that the .HA should be finalised with com-
plete analysis depth and complete scope of rooms. Dur-
ing .HA Phase 1, 531 room with elements of safety re-
lated systems that provide reactor shut down, residual
heat removal and limit of radioactive release into the
environment were analysed. External fire influence on
main building 101/1 and rooms with elements of sys-
tems analysed, ie adjacent rooms and rooms with high
fire load (warehouses, rooms with combustible lubricants,
etc) were not considered.
The main goal of the .HA for Unit 1 was to analyse
and make conclusions on the sufficiency of the existing
fire protective measures to guarantee nuclear safety at
NPP. In order to do this, it was necessary to determine
what safety functions are needed and what components
provide that and their internal relations; and to analyse
whether a sufficient degree of safety functions remains
in case of any creditable fire in combination with any
single failure.
In Phase 2, performed by LEI (Kaunas) in coopera-
tion with SwedPower (Sweden), complex analysis has
been performed of all remaining rooms with SS and
Safety related systems (SRS) elements (including rooms
for Phase 1), rooms adjacent to all main rooms, and
rooms with fire risk (stores, rooms with combustible lu-
bricants, etc). Phase 2 is a continuation of Phase 1, but
not repetition.
The combination of highly skilled staff able to per-
form the engineering work as database development,
screening of rooms, analyses and conclusions combined
with a very good local knowledge facilitating collection
of plant specific data has made LEI a very appropriate
to do the .ire Hazard Analysis Phase 2.
The INPP project management has not only acted
as the client but has taken very active part throughout
the project in supporting LEI during the collection of
specific data for more than 2000 rooms. INPP has also
made important contributions in explaining the plants safe
shutdown paths and thereby giving the basis for conclu-
sions what systems and functions can be considered as
safety related equipment in the .ire Hazard Analysis.
SwedPower has been engaged in the project for
implementing Western standards in the study regarding
reactor safety strategies and fire analysis engineering.
SwedPower is a wholly owned subsidiary of Vattenfall,
the major Swedish power company and SwedPower is one
of the largest consultants in Sweden in the energy sector
with a lot of experience from the Swedish NPPs, includ-
ing competence in the field of reactor safety analysis.
The basis for realisation .HA was the Technical
specification for .ire Hazard Analysis for Power Unit
1 of INPP (Phase 2), which was prepared by INPP and
coordinated with VATESI.
The reasons for performing the second Phase of the
.HA have been:
the SAR for Unit 1 included recommendations to
perform a complete .HA. INPP therefore initiated
Phase 1 within the SIP-2 programme. It was con-
cluded that a second Phase is needed to fulfil the
initial ambitions;
171
license requirements for INPP Unit 1 included re-
quirements to complete the .HA as originally
planned.
The work mainly concentrates on demonstration of
nuclear safety in fire event and therefore treats such items
as:
sufficient knowledge of IAEA, Lithuanian and other
(international as well) normative documents, related
to fire safety for NPP;
scope and necessity to perform additional measures
to improve fire protection;
Sufficiency of the existing fire protection measures
to demonstrate nuclear safety of the NPP from fire point
of view.
.HA Phase 2 according to Technical specification
includes complex analysis of all rooms of Unit 1 with
SS and SRS. Thus the analysis includes both the rooms
from Phase 1, where some equipment changes were per-
formed, and additional safety related rooms. The studied
components safety is to directly shut down, cool down
the reactor facility, prevent and monitor of radioactive
releases into environment. The complex analysis also
includes rooms, adjacent to all main rooms with elements
of systems analysed and rooms with fire risk (store rooms,
rooms with combustible lubricants, etc), in which fire
directly or indirectly can make an influence on the rooms.
The purpose of the second Phase was to prepare
the final .HA based on data, results and conclusions
made under the second Phase as well as results gained
under Phase 1. The level of compliance with national
and international regulations was also analysed. Phase 2
make necessary updates of the Phase1 documents, com-
plete the in depth analyse of rooms included in Phase 1
as well as extend the surveys and in depth analysis to all
relevant premises of unit 1. The analytical approaches
and criteria, which are most well known in the world,
were to be used.
.ire Hazard Analysis contains the following sepa-
rate goals:
to identify items important to safety and their ar-
rangement; treated in the analysis of safety elements
of this .HA project;
to analyse the anticipated fire growth and the con-
sequences of the fire with respect to items impor-
tant to safety; treated in the screening and further
analysis of this .HA project;
to determine the required fire resistance of fire bar-
riers; treated in the screening analysis of this .HA
project. To determine the type of A.A detectors and
necessary protection means; treated in the screening
and further analysis of this .HA project;
to identify cases where additional fire separation or
fire protection is required, especially for common
mode failures, in order to ensure that items impor-
tant to safety will remain functional during and fol-
lowing credible fire, treated in the screening and
detailed analysis of this .HA project.
3. Detailed analysis of .HA
3.1. .HA structure
.HA work includes:
Safety function analysis imaging on what safety
functions/paths were used in Phase 1 and updating them
due to additional functions and changes in the plant. This
analysis included study of systems down to individual
component and electrical connection/cable level.
Data collection concerning all safety related com-
ponents and fire protection installations, and fire relevant
data for building structures and elements includes the
following items:
preliminary design data collection and putting it into
check list;
preparation of floor plans and sketches;
surveillance of checking the data and adding the new
data into check lists;
revision of checklists filling accuracy performed by
experts;
introduction of information into the database and
floor plans.
Surveys of rooms with safety related components,
adjacent rooms and rooms with fire risk, as well as re-
freshment of data for rooms from Phase 1 without sur-
vey and recording of all data into checklists and data-
base.
Screening of all the above rooms including devel-
opment of screening criteria and associated assessment
tools.
.urther analysis of rooms, which were not pos-
sible to be screened out, including additional surveys and
methodology development.
System analysis of fire protection systems credited
and assessment of capability to prevent fires and fight
manually.
Overall assessment of the fire/nuclear safety level
including assessment of adequacy and compliance, as well
as generation and priorities of recommendations.
The IAEA practice stipulates that all premises of
the plant shall be analysed and documented. This is to
ensure that all actual conditions of the plant are known
and to ensure that a fire starting at a remote room can-
not propagate to the safety paths or spread to a major
fire in general.
Phase 2 of the .HA for Unit 1 (.HA1/2) is there-
fore mainly aimed at making the analysis complete, par-
ticularly resolving the remarks and meet the requirements
from VATESI [1314].
.orest fire was found to be one of the most prob-
able external events to occur. Although it cannot have
direct impact on Ignalina NPP safety, but operation can
be disturbed by a smokescreen. This can also cause some
difficulties for the personnel as venting system takes fresh
air from outside and has no protecting sensors for out-
side smoke. Ignalina NPP has been built in region, where
30 % are occupied by forests. The edge of the closest
172
forest is apart less than one kilometre from territory of
Ignalina NPP. On the territory of the NPP, there are only
individual trees and grass. The global fire of a forest
with a high wind to the NPP side can cause the smoke
cover on the territory of INPP. The smoke does not in-
fluence the work of reactor machinery, but the smoke in
the NPP territory can complicate the staff work. .orest
fire probability in the INPP surroundings was calculated
directly from statistical data on forest fires in Lithuania
for 19931997.
Number of fires in different areas during 19931997
.ire area in hectares Number of fires
<5 3161
5-10 4
10-20 8
20-50 2
>50 1
.orest fire was indicated as the highest probability
having external effect. However, its impact on safety level
of the Ignalina NPP is limited and therefore no protec-
tive actions required. The probability of a forest fire in
places more than 20 ha in 10 km zone around Ignalina
NPP is equal to 2,7E-03 1/year. This probability is size-
able, but any external fire cannot essentially influence
the safety of a reactor (.ig 1).
The .HA report was reviewed by VATESI special-
ists and Western experts (NNC from UK and Enprima/
.ortum from .inland) and results of this analysis are
presented in report with recommendations.
Each section of the .HA was reviewed against the
Lithuanian requirements, IAEA criteria, the Technical
Specification and NNC/Enprima/.ortum experience to
ensure that a thorough analysis was carried out. Consid-
ering that works performed were evaluated positively, the
recommendations and comments were worked out, which
also were categorised according to their importance:
Category 1. These are recommendations which are
essential to the plant safety;
Category 2. These are recommendations which are
important to the plant safety;
Category 3. These are recommendations, which are
unlikely to affect the plant safety.
It is recommended by fire-technical expertise to re-
view room categories according to hazard of fire arise
and explosion considering Lithuanian law norms; to fin-
ish off check-lists and methodology of their infill, to
develop the computerised version of check-lists; to add
final .HA report with description of fire water supply
system, automatic fire detection system, fire extinguish-
ing and fire localisation system.
Based on this recommendations Technical specifi-
cation is developed for .HA Phase 2 on INPP, which
was co-ordinated with VATESI [1518].
3.2. Example: Room 206/5
This room is a compartment of electrical engineers,
which contains equipment for emergency electricity sup-
ply of some systems. This room has redundancy in two
different compartments. The room has dimensions and
properties: length, width and height. All the surrounding
properties are made of concrete and brick. Given room
was not analysed in detail during .HA for Unit 1.
The heat flux to the target, the flame height and
temperature above flame centreline are calculated.
Temperature in a metal construction can be calcu-
lated with a lumped heat capacity method when material
is thin. The material is thin when:
1 , 0
k
d h
, (1)
where h is heat transfer coefficient (W/(m
2
K)), d is the
thickness of the plate (m), k thermal conductivity
(W/(mK)).
Time t to reach a certain temperature can be calcu-
lated by solving:
,
_
c
t
t
g o
g
e
T T
T T
, (2)
A h
V c
t
p
c

, (3)
where
p
c is specific heat ((J/(kgK)), is density
(kg/m
3
), h is heat transfer coefficient (W/(m
2
K)), A
area (m
2
), V is volume of metal shield (m
3
),
g
T is hot
gas temperature (C), is surrounding temperature (C).
The fire duration can be calculated from equation:
Qdt H m
c
, (4)
where
is rate of combustion, m is mass of combus-

tible material (kg),
c
H is heat of combustion (kJ/kg),
Q is heat release rate (kW).
When calculating extinguishing possibilities, such
equations can be used:
f
e
m
m
REMP
, (5)
.ig 1. .ire area, hectares
173
where REMP is the characteristics of the extinguisher,
e
m is the flow of extinguishing media (kg/s),
f
m is the
flow of flammable gas (kg/s).
.low of flammable gas
f
m (kg/s) is found from
the formula:
c
f
H
Q
m
, (6)
where Q is heat release rate kW,
c
H is heat of com-
bustion (kJ/kg).
.low of available extinguishing media
a
m is calcu-
lated from the following expression:
a
c
a
t
S
m
, (7)
where
c
S is the mass of extinguishing media (kg),
a
t is
the activation time of the extinguisher (s).
Time for A.A to response is calculated using soft-
ware DETACT-T2. Temperature in the room is calcu-
lated with software C.AST. In the figure sketch of the
room is shown (.ig 2).
3.3. Scenarios used for .HA Room 206/5
Scenario 1, .ire in electrical cabinets
There are seven big electrical cabinets with com-
bustible material located in the room and one smaller
cabinet with a combustible material near the northern
wall. .ire can occur inside these cabinets and there is a
possibility for fire to spread to adjacent electrical cabi-
nets. In case of fire in the electrical cabinets the metal
shield will be exposed to flames and get heated. The hot
metal shield will influence the adjacent equipment due
to heat radiation and the adjacent equipment can get
damaged. There are three rows of electrical cabinets in
the room. Each cabinet in the row contains a combus-
tible material. It was assumed that heat release in a cabi-
net is constant and never exceeds 200 kW.
Two different gas temperatures, distance between
cabinets and distance between rows of the cabinets are
used in calculations. The influence of fire in this cabinet
on the adjacent equipment is also considered.
Scenario 2, .ire in cable tray
The room contains cables. Horizontal cables are
under floor and coated with .C. .ire in these cables is
not further analysed. The room contains one vertical cable
tray. All cables in this tray are in metal casing, but fire
can still develop due short-circuit, overheating or flash-
over. The cables may thus be involved in a fire, but it
seems that metal casing will prevent adjacent equipment
from fire and it also seems that heat of radiation from
metal casing will not be able to damage adjacent equip-
ment. There is one vertical cable tray in the room. The
design fire is based on a constant heat release rate of
186 kW/m
2
and a maximum burning length of 2 metres.
The distance between tray and nearest electrical cabinet
is used for calculations.
The room is equipped with automatic fire alarm
system but does not have automatic fire suppression sys-
tem. Calculations are performed for the alarm system,
which is not so sensitive.
.ire growth rate in the room is assumed to be me-
dium. Spacing between detectors is used for calculations.
Smoke detectors will be activating in a short time after
the fire beginning.
Scenario 3, .ire in the adjacent rooms
Analysing compartment from the adjacent rooms is
separated by concrete and brick walls, which at least are
rated for 150 min.
It can be concluded that the room is well protected
from fire from outside. High temperatures and heat of
radiation can damage cabinets. The level of damage will
also depend on the exposure time. The damage thresh-
old is approximately to 200C or 8 kW/m
2
.
Room 206/5 door leads to a corridor. During work-
ing hours usually people walk there. At night the corri-
dor is also inspected by operators. Since the compart-
ment is often inspected, we assume that staff will reach
the analysing room in case of fire in 3 min after detector
response. There are two OU-5, CO
2
extinguishers in the
room and one fire extinguishing hose outside room.
Results. Since the flow of available extinguishing
media is less than that of extinguishing media, fire can
not be suppressed with one extinguisher.
But there are two extinguishers in the room. Such
fire could be suppressed with both of them even if the
efficiency is 65 %; this means that two persons have to
be at the same time in the room. Since cabinets in this
room cannot be damaged by smoke, this case is not fur-
ther analysed, but it should be taken into account during
fire extinguishing in the room.
Room analysis with C.AST. According to the check-
list, information cabinets are vulnerable only to high tem-
peratures (and water). The temperature in the room is
determined by software C.AST. It appeared that critical
.ig 2. Room 206/5 sketch
174
temperature for cabinets in the room will be reached when
two cabinets will be burning at the same time. Since it
takes some time while one burning cabinet will ignite
the adjacent, and probably till that time personal will
arrive, then the probability that critical temperature will
be reached is low.
The analysis shows that the room is well protected
against fire from outside, thus fire will not spread from
the adjacent compartments and also fire from outside will
not damage the equipment placed in the analysing room,
due to a high fire resistance of room surroundings and
doors.
If fire starts in the electrical cabinets, the adjacent
row will be safe until burning cabinet remains closed.
To avoid fire spread between equipment located in the
room, personal has to react as fast as possible when the
signal that fire has begun in room will be received.
If the room will be lost in case of fire, there is re-
dundancy in two other compartments.
4. Summary and future perspectives
The present VATESI Regulatory Guidelines for the
Safety Related Systems of .HA are issued in co-opera-
tion with VATESI and Ministry of Internal Affairs and
some other authorities. The regulatory basis for .HA
Stage 2 is the requirement in the present operating li-
cense to complete Unit .HA. In the process of accept-
ing Stage 1, VATESI also presented some requirements
and recommendations for improving analysis. VATESI
has also agreed with the project about its requirements
for .HA. These requirements have been brought together
with project internal requirement and guidance to the
Technical Specification (TS) of .HA. VATESI on-site
representative has signed the TS after review and ap-
proval of Licensing Department (HQ) specialist.
These guidelines have been anticipated in the TS
and Project Manual of .HA.
Stage 1 started in 1997 with a limited scope (.ire
Protection system (.PS)) descriptions, and limited analy-
sis of safety systems and their support systems located
in 531 rooms. VNIPIET was acting as the main contrac-
tor assisted by INPP and SIP (Sweden) and AEAT (UK).
Stage 2 started last January and the .inal Report
will be issued in July. VATESI will review the report
and a revised report is issued by the end of 2002. .HA1/
2 Lithuanian Energy Institute (LEI) Laboratory of Nuclear
Engineering is the main contractor for INPP and SIP
assists it in a project organisation.
VATESI requires INPP to perform a .HA for Unit
2 for Licensing of the unit. It is based for the most part
on the .HA1/2 results. Similarities and differences be-
tween the units assessed and required complementary
studies accomplished.
The .HA 2 report was required by VATESI to be a
stand-alone document, but in practice, it is heavily de-
pendent on other documentation, in particular, the previ-
ous .HA 1/1 and .HA 1/2, the Engineering Assessment
for the .ire Protection System and the Single .ailure
Analysis for the .ire Protection System within Tasks 1
& 4 of SAR-2.
Namely, the .HA for Unit 1 was reviewed by
VATESI and its TSOs (NNC, UK), and conditions were
imposed such that a more detailed .HA, known as
.HA1/2 (Phase 2), be completed by INPP. This work is
now being concluded, and the recommendations from
.HA 1/2 were reviewed by VATESI and its TSOs
(VGTU and Enprima/.ortum from .inland), and from
the original Unit 1 .HA1/1 have been assessed in the
Unit 2 .HA.
Unit 2 .HA was submitted by INPP. NNC and
VGTU as a consultant supported VATESI in reviewing
the Unit 2 .HA reports. This support is largely consul-
tative in nature, but a review report prepared by NNC/
VGTU and submitted to VATESI.
The NNC review of the Unit 1 .HA 1/1 was car-
ried out against the IAEA standards. Since that review,
a new IAEA Safety Standard has been issued in 2000,
NS-G-2.1, .ire safety in the Operation of Nuclear Power
Plants. However, the standard is mainly concerned with
practical protection measures and the continuous upkeep
of the .HA. It also refers directly to the requirements of
50-P-9 as its basis, and is therefore considered unlikely
to place any additional requirements on the .HA itself
as applied to Unit 2.
.ire Hazard Analysis of Ignalina NPP adequately
presents the following IAEA criteria for nuclear safety
in the event of fire:
safe shutdown of the reactor;
residual heat removal;
prevention of radioactive releases above prescribed
limits.
The screening process is based on the single failure
criteria and also adapted to the fire protection measures.
In order to give a reasonable approach, the rooms are
categorised in different types depending on which im-
pact they can achieve on reactor safety in case of fire.
.rom the fire protection point of view changes in
the management and technical policy, in the emergency
preparedness and training of the INPP staff is an impor-
tant tool for operators to assure the reliability and safety
of Ignalina NPP. Moreover, it reduces the consequences
of possible events.
The fire protection at the INPP is clearly developed
to meet the highest standards, the .ire Hazard Analysis
report demonstrates that a high level of protection against
fires is provided and given that the recommendations
proposed in the report are implemented and the safety
issues discussed in this peer review report are addressed
the intent of the IAEA guidelines will be achieved.
The current .HA includes the complex analysis of
Unit 1 rooms with SS, SRS, NOSRS and .ES, safety of
which directly affects the shutdown and cooling of reac-
tor facility, prevent and control of radioactive substance
175
releases into the environment, and also adjacent rooms
to all basic rooms and rooms with fire risk (stores, rooms
with combustible lubricants, etc), fire in which directly
or indirectly can affect main rooms. Therefore it can be
concluded that the IAEA intention has been met to the
most extent.
The results demonstrated:
the adequacy of the existing fire protection mea-
sures to ensure the nuclear safety of the INPP (tak-
ing into account the improvement measures already
committed to);
the necessity for carrying out additional measures
to improve the general level of fire protection.
After comparison of fire protection systems of Units
1 and 2, analysis of performed measures for fire protec-
tion improvement on Unit 2 for the last 8 years, analysis
of recommendations of Unit 1 for their implementation
in Unit 2 and detailed analysis of most important rooms
with SS elements, the following conclusions may be
drawn:
Defense in depth of Unit at Ignalina NPP performs
its main functions for prevention, detection, liquidation
and reduction of fire consequences as it is shown during
fire hazard analysis of Unit 1 (.HA1/2). It ensures ac-
ceptable fire safety for related systems, performing its
main functions (safe shutdown of the reactor, residual
heat removal, control and prevention of radioactive re-
leases above the prescribed limits into the environment)
in case of fire with sufficient level of nuclear and radia-
tion safety for a limited period of time, determined by
fulfilment of the main fire safety requirements (mini-
mum fire resistance of fire barriers, presence of A.A
and A.. or manual fire fighting means). The analysis
revealed that differences between fire protection systems
of Units 1 and 2 are not essential. These are good indi-
cations, allowing for a conclusion that the above state-
ments are also valid for Unit 2 of INPP [11];
Analysis of measures performed for fire protection
improvement at INPP has shown, that most recommen-
dations for Unit 1 were implemented on Unit 2 at the
same time. Implementations of such measures on Unit 2
at INPP indicate the substantial improvement of fire
safety on the Unit on the whole, as well as from the
point of nuclear safety.
5. Conclusion
The current .HA includes the complex analysis of
Unit 1 rooms with SS, SRS, NOSRS and .ES, safety of
which directly affects the shutdown and cooling of reac-
tor facility, prevent and control the radioactive substance
releases into the environment, and also adjacent rooms
to all basic rooms and rooms with fire risk (stores, rooms
with combustible lubricants, etc), fire in which directly
or indirectly can affect the main rooms. Therefore it can
be concluded that the IAEA intention has been met to
the most extent.
References
1. .ire safety in the operation of nuclear power plants, Safety
Guide. No NS-G-2.1, Vienna, IAEA, 2001.
2. Inspection of fire protection measures and fire fighting
capability at nuclear power plants. Safety practices. Safety
series No 50-P-6, Vienna, IAEA, 1994.
3. Evaluation of fire hazard analyses for nuclear power plants.
Safety practices. Safety series No 50-P-9, Vienna, IAEA,
1995.
4. Assessment of the overall fire safety arrangements at
nuclear power plants. Safety practices. Safety series No
50-P-11, Vienna, IAEA, 1996.
5. Preparation of fire hazard analyses for nuclear power plants.
Safety reports series No 8, Vienna, IAEA, 1998.
6. Safety Evaluation Report. Licensing of Unit 1 of Ignalina
nuclear power plant. Vilnius, VATESI, 1999.
7. General regulations for nuclear power plant safety. Man-
agement documents. VD-B-001-0-97, Vilnius, VATESI,
1997.
8. .ire safety. Basic requirements (Gaisrin sauga. Pagrindiniai
reikalavimai). RSN 133-91. Vilnius, The Ministry of Con-
struction and Town Planning of Lithuania, 1992 (in
Lithuanian).
9. .ire automatics of buildings and constructions (Pastat ir
statini priegaisrin automatika). RSN 138-92. Vilnius,
The Ministry of Construction and Town Planning of
Lithuania, 1995 (in Lithuanian).
10. United States Department of Energys Reactor Core Pro-
tection Evaluation Methodology for .ires at RBMK and
VVER Nuclear Power Plants / Report of U.S. Department
of Energy. Revision 1. No DOE/NE-0113 (REV.1), 1997.
11. Quality assurance system requirements of nuclear power
plants and other energy objects. VD-KS-02-99. Vilnius.
VATESI, 1999.
12. Unit 1/1 fire hazard analysis of Ignalina NPP, 1998.
13. Unit 1/2 fire hazard analysis of Ignalina NPP, 2003.
14. Unit 2 fire hazard analysis of Ignalina NPP, 2003.
15. Ignalina NPP unit 2 review of safety analysis report. SAR-
2 Task No 6.3 .ire protection and fire hazard analysis
review report. VGTU, Vilnius, 2003.
16. Ignalina Unit 2 fire hazard analysis (.HA). Progress meet-
ing to discuss preliminary findings of NNC peer review.
Vilnius, 2003.
17. Ignalina 1 .HA stage 2 (.HA1/2) consultation for VATESI
review. Memorandum. .ortum Engineering. Vilnius, 2002.
18. Ignalina 1 fire hazard analysis (.HA) (.HA1/2) progress
meeting of the independent review. Report. Enprima Engi-
neering. Vilnius, 2003.
177
177
2004, Vol X, No 3, 177182
ISSN 13923730
THE E..ECT O. MATERIAL STRENGTH ON THE BEHAVIOUR O.
CONCRETE-.ILLED STEEL ELEMENTS
Janis Brauns
1
, Karlis Rocens
2
1
Dept of Structural Engineering, Latvia University of Agriculture, Jelgava, 19 Academia St., LV-3001, Latvia.
E-mail: braun@latnet.lv
2
Dept of Structural Engineering, Riga Technical University, Riga, 16 Azenes St.,
LV-1048, Latvia. E-mail: rocensk@latnet.lv
Received 1 March 2004; accepted 14 May 2004
Abstract. Composite columns and beams are a combination of concrete and steel elements realising the advantages of
both types of materials. According to codes for concrete-filled column, the plastic resistance of the cross-section is given
as a sum of the components and taking into account the effect of confinement in case of circular sections. In this study
the stress state in composite column is determined taking into account non-linear relationship of the modulus of elastic-
ity and Poissons ratio on the stress level in the concrete core. It has been determined that the effect of confinement
occurs at a high stress level when structural steel acts in tension and concrete works in lateral compression. The stress
state and load bearing capacity of section in bending is determined taking into account non-linear dependence on
position of neutral axis. Because the ultimate limit state of material is not attained for all the parts simultaneously, to
improve the stress state of a composite element and to prevent the possibility of a failure the appropriate strength of
concrete and steel should be used. The safety of high-stressed composite structures can be achieved by using ultra-high-
performance concrete (UHPC).
Keywords: composite columns and beams, stress state, non-linear deformation, ultra-high-performance concrete.
1. Introduction
Composite columns and beams, formed as hollow
steel sections filled with concrete, have advantages in
different architectural and structural solutions. The com-
posite structure characterises with higher ductility than
the concrete column and advantages of steel may be suc-
cessfully used in connections. Concrete filling and rein-
forcing bars increase rigidity and load-bearing capacity
of the hollow steel section with no changing in external
dimensions of column. .illing improves the fire resis-
tance as well. Composite column with a corresponding
content of reinforcement can provide at least 90-minutes
of fire-resistance rating [1].
According to EC4 [2], for determining the resistance
of a section against bending moment, a full plastic stress
distribution in the section has been assumed. The inter-
nal bending moment resulting from the stresses is the
resistance of the section against acting moment. In order
to increase the plastic bending capacity reinforcement can
be included.
EC4 gives a simplified ultimate limit state design
method for the composite structures, which is applicable
for practical purposes. The method is based on consump-
tion that ultimate strength of material is attained simul-
taneously in all parts of the section.
In previous decades investigations regarding inter-
action between steel shell and concrete core has been
performed [37]. In general, it is concluded that for ser-
vice load practically no bond exists and that the core
and shell act as two independent materials. In the early
stages of loading, the Poissons ratio for concrete is lower
than that for steel and the steel tube has no restraining
effect on the concrete core. As the longitudinal strain
increases, Poissons ratio of concrete can remarkably
increase. Therefore, the lateral expansion of unconstrained
concrete gradually becomes greater than that of steel. A
radial pressure develops at the steel-concrete interface
thereby restraining the concrete core and setting up a
hoop tension in the tube. At this stage, the concrete core
is stressed triaxially and the steel tube biaxially. Interac-
tion between steel tube and concrete is the key issue to
understand the behaviour of this kind of column. It is
shown [8] that concrete compaction affects the proper-
ties of the core and also may influence the interaction
between the steel tube and its concrete core, and thus
influences the behaviour of the composite columns.
The original contribution lies mainly in the use of
concrete, especially UHPC, in the form of slender com-
posite elements for bridges rather than massive elements
more commonly associated with concrete structures
[911].
178
178
In this study stress analysis of the composite col-
umn and beam is performed with the purpose to obtain
the maximum value of load-bearing capacity and enhance
the safety of the structure using components with the
appropriate strength properties and taking into account
the composite action. The effect of UHPC on the stress
state and load-carrying capacity of composite elements
is analysed.
2. Analytical model for stress analysis
The section presents a consideration on behaviour
of short straight concrete-filled steel column under axial
loading. The term short column refers to a compres-
sion member that can attain its ultimate capacity without
overall buckling. Disregarding the local effects at the
ends, the stress state in the cross-section of the middle
part of the column is analysed. A reinforced concrete
core with radius R
0
is included in steel tube of thickness
t (.ig 1). The reinforcement consists of symmetrically
arranged longitudinal steel bars and circumferential rein-
forcing wire located at radius R
s
from axis z.
.ig 1. Concrete-filled hollow column with reinforcement
.or radial displacements w and circumferential dis-
placements u on the contact surface concrete-structural
steel (r = R
0
) and concrete-reinforcement surface (r = R
s
)
the following equalities are valid:
; ;
c a c a
u u w w
(1)
. ;
c s c s
u u w w
(2)
Here and below, the indices a, c and s refer to the
steel tube, concrete and reinforcement. In the axial di-
rection of the column, the cross-section area of the rein-
forcement is
s
z
A , but in the circumferential direction -
s
A
, placed with the given step. It is taken into account

that the reinforcement is "spread" throughout the cross-
sectional area and, as a result, an anisotropic column is
formed.
The following relationships should be written as-
suming throughout the steel tube an in-plane membrane
stressed state:

0
z
a
z
; (3)

t
R p
o c
a
; (4)

r
u
z
w
a
a
a
a
z

const; , (5)
where
c
p the contact pressure on the boundary sur-
face of steel-tube concrete,
a
z
,
a
,
a
z
and
a

stresses and strains of the steel tube in the axial and
circumferential direction. The existing equality of stresses
in the radial and circumferential direction (
c
r
=
c
) in
concrete core is assumed.
Taking into account the compatibility relationships
(1) and (2), four equations can be written:

c
c
c
z c
r
c
r
c
zr c
z
c
z
a
a
a
a
z
a
E E E E E

1 1
; (6)

c
c
c
z c
r
c
r
c
z
c
z
c
zr a
a
a
z
a
a
E E E E E

+
1 1
; (7)

c
c
c
z c
r
c
r
c
zr c
z
c
z
s
z
s
z
E E E E

1 1
; (8)

c
z
c
z
c
z c
r
c
r
c
r c
c
s
s
E E E E

1 1
, (9)
were
a
is Poisson's ratio of the steel tube,
c
ij

Poisson's ratio values of the reinforced concrete (i, j = r,
z, ; i j). Equations (6) (9) have been written taking
into account the transverse deformation. The equilibrium
equations in the z and directions are as follows
0
0
+ +
c a s
s
s
R
t
R
A

; (10)
0 ) ( + + + +
c a s
z
a
z
a c
z
c
A A q A A A , (11)
where A
c
and A
a
the cross-sectional area of the con-
crete and structural steel accordingly, E
a
modulus of
elasticity of the steel tube,
c
z
E ,
c
r
E and
c
E
moduli
of elasticity of the reinforced anisotropic concrete core
in the directions z, r and ,
s
z
E and
s
E
moduli of
elasticity of the reinforcement in directions z and , q
a uniformly distributed axial load. It is assumed the load
is applied to the entire section.
The load-carrying capacity of composite elements
in bending can be characterised by plastic deformation
of structural steel in tension, by cracking of concrete in
compression or by both. At the same time, the stresses
J. Brauns, K. Rocens / JOURNAL O. CIVIL ENGINEERING AND MANAGEMENT 2004, Vol X, No 3, 177182
179
179
in reinforcement can be more or less of the yield limit
of steel. The stress analysis of composite elements in
bending has been performed with certain assumptions:
1) the hypothesis of flat sections has been used; 2) the
concrete lying in the tension zone of the section is as-
sumed to be cracked and is therefore neglected; 3) the
conditions of deformation continuity of structural steel
and concrete fulfil in the compression zone; 4) relation-
ship for the stress distribution in the compression zone
of concrete has been assumed in the following non-lin-
ear form:
c
n
f
c
z
c
x z ) / (
, (12)
where
z
c
stresses in concrete at distance z from neu-
tral axis;
f
c
stresses in concrete in more compressed
fibre; x depth of the compressed zone of concrete; n
c
characteristic of the stress diagram form. The cross-sec-

tion of reinforced composite beam and the stress-strain
distribution diagrams are shown in .ig 2.
.ig 2. Stress and strain distribution in a composite beam
The characteristic
c
n can be determined in the fol-
lowing way
m
c
c
f
c
c
u
c
f
E
E
n
,
_
,
_

1 1
, (13)
where experimental factor 7 , 5 m ;
c
f concrete
strength.
Based on the equilibrium conditions of the internal
forces and taking into account the hypothesis of flat sec-
tions, the position of neutral axis can be determined by
using non-linear equation:
[ ]
( ) ( )
, 0
2 2 2
) 2 4 ( ) 2 2 (
) ( 2
) / 1 (
0
2 2 2
2 2
2
+ + +
+ + + + + +
+
1
]
1
+
+
h A E
ht t h hbt E ht hbt E
A E t ht bt E t ht bt E x
E E t
E E
b f
x
s s
t
a a
s s
t
a a
t
a a
u
c c
u
c
cd
(14)
where
u
c
E and
c
E tangent and initial modulus of the
concrete,
t
a
E modulus of constructional steel in ten-
sion,
u
c
ultimate deformation of concrete in com-
pression. The ultimate bending moment
u
M was deter-
mined by using the condition of equality of the moments
in section of the beam
[ ]
). ( ) (
) ( ) ( ) (
3
2
) / (
0
0
x h bt x h A
x t h x h t t x btx
btx zdz x z b M
a
t
s s
a
t
a
a n c
f
x
u
c
+
+ + + +
+ +

(15)
where b, h and h
0
width, height and effective height of
concrete section.
3. Numerical solution and analysis
In order to perform the stress analysis in a compos-
ite column the system of six linear equations (6) (11)
was solved. In matrix form it is written as
[A] X = B. (16)
The components of the unknown vector X are
stresses in the steel tube, concrete, longitudinal and spi-
ral reinforcement, ie,
[ ]
T
, , , , ,
s s
z
c c
z
a a
z
X
. (17)
The matrix of system [A] and vector of constants B
is determined from the stress-strain relationships and
equilibrium equations as follows:
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
]
1
,
_
+

+
s c c
r
c
r
c
z
c
z
s
s
s
z
c
c
r
c
r
c
zr
c
z
s
z
c a
c
c
r
c
r
c
z
c
rz
a a
a
c
c
z
c
r
c
zr
c
z
a
a
a
E E E E
R
A
R
h
E E E E
A A A
E E E E E
E E E E E

1
0
1
0 0
0 1 0 0
0
1 1
0 0
0 0 0
0 0
1 1
0 0
1 1
A
0
; (18)
[ ]
T
0 , 0 , 0 ), ( , 0 , 0
s a
A A q + B
. (19)
The following geometrical parameters were chosen
for the analysis: diameter of circular hollow section
d = 40,6 cm, thickness of the steel tube section
t = 0,88 cm, reinforcement content = 4 % of the con-
crete section, distance from the central axis to the rein-
forcement bar axis R
s
= 15,5 cm, cross-sectional area of
the longitudinal reinforcement A
z
s
= 49 cm
2
and second-
ary reinforcement A
s
= 0,1 cm
2
(per 1 cm of the col-
umn length). The modulus of elasticity for the reinforce-
ment is E
s
= 200 000 MPa, for structural steel E
a
=
210 000 MPa and Poissons ratio
a
= 0,25.
Concrete is a heterogeneous material because of its
composite structure. Nevertheless, concrete is assumed
homogeneous and the mechanical behaviour can be ex-
pressed in the form of stress-strain relationship. This
relation is non-linear. However, with increasing compres-
sive strength, the initial slope corresponding to the modu-
180
180
lus of elasticity increases and the linear part extends to
higher stress levels. .urthermore, the ultimate strain, ie
the strain at maximum stress, increases with increasing
compressive strength [9].
In order to determine real mechanical characteris-
tics of concrete under compression stresses, prisms were
tested for uniaxial compression. The development of axial
strain
a
and lateral strain e
l
fixed in the test is shown in
.ig 3. In the first part of the loading phase, the stress-
strain relation shows an almost linear response. Due to
the concrete heterogeneity, a uniform uniaxial stress ap-
plied to a concrete specimen obviously results in local
non-uniformity. At a stress, corresponding approximately
to 40 % of the peak stress due to the difference in lat-
eral deformation of material constituents, the stress-strain
relation exhibits a visible non-linear response. Approxi-
mately at 85 % of the peak stress, the response is marked
non-linear.
In the initial stage of composite column loading, the
Poissons ratio of concrete core is lower than that of
steel. Therefore, the steel tube expands faster in the ra-
dial direction compared with the concrete core; hence,
the steel tube does not restrain the concrete core. As a
result, columns centric loaded on the entire section are
affected by the difference between the values of Poissons
ratio of the steel tube, n
a
, and the concrete core, n
c
.
In analysis the initial tangent modulus and tangent
modulus at the given stress level have been used as the
modulus of elasticity of the concrete. Based on experi-
mental results, the Poissons ratio of the concrete changes
in accordance to the stress level, starting with n
c
= 0,15
in the initial stage [12]. The moduli of elasticity of a
reinforced anisotropic concrete core E
z
c
, E
r
c
, and E
c
as
well as Poissons ratio values n
ij
c
were found on the basis
of the reinforcement theory [13] taking into account the
content of the reinforcement m and mechanical proper-
ties of the constituents.
-0,001
-0,0005
0
0,0005
0,001
0,0015
0,002
0,0025
0,003
0 5 10 15 20 25
Stress, MPa
S
t
r
a
i
n
1
2
.ig 3. Stress-strain relation for concrete prism in uniaxial
compression: 1
a
; 2
l
The stresses in material components with applied
force N for conventional concrete and UHPC are given
in Table 1. In the case of conventional concrete, stresses
in steel section, concrete and reinforcement exceed the
strength of materials. By using UHPC as column filling
the load-carrying capacity keeps enough and stress level
is satisfactory.
According to EC4, for the concrete-filled circular
hollow sections, the load-bearing capacity of the con-
crete is increased due to the prevention of transverse
strain. This effect is shown in .ig 4. The transverse com-
pression of the concrete (
r
c
) at high stress levels leads
to three-dimensional effects, which promotes the increase
of the column resistance. At the same time, the circular
tensile stresses (
q
c
) arise on the cylindrical surfaces re-
ducing its normal stress capacity.
Note that when axial stress in concrete is
z
c
0,85
f
ck
, the Poissons ratio of concrete n
c
increases and be-
comes higher than that of the steel. In the region before
fracture, the Poissons ratio is approximately 0,3 [12].
Circular tensile stresses
q
c
in the concrete core of the
column change into compression stresses due to the steel
tube behaviour and the above-mentioned features of con-
crete deformation process.
.ig 4. Stress distribution in the section of column
According to [2], the plastic resistance of the cross-
section of a composite column with axial loading is given
as the sum of the components:
sd
s
z cd
c
yd
a
Rd pl
f A f A f A N + +
.
, (20)
where A
a
, A
c
and A
z
s
are the cross-sectional areas of the
structural steel, concrete and reinforcement in the axial
direction, and f
yd
, f
cd
and f
sd
are design strength values
of the materials mentioned above.
Table 1. Stress values (MPa) in constituents of composite column for ratio d/t = 44
Strength of
concrete, MPa
Axial force N,
kN
z
a

a

z
c

z
c

z
s

s

30 3200 332 4,3 47 0,1 316 94
180 3200 255 2,0 58 0,2 243 73
181
181
By using the above-mentioned geometrical and me-
chanical properties of materials, steel grade .e275 with
characteristic yield strength f
y
= 275 MPa, concrete of
class C30/37 and relative slenderness ratio of column
0,15 the load-bearing capacity is N
pl.Rd
= 7162 kN. Tak-
ing into account the effect of confinement for circular
hollow section, the load-bearing capacity increases and
becomes N
pl.Rd
= 7935 kN. Here the analysis is performed
using the secant modulus of concrete according to the
strength class of the concrete. According to the results
of stress analysis, the axial stresses in composite column
core in case of conventional concrete are
z
c
> f
ck
, where
f
ck
is the characteristic strength of concrete for the given
strength class.
The analysis of the load-bearing capacity of the
composite column is performed taking into account the
design strength of steel and concrete as well as limit ratios
of the circular hollow sections. .ig 5 shows that the first
load limiting factors are concrete design strength f
cd
and
diameter thickness ratio d/t. Using concrete of the strength
class C35/45 and steel of grade .e235 the load-bearing
capacity of the composite column increases by 18 % in
comparison with concrete class C30/37. .or a thin wall
hollow section (d/t = 90) instead of section with
d/t = 46 the steel economy is 50 %.
The results of the stress analysis of the composite
beam are shown in .ig 6. By using non-linear approach
the position of neutral axis of reinforced concrete ele-
ment depends on ultimate deformation of concrete in
compression e
c
u
. Taking into account this peculiarity, the
stresses in structural steel s
a
in tension zone approxi-
mately are equal to the yield limit f
y
(steel grade .e235)
when e
c
u
< 110
3
. The ultimate bending moment M
u
in
this case is by 40% less than plastic moment M
pl.Rd
de-
termined according to EC4. The comparison of compos-
ite beam load-bearing capacity in the case of conven-
tional concrete and UHPC depending on ultimate com-
pressive strain of concrete and compressive depth is given
in Table 2.
Concrete filling leads to the increase of the load-
bearing capacity that is much higher than that of steel
elements and promotes the fire resistance as well. The
concrete is contained within a circular steel profile and
cannot split away, even if the ultimate strength of con-
crete is reached. Nevertheless, in order to prevent the
possibility of a failure, especially, in the case of small
.ig 5. Variation of stresses in structural steel and con-
crete in relationship of ratio d/t: 1, 2 s
z
a
and s
z
c
for
axial force N = 4000 kN; 3, 4 s
z
a
and s
z
c
for
N = 5000 kN
Table 2. Comparison of composite beams with conventional concrete and UHPC
.ig 6. Influence of ultimate concrete deformation in com-
pression on the position of neutral axis, stresses in struc-
tural steel and ultimate bending moment: 1 x/h, 2 s
a
,
3 M
u
1,0
0,5
1,0
0,5
1,0 2,0 3,0
Compressive strength of concrete, MPa
30 180

Ultimate concrete
deformation,
cu
?10
3

Relative
compressive depth,
x/h
Ultimate moment
M
u
, kNm
Relative
compressive depth,
x/h
Ultimate moment
M
u
, kNm
1
2
3
0,42
0,47
0,49
92,7
155,4
248,0
0,27
0,34
0,39
186,8
271,8
383,4
cu
10
3
182
182 J. Brauns, K. Rocens / JOURNAL O. CIVIL ENGINEERING AND MANAGEMENT 2004, Vol X, No 3, 177182
thickness of the structural steel and fire, the appropriate
strength classes of concrete and steel have to be used.
The effect depends on the type of concrete because the
strength of core affects the point at which confinement
can take place.
4. Conclusions
1. On the basis of constitutive relationships for
material components, the stress state in a composite col-
umn is determined, taking into account the dependence
of the modulus of elasticity and Poissons ratio on the
stress level in the concrete. It has been proved that the
effect of confinement acts at a high stress level when
the structural steel behaves in tension and the concrete
in compression. The main load limiting factors are con-
crete design strength f
cd
and d/t ratio. In the case of
higher concrete strength class and steel grade .e235 the
load-bearing capacity of the composite column increases
by 18%. In the case of a thin hollow section (d/t = 90),
the steel economy is for 50 %.
2. By using non-linear approach, the position of
neutral axis of composite beam depends on the ultimate
deformation of concrete in compression e
c
u
. When
e
c
u
< 110
3
the ultimate bending moment M
u
is 40%
less than plastic moment M
pl.Rd
determined according to
EC4 and the stresses in structural steel s
a
t
in tension
zone are equal to yield limit f
y
(steel grade .e235).
3. The optimization of working conditions and cross-
section area of the composite structure as well as the
prevention of the possibility of a failure in the case of
small thickness of structural steel and fire can be real-
ized by using appropriate strength of concrete and steel.
By using UHPC as steel element filling the increase of
load carrying capacity can be significant.
References
1. Bergman, R.; Matsui, C.; Meinsma, C. & Dutta, D. De-
sign guide for concrete filled hollow section columns un-
der static and seismic loading. Kln: Verlag TV Rheinland
GmbH, 1995. 68 p.
2. Eurocode 4: Design of composite steel and concrete struc-
tures, Part 1.1: General rules and rules for building. ENV
1994-1-1. 33 p.
3. Johansson, M. Composite action and confinement effects
in tubular steel-concrete columns. Thesis for the degree of
Doctor of Philosophy. Gteborg: Chalmers University of
Technology, 2002. 77 p.
4. Kvedaras, A. K.; apalas, A. Research and practice of
concrete-filled steel tubes in Lithuania. Journal of Struc-
tural Steel Research, Vol 49, No 2, 1999, p. 189196.
5. Brauns, J. Analysis of stress state in concrete-filled steel
column. Journal of Structural Steel Research, Vol 49, No 2,
1999, p. 197212.
6. Lakshmi, B.; Shanmugan, N. E. Non-linear analysis of in-
filled steel-concrete composite columns. Journal of Struc-
tural Engineering, Vol 128, No 7, 2002, p. 922933.
7. OShea, M. D.; Bridge, R. Q. Design of circular thin-walled
concrete filled steel tubes. Journal of Structural Engineer-
ing, Vol 126, No 11, 2000, p. 12951303.
8. Han Lin-Hai H.; Guo-Huang, Y. Influence of concrete
compaction on the strength of concrete-filled steel RHS
columns. Journal of Structural Steel Research, Vol 59,
No 6, 2003, p. 751768.
9. Zilch, K.; Hennecke, M. Anwendung hochfesten Betons
im Brckenbau, .orschungsbericht 99. Massivbau TUM.
Mnchen, 1999.
10. Dowd, W. M.; Dauriac, Ch. E. Reactive powder concrete.
The Construction Specifier, 4852. 1996.
11. Ma, J.; Schneider, H. Properties of ultra-high-performance
concrete. LACER, No 7, 2532, 2002.
12. Neville, A. M. Properties of concrete, London: Pitman
Publishing, 1981. 409 p.
13. Malmeister, A. K.; Tamuz, V. P. & Teters G.A. Strength of
polymer and composite materials (
). Riga: Zinatne,
1980. 571 p. (in Russian).
183
183
2004, Vol X, No 3, 183190
ISSN 13923730
BEHAVIOUR O. ALUMINIUM ALLOY STRUCTURES UNDER .IRE
Beatrice .aggiano
1
, Gianfranco De Matteis
1
, Raffaele Landolfo
2
, .ederico M. Mazzolani
1
1
University of Napoli .ederico II, .aculty of Engineering, P.le Tecchio 80, 80125 Napoli, Italia.
E-mails: faggiano@unina.it; demattei@unina.it; fmm@unina.it
2
University of Napoli .ederico II, .aculty of
Architecture, Via Monteoliveto 3, 80134 Napoli, Italia. E-mail: landolfo@unina.it
Received 18 .ebr 2004; accepted 30 June 2004
Abstract. In the paper the attention is focused on the influence of high temperatures on the mechanical properties of the
aluminium alloys selected by Eurocode 9 for structural uses. Therefore, based on the analysis of existing data taken
from technical literature, the variation of the Youngs modulus, the conventional yielding strength, the ultimate strength,
the hardening factor and the material ultimate strain are represented as a function of the temperature. A mechanical
model, based on the well-known Ramberg-Osgood formulation, which appropriately takes into account the peculiarities
of such materials at high temperatures, is provided. In particular, the combined influence of the hardening factor and
temperature on the material stress-strain relationship is considered and analysed. Then, the proposed model has been
introduced in a finite element program, devoted to the global analysis of structures under fire. .inally, the results
obtained for a simple portal frame structure, designed with different aluminium alloys, are presented, showing the
valuable effect of the material modelling on the structural behaviour of aluminium structures under fire.
Keywords: aluminium alloys, mechanical properties under fire, effect of hardening, stress-strain relationships at high
temperatures, structural analysis under fire.
1. Introduction
In recent years, the employment of aluminium al-
loys is increasing for constructing bridges, footbridges,
aerials, offshore platforms, thanks to the material light-
ness, versatility and resistance to atmospheric agents.
Therefore, there is the need for effective regulation of
the design of aluminium alloy structures under fire. This,
in fact, combines the increasing use of aluminium as a
structural material with the more and more interest of
the scientific community in the behaviour of civil and
industrial constructions under conditions of exceptional
temperature. Besides, it is worth noticing that the pre-
diction of the mechanical response of aluminium alloy
structures exposed to fire is particularly complicated for
two principal reasons: on the one hand, the intrinsic dif-
ficulty of developing accurate structural analyses in post-
elastic field, taking correctly into account the mechani-
cal features of the basic material, such as the strain-hard-
ening and the limited deformation capacity; on the other
hand, the inadequate knowledge of the material behaviour
under high temperatures.
On the basis of the above considerations, also for
aluminium alloy structures, it is necessary to develop and
apply the advanced structural analysis methods [1]. There-
fore the specific mechanical properties of the material as
a function of the temperature must be considered. Thus,
the whole stress-strain curve of the material, which al-
ready at the ambient temperature shows prominent char-
acteristics of non-linearity, must be accurately defined
as a function of the temperature, representing the varia-
tion law of the following characteristic parameters: the
modulus of elasticity (E), the elastic limit stress conven-
tionally defined as 0,2% proof strength (f
0,2
), the ulti-
mate strength (f
t
) and the ultimate deformation (e
u
). In
particular, the methods of structural analysis under fire
conditions are to be held in due account: the influence
of the shape of the whole material constitutive law and
thus the kinematic strain hardening on the global
behaviour of the structure. On the other hand, it must be
considered that for allowing practical analysis in fire
conditions of complex structures by the advanced meth-
ods, such accurate material models should be imple-
mented in finite element programmes.
In this paper, the basic problems for the design and
safety checks of the aluminium alloy structures under fire
are faced. The attention is focused on the temperature
influence on mechanical properties of the materials, with
reference to the aluminium alloys selected by the
Eurocode 9 for structural use [2]. Then, a mechanical
model for the aluminium alloys, which appropriately rep-
resents the peculiarity of such materials subjected to high
temperatures, is proposed. Such simplified constitutive
184
184 B. .aggiano, et al / JOURNAL O. CIVIL ENGINEERING AND MANAGEMENT 2004, Vol X, No 3, 183190
law has been introduced in a finite element programme
for the calculus of structures under fire [3]. .inally, the
results of the structural analyses under fire obtained for
a simple portal frame and carried out for different alu-
minium alloys are presented, clarifying the influence of
the material modelling on the global response of the struc-
ture exposed to fire, evaluated in terms of time up to
collapse and considering conventional fire scenarios [4].
2. .ire analysis of aluminium alloy structures
Generally, for the structural analysis under fire, from
the point of view of the structural safety, it is necessary
to identify adequate calculus methodologies, which al-
low the correct evaluation of the mechanical resistance
of the structure. The more significant fire effects are: (1)
deterioration of materials in strength and stiffness, which
implies a reduction of the carrying capacity and worsen-
ing the deformation state; (2) thermal deformations, which
may produce increase of second-order effects and/or in-
ternal stresses.
With regard to the fire modelling, as a simple alter-
native to real fire scenarios, reference is generally made
to the so-called standard fires (for example, ISO 834
curve), which can be simply related to possible real fires
by means of equivalence criteria. The design of struc-
tures under fire must essentially assure that in case of
fire, according to the chosen thermal programme, the
static safety is guaranteed for a pre-fixed period of time,
to be associated to the class of fire resistance imposed
by the standard rules. It is clear that the analysis under
fire of complex structures must account for all those fac-
tors that are not considered by the simplified methods,
for instance, geometric and mechanical non-linearities,
creep phenomena, residual stresses, thermal deformations
and the real evolution of the mechanical behaviour of
the material under the temperature. Evidently, the appli-
cation of such methodologies cannot avoid using the fi-
nite element programme.
At the European level, the main references for codi-
fications are given by the Eurocodes: EC 1-Part 2-2, for
the applied loads in fire conditions [5]; EC 2- Part 1-2,
for reinforced concrete structures under fire [6]; EC 3-
Part 1-2, for steel structures under fire [7]; EC 9- Part
1-2, for aluminium alloys structures under fire [2].
Such provisions can be considered exhaustive and
reliable in case of the most common materials, such as
steel and reinforced concrete, while they result more
approximated in the case of the aluminium alloys struc-
tures, mainly due to the difficulties of accurate and reli-
able material modelling. Therefore, it appears convenient
to go more deeply into the problem, checking the oppor-
tunity of applying for code purposes a better schema-
tisation of the mechanical behaviour of the material, so
as to allow the maximum exploitation of the resistance
resource of the material, which, generally, results are
highly vulnerable under fire.
3. Mechanical features of aluminium alloys at high
temperatures
Common aluminium alloys melt at about 600C and
loose the 50% of their original strength at about 200C
[8, 9]. The principal behaviour parameters, such as the
stress at the conventional elastic limit (f
0,2,T
) and the elas-
tic modulus (E
T
), as a function of the temperature, are
depicted in .ig 1, according to EC9 [2]. In particular, in
.ig 1a the conventional strength f
0,2
is given through the
reduction coefficient k
0,2,T
, which provides for each al-
loy the ratio between the elastic strength at a given tem-
perature (f
0,2,T
) and the elastic strength at ambient tem-
perature (f
0,2,
).
(a)
Reduction coefficient of the conventional yielding stress
0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1
0 100 200 300 400 500 600
T (C)
k
0
.
2
,
T
3003-O 3003-H14
5052-0 5052-H34
5083-O 5086-O
5454-O 5454-H32
6061-T6 6063-T6
7075-T6
(b)
Young modulus
0
10000
20000
30000
40000
50000
60000
70000
0 100 200 300 400 500 600
T (C)
.ig 1. Mechanical properties of the examined aluminium
alloys as a function of temperature: (a) k
0.2,T
, (b) E
T
It can be noticed that the alloys in the work harden-
ing state (H) and the ones being in the hardening state
by means of heat treatment (T) exhibit a relevant loss of
strength with temperature, which is of about 7080% at
250C. Besides, the alloys in the annealed state (O) show
a less significant decay of strength, which is of about
3050% at 250C. Also, it can be observed that the
elastic modulus E decreases as far as the temperature
increases, independently of the alloy and its treatment
(.ig 1b).
.ig 2 shows the variation laws of the conventional
yielding stress f
0,2,T
, the ultimate strength f
t,T
, the hard-
E
T

(
N
/
m
m
2
)
185
185 B. .aggiano, et al / JOURNAL O. CIVIL ENGINEERING AND MANAGEMENT 2004, Vol X, No 3, 183190
ening ratio f
t,T
/f
0,2,T
and the elongation at rupture
u
,
T
as
a function of the temperature, based on data available in
technical literature. They are related to experimental ten-
sile tests carried out in oven with specific procedures,
corresponding to exposures of 10000 hours at high tem-
peratures [10].
Such diagrams show that heat treated and work hard-
ened alloys (types T and H) are characterised by ulti-
mate strength remarkably larger than the alloys in the
annealed state (type O), only up to temperatures of about
100150C. .rom .ig 2c it can be also noted that while
at ambient temperature annealed state alloys (type O)
present a strain hardening ratio (f
t,
/f
0,2
) about twice larger
with respect to the heat treated and work hardened al-
loys (types T and H), such difference strongly reduces
as far as the temperature increases, starting from tem-
peratures of about 100150C. In particular, for no-
treated alloys the strain hardening decreases with the
increment of the temperature, while in case of the treated
alloys it increases, approaching in both cases a value
next to 1,5. Then, it can be generally observed that treat-
ments, such as the tempering and the plastic working
processes, improve the material strength, but in the mean-
time they reduce both the effect of the strain hardening
and the extent of the ultimate elongation (e
u
), which ex-
periences a revival with the increase of the temperature.
.urther data available in literature [11] show that
the resistance of the aluminium alloys, given in terms of
both conventional yielding stress and ultimate strength,
decreases as far as the exposure time to an assigned tem-
perature increases. On the contrary, the ultimate elonga-
tion, and then the material ductility, increases with the
prolonged permanence at high temperatures.
.inally, other thermal properties, such as the ther-
mal expansion, the specific heat and the thermal con-
ductivity, undergo an increment with the temperature
increase. Besides, the variation law of such properties,
as indicated in EC9, results almost independent of the
alloy type.
4. Stress-strain relations for aluminium alloys at high
temperatures
The modelling of the constitutive law of aluminium
alloys evidences remarkable difficulty already at ambi-
ent temperature, due to a high variation in the mechani-
cal behaviour among the different alloys and the adopted
fabrication processes. Besides, experimentally obtained
s-e curves are not well fitted by simplified diagrams,
like the elastic-perfectly plastic one, due to the remark-
ably continuous hardening exhibited in the plastic range.
.or the structural analysis under high temperatures,
Eurocode 9 makes implicitly reference to the adoption
of a simplified relationship of the elastic-perfectly plas-
tic type, exclusively supplying the variation of the elas-
tic modulus and of the conventional yielding stress with
the temperature (.ig 3a).
Conventional yielding stress
0
100
200
300
400
500
600
0 50 100 150 200 250 300 350 400
T (C)
f
0
.
2
,
T
3003-O 3003-H14
5052-0 5052-H34
5083-O 5086-O
5454-O 5454-H32
6061-T6 6063-T6
7075-T6
(a)
(b)
(c)
(d)
Ultimate stress
0
100
200
300
400
500
600
0 50 100 150 200 250 300 350 400
T (C)
f
t
,
T
3003-O 3003-H14
5052-0 5052-H34
5083-O 5086-O
5454-O 5454-H32
6061-T6 6063-T6
7075-T6
Hardening factor
1,00
1,50
2,00
2,50
3,00
0 50 100 150 200 250 300 350 400
T (C)
f
t
,
T
/
f
0
.
2
,
T
3003-O 3003-H14
5052-0 5052-H34
5083-O 5086-O
5454-O 5454-H32
6061-T6 6063-T6
7075-T6
Ultimate strain
0
25
50
75
100
125
150
0 50 100 150 200 250 300 350 400
T (C)
t
,
T
3003-O 3003-H14
5052-0 5052-H34
5083-O 5086-O
5454-O 5454-H32
6061-T6 6063-T6
7075-T6
.ig 2. Effect of temperature on the mechanical properties
of the examined aluminium alloys: (a) f
0.2,T
(N/mm
2
), (b)
f
t,T
(N/mm
2
), (c) f
t,T
/f
0.2,T
, (d) e
u,T
(%)
186
186
In order to interpret more correctly the evolution of
the mechanical characteristics of the material with the
temperature, it would be possible to apply the Ramberg-
Osgood model, which is already commonly used for
modelling aluminium alloy at ambient temperature for
non-linear analyses [8]. The general expression of the
Ramberg-Osgood law at ambient temperature for analy-
ses in the large deformation field can be assumed as:
n
f E

,
_
+
2 , 0
002 , 0

,
where
t
u
f
f
n
2 , 0
log
002 , 0
log
.
The n exponent measures the strain hardening of
the alloy, ruling the shape of the curve in the post-elas-
tic field. In particular, it can be observed that for values
of the n factor approaching 0 the Ramberg-Osgood law
gives an indefinitely elastic behaviour, while for large
values of n factor an elastic-perfectly plastic behaviour
is obtained (.ig 3b).
.ig 3. Typical constitutive laws for the aluminium alloys:
(a) elastic-perfectly plastic law, (b) Ramberg-Osgood rela-
tionship
The extension of the Ramberg-Osgood relationship
to high temperature is based on the introduction of the
variation law with the temperature of all the relevant
mechanical parameters, such as f
0,2,T
, e
u,T
and f
t,T
. Such
variation laws could be assumed as the ones already
drawn in .ig 2.
On the basis of the elaboration of the variation laws
for the single parameter (.ig 4), for each considered al-
loy, the values of the strain hardening factor n obtained
at different levels of temperature, included the ambient
temperature, are depicted. In particular, in order to evi-
dence the influence of the mechanical properties on the
strain hardening factor n, the value obtained considering
the actual variation of the single mechanical parameters
with temperature (n-analytical), is compared to the n
values obtained taking the ultimate elongation as a con-
stant and equal to the value at ambient temperature
(n-e
u
=cost) and the constant value of n at ambient tem-
perature.
(a) (b)
.irst of all, it can be observed that for all the alloys
the n value for high temperatures is remarkably different
with respect to that one obtained at ambient tempera-
ture. .urthermore the n(T) relationship does not present
a single trend for the different examined materials. In
particular, for not treated materials (type O) it can be
noted that at increasing temperatures the strain harden-
ing factor exhibits an increment higher than 50% with
respect to the room temperature value. On the contrary,
for treated materials (H and T types) the n value is higher
than the corresponding value at ambient temperature, only
up to a temperature of about 200C, beyond which
theres a reversal trend, with values of the n factor lower
than the ones at ambient temperature.
.ig 4. The strain hardening factor n as a function of tem-
perature
B. .aggiano, et al / JOURNAL O. CIVIL ENGINEERING AND MANAGEMENT 2004, Vol X, No 3, 183190
187
187
Moreover, it can be observed that the variation of
the elongation at collapse e
u
with the temperature does
not produce a significant influence on the strain harden-
ing factor n. As a consequence, in order to simplify the
mathematical expression of the strain hardening factor,
the ultimate elongation of the material could be actually
taken constant and equal to the one at ambient tempera-
ture.
In order to emphasise the influence of the strain
hardening factor on the constitutive law of aluminium
alloys at high temperatures (.ig 5), for every examined
alloy, the s-e curves are represented at different tem-
peratures (24C, 150C, 315C), considering different
hypotheses of hardening factor evaluation, namely: (1)
actual hardening factor n with temperature (n-analytical);
(2) constant value equal to the one at ambient tempera-
ture (n-constant); (3) constant value equal to 200 (n-200),
the latter being representative of an elastic-perfectly plas-
tic mechanical behaviour.
.ig 5. The s (N/mm
2
)-e laws for the aluminium alloys at different temperatures (continued)
188
188
.ig 5. The s (N/mm
2
)e laws for the aluminium alloys at
different temperatures (continued)
The obtained diagrams show that the mechanical
behaviour of the aluminium alloys at high temperatures
is generally significantly conditioned by the strain hard-
ening factor n. Such effect appears more evident in case
of the alloys in the annealed state (O), for which at high
temperatures the n factor (.ig 5) remarkably increases,
producing s-e curves characterised by strain hardening
much lower than the one at ambient temperatures. It is
also evident that the assumption of n constant and equal
to the value at ambient temperature is unconservative for
such type of alloys. On the contrary, for treated alloys
the influence of the n factor is generally less important,
particularly for heat-treated alloys (T), where the strain
hardening of the material is very low at ambient tem-
perature as well. .inally, it can be observed that for all
the examined alloys, for temperatures higher than 200 C,
the constitutive law can be adequately schematised by
using the elastic-perfectly plastic model.
5. Influence of material modelling on global analysis
Assessing the influence of modelling of the mate-
rial on the fire resistance of aluminium structures, here-
after the structural analysis under fire condition of a
simple plane portal frame made of different aluminium
alloys is presented. The geometric characteristics, the load
conditions, the fire event model as well as the exposure
condition to fire of the structural members are presented
in .ig 6, together with the types of considered alloys
and the result of the member sizing.
.ig 6. Study cases
Y O L L A
f
, 2 . 0 d
m m / N (
2
)
n m u l o C
B E H
m a e B
E P I
O - 3 0 0 3 5 4 0 6 1 0 5 4
4 1 H - 3 0 0 3 5 4 1 0 6 1 0 0 3
O - 2 5 0 5 0 9 0 6 1 0 3 3
4 3 H - 2 5 0 5 5 1 2 0 6 1 0 4 2
O - 4 5 4 5 7 1 1 0 6 1 0 0 3
2 3 H - 4 5 4 5 7 0 2 0 6 1 0 4 2
O - 3 8 0 5 5 4 1 0 6 1 0 0 3
O - 6 8 0 5 7 1 1 0 6 1 0 0 3
6 T - 1 6 0 6 3 8 2 0 6 1 0 4 2
6 T - 3 6 0 6 0 2 2 0 6 1 0 4 2
6 T - 5 7 0 7 7 1 5 0 6 1 0 4 2
189
189
.or the analysis of the structure under fire, the fi-
nite element program SA.IR98a [3], developed by
J. M. .ranssen of the University of Lige, has been used.
Such a software has been upgraded (2002) by imple-
menting also the material models for aluminium alloys
provided by Eurocode 9. In particular, in order to assess
the influence due to the strain hardening on the behaviour
of the whole structural complex, different material mod-
els based on the Ramberg-Osgood law have been con-
sidered, assuming (1) the actual strain hardening factor
n (n-analytical), (2) a constant value equal to the one at
ambient temperature (n-constant) and (3) n equal to 200
(n-200), this latter corresponding to the elastic-perfectly
plastic model as implicitly assumed in EC9-Part-1-2. The
used software allows to consider element models
characterised by spread plasticity.
In .ig 7 the results of the analysis, given in terms
of fire resistance of the structure (R) expressed in min-
utes, are specified. The structural resistance R has been
also normalised, considering as reference value for each
alloy the one corresponding to the material model based
on the actual strain hardening value (n-analytical). In the
above analysis, the considered collapse condition corre-
sponds to the loss of stability of the structure, which
occurs when, due to the deterioration of the mechanical
features of the material induced by the temperature, it is
not able to balance the applied external loads.
Y O L L A
) n i m ( e c n a t s i s e r e r i F
l a c i t y l a n a - n t n a t s n o c - n 0 0 2 - n
O - 3 0 0 3 6 6 4
4 1 H - 3 0 0 3 6 0 4 , 5 5
O - 2 5 0 5 0 2 , 7 8 0 2 , 5
4 3 H - 2 5 0 5 6 0 4 , 5 0 2 , 5
O - 3 8 0 5 0 4 , 7 8 0 2 , 6
O - 6 8 0 5 0 4 , 7 8 0 4 , 6
O - 4 5 4 5 0 4 , 7 8 0 4 , 6
2 3 H - 4 5 4 5 7 0 4 , 6 6
6 T - 1 6 0 6 0 2 , 5 0 2 , 5 0 2 , 5
6 T - 3 6 0 6 7 7 0 2 , 6
6 T - 5 7 0 7 0 4 , 6 0 2 , 6 0 2 , 5
.ig 7. .ire resistance for the study cases
The examination of the obtained results evidences
that the effect of the strain hardening factor on the glo-
bal behaviour of the structure is similar to that already
obtained at the material model level. Generally, the fire
resistance of the portal frame is higher in case of mate-
rial model with constant value at ambient temperature of
the factor n, while it is lower when the elastic-perfectly
plastic material model is used. In particular, it can be
observed that in most unfavourable conditions (3003-O
and 5052-O alloys), when considering the elastic-perfectly
plastic material model, a fire resistance reduction of the
structure of about 30 % is gained. Besides it can be
pointed out that not treated alloys (O) give rise to the
best behaviour under fire due to the beneficial effect of
material strain hardening.
However, it is worth noticing that the presented re-
sults are somewhat restricted, being referred to a simple
portal frame structures.
Moreover, it has to be evidenced the limited col-
lapse time, which on average is equal to about 7 min-
utes. Anyway, this limited fire resistance refers to a spe-
cific structural scheme, considering a standard fire model
and the worse exposure condition of the frame members
(unprotected, 4 sides exposure and inconvenient cross-
section shape). Consequently, considerable margins of im-
provement of the performances to fire of the structure
exist.
Nonetheless, it must be evidenced that the main aim
of the study is a more comparative evaluation of the strain
hardening at high temperatures for different aluminium
alloys for structural uses and its impact on the structural
fire resistance than the optimization of the behaviour of
structures under fire conditions.
6. Conclusive remarks
In order to set up advanced fire design methods for
aluminium alloy structures, in this paper the influence of
temperature on the mechanical properties of the material
has been analysed, taking into considerations all the al-
loys selected by Eurocode 9 for structural uses. Particu-
lar attention has been focused on material strain harden-
ing, which characterises the mechanical behaviour of the
material in plastic range. In fact, as a first result of the
study, it has been pointed out that simplified mechanical
models, such as the elastic-perfectly plastic one, gener-
ally are not able to correctly characterise the material
behaviour at high temperatures, since they disregard the
beneficial effect due to continuous material hardening,
which is somewhat effective in balancing strength decay
due to high temperatures. Therefore, in order to take spe-
cifically into account the effect of the strain hardening,
a more comprehensive mechanical model for aluminium
alloys has been proposed, based on the well known
Ramberg-Osgood law, which is able to represent in an
appropriate manner all the peculiarities of such materi-
als exposed to high temperatures. Then, such model has
190
190
been introduced in a finite element program for global
analysis of structures subjected to fire. The structural
analysis in fire conditions of a study case related to a
simple portal frame has pointed out the remarkable ef-
fect of material modelling for aluminium alloys, since
the adoption of elastic-perfectly plastic model results are
very conservative and not convenient for a material which
exhibits a rapid strength decay with high temperature.
References
1. Wald, ..; Bosiljkov, V.; De Matteis, G.; Haller, P.; Vila
Real, P. Structural integrity of buildings under exceptional
fire. In: 1st Cost C12 Seminar, Lisbon 1819 April, 2002.
2. prEN 1999-1-2. Eurocode 9: Design of aluminium struc-
tures - Part 1-2: General rules-Structural fire design, Eu-
ropean Communities for Standardisation, Brussels, 2003.
3. .ranssen, J. M. SA.IR98a - Users Manual. University of
Liege, 1998.
4. .aggiano, B.; De Matteis, G.; Landolfo, R.; Mazzolani,
.. M. On the behaviour of alluminium alloy structures ex-
posed to fire. In: Proceedings of the XIX National Con-
gress C.T.A.: III week of steel Constructions, 2830 Sept,
Genova, 2003, p. 393406.
5. EN 1991-1-2. Eurocode 1: Basis of design and actions on
structures - Part 2-2: Actions on structures - Actions on
structures exposed to fire, European Communities for
Standardisation, Brussels, 2002.
6. prEN 1992-1-2. Eurocode 2: Design of reinforced con-
crete structures - Part 1-2: General rules-Structural fire
design, European Communities for Standardisation, Brus-
sels, 2002.
7. prEN 1993-1-2. Eurocode 3: Design of steel structures -
Part 1-2: General rules-Structural fire design, European
Communities for Standardisation, Brussels, 2003.
8. Mazzolani, .. M. Aluminium alloy structures. 2nd edn.,
E&.N Spon, London, 1994, UK.
9. Lundberg, S. Design for fire resistance. In: Training in
Aluminium Application Technology (TALAT) EU-
COMETT Program, .. Ostermann Ed., Aluminium Train-
ing Partnership, Brussels: Section 2500, 1995.
10. ASM Specialty Handbook. Aluminium and aluminium al-
loys. Edited by J. R. Davis Davis &Associates, 1993.
11. Conserva, M.; Donzelli, G.; Trippodo, R. Aluminium and
its application. EDIMET, 1992.
191
191
2004, Vol X, No 3, 191197
ISSN 13923730
DAMAGE DETECTION USING PARAMETER DEPENDENT DYNAMIC EXPERI-
MENTS AND WAVELET TRANS.ORMATION
Andrzej Garstecki, Anna Knitter-Pitkowska, Zbigniew Pozorski and Krzysztof Ziopaja
Institute of Structural Engineering, Pozna University of Technology, ul. Piotrowo 5, 60-965 Pozna, Poland.
E-mail: Andrzej.Garstecki@put.poznan.pl; Anna.Knitter@ikb.poznan.pl; Zbigniew.Pozorski@ikb.poznan.pl;
Krzysztof.Ziopaja@ikb.poznan.pl
Received 1 March 2004; accepted 15 June 2004
Abstract. The paper deals with a class of problems, where localised damage is detected using static and dynamic tests.
Response of a structure is analysed employing discrete wavelet transformation as a tool for signal processing. The
localised damage in beam structures was modelled as bending stiffness reduction. The efficiency of static and dynamic
tests is studied. The minimal number of experimental measurement data and the precision of measurements required for
the successful damage localisation is discussed by several numerical examples.
Keywords: structural identification, damage detection, wavelet transformation.
1. Introduction
The problem of localisation and estimation of struc-
tural damage is one of the most important engineering
problems. It is connected with the assessment of struc-
ture safety and serviceability. This issue belongs to a
wider class of identification problems, where unknown
parameters of a system are determined by experimental
tests. Structural identification has focused much atten-
tion in the last two decades.
.irst papers devoted to damage detection were based
on the modal analysis of structural response. It was ob-
served [1, 2] that damage results in stiffness reduction,
increase of damping and decrease of natural frequency
of a structure. An extensive review of damage identifi-
cation techniques using information on changes in natu-
ral frequencies and shape modes of vibrations was pre-
sented [3]. Unfortunately, global static and dynamic re-
sponses are rather insensitive to localised damage and
often are smaller than the variations resulting from inac-
curacy in structure modelling. Hence, different ways of
improvement of experiments were proposed. Variable
location of support or additional concentrated mass en-
hancing structural response was proposed [4]. Optimiza-
tion of loads with the aim to better exhibit the discrep-
ancy between responses of the damaged and undamaged
structures was presented [57].
New class of identification methods emerged with
the development of the so-called soft methods: fuzzy sets,
genetic algorithms and neural networks. Application of
genetic algorithms in damage identification was presented
in [8], whereas prospects of neural networks was dis-
cussed in [9]. In paper [10] fuzzy sets, genetic algorithms
and neural networks were used simultaneously.
Entirely new method rooted in the mathematical
theory of signal analysis emerged in the last years,
namely, wavelet transformation. It has become one of
the most promising tools in structural identification. The
theory of wavelet transformation was developed by
Daubechies [11], Mallat [12] and Chui [13]. Newland
[14] showed the potential of this tool in vibration signal
analysis. Wavelet transformation was applied to damage
identification by Wang and Mc.adden [15]. Application
of the method in the space domain was presented [16].
Analysis of transformation parameters was performed in
[17], where the efficiency of wavelets in damage identi-
fication was demonstrated.
In this paper we continue discussion on application
of wavelet transformation in structural damage identifi-
cation. The attention will be focused on the identifica-
tion efficiency in the analyses of static and dynamic struc-
tural responses. The problem of the number of measure-
ment points and the influence of noise in measurements
are discussed.
2. Problem formulation
We use beam, frame or plate models of damaged
structures. The specific type of the structure does not
192
192
make much difference, provided that we can receive the
response for any action (not necessarily defined). Our
main task is to detect localisation of damage in the struc-
ture, if such a damage exists. This localisation will be
determined basing on signal analysis of structural re-
sponse to different actions.
We consider numerical models of beams loaded by
static and dynamic forces. In all examples the damage is
modelled as local stiffness reduction at small, prescribed
distance introduced at the point of existing damage. The
application of damage model in the form of an elastic
hinge was discussed by the authors in [17].
In the procedure of damage identification various
structural responses are analysed, namely static displace-
ments or the amplitudes of dynamic displacement and
acceleration. The structural response represented in the
form of a discrete signal is transformed using wavelet
transformation. Theoretical background of this transfor-
mation was published in [1114]. Basic information on
discrete wavelet transformation, directly connected with
the present study, will be provided in Chapter 3.
As a result of signal processing we expect evident
disturbances of transformed signal to appear in the place
of damage location. We assumed that the response of
undamaged structure is not known. The aim of the paper
is not limited to demonstration that the defect identifica-
tion is possible. Crucial problem is how to achieve this
goal with the minimum number of measurement points
and with a specified noise level representing measure-
ments inaccuracy inevitable in experiments.
3. Basis of wavelet transformation
Wavelet transformation is a method of decomposi-
tion of arbitrary signal f(x) into an infinite sum of wave-
lets at different scales according to the expansion:
( ) k x W c x f
j
j k
k j

2 ) (
,
, (1)
where W(x) is a wavelet (mother) function. Integers j
and k are dilation (scale) and translation (position) indi-
ces, respectively. The terms c
j,k
are numerical constants
called wavelet coefficients.
When j is negative, W(2
j
x-k) can always be ex-
pressed as a sum of terms ( ) k x , providing
( ) ( ) ( ) k x W c k x c x f
j
j k
k j
k
k
+

2
0
, ,

, (2)
where ( ) x is a scaling (father) function and
k
c
,
is a
new set of coefficients.
In order to set up a discrete wavelet transform al-
gorithm, it is convenient to limit the range of the inde-
pendent variable x to one unit interval so that f(x) is
assumed to be defined only for 0 x 1. Then, the
variable x is non-dimensional. Additionally, assuming that
f(x) is one period of a periodic signal, the wavelet ex-
pansion can be written in the form
( ) ( ) ( )
+
+
j k
j
k
k x W a x a x f
j
2
2
0
. (3)
The coefficients
k
j
a
+ 2
represent the amplitudes of
subsequent wavelets. The integer j describes different
levels of wavelets starting from j=0. Integer k specifies
the number of wavelets at each level, so that it covers
the range k=0 to 2
j
1.
The discrete wavelet transform, which is used in this
paper, is an algorithm for computing coefficients
k
j
a
+ 2
when f(x) is sampled at equally spaced intervals over
0 x 1. Since the number of sampled values is lim-
ited, every function f(x) is approximated by f
J
(x) using
N=2
J
discrete values. Note that the scale indicator
j=0, 1, ..., J1. Henceforth, keeping in mind (3), the
discrete signal decomposition can be written as
1 2 1 0
... ...

+ + + + + + +
J J j J
f f f f f f f

(4)
or
1 2 1
... ... D D D D D S f
m M M M J
+ + + + + + +

, (5)
where m = J j.
Each component in signal representation provides
information at the scale level j. The last terms in (4) and
(5) corresponds with the most detailed representation of
the signal (high-frequency oscillations). The preceding
representations deliver the more and more rough infor-
mation about the signal and correspond to the lower fre-
quency oscillations.
In practice DWT requires neither integration, nor
explicit knowledge of scaling and wavelet functions, gen-
erating transformation. Due to the character the above
method of signal representation it is called multi-resolu-
tion analysis (MRA).
4. Numerical analysis
4.1. Static and dynamic structural responses
To compare the efficiency of damage identification
for different signals representing static and dynamic struc-
tural responses, a propped cantilever beam model pre-
sented in .ig 1 was used.

,
,
,
,
,
.ig 1. Beam structure with the damage modelled as stiff-
ness reduction
A. Garstecki, et al / JOURNAL O. CIVIL ENGINEERING AND MANAGEMENT 2004, Vol X, No 3, 191197
193
193
The properties of the structure are: bending stiff-
ness EI = 108 kNm
2
; damage localised at x =1,40 m
measured from the clamped end; the damage is mod-
elled as stiffness reduced to EI
d
= 62,5 kNm
2
on the
portion b = 0,01 m. All static and dynamic responses
were computed using .EM (ABAQUS program). Two
classes of problems were examined: static response to
the concentrated force P=1 kN and steady-state vibra-
tions excited by the dynamic force P(t)=1cos(2pft). Three
frequencies of the dynamic excitation were assumed,
namely: f
1
= 10 Hz, f
2
= 25 Hz and f
3
= 80 Hz. The fre-
quency f
2
is close to the first eigenfrequency of the sys-
tem (32.81 Hz), whereas the frequency f
3
is near to the
.ig 2. Wavelet transformation using wavelet daublet8 of
static displacement signal for the force localisation
x=0,5 m: a) detail D
1
, b) D
2
, c) D
3
, d) D
4
second one (106.50 Hz). In both, static and dynamic
cases, the concentrated forces acting on the structure were
localised in various points: x = 0,5 m, x = 1,0 m,
x = 1,5 m and x = 2,0 m. Altogether, 16 analyses were
performed.
As a structural response signal the static vertical dis-
placements and the amplitudes of vertical displacements
and accelerations were analysed. In each case the re-
sponse signal was computed in 512 points uniformly dis-
tributed at the longitudinal axis of the beam. This data
is treated as respective discrete signals in the space do-
main and is processed using wavelet transformation.
.ig 3. Wavelet transformation using wavelet daublet8 of
static displacement signal for the force localisation
x = 1,5m: a) detail D
1
, b) D
2
, c) D
3
, d) D
4
194
194
Usually, wavelets named daublet8 were used in this trans-
formation. As a result of the transformation we obtained
evident disturbances in the place where damage was
localised in the most detailed signal representations,
namely D
1
, D
2
or D
3
.
In .igs 2 and 3 details D
1
, D
2
, D
3
, D
4
of the wave-
let transformation of static response for the force
localisation at the points x = 0,5 m and next x = 1,5 m
are presented. Note in these figures that the damage (and
also the force) localisation is demonstrated in the utmost
evident way by the detail D
3
. Defect is visible for both
force positions, though greater effect was obtained for
the case when the force was localised near the damage.
It is worth to notice that in all graphs high disturbances
at the ends of interval appeared. They increased with
successive higher order signal representations. They re-
sult from the fact that the employed wavelets exceed the
range [0,1]. Therefore in wavelet transform algorithm it
was assumed that the analysed signal is periodic. Hence,
overlapping of wavelets results in disturbances at bound-
aries.
To compare the efficiency of static and dynamic re-
sponses in damage identification, transformations of these
.ig 4. Detail 3 of displacement amplitudes transforma-
tion for the harmonic force localisation x = 1,5 m and fre-
quencies: a) f
1
= 10 Hz, b) f
2
= 25 Hz, c) f
3
= 80 Hz
responses are presented in .igs 3 and 4, respectively.
The force P was localised at x=1,5 m and vertical dis-
placements were analysed. On the graphs 4a, b and c
detail 3 for frequencies of excitation f
1
= 10 Hz,
f
2
= 25 Hz, f
3
= 80 Hz is presented, respectively.
.ig 4 demonstrates that in this case damage was
properly identified for the frequencies f
1
and f
2
. The fre-
quency f
3
has not detected the damage. Of course, it is
connected with the specified damage position.
.ig 5. Detail 3 of acceleration amplitudes transformation
for the force position x = 1,5 m and frequency: f
2
= 25 Hz
Next, we studied the usefulness of the structural
response in the form of accelerations. Note, that it is
often easier to measure in situ the accelerations than the
displacements. .ig 5 illustrates the wavelet transform of
the accelerations for the case f
2
identical to the case
shown in .ig 4b.
Unfortunately, identification based on transformation
of acceleration amplitudes has not brought expected re-
sults. It needs further studies.
Let us come back to the problem of a choice of
excitation frequency. If we change our model so that the
dynamic force is acting at the point x=2,0 m and the
damage will be defined at the point x
d
=0,75 m, proper
damage identification will occur only for the frequency
f
3
=80 Hz. This phenomenon was shown in .ig 6.
Summarising the results of the above analyses we
observe that damage is detected as well for static as for
dynamic structural responses. The application of dynamic
excitation provides more possibilities in planning the ex-
periment. On the other hand, it can lead to some diffi-
culties in the valuation of dynamic response.
4.2. Number of measurement points
One of the key problems in the damage identifica-
tion is the prediction of the minimal number of measure-
ments required for proper data processing. .ig 7 pro-
vides the first insight into this problem. On the succes-
sive graphs 7a, b, c, the wavelet transformations of the
static response represented by 128, 64 and 32 uniformly
distributed measurement points are depicted.
Basing on .ig 7 we can conclude that in our case
the minimal number of measurements was 64. However,
it is difficult to define arbitrarily the number of experi-
195
195
.ig 6. Detail 3 of displacement amplitudes transformation
for the damage position x
d
=0,75 m, force localisation
x=2,0 m and frequencies: a) f
1
=10 Hz, b) f
2
=25 Hz, c)
f
3
=80 Hz
.ig 7. Wavelet transformation of static displacement sig-
nal for the damage position x
d
= 1,4 m and force
localisation x=1,5 m; number of measurement points: a)
128, b)64, c) 32
mental data, which is indispensable for damage identifi-
cation. This number depends on many factors, for ex-
ample, damage quantity measured relatively to the
undammaged part of the structure. .uture research will
tend towards minimisation of the number of measure-
ments and improvement of tools of signal processing.
4.3. Noise in measured data
Inevitable element of any experiment is measure-
ment inaccuracy. In this part of the paper we want to
quantify the level of measurement inaccuracy which made
damage identification impossible. To model the inaccu-
racy we used white noise generator.
.or the damage localised at the point x
d
=1,4m and
the force position x=1,5 m, static displacements of the
beam structure were analysed. The noise was superposed
on the signal for N=512 and 128 measurement points.
Details of wavelet transformation are presented in .igs 8
and 9.
The performed analysis demonstrated that proper
damage identification became practically impossible with
the noise level
7
10 3

m and
7
10 5

m, respectively. It
corresponded to measurement precision of the displace-
ment in the damage position (
3
10 5056 , 1

d
u m).
5. Concluding remarks
The application of wavelet transform to damage
localisation was discussed in the paper by the way of
several numerical examples. The analyses were based on
both, static and dynamic responses of damaged structures.
Information about undamaged structure was not used.
Presented examples showed that correct damage
identification was possible using transformation of sig-
nals issued from both, static and dynamic experiments.
Dynamic experiments provided more possibilities for
damage identification, however, in this case the valua-
tion of the experimental data could be more complicated
than in the static case.
The examples demonstrated that damage was proper
localized also with reduced number of experimental data
and with a specified level of noise, representing mea-
surement inaccuracy. The effectiveness of transformation
196
196 A. Garstecki, et al / JOURNAL O. CIVIL ENGINEERING AND MANAGEMENT 2004, Vol X, No 3, 191197
d
x=1,5 m and 512 measurement points; noise level
a)
8
10 2

, b)
7
10 1

, c)
7
10 3

m
d
x=1,5 m and 128 measurement points; noise level
a)
8
10 5

, b)
7
10 1

, c)
7
10 5

m
with the use of common types of wavelets is limited by
disturbances, which are induced at the ends of the space
domain of the transformed signal.
The great advantage of wavelets is that information
about the undamaged structure is not required in defect
identification process. The examples discussed in the
paper confirmed that wavelet transformation is a very
promising tool in structural identification problems. The
study will be continued in the direction of implementa-
tion of wavelet transformation to the identification of
damage in real life engineering structures.
Acknowledgement
.inancial support by State Committee for Scientific
Research, grant 5 T07 E 045 22 is kindly acknowledged.
References
1. Adams, R. D.; Cawley, P.; Pye, C. J.; Stone, B. J. A vibra-
tion technique for non-destructively assessing the integrity
of structures. J. of Mech. Engr. Sci. 20, 1978, p. 93100.
2. Cawley, P.; Adams, R. D. The location of defects in struc-
ture from measurements of natural frequencies. J. Strain
Anal., Vol 14, 1979, p. 4957.
3. Doebling, S. W.; .arrar, C. R.; Prime, M. B.; Shevitz, D. W.
Damage identification and health monitoring of structural
and mechanical systems from changes in their vibration
characteristics: A literature review. Report LA-13070-MS,
UC-900, Los Alamos National Laboratory, New Mexico
87545, USA, May, 1996.
4. Dems, K.; Mrz, Z. Identification of damage in beam and
plate structures using parameter dependent frequency
changes. Engineering Computations, Vol 18, 1/2, 2001.
p. 96120.
5. Lee, J.; Haftka, R. T.; Griffin, O. H. Jr; Watson, L.T.;
Sensmeier, M. D. Detecting delaminations in a composite
beam using anti-optimization. Struct. Optim., 8, 1994.
p. 93100.
6. Mrz, Z.; Garstecki, A.; Optimal loading in identification
and design of structures. In: Proc. of 2
nd
WCSMO,
Zakopane, (W. Gutkowski, Z. Mrz Ed.), I.TR, 1997,
p. 279284.
197
197 A. Garstecki, et al / JOURNAL O. CIVIL ENGINEERING AND MANAGEMENT 2004, Vol X, No 3, 191197
7. Mrz, Z.; Garstecki, A. Optimal loading conditions in de-
sign and identification of structures. Part 1. Discrete for-
mulation, Stuct. & Multidiscipl. Optim., 2004 (in print).
8. Burczyski, T.; Orantek, P. Hybrid evolutionary algorithms
aided by sensitivity information in identification and struc-
tural optimization. In: Proc. of 2
nd
ECCM, Krakw, 2001.
9. Waszczyszyn, Z.; Ziemiaski, L. Neural networks in me-
chanics of structures and materials new results and pros-
pects of applications. Computers and structures, Vol 79,
Issue 2225, Sept 2001, p. 22612276.
10. .uruta, H.; He, J.; Watanabe, E. A fuzzy expert system for
damage assessment using genetic algorithms and neural
networks. Microcomputers in Civil Eng., 11, 1996, p. 37
45.
11. Daubechies, I. The wavelet transform, time-frequency lo-
calization and signal analysis. IEEE Trans. on Informa-
tion Theory, 36, 1990, p. 9611005.
12. Mallat, S. A theory for multiresolution signal decomposi-
tion: the wavelet representation. IEEE Trans. Pattern Anal.
And Machine Intell. 11, 1989, p. 674693.
13. Chui, C. K. An introduction to wavelets. Academic Press.
Inc., San Diego, 1992.
14. Newland, D. E. Random vibrations, spectral and wavelet
analysis. 3
rd
ed Longman, Harlow and John Wiley, New
York, 1993.
15. Wang, W. J.; Mc.adden, P. D. Application of wavelets to
Gearbox vibration signals for fault detection, J. of Sound
and Vib. 192, 1996, p. 927939.
16. Wang, Q.; Deng, X. Damage detection with spatial wave-
lets. J. Solid & Struct., 36, 1999, p. 34433468.
17. Knitter-Pitkowska, A.; Pozorski, Z.; Ziopaja, K.; Garstecki,
A. Damage detection using dynamic approach and sensi-
tivity analysis In: Proc. CMM-2003, June 36, Gliwice/
Wisa, Poland.
199
199
2004, Vol X, No 3, 199208
ISSN 13923730
ADVANCED ANALYSIS O. STEEL .RAMES WITH E..ECTS O. JOINT
DE.ORMABILITY AND PARTIAL STRENGTH ACCOUNTED .OR
Marian A. Gizejowski
1
, Czeslaw J. Branicki
2
, Anna M. Barszcz
3
, Pawel Krol
4
1,2
University of Botswana, P. Bag 0061, Gaborobne, Botswana.
1
E-mail: gizej@mopipi.ub.bw,
2
E-mail: branic@mopipi.ub.bw,
3,4
Warsaw University of Technology, Armii Ludowej 16, 00-637 Warsaw, Poland.
3
E-mail: abar@il.pw.edu.pl,
4
E-mail: p.krol@ il.pw.edu.pl
Received 1 Dec 2003; accepted 21 May 2004
Abstract. The paper summarises the current progress in methods of advanced analysis for design of frames with semi-
rigid joints. The methods presented in the paper belong to general second-order refined plastic-hinge methods that allow
for the combined effects of joint stiffness degradation and distributed plasticity along the member length as well as
across the member sections. The advanced analysis for steel frame design, proposed by the authors, is based on the
spring-in-series model. The effect of joint semi-rigidity and partial strength is taken care of by specifying certain values
of the initial stiffness, ultimate moment and the shape factor of the moment-rotation characteristic for the spring repre-
senting the joint. The effect of imperfections affecting the performance of imperfect structural members in compression
is modelled by the application of a simplified tangent modulus concept combined with the reduction of the initial value
of the elasticity modulus. The effect of residual stresses is taken care of by specifying certain values of the shape
parameter for the moment-rotation characteristic of the spring representing the gradual yielding of the member. It is
dependent upon the cross-section type and fabrication method (ie upon the residual stress pattern resulting from rolling
or welding processes). A case study analysis is presented. Concluding remarks referring to the application of advanced
analysis in design, pertaining to the study case considered, are drawn.
Keywords: steel frame, semi-rigid joint, advanced analysis, spring-in-series model.
1. Introduction
Over the past 25 years, extensive research has been
made to develop and validate several advanced analysis
methods that can capture the frame ultimate strength
through the geometrically and materially non-linear analy-
sis carried out in such a way that member local and over-
all buckling capacity checks are not required. Second
order plastic-zone, conventional plastic-hinge and refined
plastic-hinge methods may be used in advanced analy-
sis. These methods differ as to the degree of refinements
in representing the plastic yield effects of joints and
members. The plastic zone method allows for the de-
tailed modelling of the gradual yielding effects across
the member sections and along the member length, in
contrast to the conventional plastic-hinge method. The
latter one does not allow for any refinements since a
rigid-perfectly-plastic relationship for the behaviour of
member sections and an elastic-perfectly-plastic relation-
ship for the behaviour of joints are used. The refined
plastic-hinge method is somewhere in between the two
above-mentioned methods but it may be calibrated in such
a way that it reproduces, with the accuracy acceptable
for practical application, the behaviour of real structures
(ie imperfect structures).
Refined plastic-hinge methods for advanced frame
analysis are based on different assumptions used in mod-
elling the stiffness degradation effects. Two methods
gained popularity, namely the method based on a two-
surface degradation model [1] and the method based on
a spring-in-series model [2]. In the former advanced
analysis, two element end springs representing the joint
flexibility are used and combined with the introduction
of two stiffness reduction factors representing the effect
of member inelastic deformations. The reduction factors
are applied directly to flexural terms of the element stiff-
ness matrix. In the latter method of advanced analysis,
two-element end springs are used in order to represent
the effects of joint flexibility and member yielding. This
method is further developed by the authors in order to
account for the effect of imperfections on the behaviour
of structural systems.
A brief summary of the spring-in-series model is
followed by the detailed discussion of issues related to
the influence of imperfections on the frame load deflec-
tion characteristic and the frame ultimate state. The pro-
posed values of model parameters are verified. The ex-
ample of a sway frame is considered in order to show
how the different values of model parameters affect the
200
200 M. A. Gizejowski, et al / JOURNAL O. CIVIL ENGINEERING AND MANAGEMENT 2004, Vol X, No 3, 199208
frame load deflection characteristic, the ultimate load
capacity and the magnitude of deflection at the frame
ultimate state. .inal conclusions are drawn.
2. Implementation of the spring-in-series model
Let us consider a finite element (.E) of discretised
structure, consisting of the member with two springs at
each node and subjected to flexural deformations only
[2]. The arrangement of components in this finite ele-
ment is depicted in .ig 1. The finite element representa-
tion of the incremental equilibrium equations in the lo-
cal co-rotational coordinates can be summarised as fol-
lows
( ) q k k k k k Q
j2 m2 m1 j
+ + + +
1
, (1)
where the set of equations (1) is expressed in terms of
internal and external static and kinematic variables, as-
sociated with the member nodes, joint nodes and finite
element nodes. All the node types are given in .ig 1,
with reference to the origin (denoted by 1) and the end
(denoted by 2).
.E
node 1
.E
node 2
Member
node 1
Joint
node 1
Joint
node 2
Member
node 2
Member
Member springs
Joint springs
.ig 1. .inite element with internal degrees of freedom
The incremental nodal forces and displacements in
equation (1) are expressed as follows:
1
1
1
1
1
1
1
]
1
2
2
2
1
1
1
M
M
M
M
M
M
j
m
m
j
Q ,
1
1
1
1
1
1
1
]
1
2
2
2
1
1
1

j
m
m
j
q , (2)
where
2 1
, finite element degrees of freedom (ex-
ternal DO.s),
2 1
,
j j
joint spring degrees of free-
dom (internal DO.s),
2 1
,
m m
member spring
degrees of freedom (internal DO.s).
The components of the finite element stiffness ma-
trix in equation (1) are expressed as follows:
the contribution of the joint instantaneous stiffness at
the origin:
1
1
1
1
1
1
1
]
1
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 1 1
0 0 0 0 1 1
1 j
S
j1
k (3)
the contribution of the joint instantaneous stiffness at
the end;
1
1
1
1
1
1
1
]
1
1 1 0 0 0 0
1 1 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
2 2 j
S
j
k
(4)
the contribution of member imperfections and plastic
zones developing from the origin onward;
1
1
1
1
1
1
1
]
1
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 1 1 0
0 0 0 1 1 0
0 0 0 0 0 0
1 m
S
m1
k
(5)
the contribution of member imperfections and plastic
zones developing from the end backward;
1
1
1
1
1
1
1
]
1
0 0 0 0 0 0
0 1 1 0 0 0
0 1 1 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
2 m
S
m2
k
(6)
and finally, the contribution of the member elastic stiff-
ness with the second-order effects taken into account;
1
1
1
1
1
1
1
]
1
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
22 21
12 11
k k
k k
L
EI
k
(7)
The terms of the member elastic stiffness component
given by equation (7) are written in the standard form:
2
22 11
15
2
4 + k k
,
2
21 12
30
1
2 k k
, (8)
where the member axial load factor
201
201 M. A. Gizejowski, et al / JOURNAL O. CIVIL ENGINEERING AND MANAGEMENT 2004, Vol X, No 3, 199208
EI
NL
2
2

, (9)
and: N the member axial force, L the element length,
E Young modulus, I sectional moment of inertia.
The contribution of the joint stiffness is evaluated
on the basis of joint tangent stiffnesses S
j1
(calculated
for the spring placed at the origin) and S
j2
(calculated
for the spring placed at the end). They depend upon the
magnitude of total moments in the current incremental
stage M
j1
=M
1
and M
j2
=M
2
, respectively. Adopting the
Kishi-Chen power model for the modelling of the stiff-
ness degradation process [1], the joint tangent stiffness
S
j
can be expressed as follows:
j
j
j n
n
n
R j
j
ini j j
M
M
S S
1
,
,
1
+
1
1
]
1
,
_

, (10)
where the joint properties are: S
j,ini
initial stiffness,
M
j,R
ultimate moment, n
j
shape factor of the mo-
ment-rotation characteristic accounting for the effect of
joint gradual yielding.
The contribution of the member imperfections af-
fecting the flexural behaviour and plastic zones is evalu-
ated on the basis of tangent stiffnesses S
m1
(calculated
for the spring placed at the origin) and S
m2
(calculated
for the spring placed at the end). They depend upon the
magnitude of total moments in the current incremental
stage M
m1
=M
1
and M
m2
=M
2
, respectively. In order to
describe this effect, the Kishi-Chen power model is also
adopted, so that the tangent stiffness S
m
can be expressed
in the way similar to equation (10):
m
m
m n
n
n
R m
m
ini m m
M
M
S S
1
,
,
1
+
1
1
]
1
,
_
, (11)
where the spring properties are: S
m,ini
initial stiffness
representing the effect of geometric imperfections, M
m,R
ultimate moment of the member section (if necessary

reduced in the presence of a non-zero axial force), n
m

shape factor of the moment-rotation characteristic ac-
counting for the effect of residual stresses.
Recognising that
1 1 1
M M M
m j
and
2 2 2
M M M
m j
, and condensing the set of equa-
tions (1) in order to eliminate internal degrees of free-
dom, the following condensed set of equations for the
finite element, referred from here on to as the finite
superelement (.SE), is obtained:

q k Q
T
, (12)
where the tangent stiffness matrix
1
]
1
22
12 11
.

T
T T
T
k symm
k k
L
EI
k
(13)
and the condensed vectors of nodal forces and displace-
ments
1
]
1
2
1
M
M

Q ,
1
]
1
2
1

q . (14)
The terms of the stiffness matrix in equation (12) can be
expressed as
( ) [ ] ( ) { }
( ) [ ] ( ) [ ] ( )( )
2
12 2 1 22 2 2 11 1 1
1
2
12 2 22 2 2 11
11
1 1 1 3 1 3
1 1 3 3
k r r k r r k r r
r k r k r r k
k
T
+ +
+
,
(15)
( ) [ ] ( ) { }
( ) [ ] ( ) [ ] ( )( )
2
12 2 1 22 2 2 11 1 1
2
2
12 1 11 1 1 22
22
1 1 1 3 1 3
1 1 3 3
k r r k r r k r r
r k r k r r k
k
T
+ +
+
,
(16)

( ) [ ] ( ) [ ] ( )( )
2
12 2 1 22 2 2 11 1 1
2 1 12
12
1 1 1 3 1 3
9
k r r k r r k r r
r r k
k
T
+ +
.
(17)
The fixity factors in equations (15)-(17) are calcu-
lated on the basis of the spring-in-series model. .ig 2
shows the finite superelement after the condensation pro-
cess.

Member
node 1
.SE
node 1
.SE
node 2
Member
node 2
Member
Integrated
springs
.ig 2. .inite superelement with external degrees of free-
dom
Two springs at each member end are integrated into
one spring that has the integrated spring properties given
by:
tangent stiffnesses
1 1
1
1 1
1
m j
S S
S
+
, (18)
2 2
2
1 1
1
m j
S S
S
+
(19)
ultimate moments
( )
1 , 1 , 1 ,
; min
R m R j R
M M M
, (20)

( )
2 , 2 , 2 ,
; min
R m R j R
M M M
. (21)
Thus the fixity factors
L S
EI
r
1
1
3
1
1
+
,
L S
EI
r
2
2
3
1
1
+
. (22)
By the transformation of the set of equations (12)
in order to express it in the coordinates associated with
202
202
the last known configuration and adding the contribution
of the axial deformations, the final set of finite element
local equations can be established. It has to be noted
that the vectors of nodal forces and displacements of the
final set of equations are referred to the external nodes
of the superelement (nodes of the discretised structure).
This is a set of equations for the element with the single
spring at each node but with the integrated stiffnesses
incorporated as a result of member and joint inelastic
deformations, with the influence of imperfections taken
into account.
3. Adopted solution procedure
Solution procedures for frame non-linear equilibrium
equations can be one of the following:
incremental-iterative methods that satisfy equilibrium
equations at each incremental step of analysis, thus they
allow to trace the equilibrium path accurately (within the
degree of accuracy made possible by assumptions adopted
in modelling),
simple incremental methods that do not iterate for the
equilibrium position at incremental steps, hence they pro-
duce a drift-off error in the prediction of the frame load
deflection characteristic.
.or practical applications, the simple incremental
method is preferred since it is straightforward, it posesses
the computational efficiency and, what is the most im-
portant, it can be treated as a natural extension of the
linear analysis into the region of structural non-linear
behaviour [1]. The computer implementation of advanced
analysis presented herein is therefore based on the simple
incremental procedure with a prescribed increment of the
applied loads. In this procedure, at each incremental step,
the values of member nodal forces and displacements
are computed as sum of the incremental values. The en-
tries to the stiffness matrix from each superelement are
updated at the end of current increment, and only then a
new load increment is applied. .ig 3 shows how the in-
tegrated spring properties are updated from the incre-
mental step i to the next step (i+1). The tangent
stiffnesses of separate springs S
ji
and S
mi
are calculated
on the basis of the total moments M
ji
= M
mi
= M
i
. The
stiffness S
i
of the integrated spring in the step i is calcu-
lated according to (18) or (19), depending on the end
number taken into consideration. The resultant stiffness
S
i
is kept constant for the incremental step (i+1). The
axial force entry to the superelement stiffness matrix is
also updated and the axial force N
i
from the step i used
for the prediction of incremental forces and displacements
in the step (i+1). When under the constant load incre-
ment the force state exceeds that resulting from the ulti-
mate moments (20) or (21), depending on the end num-
ber taken into consideration, a suitable scaling scheme
is used to calculate the reduced load increment in order
to avoid plastic hinges from forming within a constant-
stiffness load increment (see .ig 3). Load step sizes less
then or equal to the prescribed constant-stiffness load
increment magnitude are computed directly, since the
relations within the incremental step are linear. This is
combined with a suitable search procedure for the for-
mation of multiple hinges at one incremental step. Plas-
tic hinges may therefore form only after the load incre-
ment.
M
S= (S
j
-1
+S
m
-1
)
-1
incr ement (i- 2)
adopted appr oximation
actual
plastic
hinge
for mation
M
i-2
M
i+1
M
j,R
=M
R
M
i
M
i-1
M
m,R
(i)
(i)
(i+1)
( i+1)
.ig 3. Constant-stiffness load increment scheme
4. Calibration of model parameters
4.1. Stiffness degradation in compression and tension
The axial force may have a great impact on the flex-
ural stiffness of members. The P- effect is taken care
of by the introduction of second-order terms in defining
the flexural stiffness of the member element see equa-
tions (8). The effect of imperfections on member inelas-
tic deformations, such as residual stresses and out-of-
straightness, is not explicitly reflected in equilibrium
equations summarised in Chapter 3. The concept of tan-
gent modulus was used in [1] for modelling the buckling
behaviour of members in tension and compression. A
similar concept is adopted for the advanced analysis pro-
posed in this paper so that the Young modulus E is re-
placed in equation (13) by the tangent modulus E
T
. The
tangent modulus is constructed for a hypothetically ideal
member in such a way that in case of compression it
reproduces the behaviour of imperfect strut through the
buckling curves of design specifications. The following
formula is adopted
n
n
n
R
T
N
N
E E
1
0
1
+
1
1
]
1
,
_
, (23)
where E
0
reduced elasticity modulus representing the
effect of geometric imperfections for compression of slen-
der members (for tension members E
0
=E), N
R
com-
pression or tension capacity of the section, n shape
factor modelling the effect of residual stresses
r
on the
stress-strain curve of the axially loaded member.
M. A. Gizejowski, et al / JOURNAL O. CIVIL ENGINEERING AND MANAGEMENT 2004, Vol X, No 3, 199208
203
203
The tangent modulus given by equation (23) repro-
duces the buckling resistance N=N
Rb
through the equa-
tion
0
2
1
E
E
N
N
T
R
Rb
, (24)
where: compression member relative slenderness
ratio
A E
N
R
0
, (25)
and: l member slenderness, A cross-sectional area.
(1) D/B1,2 (2) D/B>1,2 (3) All
D

B B
d

0,5
0,5
0,5
0,5
0,3
0,3
0,3
0,3
B
0,075B
0,075d
0,125d
0,125B
1,0
1,0
0,25
0,25
+
+
+
+
+
+
+
+
+
- -
- -
- -
- -
-
.ig 4. Standard residual stress patterns
The shape parameter, reflecting the effect of residual
stresses s
r
, depends on the type of the cross-section.
Generally three different types of I-sections and associ-
ated with them standard residual stress patterns are used
(s
r
/f
y
in .ig 4). .or sections bent about the stronger axis,
the buckling curve is taken from Table 1[3].
Table 1. Buckling curves for typical I-sections
Technology Heat treatment D/B >1,2 D/B 1,2
Rolled a b
Not-heat-treated a Welded
Annealed b
An exercise is carried out to find the best-fit shape
parameters n for curves a and b. They are found to be
n=4 and n=3 for curves a and b respectively. The ef-
fect of geometrical imperfections on the stiffness of com-
pression members is approximately taken care of in such
a way that the reduced elasticity modulus is calculated
as the Young modulus divided by the partial safety fac-
tor g
c
applied to the buckling resistance of slender mem-
bers. In [1] the value 1/g
c
=0,85 is suggested. The Pol-
ish code [3] uses for design values of the buckling resis-
tance 1/g
c
= 0,75. The European pre-code [4] uses
1/g
c
=0,91. If the reference is made do the American
LR.D code [5], one can find out that the value of
1/g
c
= 75 , 0 85 , 0 877 , 0 is used. This agrees with the
recommendations of the Polish code [3].
It is worth noting that the method of reduced elas-
ticity modulus is a very convenient method to account
for the effect of geometrical imperfections, if compared
with the method of explicit imperfection modelling or
the equivalent notional load method [1]. If the latter two
methods are used, all the members have the elasticity
modulus E, regardless whether they are under tension or
compression.
4.2. Stiffness degradation in bending
Chen and Chui [2] proposed the following stiffness
degradation relationship for the member spring

m
S for
pr m
M
M
,
<
pr m
er m
M
M
,
,
, (26)
pr m
er m
pr m
pr m
m
M
M
M
M
M
M
L
EI
S
,
,
,
,
1
6
for 1
, ,
,

pr m pr m
er m
M
M
M
M
,
(27)
where M
m,pr
plastic moment of the cross section in the
presence of axial force, M
m,er
first yield moment, re-
duced in the presence of residual stresses, given by
y
R y
r
er
Zf
N
N
f
M
,
_

1
, (28)
and: Z elastic section modulus, f
y
yield stress,
r

maximum value of the residual stress coordinate that
for the standard cases can be taken from .ig 4.
In the computer implementation of the advanced
analysis developed in [2], a very large value of 10
+10
EI/L was assigned in the elastic region equation (26),
and a very small value of 10
10
EI/L in the post-yield
region (when M = M
m,pr
). The model is therefore some-
what complicated since is based on a discontinuous rela-
tionship, requiring artificial numbers to be used for avoid-
ing numerical difficulties.
A different model is developed for the proposed
advanced analysis. The effect of stiffness degradation in
bending is taken care of by the adoption of equation (10).
The model parameters are calibrated against the Chen-
Chui proposal [2] for the case of pure bending. When
integrating equations (26) and (27), the following sec-
tion spring moment-rotation characteristic is obtained
0 for
pr m
M
M
,
<
pr m
er m
M
M
,
,
, (29)
'
,
_
,
_

pr m
er m
pr m
pr m
M
M
M
M
EI
L M
,
,
,
,
1 1
6
1
1
]
1
,
_
,
_
+
pr m pr m
er m
M
M
M
M
, ,
,
1 ln 1 ln 1
for 1
, ,
,

pr m pr m
er m
M
M
M
M
. (30)
204
204
The average values of section plastic reserve shape
factors Z S
p
/ , where S is the section plastic modu-
lus for bending about the strong axis, and the maximum
residual stress factors
y r r
f / are used in the cali-
bration (Table 2).
Table 2. Parameters used in calibration
Section proportioning
Technology Parameter
D/B1,2 D/B>1,20
p

1,10 1,15
Rolled
r
0,50 0,30
p

1,20
Welded
r
1,00
The initial stiffness representing the effect of geo-
metric imperfections (both out-of-flatness of sectional
walls and member out-of-plane) are taken into consider-
ation assuming the value of ten times greater than the
stiffness multiplier in equation (27)
L
EI
S
ini m
60
,

. (31)
The lower-bound approximation is used for the calibra-
tion of shape factor n
m
. It is chosen in such a way that
the section spring moment-rotation characteristic of
present formulation, for larger values of the rotation,
approaches that proposed by Chen and Chui [2]. The
values of n
m
parameter are summarised in Table 3 and
the resultant characteristics are compared in .ig 5 with
those of Chen and Chui [2]. Numbers in brackets refer
to the standard residual stress patterns given in .ig 4.
Table 3. Shape factors n
m
of the moment-rotation characteristic
Section proportioning
Technology
D/B1,2 D/B>1,20
Rolled (2) 1,25 (1) 1,60
Welded (3) 0,75

0 0,0002 0,0004 0,0006 0,0008 0,001

0,0
0,2
0,4
0,6
0,8
1,0
p r e s e n t

s t u d y
f r o m

[ 2 ]
( 1 )
( 2 )
( 3 )
.ig 5. Comparison of section spring characteristics
4.3. Stiffness degradation of joints
The moment-rotation characteristic of joints adopted
in the present study is of the same nature as the relation-
ship for the modelling of member stiffness degradation
in bending. The stiffness degradation in the adopted
model is described by the continuous function equa-
tion (9), which appears to be different from the approach
recommended in European standardisation documents [4,
6]. The joint stiffness degradation used in [6] is a piece-
wise continuous function with the points of discontinuity
at the levels of 2/3 of the ultimate moment and at the
ultimate moment itself
ini j j
S S
,
for
R j
M
M
,
<
3
2
, (32)
( )
,
_
,
_
R j
ini j j
M
M
S S
,
,
1
1
3
2
for 1
3
2
,

R j
M
M
, (33)
0
j
S for
R
>1, (34)
where: y parameter equal to 2,7 for end-plate joints
and 3,1 for flange-cleated joints, and the rotation at the
ultimate moment is given by
ini j
R j
R
S
M
,
,
2
3

,
_
. (34)
The values of the joint stiffness at the discontinuity points
are given in Table 4.
Table 4. Change in values of joint normalised stiffness
Joint type Point
R j
M M
,
/
Change in
ini j j
S S
,
/
End-plate .lange-cleated
from 1,00 1,00 0,67
to 0,27 0,24
from 0,09 0,07 1,00
to 0,00 0,00
It can be noted from Table 4 that the parameter y
does not have any impact on the stiffness degradation in
end-plate and flange-cleated joints. In order to obtain an
objective evaluation of the shape factor n
j
of the pro-
posed M- characteristic, and to maintain a similar reli-
ability level as used in [6], the areas between the
ini j j
S S
,
/ curve from the European code [6] and that
proposed herein are considered. The positive sign A
(+)
is
assigned when the proposed curve lies below that from
[6] and the negative sign A
(-)
in the opposite case. By
considering the resultant parameter DA= A
(+)
+A
()
, and
predicting the value of n
,
factor in such a way that
205
205
DA @ 0, the best fit solution is obtained. On the average
of end-plate and flange-cleated joints, the value n
j
= 1,5
is selected. The comparison of the joint M-f character-
istic adopted in the proposed advanced analysis and those
from the European code [6] is given in .ig 6.

0 0,002 0,004 0,006 0,008 0,01

0,0
0,2
0,4
0,6
0,8
1,0
M
/
M
j
R

" f l a n g e - c l e a t e d

j o i n t s "
" e n d - p l a t e - j o i n t s "
p r e s e n t

s t u d y
.ig 6. Comparison of joint M-f characteristics
The joint physical parameters like initial stiffness
and ultimate moment may be calculated by the compo-
nent method given in [6]. The method requires the iden-
tification of the joint active components, calculation of
the stiffness and strength of each component and finally
the calculation of joint properties for the joint assembled
from components. Since the procedure for calculating the
joint physical properties is generally not friendly for hand
calculations, the use of computer-based calculations is
advisable [1, 7]. The proposed advanced analysis will
be eventually based on the joint type catalogue from
which the type of the joint can be identified and then its
mechanical properties calculated automatically.
5. Computer programme and its applications
5.1. Computer code LILANN
The computer implementation of the proposed ad-
vanced analysis is based on the conventional geometri-
cally linear plastic-hinge programme LILAN that has been
used in engineering education at the Gdansk University
of Technology and the University of Botswana (author
C.J. Branicki [8]). This ensures that the programme
LILANN being developed for the advanced analysis is
of the same complexity as its predecessor. Thus, con-
ventional plastic-hinge solving techniques and their nu-
merical implementations for the geometrically non-linear
analysis remain unchanged and may be applied within
the refined plastic-hinge concept based on the spring-in-
series model. Unlike other commercially available pack-
ages, the LILANN programme can be used with a great
confidence by structural engineers who are familiar with
the concepts and methods used in the programme be-
cause these concepts and methods have been taught dur-
ing the university education.
The research version of the computer programm
LILANN allows for the following analyses:
LL: geometrically linear (first-order) analysis with bi-
linear joint characteristics,
NL: geometrically non-linear (second-order) analysis
with bilinear joint characteristics,
NN: geometrically non-linear (second-order) analysis
with curvilinear joint characteristics of Chen-Kishi type.
The analysis of type NN is to be finally implemented
as advanced analysis for design purposes.
5.2. .rame for case study
In order to illustrate the differences between the appli-
cation of analyses implemented in the research version of
LILANN and the implication of different ways of model-
ling the geometric imperfections, the example of three-
bay six-story frame shown in .ig 7 is considered.

15,2 kN
15,2 kN
15,2 kN
15,2 kN
15,2 kN
8,25 kN
8 x 3 m = 24 m
6

x

3
,
6

m

=

2
1
,
6

m
42,5 kN
106kN
7 N 85 kN
106kN
42,5 kN
7 N 117

kN
.ig 7. .rame geometry and loading
The unstiffened extended end-plate beam-to-column
joints are used so that their semi-rigid partial-strength
properties have to be considered in the frame analysis
and design. Sizes of frame members as well as joint end-
plate thicknesses and bolt diameters are taken the same
as reported in [9]. The design was carried out automati-
cally using the commercial package ROBOT Millennium.
The options for elastic second-order analysis, non-linear
joint M-f characteristic and design according to the Pol-
ish steel code [3] were selected. .ig 8 shows the mem-
ber sizes and the joint types yielding from the design.

H
E
B

2
4
0
H
E
B

1
8
0
1 2 2 2 2 2 2 1
5 6 6 6 6 6 6 5
3 4 4 4 4 4 4 3
3 4 4 4 4 4 4 3
5 6 6 6 6 6 6 5
5 6 6 6 6 6 6 5
IPE 270

IPE 330
.ig 8. Member sizes and types of semi-rigid joints
The mechanical properties of beam-to-column joints
are listed in Table 5.
206
206
Table 5. Mechanical properties of beam-to-column joints
Mechanical properties
Joint
type
Connected profiles
Initial
stiffness
[kNm/rad]
Ultimate
moment
[kNm]
1 IPE 270 to HEB 180 16100 65
2 IPE 270 to HEB 180 54500 80
3 IPE 330 to HEB 180 25700 83
4 IPE 330 to HEB 180 84500 105
5 IPE 330 to HEB 240 31700 99
6 IPE 330 to HEB 240 79300 112
The classification of beam-to-column joints accord-
ing to stiffness is conveniently done with use of the fix-
ity factors. .ig 9 shows the bar graph for all the joints
listed in Table 5. It is clear that joints have generally to
be considered as semi-rigid. Examining the joint strength
one can conclude that all the joints are of partial-strength.

0 1 2 3 4 5 6
0
0,2
0,4
0,6
0,8
1
.
i
x
i
t
y

f
a
c
t
o
r

R i g i d j o i n t s
n o m i n a l l y r i g i d j o i n t s
n o m i n a l l y p i n n e d j o i n t s
p i n n e d j o i n t s
s e m i - r i g i d j o i n t s
.ig 9. Classification of joints according to stiffness
The influence of geometric imperfections is simu-
lated in such a way that three cases are considered using
the following notations for types of analysis:
1) the modulus of elasticity of members with a com-
pressive force is reduced to 0,85E and for members with
a tensile and zero force kept E, while the notional hori-
zontal loads are not added to the applied horizontal loads;
pressive force is kept the same as for members with a
tensile and zero force, and equal to E, while the notional
horizontal loads are added to the applied horizontal loads;
pressive force is reduced to 0,85E and for members with
a tensile and zero force kept E, while the notional hori-
zontal loads are also added to the applied horizontal
loads.
5.3. Results of the analysis of LL type
.ig 10 shows the frame load deflection characteris-
tics (the load factor vs the sway displacement of the frame
top node) LL-1, LL-2 and LL-3, corresponding to three
ways of modelling of geometric imperfections mentioned
in the previous chapter. The results show that the ulti-
mate load factor is the same for all the analyses. The
difference occurs in displacements. The maximum displace-
ment is for the analysis LL3, the smaller for the analysis
LL-2 and the smallest for the analysis LL-1. The history
of plastic hinge formation is shown in .ig 11 for the analy-
sis LL-1, .ig 12 for LL-2 and .ig 13 for LL-3.

0 5 10 15
Sway displacement [cm]
0,0
0,2
0,4
0,6
0,8
1,0
1,2
1,4
L
o
a
d

f
a
c
t
o
r

L L - 1
L L - 2
L L - 3
.ig 10. Load deflection characteristics for analyses LL
.ig 11. Plastic hinge formation history for analysis LL-1
207
207
5.4. Results of the analysis of NL type
tics (the load factor vs the sway displacement of the top
node of the frame) NL-1, NL-2 and NL-3, correspond-

0 5 10 15 20 25
0,0
0,2
0,4
0,6
0,8
1,0
1,2
1,4
N N - 1
N N - 2
N N - 3
L
o
a
d

f
a
c
t
o
r

.ig 14. Load deflection characteristics for analyses NL
.ig 17. Plastic hinge formation history for analysis NL-3
ing to three ways of modelling of geometric imperfec-
tions mentioned in Chapter 5.2.
The highest ultimate load factor is for the analysis
NL-1, a lower for the analysis NL-2 and the lowest for
the analysis NL-3. The difference occurs also in displace-
ments. The maximum displacement is for the analysis
NL3, the smaller for the analysis NL-2 and the smallest
for the analysis NL-1. The history of plastic hinge for-
mation is shown in .ig 15 for the analysis NL-1, .ig 16
for NL-2 and .ig 17 for NL-3.
5.5. Results from analysis of NN type
tics (the load factor vs the sway displacement of the top
node of the frame) NN-1, NN-2 and NN-3, correspond-
ing to three ways of modelling the geometric imperfec-
tions mentioned in Chapter 5.2. The tendency in the ul-
timate load factor and in the considered displacement is
the same as in the analysis of type NL. The history of
plastic hinge formation is shown in .ig 19 for the analy-
sis NN-1, .ig 20 for NN-2 and .ig 21 for NN-3.
6. Summary and concluding remarks
The conclusions drawn below pertain only to par-
ticular frame analysed in this paper. Summary of all the
calculations is given in Table 6. .or comparison, the
results from PHINGE_R [10] are also provided so that
one can see how the results change when different soft-

0 5 10 15 20 25
0,0
0,2
0,4
0,6
0,8
1,0
1,2
1,4
L
o
a
d

f
a
c
t
o
r

N N - 1
N N - 2
N N - 3
.ig 18. Load deflection characteristics for analyses NN
.ig 19. Plastic hinge formation history for analysis NN-1
208
208
ware is used. It can be concluded that though all analy-
sis types predict ultimate load within a 4% accuracy,
the LL analyses predict the displacements that are 60%
underestimated. Since partial-strength joints are used, the
member inelastic deformations may not be advanced in
structural members so that the difference in results for
NL and NN analyses is not observed. .or the frame used
in this study, both types of analysis may be recommended.
When notional forces and the reduced elasticity modulus
for elements with compression are used, the results are
too conservative (results for NN-3 in Table 6). Compar-
ing the results for analyses NN-1 and NN-2 one can see
that the analysis NN-2 leads to the slightly lower load
factor but of the meaningful higher displacements than
the analysis NN-1.
The conclusion is that for achieving the same level
of reliability in advanced analyses using notional loads
or reduced elasticity modulus of structural members, a
higher reduction of the elastic modulus of compression
members is required. The suggestion is that the value
close to that adopted in the American LR.D code [5]
and in the Polish code [3] should be preferred for the
frame considered. It implies that here the value
E
0
=0,75E should be used instead of E
0
=0,85E, as sug-
gested for advanced analysis [1].
The present study indicates a necessity for further
research in order to calibrate model parameters for the
proposed method in such a way that they reflect better
the behaviour of real structures (imperfect structures).
References
1. Chen, W. ..; Kim, S. E. LR.D steel design using advanced
analysis. Boca Raton & New York: CRC Press, 1997.
2. Chan, S. L.; Chui, P. P. T. Non-linear static and cyclic
analysis of steel frames with semi-rigid connections.
Amsterdam et al.: Elsevier, 2000.
3. PN-90/B-03200. Steel structures: Static calculations and
design. Warsaw: PKNMiJ, 1990.
4. ENV 1993-1-1. Design of steel structures. Part 1: General
rules and rules for buildings. Brussels: CEN, 1992.
5. AISC. Load and resistance factor design specifications.
Chicago: American Institute of Steel Construction, 1994.
6. ENV 1993-1-8. Design of steel structures. Part 1.8De-
sign of joints. Brussels: CEN, 2001.
7. .ealla, C.; Pilusi, V.; Rizzano, G. Structural steel semi-
rigid connections. Boca Raton & New York: CRC Press,
2000.
8. Branicki, C. J. LILAN: Program for plastic analysis of pla-
nar frames. CAL/CAT Structural analysis software. Uni-
versity of Gdansk-University of Botswana, 2002.
9. Brodka, J.; Barszcz, A.; Gizejowski, M. & Kozlowski A.
Stiffness and strength of sway frames with semi-rigid joints.
Rzeszow-Warszawa: Rzeszow Univ. Publishers, 2004 (in
Polish) [in press].
10. Gizejowski, M. Computational models of steel plane frames
with semi-rigid joints. Series: Civil Engineering, Vol 136.
Warsaw: Warsaw University of Technology Press: Habil.,
2000 (in Polish).
Table 6. Comparison of load factors and sway displacements
Software
Analysis
type
Model
type
Load
factor
Sway
[cm]
1 9,6
2 11,4 LL
3
1,384
13,2
1 1,374 16,1
2 1,344 16,7 NL
3 1,326 18,3
1 1,351 15,1
2 1,336 17,1
LILANN
NN
3 1,317 20,6
1 1,331 10,8
2 1,341 14,6 PHINGE_R NN
3 1,346 17,1
209
209
2004, Vol X, No 3, 209215
ISSN 13923730
.LEXURAL CAPACITY O. CONCRETE BEAMS REIN.ORCED WITH STEEL AND
.IBRE-REIN.ORCED POLYMER (.RP) BARS
Hau Yan Leung
Dept of Civil Engineering, Chu Hai College, Yi Lok Street, Riviera Gardens, Tsuen Wan, N.T., Hong Kong.
E-mail: brianleung@chuhai.edu.hk
Received 4 March 2004; accepted 21 Apr 2004
Abstract. Although much research on concrete beams reinforced with fibre-reinforced polymer (.RP) rods has been
conducted in recent years, their use still does not receive the attention it deserves from practicising engineers. This is
attributed to the fact that .RP is brittle in nature and the collapse of .RP-reinforced concrete member may be cata-
strophic. A rational beam design can incorporate a hybrid use of .RP rods and steel rods. Current design codes only
deal with steel-reinforced or .RP-reinforced concrete members. Therefore in this study some design charts and equa-
tions for concrete beam sections reinforced with .RP rods and steel rebars were generated. Results from the theoretical
derivations agreed well with experimental data.
Keywords: concrete, beam, flexural capacity, .RP, steel.
1. Introduction
The ultimate capacity of the concrete beam has been
adopted as one of the design criteria in some design codes
[13] for many years. These codes provide design equa-
tions and charts for steel reinforced concrete members
only. Today, a novel material known as fibre-reinforced
polymer (.RP) composite has been developed. .RP is
manufactured through a pultrusion process and it usually
contains a bundle of artificial fibres embedded in a resin
matrix. Three common kinds of fibres are carbon, glass
and aramid. Being used together with different resins,
such as epoxy and vinyl ester, and different surface pro-
files, such as sand-coated and fibre wrapping, a variety
of .RP products is thus commercially available. The
excellent properties of .RP include high strength-to-
weight ratio, good fatigue resistance, non-corrosive, non-
electromagnetic and high resistance to chemical attack
and due to these superiorities over a conventional steel,
.RP has been widely suggested by researchers to be in-
corporated in concrete design. A lot of research results
pertaining to the investigation of concrete beams rein-
forced with .RP rods has emerged recently [411].
Nonetheless, it has been reckoned that the brittle-
ness of .RP hinders further developments in concrete
structures. A more rational use of .RP in concrete beams
is to adopt a mix of .RP rods and steel rebars, so that
the beam ductility at its ultimate can be provided by the
steel rebars. Since the characteristics of .RP are differ-
ent from those of steel, it can be conceived that current
design formulas for reinforced concrete (RC) with steel
bars may not be applicable to RC with .RP bars. In
order to work out the flexural strength of concrete beams
reinforced with .RP rods and steel rebars, some design
charts or equations are necessary. The present work
describes a prediction model of the flexural strength of
concrete beams embedded with .RP rods and steel rebars
and the results from the prediction model agreed with
the experimental data with reasonable accuracy.
2. Theoretical derivations
.our major assumptions in the ensuing derivations
are:
Linear strain variation across the section;
Tension concrete ignored;
The non-linear stress profile of concrete is modelled
by an equivalent rectangular stress block;
The stresses of steel rebars and .RP rods can be
derived from their corresponding stress-strain curves.
By adopting two kinds of reinforcement inside a
beam section, it is more appropriate to place the .RP
rods closer to the bottom of the section. It is because
.RP has a higher resistance to external attack, such as
corrosion and chemical reaction; and more importantly,
a higher degree of protection to steel can be achieved
when the steel rebars are located further from the con-
crete surfaces. With this rationale in mind, the steel rebar
layer is likely to be at a smaller lever arm than that of
.RP rods.
210
210
2.1. Balanced section
.ig 1. Balanced ultimate condition
.ig 1 illustrates a concrete beam section internally
reinforced with steel rebars and .RP rods and their cross-
sectional areas are
st
A and
f
A respectively. Since the
.RP rods are used as the primary reinforcement and steel
as secondary, the depth of the steel rebars ( )
st
d is thus
proposed to be less than that of .RP rods ( )
f
d , ie .RP
rods are placed closer to the bottom of the beam. When
the beam section in flexure reaches a balanced ultimate
condition, crushing of compression concrete and rupture
of .RP rods occur simultaneously (.ig 1). If a linear
strain distribution across the section is assumed, the neu-
tral axis position ( ) c at balanced ultimate condition can
be easily determined by (1).
fu cu
cu
f
d
c

, (1)
where
cu
ultimate strain of compression concrete and
fu
ultimate strain of .RP rods. The linear strain pro-

file also leads to (2).
cu
st cu st
c
d
, (2)
where
st
is the strain of steel rebars. By combining (1)
and (2), it gives
fu cu
st cu
f
st
d
d

+
+
. (3)
Equation (3) indicates that the depth of steel rebars
is governed by the strain values of concrete, .RP rods
and steel rebars. It is apparent that when
st
d is smaller
than c, the steel rebars will be subject to compression.
Therefore in order to avoid compressive steel rebars,
c d
st
must be satisfied and the limiting value of
st
d
is c. If the limiting case is considered, the steel rebars
are in fact located at the neutral axis and 0
st
re-
sults. Equation (3) then reduces to (1). On the other
hand, when
st
d is increased,
st
increases as well. At
the onset of steel yielding,
st
reaches the yield strain
of steel ( )
y
and the value of
st
d to cause steel yield-
ing can be obtained by (4)
fu cu
y cu
f
st
d
d

+
+
. (4)
By changing the value of
fu
, the limiting ratios of

st
d and
f
d for both zero steel strain and onset of steel
yielding can be easily calculated by (1) and (4). In the
ensuing derivations, 0035 , 0
cu
is assumed.
As shown in .ig 2, (1) and (4) divide the
f st
d d
fu
space into three different regions. In region 1, the
ratio of
f st
d d is larger than its value at the onset of
steel yielding for a particular value of
fu
, the steel
rebars thus yield at the balanced ultimate condition of
the beam section. However, in region 2, the
f st
d d ratio
is less than the yielding value of ,
f st
d d the steel rebars
remain elastic at ultimate. In region 3, c d
st
< and com-
pression steel rebars are obtained. In the present study,
the beam section in the first two regions will be exam-
ined.
2.2. Classification of section
.ig 2.
f
st
d
d
against
fu
at balanced ultimate condition
Having established the value of
fu
, the value of
f st
d d ratio in the region 1 can then be easily chosen
from .ig 2. It should be noted that
y st
> and the
steel stress ( )
st
f equals its yield stress ( )
y
f at the ulti-
mate stage of the balanced beam section (.ig 3). Inter-
nal force equilibrium of the section leads to
( ) ( )
fu f y st c
f A f A b c f +
1
85 , 0 , (5)
where
c
f concrete cylinder compressive strength (in
MPa),
1
ratio of the equivalent rectangular stress
block depth to the neutral axis depth, b beam breadth
and
fu
f
ultimate tensile strength of .RP rods.
Rearrangement of (5) gives
y
f
st
st
st
fu
f
f
f
c
f
d
d
bd
A
f
bd
A
d
c
f
,
_
,
_
,
_
,
_
1
85 , 0
. (6)
H. Y. Leung / JOURNAL O. CIVIL ENGINEERING AND MANAGEMENT 2004, Vol X, No 3, 209215
f
s
t
d
d
(%)
fu
beam
section
strain
diagram
b
c
f
A
st
A
f
d
st
d
cu
fu
st
N.A.
211
211
.ig 3. Balanced beam section design (region 1)
.ig 4. Balanced beam section design (region 2)
It should be noted that (6) describes the relation-
ship between ( )
f f
bd A and ( )
st st
bd A for a balanced
beam section in region 1.
In contrast, the steel rebars inside the beam section
in region 2 of .ig 2 remain in its elastic range and steel
yielding does not occur. By considering the internal force
equilibrium again (.ig 4), it gives
( ) ( )
fu f st st c
f A f A b c f +
1
85 , 0 . (7)
It is noteworthy that
st
f in (7) is smaller than
y
f
. Re-
arrangement of (7) gives rise to
st
f
st
st
st
fu
f
f
f
c
f
d
d
bd
A
f
bd
A
d
c
f
,
_
,
_
,
_
,
_
1
85 , 0 . (8)
The elastic behaviour of steel rebars can be de-
scribed by
st st st
E f , (9)
where
st
E denotes the elastic modulus of steel rebars.
.rom the strain diagram in .ig 4, the steel strain can be
calculated by
cu
st
st
c
c d

,
_

. (10)
Substitute (10) into (9), it gives

cu
f
f
st
st st
d
c
d
d
E f
1
1
1
1
]
1
,
_
,
_
1
. (11)
Then, by putting (11) into (8), the variation of
( )
f f
bd A against ( )
st st
bd A for a balanced beam sec-
tion in region 2 is obtained:
( ) . 1
85 , 0
1
1
1
1
1
]
1
,
_
,
_
,
_
,
_
,
_
,
_
f
f
st
cu st
f
st
st
st
fu
f
f
f
c
d
c
d
d
E
d
d
bd
A
f
bd
A
d
c
f
(12)
By using (7) and (12), the variation of
( )
f f
bd A
against ( )
st st
bd A for any balanced beam sections in
regions 1 and 2 of .ig 2 can be calculated. To initiate
the generation of such a plot, % 5 , 1
fu
is selected.
.rom .ig 2, the limiting value for compression steel,
19 , 0
f
d c and the value for steel yielding,
31 , 0
f st
d d are obtained. Generation of the
( )
f f
bd A ( )
st st
bd A plot requires other input param-
eters and they include 40
c
f MPa,
( ) 7 , 6 / 6 , 27 05 , 0 85 , 0
1

cu
f , 600
fu
f MPa,
460
y
f MPa and 200
st
E GPa. Using this data, a
typical
( )
f f
bd A
( )
st st
bd A plot can then be devel-
beam
section
b
f
A
st
A
f
d
st
d
N.A.
strain
diagram
c
cu
fu
stress
diagram
c
f c
f 85 , 0
1
c
fu
f
fu f
f A
force
diagram
st
f
st st
f A
v st
<
beam
section
b
1
f
A
st
A
f
d
st
d
N.A.
strain
diagram
c
cu
fu
v
>
stress
diagram
c
f c
f 85 , 0
1
c
fu
f
fu f
f A
force
diagram
v
f v st
f A
212
212
.ig 5.
f
f
bd
A
against
st
st
bd
A
oped and it is given in .ig 5. In .ig 5, the values of
( )
f f
bd A and ( )
st st
bd A for different ratios of
f st
d d are shown. The solid lines ( ) 31 , 0 >
f st
d d in-
dicate that the steel rebars yield at the ultimate state of
the balanced beam section whereas the dotted lines
( ) 31 , 0 <
f st
d d show that the steel rebars are in its elas-
tic range. It is clear from .ig 5 that there is an inverse
relationship between ( )
f f
bd A and ( )
st st
bd A for ev-
ery value of .
f st
d d To maintain a balanced beam sec-
tion, an increase in ( )
st st
bd A always leads to a reduc-
tion in ( )
f f
bd A and vice versa. By using this plot,
one may manipulate the combinations of ( )
f f
bd A and
( )
st st
bd A so as to achieve a balanced section in region
1 or region 2.
In .ig 5, each line represents a balanced beam sec-
tion in which the steel yields or does not yield at the
ultimate state. .or example, when the solid line for
4 , 0
f st
d d is considered, this line gives the balanced
ratio of ( )
f f
bd A once ( )
st st
bd A has been deter-
mined. When a higher ratio of ( )
f f
bd A is used, the
section becomes over-reinforced whereas an under-rein-
forced section will be obtained when a reduced value of
( )
f f
bd A is adopted. It is interesting to observe that
when % 6 , 2 >
st st
bd A , the section will become over-
reinforced irrespective of the ( )
f f
bd A ratio. When
3 , 0
f st
d d is chosen, the limiting value of
st st
bd A
becomes 4,0 % approximately. One may therefore con-
clude that the ( )
f f
bd A ( )
st st
bd A plot is useful in
that the classification of designed beam sections can be
readily obtained.
2.3. .lexural Capacity
To determine the nominal flexural capacity of the
beam section at ultimate state, one may need to consider
over-reinforced and under-reinforced sections in regions
1 and 2 separately.
In region 1, when an over-reinforced beam section
is adopted, ie concrete crushes and steel yields at ulti-
mate. Internal force equilibrium and linear strain distri-
bution across the section lead to
( ) ( ) , 85 , 0
1 f f y st c
f A f A b c f +
(13)
cu
f
f f f f
c
c d
E E f
,
_

. (14)
Substitute (14) into (13) and then solve a quadratic
equation for c. The nominal value of the flexural capac-
ity ( )
n
M can be calculated by
( ) ( )
( ) ( ).
2
85 , 0
1
1
c d f A c d f A
c
c b c f M
f f f st y st
c n
+
+
,
_
(15)
In contrast when an under-reinforced section is
designed, the .RP rods are expected to break and yield-
ing of steel occurs at ultimate state. Internal force equi-
librium gives rise to
( ) ( )
fu f y st c
f A f A b c f +
2
85 , 0 . (16)
And thus

b f
f A f A
c
c
fu f y st
2
85 , 0
, (17)
where
2
is the ratio of the depth of the equivalent stress
block to the neutral axis depth when the maximum con-
crete strain is less than its ultimate value. When a linear
stress profile of concrete is used,
2
can be calculated
by equating the force of concrete stress profile and that
of equivalent stress block. Having determined the value
for c,
n
M can be determined by
( ) ( )
( ) ( ).
2
85 , 0
2
2
c d f A c d f A
c
c b c f M
f fu f st y st
c n
+
+
,
_
(18)
However, when over-reinforced beam section in re-
gion 2 is used (ie concrete crushes, steel remains in elastic
range at ultimate state), internal force equilibrium results
in
( ) ( )
f f st st c
f A f A b c f +
1
85 , 0 . (19)
Linear strain distribution is assumed and thus gives rise
to (20) and (21).

cu
st
st st st st
c
c d
E E f

,
_

, (20)
(
%
)
/
f
f
b
d
A
(%) 5 , 1
fu
(%) /
st st
bd A
213
213

cu
f
f f f f
c
c d
E E f
,
_

. (21)
Substitution of (20) and (21) into (19) leads to a
quadratic equation of c. By solving the value of c,
n
M
can be calculated by
( ) ( )
( ) ( ).
2
85 , 0
1
1
c d f A c d f A
c
c b c f M
f f f st st st
c n
+
+
,
_
(22)
If the beam section is under-reinforced (ie .RP
breaks and steel in elastic range), (19) becomes (23):
( ) ( )
fu f st st c
f A f A b c f +
2
85 , 0 . (23)
The steel stress can be calculated by Eq (24):
fu
f
st
st st st st
c d
c d
E E f
,
_

. (24)
.inally the equation for
n
M is formed:
( ) ( )
( ) ( ).
2
85 , 0
2
2
c d f A c d f A
c
c b c f M
f fu f st st st
c n
+
+
,
_
(25)
Non-dimensional plots of ( )
2
f c n
bd f M

against
( )
f f
bd A for a beam section in regions 1 and 2 are
given in .igs 6 and 7 respectively. To generate these
plots, the values used in addition to the previously as-
sumed data are 4 , 0
f st
d d for .ig 6 and 3 , 0
f st
d d
for .ig 7.
As in the two design charts (.igs 6 and 7), the non-
dimensional flexural capacity increases with the non-di-
mensional .RP area for any beam section in region 1 or
.ig 7. ( )
2
f c n
bd f M

against ( )
f f
bd A for a beam
section in region 2
.ig 6. ( )
2
f c n
bd f M

against ( )
f f
bd A for a beam
section in region 1
2. The dotted lines show under-reinforced beam sections,
whereas solid lines indicate over-reinforced sections.
3. Comparisons with experiments
The theoretical flexural strengths are compared with
the results of an experimental study. A total of eight
concrete beams internally reinforced with steel rebars and
G.RP rods were tested in which four beams were cast
with low-strength concrete (~35 MPa) and the rest were
made of medium-strength concrete (~ 60 MPa). Ordi-
nary Portland cement was used to make the concrete.
The detailed mix proportions are given in Table 1. All
beams were cast horizontally.
Table 1. Concrete mix proportions
Materials quantity (kg/m
3
) Mix 1 Mix 2
Ordinary Portland cement 340 530
20 mm aggregates 720 850
10 mm aggregates 370 365
Sand 720 450
Water 210 185
Superplasticiser (litre/m
3
) -- 1.62
Actual concrete cube strength (MPa) 35,6 61,0
Two different kinds of reinforcement were adopted
at two different levels in the tension zone of all beam
specimens. High-yield steel deformed bars and glass fi-
bre-reinforced polymer (G.RP) composite rods were
used. The high-yield steel reinforcing rebars were of di-
ameter 10 mm and they had a yield strength and elastic
modulus of 460 MPa and 200 GPa respectively. The two
kinds of G.RP rods adopted were manufactured by
(%) /
f f
bd A
30 , 0 /
f st
d d
2
f
c
u
n
b
d
f
M
(%) /
f f
bd A
40 , 0 /
f st
d d
2
f
c
u
n
b
d
f
M
214
214
Hughes Brothers Inc. USA and Pultrall, ADS Compos-
ites Group, Canada. These G.RP composite rods were
made from high-strength E-glass fibres and a durable
resin matrix through a pultrusion process. A wide range
of bar diameter is available in the market and a diameter
of 3/8 inch was selected because its nominal size
(9,53 mm) was comparable to that of the steel rebar
adopted. Tensile tests were performed and results indi-
cated that the tensile strength and elastic modulus of the
two G.RP rod types were 760 MPa and 40,8 GPa and
703 MPa and 41,7 GPa respectively.
The total depth of the beam section was 200 mm
and the distances between the bottom surface of the beam
and the centreline of the steel rebars and G.RP rods
were 70 and 40 mm respectively. In addition to the hy-
brid flexural reinforcements, minimum shear reinforcing
stirrups was also designed at two shear spans.
Testing of specimens was conducted using a four-
point bending setup. Each of the 2,5 m long specimens
were tested over a total span of 2,2 m with a shear span
of 0,7 m at both ends, leaving a constant bending region
of 0,8 m.
Some input data is listed below and more details
can be found in Table 2.
150 b mm 130
st
d mm
1 , 157
st
A mm
2
( ) % 81 , 0
st st
bd A
460
y
f Mpa 200
st
E GPa
160
f
d mm
81 , 0
f st
d d
According to the beam details, the theoretical mo-
ment capacity for each beam was calculated (Table 2).
It is clear that the theoretical ultimate moments match
with the experimental ultimate moments with reasonable
accuracy. Without taking into account the effects from
tension stiffening and compression steel, an underesti-
mated ultimate moment is found.
Additionally, it was observed that the strain of steel
for all beams exceeded its yielding value and the beam
failure modes were largely governed by crushing of com-
pression concrete except for the last two specimens in
which rupture of G.RP rod occurred. This confirms that
the all beams are categorised in region 1 and increase in
concrete compression capacity in the last two beam-sec-
tions lead to brittle G.RP-rupture.
4. Conclusions
.rom the present study, the following conclusions
can be drawn.
Since a balanced beam section is governed only by
the ultimate strains of compression concrete and outer
.RP rods, the depth of the internal steel rebars plays an
important role in its ultimate behaviour. Three distinct
regions in a ( )
f st
d d
fu
plot can be developed and
the steel rebars may (1) yield; (2) remain in its elastic
state; or (3) be subject to compression at the ultimate
stage of the section.
A simple chart which allows a quick check of the
reinforcement areas for a balanced beam section has been
developed. It is found that when the .RP rods are lo-
cated closer to the bottom of the beam, the steel ratio
( )
st st
bd A is inversely proportional to the .RP ratio
( )
f f
bd A .
There exists a limiting value of ( )
st st
bd A for a
particular ratio of ( )
f st
d d . When the designed value
of ( )
st st
bd A is greater than this limiting value, the beam
section must be over-reinforced irrespective of the
( )
f f
bd A values.
Design charts of ( )
2
f c n
bd f M

for any combina-
tions of ( )
f f
bd A and ( )
st st
bd A have been developed.
It allows a quick estimate of the flexural capacity of the
designed beam section.
References
1. American Concrete Institute. Building code requirements
for reinforced concrete ACI 318-95 and commentary (318R-
95), .armington Hills, Michigan, 1995.
Table 2. Comparisons of ultimate moments
Beam number
cu
f
(MPa)
f
A
(mm
2
)
fu
f
(MPa)
f
E
(GPa)
f
f
bd
A

(%)
(exp) n
M
(kNm)
) (theo n
M
(kNm)
% 100
(exp)
) (
n
theo n
M
M

L2 35,6 142,7 760 40,8 0,59 22,2 17,6 79,3 %
L4 35,6 142,7 703 41,7 0,59 20,7 17,8 86,0 %
L5 35,6 214,0 760 40,8 0,89 23,1 20,0 86,6 %
L7 35,6 214,0 703 41,7 0,89 23,0 20,1 87,4 %
H2 61,0 142,7 760 40,8 0,59 21,1 20,6 97,6 %
H4 61,0 142,7 703 41,7 0,59 21,4 20,6 96,3 %
H5 61,0 214,0 760 40,8 0,89 27,1 23,7 87,4 %
H7 61,0 214,0 703 41,7 0,89 27,0 23,9 88,5 %
215
215
2. British Standards Institution. Structural use of concrete.
Part 1. Code of practice for design and construction BS
8110-1. Part 1:1997, BSI, 389 Chiswich High Road, Lon-
don W4 4AL, 1997.
3. American Concrete Institute. Guide for design and con-
struction of concrete reinforced with .RP bars ACI 440.1R-
01, .armington Hills, Michigan, 2001.
4. Granga, Rao H. V. S.; .aza, S. S.; and Anderson, A. De-
signing concrete beams with .RP rebars. Proceedings of
the Concrete 2000 Conference, 1993, p. 18471856.
5. Benmokrane, B.; Chaallal, O.; and Masmoudi, R. Glass-
fibre reinforced plastic (G.RP) rebars for concrete struc-
tures. Construction and Building Materials, Vol 9, No 6,
1995, p. 353364.
6. Theriault, M.; and Benmokrane, B. Effects of .RP rein-
forcement ratio and concrete strength on flexural behaviour
of concrete beams. Journal of Composites for Construc-
tion, ASCE, Vol 2, No 1, 1998, p. 716.
7. Grace, N. ..; Soliman, A. K.; Abdel-Sayed, G.; and Saleh,
K. R. Behavior and ductility of simple and continuous .RP
reinforced beams. Journal of Composites for Construction,
ASCE, Vol 2, No 4, 1998, p. 186194.
8. Masmoudi, R.; Theriault, M.; and Benmokrane, B. .lex-
ural behavior of concrete beams reinforced with deformed
fiber reinforced plastic reinforcing rods, ACI Structural
Journal, Vol 95, No 6, 1998, p. 665676.
9. Alsayed, S. H.; Al-Salloum, Y. A.; and Almusallam, T. H.
Performance of glass-fiber reinforced plastic bars as a re-
inforcing material for concrete structures. Composites Part
B: Engineering, Vol 31, 2000, p. 555567.
10. Peece, M.; Manfredi, G.; and Cosenza, E. Experimental
response and code models of G.RP RC beams in bend-
ing. Journal of Composites for Construction, ASCE, Vol 4,
No 4, 2000, p. 182190.
11. Toutanji, H.; and Saafi, M. .lexural behavior of concrete
beams reinforced with glass fiber-reinforced polymer
(G.RP) bars. ACI Structural Journal, Vol 97, No 5, 2000,
p. 712719.
217
217
2004, Vol X, No 3, 217226
ISSN 13923730
TRUSS OPTIMIZATION UNDER COMPLEX CONSTRAINTS AND RANDOM LOADING
Arnoldas Norkus
1
, Romanas Karkauskas
2
Vilnius Gediminas Technical University, Saultekio al. 11, LT-10223 Vilnius-40, Lithuania.
E-mail:
1
arnoldas.norkus@st.vtu.lt;
2
rokark@st.vtu.lt
Received 2 .eb 2004; accepted 17 Apr 2004
Abstract. An actual design of light-weight structures must evaluate strength, stiffness and stability constraints as well as
the nature of external loading. A designed structure must satisfy optimality and safety criterions per prescribed mainte-
nance period. One faces the known difficulties when trying to implement several from the above-mentioned require-
ments into optimization problem for further successful numerical realisation. A method to formulate the optimization
problem, incorporating all above described criterions, the mathematical model and algorithm to solve it numerically,
taking into account the stochastic nature of external loading, are presented for elastic-plastic truss-type structure.
Keywords: elastic-plastic structure; optimization; stiffness and stability constraints; stochastic loading process, finite
element discrete model.
1. Introduction
The aim of structural design is a creation of a struc-
ture to satisfy optimality and safety criterions in respect
of maintenance (strength, stiffness, stability) requirements.
It is important to note that stiffness and stability con-
straints often dominate versus strength conditions in ac-
tual design of optimal structures. When solving the limit
equilibrium problem [13], the structural parts deforma-
tions or nodal displacements can exceed the fixed ad-
missible magnitudes and/or fail due to stability loss.
Actually, the structure is designed for a certain period
for prescribed reliability level. Therefore one must also
seek to evaluate the loading process peculiarities (peri-
ods of activity and absence of separate loads when com-
bining their common influence on structural response per
reference period of time) in the practical design. An
employment of elastic-plastic structural behaviour model
allows to evaluate the residual response effects. The ap-
peared self-equilibrated internal forces allow to adapt the
prescribed reliability level optimal structure to variable,
random in general, process of external loading [4]. Such
an approach allows to evaluate the alternating plasticity
case and thus the structure fail mode due to low-cycle
fatigue. This failure case is the actual one in structural
analysis and design [5, 6].
Thus, the main optimal structural design problem is
to be stated as the structural optimization problem under
presence of stiffness, stability requirements and gener-
ally, the stochastic process of loading.
The designed optimal structure is in the state prior
to plastic failure, resulting from stiffness and stability
constraints. During the optimization procedures the truss-
type discrete structure tensile members are assumed to
be loaded up to the yield limit, the compressive mem-
bers up to the critical limits, combining yielding and
buckling modes, the stiffness constraints are represented
via the nodal displacements restrictions in prescribed
directions. The problem solution realises the certain op-
timization cycles, in each cycle employing the set of
members limit forces, identified in the previous one. The
optimization cycles are continued until a certain conver-
gence.
The structure limit states represent strength, stiff-
ness and stability loss failure modes. The total structure
failure probability
rf
P is a union of the above modes.
Taking the structure physical and geometrical properties
to be described as random values, the structure
rf
P evalu-
ation (reliability analysis) problem is rather complicated
even applying the constant random loads and evaluating
strength conditions only (ie limit states correspond to
different plastic collapse mechanisms modes) [7, 8]. The
reliability evaluation problem, stated as the mathemati-
cal non-convex mathematical programming problem, so-
lution process results in the selection of the most reli-
able plastic mechanisms, corresponding to local
extremums, in order to calculate
rf
P versus plastic col-
lapse (note that different plastic mechanisms can be se-
lected in each optimization cycle). The number of plas-
tic mechanisms (failure modes) increases considerably
with the number of redundancy.
An additional evaluation of displacement limitations,
that of stability loss limit states in concert with stochas-
218
218
tic process of loading makes the main optimal struc-
tural design problem rather complicated even for the el-
ementary structures and is more interesting in theoretical
versus practical usage.
Let the physical and geometrical properties of struc-
tural members and the loading application points repre-
sent as determinate ones. The actual variation of carry-
ing structural members properties in principle must be
low, therefore under certain restrictions can be taken as
determinate ones, fixed with sufficiently high reliability
level. Then the actual elastic response of elastic-plastic
structure (in terms of internal forces and displacements),
subjected by stochastic loading process can be stated as
the above extreme (due to certain reliability per refer-
ence period of time) values identification problem. After
when these identified values are fixed as determinate
ones, the main optimization problem of the structure in
terms of determinate values is to be solved. The similar
approach, taking into account only strength conditions,
was applied [4].
A direct application of the mathematical program-
ming method [9] for optimization problem solution does
not ensure the obtained extremum to be the global one.
The way to overcome the latter lack is an application of
mathematical programming theory in concert with ex-
treme energy principles. The obtained optimization prob-
lem, containing the complementarity condition, is the
multi-extremum one, as its solution results in a certain
number of discrete points. Incorporation of the comple-
mentaritys condition into the objective function of opti-
mization problem [1012] ensures the admissible set to
become the convex one in case of convex yield condi-
tions, having the unique optimal solution. But one must
note that problem numerical solution via mathematical
programming methods is very complicated and in some
cases unavailable.
The task of the present investigation is further de-
velopment of the reliability-based optimization under
presence stiffness and stability constraints problem solu-
tion methods, applying the extreme energy principle [2,
12] in concert with certain optimization procedures for
elastic-plastic truss-type structure (eg truss), subjected by
stochastic process of loading.
2. Truss optimization problem mathematical model
and algorithm
The truss optimization problem under the presence
of stiffness, stability constraints and stochastic process
of loading consists of 3 principal parts of optimization
cycles (iterations):
1. Determining truss variation bounds of elastic forces
+
e
S ,
e
S and displacements
+
e
u ,
e
u per time pe-
riod T , satisfying prescribed safety reliability
rf rs
P P 1
.
2. Defining the actual stress and strain state (SSS) of
the structure, prior to plastic collapse, correspond-
ing to fixed variation bounds of internal forces and
displacements (analysis problem).
3. Optimizing the areas of members (conditioning the
limit
0
S and that of critical
cr
S forces) to satisfy
strength, stiffness, stability and constructional con-
straints.
Each of the parts consists of separate problems, to
be solved individually one after another during each
optimization cycle and continued until certain conver-
gence. The iterative optimization procedure is conditioned
by the circumstance that elastic forces and displacements
from one side and limit and critical forces from the other
side depend on the actual cross-sectional areas of bars,
being as input and output data of optimization cycle parts
1 and 3.
2.1. Evaluation of elastic forces and displacements
caused by stochastic loading process
Let the loading process to be described by action
of individual loads ( ) ( ) ( ) ( ) ( ) .
~ ~ ~ ~ T
m 2 1
t F ..., , t F , t F t F Then
the j-th elastic axial force and the i-th elastic displace-
ment of the truss are expressed, respectively, by:
( ) ( ) ( ) ( )
( ) ( ) ( )
( ) n, 1,..., j , t F
t F ... t F t F
t S ... t S t S t S
m
1 i
i ji
m jm 2 j2 1 j1
j em, j e2, j e1, j e,

+ + +
+ + +
~
~ ~ ~
~ ~ ~ ~
( ) ( ) ( ) ( )
( ) ( ) ( )
( )
+ + +
+ + +
m
1 i
i ii
m im 2 i2 1 i1
i em, i e2, i e1, i e,
, t F
t F ... t F t F
t u ... t u t u t u
~
~ ~ ~
~ ~ ~ ~

(1)
where
jm
and
im
are the truss finite element method
discrete model j-th elastic axial force and i-th elastic
displacement influence determinate coefficients versus
m-th load process ( ) t F
~
m
, respectively.
To identify the maximum and the minimum magni-
tudes of processes (1) per reference period of time
T due to prescribed design safety reliability
rf rs rd
P 1 P P for all internal forces, selected into
the vectors
( ) ( )
T
n , e , e , e
T
j , e e
S ..., , S , S S
+ + + + +

2 1
S ,
( ) ( )
T
n e, e,2 e,1
T
j e,
S ..., , S , S S

e
S ,
and that of nodal displacements
( ) ( )
T
m , e , e , e
T
i , e e
u ..., , u , u u
+ + + + +

2 1
u
,
( ) ( )
T
m , e , e , e
T
i , e e
u ..., , u , u u

2 1
u
,
A. Norkus, R. Karkauskas / JOURNAL O. CIVIL ENGINEERING AND MANAGEMENT 2004, Vol X, No 3, 217226
219
219
one must solve the following problem:
[ ]. , 0
,
) (
; ) (
P
e
~
e
e e
r
T t
P
t
t
rd
,
_

+
+
u u u
S S S
~
(2)
An application of the exact methods (eg random
crossing theory) for processes combination (Eqn (1))
problem seems to be rather complicated one, dealing with
multiple convolution integrals calculation problems to
build the above values distribution and density func-
tions. The latter are required for elastic response extreme
magnitudes identification due to fixed design probabil-
ity
rd
P [13, 14]. The number of such processes, to be
investigated, is equal to n + m, ie to the total number of
components of internal forces and displacement vectors
(2). The number of these operations (recalculations of
the above-mentioned elastic response magnitudes, pro-
vided in the first part of each optimization cycle), also
depends on the successful problem solution starting point,
chosen for faster convergence of the main optimization
problem.
Thus, the exact evaluation methods of extreme elas-
tic response values seems to be more interesting from
the theoretical point of view, being in contradiction when
applying them for practical usage to solve real main struc-
tural optimization problems.
Taking into account peculiarities of the investigated
problem, the loading process approximation (load com-
bination) problem useful approach would be the selec-
tion of process finite number time points, to define the
extreme influence of loading to structure elastic response
values. The Turkstra's load combination rule, applied for
mutually independent load processes, proposes the num-
ber of such points to be equal to the number of indi-
vidual loads processes. The employed rule, being imple-
mented in design codes of several countries, provides
sufficient lower bound probability evaluation to identify
the extreme magnitudes of considered values. But one
must note that the Turkstra's rule accuracy is also lim-
ited by certain features of loads processes to be com-
bined [15].
.ollowing the Turkstra's rule idea to solve the prob-
lem (1) numerically, the aim is to represent the loading
process by certain number of individual loads determi-
nate magnitudes combinations, sufficient to identify the
extreme loading influence to all structural elastic response
values. The identified combinations of individual loads
in terms of determinate magnitudes must be selected to
the loading matrix (matrixes), being constant during all
iterative optimization procedures. The loading matrices
represent the possible loading states combinations of
individual loads magnitudes, corresponding to prescribed
reliability level and time period under consideration.
Let all the individual loads be mutually indepen-
dent, stationary processes. Applying the Borges-
Castanheta load model, the load process is approximated
as a sequence of rectangular pulses of fixed duration t,
taking into account probability of load absence during
the pulse. Then the load distribution function for pulse
reads:
( )
( )
'
+
<
> +
1, q p
0, F , F P q
0, F , F P q p
P
* *
F
*
F
* *
F,
when
when
(3)
where ( ) F P
F
is the load distribution function during
its activity period; p* the load absence probability
during pulse t.
The load magnitude, corresponding to certain prob-
ability
, F
P is found from Eqn (3), taking into account
the load absence probability, ie:
( )
( ) ( )
( ) ( ) ( )
( )
( )
( )
'
,
_
,
_
1. F P
0 P
q p ,
q
p F P
P
, 0 P
q p F P 0 P
q 0,
, 0 P
q F P 0 ,
q
F P
P
P
F,
F
* *
*
*
F, 1
F
F
* *
F, F
*
F
*
F,
*
F, 1
F
F,
when
when
when
(4)
Then the load extreme magnitudes distribution func-
tions per reference period T read:
. T/ v
, (F))) (P (1 1 (F) P , (F)) (P (F) P
v
F, F minx,
v
F, F max,

(5)
The Turkstra's rule does not limit the number of
loads when combining their extreme effect, but usually
it is enough to evaluate three time-variable forces (when
required, the preliminary reliability investigation of com-
mon activity periods of more variable loads must be
provided). Then the extreme, for instance, internal force
(Eqn (1)) magnitude with required
rd
P per reference
period T, is achieved, analysing the following loading
process time points:
( )
( )
( ) ( ) ( )
( ) ( ) ( )
( ) ( ) ( )
( ) ( ) ( )
'
+ +
+ +
+ +
+ +
, P P 0,5 P 0,5 P
, 0,5 P P P 0,5 P
, 0,5 P 0,5 P P P
, 0,5 P 0,5 P P P
P P
ekstr rd,
1
ekstrS
1
ekstrS
1
ekstrS
1
ekstrS ekstr rd,
1
ekstrS
1
ekstrS
1
ekstrS
1
ekstrS ekstr rd,
1
ekstrS
1
ekstrS
1
ekstrS ekstr rd,
1
ekstrS
ekstr rd,
1
Sei ekstr
T e1, e1,1 e1,1
e1,1 T e1, e1,1
e1,1 e1,1 T e1,
e1,1 e1,1 T e1,
3
1 i
(6)
where t
1
> t
2
> t
3
.
When calculating determinate maximum and mini-
mum magnitudes, it is proposed to select the quintiles
by:
P
rd
= P
rd(max)
P
rd(min)
(7)
220
220
and 0,5, as it fit sufficiently with magnitudes, identified
by exact, random crossing methods [16, 17].
When applying the described loading process ap-
proximation method to evaluate all extreme elastic re-
sponse values, the loading matrices, selecting the required
loads determinate magnitudes combinations, (eventual
loading process extreme states) are:
[ ]
( ) ( ) ( ) ( )
( ) ( ) ( ) ( )
( ) ( ) ( ) ( )
[ ]
( ) ( ) ( ) ( )
( ) ( ) ( ) ( )
( ) ( ) ( ) ( )
,
1
1
1
1
1
1
]
1
1
1
1
1
1
1
]
1
+
t F min t F min t F min t F min
t F t F min t F t F min
t F t F t F min t F min
F
,
t F max t F max t F max t F max
t F t F max t F t F max
t F t F t F max t F max
F
3
T
3
2 2
T
2 2
1 1
T
1
T
-
3
T
3
2 2
T
2 2
1 1
T
1
T
2 1 2
2 2
1
1 1
2 1 2
2 2
1
1 1
(8)
where determinate values correspond to extreme P
rd(max)
,
P
rd(min)
(Eqn (7)) and 0,5 quantiles.
Note, that taking into account Eqns (4) and (5),
the extreme magnitudes per considered period of time
are calculated by improved design probabilities:
( )
( ), P P
q
P 1 1
P F (t) F min
, P P
q
p P
P F (t) F max
*
r(min)
1
F
*
v
r(min)
1
F
F i
i T,
*
r(max)
1
F
*
*
v
r(max)
1
F
F i
i T,
,
_
,
_

j
i
i
T
v
or
for
j i
> , j = 1, 2, 3. (9)
.ind that the quantiles, resulting the different sign ex-
treme magnitudes of loads (.ig 1), are constrained by
( )
( ) ( )
( )
( ) ( ) 0. t minF 0 P q (1 1 P
0, t maxF 0 P q p P
i
v
F
* *
min r
i
v
F
* * *
max r
>
,
_
<
>
,
_
+ >
for and
for
(10)
The vectors of extreme elastic response magnitudes then
are formed from obtained elastic solution in terms of
matrices
[ ] [ ] [ ] [ ] [ ]
[ ] [ ] [ ] [ ]
[ ] [ ] [ ] [ ] [ ]
[ ] [ ] [ ] [ ]
+
+ + +
+
+ + +
+
+
+
+
F u
, F F u
, F S
, F F S
e
e
e
e
(11)
by selecting certain extreme magnitudes from correspond-
ing matrices lines.
The above elastic response influence matrices are
split to the ones, consisting of positive and zero or nega-
tive and zero magnitudes [2], to satisfy:
[ ] [ ] [ ]
+
+ and [ ] [ ] [ ].
+
+ (12)
Note, that matrices (8) in coordinate space of indi-
vidual loads represent the convex polyhedron (created
according to fixed reliability
rd
P ) outer surface apexes,
corresponding to stochastic loads possible common ac-
tion states. An assemblage of these apexes to matrix
[ ] F provides all possible extreme structural response
causing loading cases:
[ ] [ ] [ ] F S
e
and [ ] [ ] [ ] F u
e
. (13)
Then the extreme elastic response magnitude vec-
tors are formed by selecting certain extreme magni-
tudes from corresponding above matrices lines. This ap-
proach versus to that of Eqn (11) requires more efforts
when selecting all apexes and is applicable for less
number of individual forces cases, when their action is
prespecified by non-alternating direction processes, other
cases.
2.2. Mathematical model of analysis problem
The structure is subjected to the known external
loading, the areas and physical-geometrical properties of
structural members are fixed. The structure, prior to plas-
tic collapse, SSS evaluation problem is realised by the
following mathematical model [12, 18] for discrete truss-
type structure finite element model:
find
[ ] [ ] [ ] ( )
( ) max
2
1
2
1
2
0 1 2 2 2 1 1 1

+ + +
+
cr e
T
e
T T T T
G G G
S S
S S
subject to . ,
2 1
0 0 (14)
The quadratic programming problem (14) contains
the following values:
( ) ( )
+ + + + +
n , e , e , e
T
j , e e
S ..., , S , S S
2 1
S vector of structural
extreme positive axial forces;
221
221
PF(F)
1,0
-F
PF,(F)
+
-
PF(F)
PF(F)
p*+q*PF(0)
^
1-(1-q*PF(0))
^
^
p* p*
(p*+q*PF(0))
^
^
q*PF(0)
^
(q*PF(0))
^
^
0 0
PF(F)
^
^
1-(1-(p*+q*PF(0)))
Pr
1,0
Pr
-
PF,(F)
+
F
.ig 1. Distribution functions for extreme load magnitudes
( ) ( )

n , e , e , e
T
j , e e
S ..., , S , S S
2 1
S vector of structural
extreme negative axial forces;
( )
T
j , 1 1
vector of Lagrange multipliers of comple-
mentarity conditions for tensile bars, reading
( ) 0
0 1
+
+
S S S
j , e j , r j ,
; (15)
j r
S
,
the j-th finite element (bar) residual force in the
running optimization cycle;
j
S
, 0
j-th finite element
limit axial force (yielding force),
y j y j ,
, A S
0
the material yield limit,
j
A j-th finite element area,
arbitrarily fixed or identified via previous optimization
cycle; ( )
T
j , 2 2
the Lagrange multipliers of comple-
mentarity's conditions for compressive bars, reading
( ) 0
2

j , cr j , e j , r j ,
S S S , (16)
where
j cr
S
,
j-th finite element limit buckling-yielding
axial force, arbitrarily fixed or identified via previous
optimization process; [ ] G constant per optimization
process structure finite element discrete model ( ) n n -
dimensional influence matrix for residual internal forces,
reading
[ ] [ ][ ][ ][ ] [ ] [ ][ ] [ ] [ ] K
T T 1
B K R K R K B G . (17)
Here:
[ ] K tension-compression diagonal stiffness
j
j
l
A E
ma-
trix of structural elements j = 1, 2, , n ((n total
number of bars;
j
l , E the material elasticity modulus
and the j-th bar length, respectively); [ ] K the struc-
ture finite element ( ) m m -dimensional discrete model
global stiffness matrix, where m number of global dis-
placements;
[ ] B ( ) n n 6 -dimensional configuration matrix, contain-
ing either zero elements or configuration submatrices for
internal forces
[ ] ( ) , j ..., , j , j k , b B
k , j j
6 4 6 5 6 (18)
expressed via direction cosines in the global coordinate
system:
[ ] ( )
j j j j
cos cos cos B 0 0 0
; (19)
[ ] K
quasidiagonal ( ) n n 6 6 -dimensional matrix, con-
taining j-th space finite element stiffness diagonal
submatrices
[ ]
[ ] [ ]
[ ] [ ]1
1
]
1
j j
j j
j
j
j
k k
k k
l
A E
K
(20)
in the global coordinate system, where:
[ ] ,
1
1
1
1
]
1
j
2
j j j j
j j j
2
j j
j j j j j
2
j
cos cos cos cos cos
cos cos cos cos cos
cos cos cos cos cos
k (21)
[ ] ( ) m n R 6 dimensional configuration matrix of local
and global displacements, containing unit and zero com-
ponents.
Note that introducing the above finite element dis-
crete model notations, the internal force and nodal dis-
placement influence coefficients and in (1) are
obtained from the matrices
[ ] [ ] G , [ ] [ ] [ ] [ ] [ ] .
1
]
1
T T 1
B K R K (22)
222
222
The solution of the quadratic programming prob-
lem (14) yields the Lagrange multipliers
1
and
2
mag-
nitudes. The components of residual internal forces
( ) ( )
T
n r, r2 r,1
T
j r,
S ..., , S , S S

r
S and that of residual
displacements ( ) ( )
T
m r, r,2 r,1
T
i r,
u ..., , u , u u
r
u are de-
termined by
[ ]( ),
2 1 r
S G [ ]( ).
2 1 r
u (23)
Note, that one cannot state the above obtained re-
sidual structural response values to be the actual ones.
Generally, the exact distribution of residual values de-
pends upon the loading history, observed/simulated from
individual loads processes [19]. Actually, in case of sto-
chastic loading, the loading history prediction practically
is unavailable. Thus, an employment of mathematical
model (14) yields only the possible distribution of re-
sidual structural response values, ensuring adaptation of
optimal structure to most dangerous eventual stochastic
loading states corresponding to prescribed reliability level
per reference period of time.
Having identified the above-mentioned values of
residual response, the extreme magnitudes of internal
forces and displacements are obtained by
. ,
; ,
+ +
+ +
+
+
e r e r
e r e r
u u u u u u
S S S S S S
(24)
2.3. Mathematical optimization model of structure
bars cross-sectional areas
The structure bars cross-sectional areas optimization
mathematical model under stiffness and stability con-
straints consists of:
1. Axial strength conditions expressed via areas of
optimised bars;
2. Strength conditions versus buckling of bars;
3. Displacements limitations constraints;
4. Constructive limitations for bars areas.
The structure optimality criterion expresses total
material minimum weight of the bars.
Thus, the structure optimization model under pres-
ence of all the above-mentioned conditions is as follows:
find min l A W
n
k
n
r
r k
k

0
1 1
subject to
, A A
, m ..., 2, 1, t , u
A E
l S S
u
, u
A E
l S S
u
, n ..., 2, 1, k , S A
, S A
min k, k
t min t,
n
1 j
j k,
j t,
min
j t,
t
max t,
n
1 j
j k,
j t,
max
j t,
t
0 j j k, cr
j j k, y
j
j
+
(25)
where:
k
A is the optimized cross-sectional area of the
k-th group of bars;
k
n is the number of bars in k-th
group;
r
l is the total length of bars, corresponding to
k-th group; n n ... n n
n
+ + +
0
2 1
;
0
n is the number of
optimized parameters ie the number of different group
of bars; is the structure material density;
y
,
cr
are material yield and critical stresses for tensile and

compressive-buckled members, respectively;
t
u is the dis-
placement, being constrained along the certain direction
t;
t
m is the number of constrained displacements;
0 < >
min t, max t,
u 0, u are the upper and the lower
bounds for displacement magnitudes to be constrained;
min , k
A is the lower bound of cross-sectional area
k
A
magnitude (constructive limitation).
Applying the virtual displacement principle, the dis-
placement
t
u can be expressed by
n
j
j , k
j , t
ext
j , t
t
A E
l S S
u
j
1
, (26)
where
ext
j , t
S is j-th bar total internal force, corresponding
to loads combination, resulting in the maximal (minimal)
displacement
t
u magnitude;
j , t
S is j-th bar internal force
caused by unit force applied onto the node along the
direction of nodal constrained displacement
t
u of the
truss, being in the state prior to plastic collapse.
The structural member design stress versus buck-
ling due EN3 [20] is calculated by:
y cr
, (27)
where
is certain bar reduction coefficient, depending

on compressive member dimensionless slenderness .
The dimensionless slenderness is prescribed by ratio:
E
.
Here the unit slenderness is defined by i l
b
, where
b
l is an actual length of the buckled member, i is the
radius of gyration,
y
E
E

2
is the Euler's slen-
derness. .or example, the tube cross-sectional reduction
coefficient analytically is described by the relation
1
1
2 2
+
, where ( ) [ ]
2
2 0 1
2
1
+ + , ,
with being the variance coefficient (eg 21 0, for
hot laminated pipes, 34 0, for cold laminated pipes).
The truss-type structural members cross-sectional
areas optimization problem is the non-linear mathemati-
cal programming problem, having one extremum. Two
solutions cases due to possible admissible sets of vari-
ables are shown in .igs 2 and 3 in case of two opti-
mized parameters.
223
223
The .ig 2 presents an optimal solution in the case
when the displacement constraints are not activated. The
optimization process is preconditioned by strength/sta-
bility constraints.
.ig 3 presents the case when the optimal solution is
preconditioned also by stiffness constraints. In consid-
ered case the optimal solution contains the satisfied as
equality leastwise one strength/stability condition in con-
cert with stiffness constraint. It is obvious that the opti-
mal solution objective function for the case results the
more structural weight magnitude, when compared with
the obtained one, presented in .ig 2.
Analysing the above .igures, one can obviously find
that the optimization process converges to unique solu-
tion in both possible cases.
2.4. Analysis and optimum design algorithm main steps
The structure stochastic load combination problem
in terms of created, constant per optimization process
matrices/matrix, to predefine structural elastic response
extreme values, is already provided. The minimum weight
elastic-plastic truss-type structure optimization consists
of the following eight steps:
1. Create the configuration matrices [ ] B and [ ] R .
2. Determine the limit axial forces
0
S and
cr
S .
(Choose primary cross-sectional areas of bars for
the start of optimization procedures, introduce the
new ones, corresponding to the previous iteration
optimization result, for the running optimization
iteration).
3. Create structure discrete model quasidiagonal stiff-
ness matrix for tensile-compressive elements [ ] K
(Eqn (20)) and global stiffness matrix [ ] K .
4. Determine
+
e
S ,
e
S and
+
e
u ,
e
u (Eqns. (11) or
(13)), applying the loading matrices or matrix.
A2
A1
0
Admissible set
Optimal solution
W(A1, A2)=const.
min
A =(A1, A2)
* * *
f2(S , S , S0, Scr, A2)=0
+ -
f3(u , u , ut)=0
+ -
f1(S , S , S0, Scr, A1)=0
+ -
.ig 2. Piecewise-linear admissible set
5. Create influence matrix of residual internal forces
[ ] G (Eqn (17)).
6. Solve the SSS evaluation quadratic programming
problem to find residual structural response values
(problem (14)).
7. Solve structure bars areas optimization problem
under the presence of strength, stiffness and stabil-
ity constraints (problem (25)).
8. Check the optimization problem prescribed conver-
gence criterion in respect of previous iteration re-
sult. When it is not satisfied, repeat steps 28. The
prescribed criterion can be the fixed admitted toler-
ance of structure weight function, that of the bars
areas (both criterions are correlated).
Note, that the number of iterations of optimization
problem solution depends on successful starting point (ie
chosen for 2-nd step primary input data (bars areas)).
They can be chosen arbitrarily, solving eg the constant
loading problem, or any similar one, being naturally close
to actual constraints of the problem under investigation.
2.5. Numerical example
Consider the space steel structure of 25 bars (.ig 4),
subjected to 3 different mutually independent stochastic
load processes.
The first load ) (
1
~
t F is applied to the 1-st node (di-
rection cosines in respect of structure global axes x, y, z
are: , 8146 , 0 cos
x
2037 , 0 cos
y
and . 5431 , 0 cos
z
The second load ) (

2
~
t F is applied to the 2-nd node (di-
rection cosines in respect of structure global axes x, y, z:
, 8304 , 0 cos
x
2491 , 0 cos
y
and . 4983 , 0 cos
z
The third load ) (

3
~
t F consists of simultaneous (fully cor-
related) action of two loads, being parallel to the axis z
A2
A1
0
Admissible set
Optimal solution
A =(A1, A2)
* * *
W(A1, A2)=const.
min
f1(S , S , S0, Scr, A1)=0
+ -
f3(u , u , ut)=0
+ -
f2(S , S , S0, Scr, A2)=0
+ -
.ig 3. Non-linear admissible set
224
224
and applied to the 3-rd and 6-th nodes, respectively. All
three loads are prespecified by non-alternating direction
processes (no direction changes). The load processes are
described by stationary pulses (5). The load distribution
function during activity periods satisfy the Gumbel's law:
( ) ))), ( exp( exp( u F a F P
F

then
( ) )). ln (ln(
1
1
r
F
P
a
u F P
The loads processes are described by the following

parameters:
: ) (
1
~
t F
1
= 91 day,
1
u = 141,62 kN,
1
a = 1,90kN,
, 2493 , 0
*
1
q
*
1
p = 0,7507;
) (
2
~
t F :
2
= 91 day,
2
u = 115,69 kN,
2
a = 2,10 kN;
, 1032 , 0
*
2
q
*
2
p = 0,8968;
.ig 4. Truss design scheme
) (
3
~
t F :
3
= 0,26 day,
3
u = 19,82 kN,
3
a = 1,23 kN;
, 0039 , 0
*
3
q
*
3
p = 0,9961.
The truss service term is taken T = 50 years, design prob-
ability 999 , 0 1
rf rs rd
P P P .
Taking into account the problem features (direc-
tion nature of loads, design probability magnitude, ser-
vice term, duration of pulses and probabilities of activ-
ity per pulses) the stochastic loading combinations
(apexes) of polyhedron, describing the loads common
action possible set, can be described by the following
five combinations. Theses apexes of polyhedron, indi-
cated by solid lines in .ig 5, represent the loads pos-
sible variation volume, created due to design probability
.
rd
P The dashed lines indicate the loads variation bounds
in case of repeated-variable loading in terms of extreme
loads variation bounds. The polyhedron apexes assembled
into the loading matrix, read:
[ ]
1
1
1
1
1
1
1
1
1
]
1
T
T
T
T
T
T
t F
t F t F
t F
t F t F
t F
F
) ( max 0 0
) ( max ) ( max 0
0 ) ( max 0
) ( max 0 ) ( max
0 0 ) ( max
3
2
3 2
2
1
3 1
1
.
.ig 5. Loading process polyhedron
Taking into account loads constant directions, the
fixed by (9) design probabilities are to be chosen as
P
rd
= P
rd(max)
, P
rd(min)
= 0 and the other quintiles to be
equal 0,5 (Eqn (8)). Then applying (11) and (12) to cal-
culate determinate loads magnitudes, the loading matrix
finally reads:
[ ]
1
1
1
1
1
1
]
1
00 , 30 0 0
10 , 20 42 , 120 0
0 42 , 120 0
10 , 20 0 31 , 147
0 0 31 , 147
T
F
.
Eight groups (ie 8 different truss areas) of pipe type
cold laminated bars, (.ig 3) are optimized to ensure truss
minimal weight under presence strength, stiffness, stabil-
ity and constructional constraints. The material proper-
ties are: yield limit
y
240 MPa, elasticity modulus
E = 207 GPa. Truss nodes 1 and 2 extreme displace-
ments are constrained in directions x and z by 1 cm. The
225
225
constructional requirement introduces minimal bars area
to be 2 cm
2
.
The optimization process was realized per 13 itera-
tions (see Table). As starting point input data (see row
0 of the Table) was taken an optimization solution of
the considered truss, loaded by the same three loads as-
suming them to be applied as constant values with ex-
treme magnitudes, presented in the above loading ma-
trix, ie: 174,31 kN; 120,42 kN; 30,00 kN for the first,
second and third loads respectively. The starting point
(optimal solution) was obtained in 12 iterations, apply-
ing the same algorithm, for its starting point taking all
areas to be equal to 10 cm
2
ones.
The stochastical loading process subjected truss
optimal solution was reached in 13 iterations (the last
row of the Table) with 0,017 % tolerance for 3rd and
7th areas and 0 % tolerance for remaining areas and to-
tal weight of the structure.
Some notes on optimal truss behavior, being adapted
to considered loading process. The 1st node reached it
allowable 1 cm extreme magnitude, while the other dis-
placements do not achieve this magnitude. The 4th and
5th areas reached their minimum equal to 2 cm
2
area
magnitudes. The 1-2, 1-4, 4-9 truss members are loaded
up to the critical values. Critical states for tensile mem-
bers are not achieved. A deformable behaviour compari-
son of the optimized truss with the one, obtained for the
constantly loaded truss (ie with the starting point repre-
senting optimization result row 0 of the Table). The
constantly loaded truss 1-2, 1-4 members were loaded
up to the critical values. Critical states for tensile mem-
bers were not achieved.
3. Conclusions
1. The performed numerical experiment illustrates
the efficiency of proposed algorithm, when applying the
proposed optimization under presence stiffness and sta-
bility constraints method for the truss, subjected to sto-
chastic loading process.
2. The convenient loading approximation method
required for iterative structural optimization to identify
extreme magnitudes of structural response with prescribed
reliability per reference period of time is presented. The
loading process approximation principles can be also
applied for other structural analysis and design cases.
3. The new truss structural optimization model is
presented. The result reliability is ensured by the appli-
cation of the extreme energy principle for analysis prob-
lem to fix the actual structural stress and strain state in
each optimization cycle.The proposed optimization
method principles can be implemented for other types
of structures.
4. The evaluation of stochastic loading process na-
ture, implemented into structural optimization process,
ensures material resources savings, when compared with
the optimal result, obtained when representing the load-
ing process either by repeated variable process or con-
stant loads in terms of extreme magnitudes, achieved in
the considered period of time.
References
1. Atrek, E.; Gallager, R. H.; Ragsdell, K. M. and
Zienkiewicz, O. C. New directions in optimum structural
design. New-York: John Wiley, 1984. 592 p.
2. Cyras, A. A. Mathematical models for the analysis and
optimization of elastoplastic structures. Chichester: Ellis
Horwood Lim., 1983. 121 p.
3. Levy, R. and Lev, O. E. Recent developments in structural
optimization. J. Struct. Engng. ASCE 112, 1987, p. 1939
1962.
4. Cyras, A. and Norkus, A. On optimization of elastoplastic
structures under random loading. Mech. Res. Comm.,
Vol 20, No 4, 1993, p. 293299.
Optimization problem solutions per iterations
Truss weight
(kg)
Iteration
number
A
1

(cm
2
)
A
2

(cm
2
)
A
3

(cm
2
)
A
4

(cm
2
)
A
5

(cm
2
)
A
6

(cm
2
)
A
7

(cm
2
)
A
8

(cm
2
)

0 2,00 8,089 12,540 2,000 4,163 5,383 2,759 17,817 519,35
1 2,826 7,434 13,414 2,000 2,000 5,418 6,513 15,869 546,80
2 3,565 7,791 11,726 2,000 2,000 2,767 5,515 10,920 435,63
3 3,373 7,642 11,543 2,000 2,000 2,696 5,636 11,072 432,79
4 3,737 7,760 11,507 2,000 2,000 2,682 5,604 10,990 432,74
5 3,106 7,658 11,519 2,000 2,000 2,694 5,634 11,057 432,15
6 3,686 7,780 11,495 2,000 2,000 2,681 5,603 10,988 432,71
7 3,183 7,670 11,511 2,000 2,000 2,695 5,634 11,061 432,37
8 3,702 7,760 11,500 2,000 2,000 2,684 5,608 11,000 432,78
9 3,160 7,682 11,562 2,000 2,000 2,703 5,732 11,091 434,74
10 3,696 7,732 11,573 2,000 2,000 2,696 5,746 11,020 435,51
11 3,432 7,701 11,582 2,000 2,000 2,697 5,762 11,010 435,00
12 3,429 7,702 11,576 2,000 2,000 2,697 5,764 11,010 435,00
13 3,429 7,702 11,578 2,000 2,000 2,697 5,765 11,010 435,00
226
226
5. Macke, M. A reliability-based design format for fatigue-
sensitive steel structures. In: Proc. of 8th Int. Conf. on
Applic. on Statist. and Probab., Sidney, Australia, 1999,
Roterdam: Balkema, 2000.
6. Sobzyk, K. Spenser, .. Random fatigue: from data to
theory. Boston: Academic Press, 1992. 254 p.
7. Simoes, L. M. Stochastically dominant modes of frames
by mathematical programming. J. Struct. Engng. ASCE 116,
1990, p. 10401060.
8. Ellis, J. H.; Zimmerman, J. J.; Corotis, R. B. Stochastic
programs for identifying critical structures collapse mecha-
nisms. Appl. Math. Modelling, Vol 15, No 7, 1991, p. 367
373.
9. Khan, M. R. Optimality criterion techniques applied to
frames having general cross-sectional relationships. AIAA
J. 22(5), 1984, p. 669676.
10. Jarmolajeva, E.; Atkoinas, J. Shakedown loading opti-
mization under residual displacements formulation and
solution for circular plates. J. of Civ. Engng. and Manag.,
Vol VIII, No 1. Vilnius: Technika, 2002, p. 5467.
11. Kalanta, S. New formulations of optimization problems of
elasto-plastic bar structures under displacement constraints.
Mechanics (Mechanika), 5(20). Kaunas: Technologija,
1999, p. 916 (in Russian).
12. Karkauskas, R. and Norkus, A. Optimization of geometri-
cally nonlinear elastic-plastic structures under stiffness con-
straints. Mech. Res. Comm., Vol 28, No 5, 2001, p. 505
512.
13. Bucher, C. Some recent software developments for stochas-
tic structural analysis. In: Proc. of 7th Int. Conf. on
Structur. Safety and Reliability (ICOSSAR), Kioto, Japan,
1997, Roterdam: Balkema, 1981.
14. Ditlevsen, O. and Madsen, H. O. Probabilistic modelling
of man-made load processes and their individual and com-
bined effects. In: Proc. of 3rd Int. Conf. on Struct. Safety
and Reliability (ICOSSAR), Kioto, Japan, 1997, Roterdam:
Balkema, 1981.
15. Mori Yasuhiro, Kato Takahiro. Probabilistic models of com-
binations of stochastic loads for limit state design. J. Struct.
Safety 25, 2003, p. 6997.
16. Larrabee, R. D. Approximate stochastic analysis of com-
bined loading. Research report R78-28 of Dept. of Civ.
Engng., 1978.
17. Madsen, H. O. On stochastic load combination. Report
No 30, Solid Mech. Div. Univ. of Waterloo, Ontario,
Canada, 1978.
18. Karkauskas, R. Analysis of geometrically nonlinear
elastoplastic structures. Civil Engineering (Statyba), 4 (1).
Vilnius: Technika, 1998, p. 3642.
19. Atkoinas, J. Design of elastoplastic systems under re-
peated loading (
). Vilnius: Mokslo ir enciklo-
pedij leidykla, 1994. 146 p. (in Russian).
20. EC3. Eurocode No 3. Design of Steel Structures, Vol 1,
Nov 1990, Vol 2, July 1990.
227
227
2004, Vol X, No 3, 227233
ISSN 13923730
DILATANCY O. CEMENT SLURRIES WITH CHEMICAL ADMIXTURES
Gintautas Skripkinas
1
, Mindaugas Daukys
2
Dept of Building Materials, Kaunas University of Technology, Student g. 48, LT-51367 Kaunas, Lithuania.
1
E-mail: Gintautas.Skripkiunas@ktu.lt;
2
E-mail: Mindaugas.Dauksys@ktu.lt
Received 21 March 2004; accepted 7 June 2004
Abstract. The results of experimental investigation of plasticising, air-entraining and viscosity-modifying admixtures
with different chemical composition influence on rheological properties and dilatancy of cement slurries are presented in
the article. Rheological properties of cement slurries were tested by a rotation viscometer with coaxial cylinder, at
different gradients of velocity. Dilatancy of cement slurries was determined by the distortion of the rheological curve
and index of dilatancy D was calculated according to proposed methods. The results of investigation show that the
properties of liquid disperse medium of structural system influence slightly the dilatancy of cement slurries. Plasticising
admixtures with a different chemical composition slightly increases but air-entraining and viscosity-modifying admix-
tures slightly reduces the dilatancy of cement slurries.
Keywords: cement slurry, chemical admixtures, rheology, yield stress, viscosity, dilatancy, viscometer.
1. Introduction
Cement slurries are among the so-called structural
systems that have certain rheological properties. Accord-
ing to the shape of the rheological curve structural sys-
tems are divided into some groups: Newtonian fluid,
pseudoplastic fluid, dilatantic fluid, Bingham fluid, plas-
tically dilatantic body and pseudoplastic body [1, 2].
Graphical expression of shear rate dependency upon shear
stress is called flow curve (.ig 1).
.ig 1. .low curves of different structural systems: 1
Newtonian fluid; 2 pseudoplastic fluid; 3 dilatantic
fluid; 4 Bingham fluid; 5 plastically dilatantic body;
6 pseudoplastic body
Cement slurries and concrete mixtures are usually
defined as Bingham fluids (systems); they have two main
rheological characteristics yield stress
0
and plastic
viscosity [35].
The dilatant behaviour increases the viscosity of
cement slurries with the increasing shear stress. The di-
latancy has negative influence on cement slurries and
concrete mixtures pumping by concrete pumps. Various
chemical admixtures were tested to change rheological
properties and to reduce dilatancy of cement slurries for
improving concrete pumping.
2. Literature review
Chemical admixtures are often used for modifica-
tion of rheologic properties of fresh concrete. Plasticiz-
ing and superplasticising admixtures with different chemi-
cal composition are used for changing rheologic proper-
ties of concrete mixtures [6, 7]. Chemical composition
of admixture influences the plasticising effect of admix-
tures [8, 9]. Rheologic properties of concrete mixtures
are influenced by air-entraining admixtures that are used
for increasing freezing-thawing resistance of hardened
concrete. The use of viscosity-enhancing admixtures in-
creases the homogeneity of cement slurries and concrete
mixtures and leads to a greater uniformity of hardened
properties. These admixtures greatly influence the thix-
otropy of concrete mixtures [10, 11]. Viscosity-enhanc-
ing admixtures are used for increasing flowability of
concrete using concrete pumps.
.or testing the rheologic properties of cement slur-
ries and concrete mixtures different viscometers are ap-
plied [1214]. Comparing results obtained by viscom-
eters with coaxial cylinders of different dimensions, the
228
228
influence of geometrical dimensions on the results of
rheologic properties testing was observed [15]. .resh
concrete rheologic properties tested with different type
viscometers show some differences in viscosity testing
results [16]. Therefore some investigators propose to use
the relative plastic viscosity instead of plastic viscosity
for comparing the results obtained with different viscom-
eters [17]. The proposed relative plastic viscosity of fresh
concrete is the ratio of viscosity of concrete mixture to
the viscosity of mortar or matrix. It eliminates the influ-
ence of geometrical parameters of equipment on test re-
sults.
New testing methods show the influence of mixing
on the rheologic properties of cement slurries [18]. The
hysteresis loop was observed from shear-induced struc-
tural breakdown in cement paste. Plastic viscosity in this
investigation was determined using the slope of the hys-
teresis down curve on the range of shear rates in some
limits. It was determined in this investigation that mix-
tures obtained in high shear mixer are more flowable.
Cement slurries and concrete mixture testing results
demonstrate that Bingham model is not correct for de-
scribing the cement slurries flow. .low curves of cement
slurries are not straight lines like in the Bingham model
but with the increasing of shear stress they became the
distort lines [1921]. The distortion of flow curve indi-
cates the dilatancy of cement slurries, ie the viscosity of
system is increasing with the shear stress increase. There
are no characteristics for quantitative estimation of dila-
tancy of mixtures.
The first investigation into dilatancy phenomenon
was carried out by O. Reynolds. According to the au-
thor, the increase of system viscosity can be explained
by the increase of disperse system volume after the
change of solid particles volume displacement caused by
the movement of particles to each other. The disperse
system volume increase causes the relative decrease of
disperse medium volume and the increase of system vis-
cosity [22]. In .ig 2 spheric solid particles have the
pyramidic volume distribution (left) and the cubic one
(right). Solid pyramidic particles distribution take up a
minimum volume. After affect of shear force the volume
distribution of solid particles can be changed to cubic
distribution that will increase the volume occupied by
solid particles by 1,41 time.

.ig 2. Volume change of spheric particles system caused
by shear force (pyramidic distribution on the left; cubic
distribution on the right)
.. S. Kaplan, J. E. Pivinskij and A. N. Saprykin
investigate rheological properties and dilatancy of quartz
glass suspensions [23]. They have noted that dilatancy is
observed in high-concentration disperse systems and pro-
posed a rheological equation for dilatant fluids:
n

, (1)
where shear stress; shear rate; plastic
viscosity; n degree of dilatancy.
.or Newtonian fluid n = 1, for dilatant fluid n > 1,
for pseudoplastic fluid n < 1. This investigation shows a
great influence of liquid phase pH on dilatancy of sus-
pension [23]. The degree of dilatancy depends on solid
particles concentration, shape and dimensions of particles,
viscosity and composition of disperse medium [23].
In the investigation the superplasticisers effects on
viscosity and yield stress of cement mixtures K. Asaga
and D. M. Roy found that dilatant behaviour is observed
in all cases with sulphonated naphthalene and melamine
formaline condensates admixtures concentrations higher
than 0,75 % [6]. Investigation of superplasticised cement
pastes containing mineral additives carried out by ..
Curcio and B. A. DeAngelis show that dilatant behaviour
is strongly dependent on the water/binder ratio, on the
level of cement replacement by mineral additives and on
the fineness of the latter. Dilatancy is caused by angular
and platelike particles [24].
Ch. Hu and .. de Larrard determined that dilatancy
is related to the tested material and testing method. Any
relationship does not exist between the dilatancy and the
Bingham characteristics yield stress and plastic vis-
cosity [14]. V. Ukraincik defined that the phenomena of
shear thickening and thinning obtained on flow curves
can partly be explained as being due to dilatancy and to
related changes in the fluid phase to pore air ratio in
fresh concrete. Shear-box tests with simple shear have
shown the proportionality of the dilatancy to the square
of shear stress [25].
3. Experimental procedure
The experimental investigation was carried out to
determine the dilatancy of cement slurries and influence
of chemical admixtures on it. Plasticising, air-entraining
and viscosity-modifying admixtures were tested. Yield
stress and viscosity of cement slurries were determined
and dilatancy was estimated by the distortion of flow
curve.
Cement slurries were mixed by hands for 5 min.
All slurries were prepared with the same W/C ratio
(0,55). Plasticising admixtures were mixed with water,
used for slurries preparing. Air-entraining admixtures
were added separately after slurries mixing. Total amount
of admixtures was in the range from 0,2 to 1,0 % of
cement mass.
In this investigation Portland cement CEM I 42,5
produced by the AB Akmens cementas was used.
G. Skripkinas, M. Daukys / JOURNAL O. CIVIL ENGINEERING AND MANAGEMENT 2004, Vol X, No 3, 227233
229
229
Water requirement for normal consistency of cement
slurry was 27,5 %, the specific surface of cement
particles 353 m
2
/kg, the cement particles density
3110 kg/m
3
.
Plasticising admixtures used: plasticiser SMR based
on lignosulphonates, solution density 1,14 kg/l;
superplasticiser Glenium 127 based on polycarboxyl poly-
mers, the solution density 1,10 kg/l. Air-entraining ad-
mixture Amex SB 22 (LP) made of liquid surface ten-
sion increasing materials, was used, the solution density
1,01 kg/l. Starvis 2006 L was used as viscosity-modify-
ing admixture (synthetic polymers with a big molecular
mass, the solution density 1,01 kg/l).
Rheologic properties of cement slurries were tested
with rotational viscometer BCH-3 (.ig 3).
.ig 3. Scheme of rotational viscometer with coaxial cyl-
inders: 1 cylinder connected with measuring scale; 2
internal (rotating) cylinder; 3 external (fixed) cylinder;
4 cement slurry
The cement slurry was poured into the cylindrical
vessel (3) (external cylinder) fixed to the viscometer.
Internal cylinder (2) was rotated in the external cylinder
with the slurry. The friction between the rotating inter-
nal cylinder and the cylinder, which is inside it (1) and
connected to the measuring scale turns internal cylinder
(1) connected to the measuring scale due to cement slurry
(4) inside friction.
Viscosity of cement slurry was estimated using dif-
ferent internal cylinder rotation rates (600, 400, 300, 200
1/min). Shear force . (mN) which turns cylinder con-
nected to the measuring scale, was calculated by the
measuring scale readings using the calibration line. Such
lines were evaluated experimentally and depend on the
elasticity of the springs in the device. A more elastic
spring is suitable for measuring viscosity of cement slur-
ries with low W/C ratio, when there are bigger shear
stresses.
Shear rate
for the different rotation rate of the

cylinder (2) was calculated by the equation (2):
, ,
2
1
s
d
n r
s
i
i

(2)
where r radius of the cylinder connected to the mea-
suring scale, m;
i
n rotation rate of the internal cylin-
der ( 4 , 1 i ), 1/s;
s
d distance between (1) and (2)
cylinders, m.
Shear stress of cement slurry for different shear
rates was calculated by equation (3):
Pa,
h r
F
i
i
,
2
(3)
where
i
F shear force calculated by the readings in
measuring scale using calibration line, mN; r and h
radius and height of the cylinder connected to the mea-
suring scale, m.
Viscosity of cement slurries for the different shear
rates was calculated by the (4) equation:
s, Pa
i
i
i

,
0
(4)
where
0
yield stress, determined by curve
point were the curve cross the axis.
4. Experimental results
.low curves of cement slurries (W/C ratio 0,55)
without and with chemical admixtures dosage 0,2 % of
the cement mass are presented in .ig 4.
.ig 4 curves in show that the superplasticizer based
on polycarboxyl polymers the biggest reduction effect
on yield stress of cement slurries (about 65 % in com-
parison with cement slurry without admixture) produces
when used in the dosage 0,2 % of cement mass.
.ig 4. .low curves of cement slurries with chemical ad-
mixtures dosage 0,2 % of the cement mass: 1 without
admixture; 2 with plasticiser (lignosulphonates); 3 with
superplasticiser (polycarboxyl polymers); 4 with air-en-
training admixture; 5 with viscosity-modifying admix-
ture (synthetic polymers with big molecular mass)
Less efficient in yield stress reduction is a plasticiser
on the base of lignosulphonates (about 30 % in com-
parison with cement slurry without admixture). Air-en-
training admixture opposite to plasticising admixtures
increases the yield stress of cement slurry. Viscosity-
modifying admixture increases the yield stress of cement
230
230
Table 1. Yield stress and viscosity of cement slurries
slurry more than air-entraining admixture. Yield stress
increasing makes mixtures more stable, ie reduces sedi-
mentation of aggregates and water bleeding in mixtures.
Superplasticiser and plasticiser insignificantly reduce vis-
cosity of cement slurry (Table 1). Air-entraining admix-
ture increases the viscosity of cement slurry (Table 1).
mixtures dosage of 0,4 % of cement mass: 1 without
ture (synthetic polymers with a big molecular mass)
The dependence of shear rate on the shear stress
for the cement slurries without and with chemical ad-
mixtures dosage of 0,4 % of the cement mass are pre-
sented in .ig 5. Superplasticiser and plasticiser in the
dosage of 0,4 % of the cement mass reduces yield stress
of cement slurry more efficient than in a less dosage
(about 45 % and 75 % correspondingly) (.ig 5) and re-
duces viscosity of cement slurry more than in the less
dosage (Table 1). Dosage of air-entraining admixture of
0,4 % of the cement mass has no effect on the yield stress
and viscosity of cement slurry compared with cement
slurry with no admixture. An increase of viscosity-modi-
fying admixture dosage to of 0,4 % of cement mass more
increases the yield stress of cement slurry than less dos-
ages of this admixture (.ig 5).
.low curves of cement slurries with chemical ad-
mixtures dosages of 0,6 % and 1,0 % of mass cement
are presented in the .igs 6 and 7.
mixtures dosage of 0,6 % of cement mass: 1 without
.igs 6 and 7 show that with an increase of plasticiser
and superplasticiser dosage yield stress of cement slurry
decreases. Effect of big dosages (0,6 and 1,0 % of the
cement mass) of air-entraining and viscosity-modifying
admixtures on yield stress and viscosity of cement slur-
ries is similar to the effect of 0,4 % dosage of these
admixtures.
Plasticisers and superplasticisers change the dilatancy
of cement slurry. This can be seen b increasing the dis-
tortion of the flow curves (.igs 47). Viscosity-modify-
Shear rate , 1/s
Composition of cement slurry
Dosage of
admixture, %
Viscosity

205 315 420 630

0
,

Pa
W/C = 0,55 without admixture
i
, Pas 0,0245 0,0256 0,0296 0,0363 12,00
0,20 % 0,0180 0,0242 0,0285 0,0300 8,33
0,40 % 0,0155 0,0225 0,0257 0,0282 6,67
0,60 % 0,0147 0,0213 0,0253 0,0276 5,74
W/C = 0,55
with plasticiser
1,00 %

i
, Pas
0,0128 0,0201 0,0229 0,0249 4,81
0,20 % 0,0129 0,0208 0,0250 0,0298 3,93
0,40 % 0,0109 0,0174 0,0235 0,0274 3,04
0,60 % 0,0091 0,0170 0,0210 0,0251 2,32
W/C = 0,55
with superplasticser

1,00 %

i
, Pas
0,0070 0,0143 0,0180 0,0220 1,43
0,20 % 0,0289 0,0243 0,0281 0,0360 14,14
0,40 % 0,0240 0,0267 0,0309 0,0365 11,44
0,60 % 0,0219 0,0232 0,0262 0,0330 9,70
W/C = 0,55
with air-entraining admixture

1,00 %

i
, Pas
0,0252 0,0233 0,0247 0,0331 12,08
0,20 % 0,0232 0,0255 0,0290 0,0314 16,40
0,40 % 0,0246 0,0264 0,0291 0,0326 17,85
0,60 % 0,0224 0,0257 0,0265 0,0301 17,65
W/C = 0,55
with viscosity-modifying admixture

1,00 %

i
, Pas
0,0218 0,0246 0,0262 0,0296 16,25

231
231
ing admixture has the opposite effect on the cement slurry
dilatancy. The flow curves of cement slurry with this
admixture are similar to the straight lines.
The influence of dosage of different chemical ad-
mixtures on yield stress of cement slurry with W/C ratio
0,55 is presented in .ig 8.
.ig 8. Influence of dosage of chemical admixtures on yield
stress of cement slurry: 1 with plasticiser (lignosulpho-
nates); 2 with superplasticiser (polycarboxyl polymers);
3 with air-entraining admixture; 4 with viscosity-modi-
fying admixture (synthetic polymers with a big molecular
mass)
The results plotted in .ig 8 show that a super-
plasticiser on the base of polycarboxyl polymers more
effectively by reduces the yield stress of cement slurry
with increasing the admixture dosage. Big dosages (1,0 %
and more) of this admixture reduce yield stress of ce-
ment slurry almost to zero. The effect of a plasticiser on
the base of lignosulphonates dosage is the same like a
superplasticiser but with a less reduction of yield stress
of cement slurry. Small dosages (up to 0,2 % of cement
mass) of these admixtures more effective by reduce yield
stress of cement slurry than the largest dosages. An air-
entraining admixture has no significant effect on yield
stress of cement slurry, but small dosages of this admix-
ture (0,2 %) show little increase in cement slurry yield
stress. Viscosity-modifying admixture on the base of syn-
thetic polymers with a big molecular mass increases the
yield stress of cement slurry in dosages up to 0,4 % and
keep it stable in dosages above 0,4 % of the cement
mass.
The influence of dosage of different admixtures on
viscosity of cement slurry tested at 205 1/s shear rate is
presented in .ig 9.
.ig 9. Influence of dosage of chemical admixtures on vis-
cosity of cement slurry tested at 205 1/s shear rate: 1
with plasticiser (lignosulphonates); 2 with superplasticiser
(polycarboxyl polymers); 3 with air-entraining admix-
ture; 4 with viscosity-modifying admixture (synthetic
polymers with a big molecular mass)
A superplasticizer on the base of polycarboxyl poly-
mers and plasticizer on the base of lignosulphonates re-
duces the viscosity of cement slurry in the small range
with increasing the admixture dosage. Superplasticizers
are more efficient in reduction of viscosity of cement
slurry than plasticizers. An air-entraining admixture neg-
ligibly increases the cement slurry viscosity in the range
of dosage up to 0,2 % from the cement mass and has no
effect on viscosity with later increasing the dosage. Vis-
cosity-modifying admixture has no significant effect on
viscosity of cement slurry in the range of all used dos-
ages.
.or the evaluation of dilatancy of cement slurries
the index D was proposed. Index of dilatancy D was
calculated by the equation (5):
, ,
Pa
s Pa
D

(5)
where the change of the viscosity (Pas) of ce-
ment slurry at shear rates 630 1/s and 205 1/s:
s, Pa ,
205 630
(6)
is the change of the shear stress (Pa) of cement slurry
at shear rates 630 1/s and 205 1/s:
Pa ,
205 630
. (7)
mixtures of 1,0 % of cement mass: 1 without admix-
ture; 2 with plasticiser (lignosulphonates); 3 with a
232
232
The dilatancy index D for the cement slurries with
different quantity of the admixtures is presented in
Table 2.
Table 2. Index of dilatancy D of cement slurries with different
admixture
Composition of cement
slurry
Dosage of
admixture,
%
D10
-3
,
Pas/Pa
W/C = 0,55
without admixture
0,660
0,20 % 0,787
0,40 % 0,870
0,60 % 0,898
W/C = 0,55
with plasticiser
1,00 % 0,926
0,20 % 1,049
0,40 % 1,098
0,60 % 1,149
W/C = 0,55
with superplasticiser

1,00 % 1,209
0,20 % 0,424
0,40 % 0,692
0,60 % 0,680
W/C = 0,55
with air-entraining
admixture
1,00 % 0,504
0,20 % 0,546
0,40 % 0,517
0,60 % 0,536
W/C = 0,55
with viscosity-modifying
admixture
1,00 % 0,551
.ig 10. Index of dilatancy D of cement slurries with 1,0 %
of admixtures: 1 without admixture; 2 with plasticiser
(lignosulphonates); 3 with superplasticiser (polycarboxyl
polymers); 4 with air-entraining admixture; 5 with
viscosity-modifying admixture (synthetic polymers with a
big molecular mass)
.ig 10 presents the index of dilatancy of the ce-
ment slurries with 1,0 % of admixtures. It shows that the
dilatancy of cement slurry is higher with different chemi-
cal composition plasticising admixtures, compared with
the cement slurry without an admixture. The dilatancy
of the cement slurry is much higher with a superplasticiser
based on polycarboxyl polymers. The air-entraining and
viscosity-modifying admixtures are slightly decreasing the
cement slurries dilatancy in comparison to the cement
slurry with no admixture.
Chemical admixtures (except for air-entraining and
viscosity-modifying admixtures) which are used in prac-
tice at the moment, are changing rheologic properties of
concrete mixture but they are not improving mixtures
flowing in the conditions of increasing shear stress be-
cause of its higher dilatancy and so they are not useful
for concrete mixture transportation with concrete pumps.
5. Conclusions
1. Cement slurries and concrete mixtures are dila-
tant fluids. The dilatancy of concrete mixtures must be
taken in to account in case of usage concrete pumps.
2. Plasticising admixtures with different chemical
composition decrease the yield stress and viscosity of
cement slurry but they increase the dilatancy of cement
slurry.
3. Superplasticisers based on polycarboxyl poly-
mers effective by decrease the yield stress, but signifi-
cantly increase the cement slurry dilatancy.
4. Air-entraining admixtures result in little reduc-
tion of dilatancy of cement slurry and are useful for
pumpable concrete.
5. A viscosity-modifying admixture increases the
yield stress and stability of cement slurry but has no sig-
nificant dilatancy reduction effect.
6. The dilatancy of cement slurry and concrete
mixture can be estimated by index of dilatancy D, ie
ratio of change of viscosity to change of shear stress at
different shear rates.
References
1. Kruglickij, N. N. .undamentals of physical-chemical me-
chanics. Part 1 ( - .
I). Kiev: Vishaja shkola, 1975. 207 p. (in Russian).
2. Ovchinikov, P. ..; Kruglickij, N. N.; Michailov, N. V.
Rheology of thixotropy systems (
). Kiev: Naukova dumka, 1972. 120 p. (in Rus-
sian).
3. Tang, C. W.; Yen, T.; Chang, C. S.; Chen, K.-H. Optimiz-
ing mixture proportions for flowable high-performance
concrete via rheology tests. ACI Materials Journal, Vol 98,
No 6, 2001, p. 493502.
4. Pivinski, J. E. Rheology in technology of ceramics and
refractories. The basic rules and models of rheology. Re-
fractories (), No 3, 1994, p. 715 (in Russian).
5. Banfill, P. .. G. The rheology of fresh cement and con-
crete A Review. In: Proc. of 11th International Cement
Chemistry Congress, Durban, May 2003, p. 136149.
6. Asaga, K.; Roy, D. M. Rheological properties of cement
mixes: effects of superplasticisers on viscosity and yield
stress. Cement and Concrete Research, Vol 10, No 2, 1980,
p. 287295.
7. .erraris, Ch. ..; Gaidis, J. M. Connection between the
rheology of concrete and rheology of cement paste. ACI
Materials Journal, Vol 88, No 4, 1992, p. 388393.
233
233
8. Aitcin, P. C.; Jiang, S.; Byung-Gi, K.; Nikinamubanzi, P. C.
Cement/superplasticizer interaction. The case of poly-
sulfonates. Bulletin des Laboratories des Ponts et
Chaussees-233, Ref 4373, 2001, p. 8999.
9. Sakai, E.; Yamada, K.; Ohta, A. Molecular structure and
dispersion-adsorption mechanisms of comb-type super-
plasticizers used in Japan. Journal of Advanced Concrete
Technology, Vol 1, No 1, 2003, p. 1625.
10. Khayat, K. H.; Saric-Coric, M.; Liotta, .. Influence of
thixotropy on stability characteristics of cement grout and
concrete. ACI Materials Journal, Vol 99, No 3, 2002,
p. 234241.
11. Khayat, K. H.; Guizani, Z. Use of viscosity-modifying
admixture to enhance stability of fluid concrete. ACI Ma-
terials Journal, Vol 94, No 4, 1997, p. 332340.
12. Uriev, N. B.; Michailov, N. V. Rheological properties of
cement slurries under vibration conditions. Reports of
Academy of Science of SSSR. Technical Physics
( ), Vol 153, No 4, 1963, p. 828831
(in Russian).
13. Saak, A. W.; Jennings, H. M.; Shah, S. P. The influence
of wall slip on yield stress and viscoelastic measurements
of cement paste. Cement and Concrete Research, Vol 31,
No 2, 2001, p. 205212.
14. Hu Ch.; De Larrard, .. The rheology of fresh high-perfor-
mance concrete. Cement and Concrete Research, Vol 26,
No 2, 1996, p. 283294.
15. Lapasin, R.; Papo, A.; Rajgelj, S. .low behavior of fresh
cement pastes. A comparison of different rheological in-
struments and techniques. Cement and Concrete Research,
Vol 13, No 3, 1983, p. 349356.
16. .erraris, C. ..; Brower, L. E., editors. Comparison of con-
crete rheometers: International tests at LCPC. In: NISTIR
6819, Nantes, .rance, October 2000. 147 p.
17. .erraris, C. ..; Martys, N. S. Relating fresh concrete vis-
cosity measurements from different rheometers. Journal of
Research of the National Institute of Standards and Tech-
nology, Vol 108, No 3, 2003, p. 229234.
18. Williams, D. A.; Saak, A. W.; Jennings, H. M. The influ-
ence of mixing on the rheology of fresh cement paste.
Cement and Concrete Research, Vol 29, 1999, p. 1491
1496.
19. Skripkinas, G.; Daukys, M. Investigation of rheological
properties and dilatancy of cement slurries. In: Concrete
and Reinforced Concrete. Proceedings of conference
(Betonas ir gelbetonis. Konferencijos praneim media-
ga). Kaunas: Technologija, 2003, p. 2835 (in Lithuanian).
20. Daukys, M. Influence of cement tipe on rheological prop-
erties of cement slurries. In: Book of Abstracts of the 5th
International Summer School - Conference. Advanced
materials and technologies, Palanga, Lithuania, 25-29 Au-
gust 2003, p. 31.
21. Skripkinas, G. Optimization of concrete macrostructure
according to its technological and servise properties and
raw materials resources (Betono makrostruktros optimi-
zavimas, atsivelgiant jo technologines ir eksploatacines
savybes bei aliav iteklius). PhD thesis. Kaunas Univer-
sity of Technology, 1993. 185 p. (in Lithuanian).
22. Reiner, M. Deformation, strain and flow. An elementary
introduction into rheology ( .
). Moscow: Gostoptechizdat, 1963.
381 p. (in Russian).
23. Kaplan, .. S.; Pivinskij, J. E.; Saprykin, A. N. About fea-
tures dilatancy of consolidation dispersion of a quartz glass.
Journal of Colloidals ( ), Vol L,
N 6, 1988, p. 10921098 (in Russian).
24. Curcio, ..; DeAngelis, B. A. Dilatant behavior of
superplasticised cement pastes containing metakaolin.
Cement and Concrete Research, Vol 28, No 5, 1998,
p. 629634.
25. Ukraincik, V. Study on fresh concrete flow curves.
Cement and Concrete Research, Vol 10, No 2, 1980,
p. 203212.
235
235
2004, Vol X, No 3, 235239
ISSN 13923730
CHANGES IN SUBJECTIVE ACOUSTICAL INDICATORS IN HALLS WITH LONG
AND SHORT REVERBERATION TIMES
Vytautas J. Stauskis
Dept of Building Structures, Vilnius Gediminas Technical University,
Trak g. 1/26, LT-01132 Vilnius, Lithuania. E-mail: stauskis@ar.vtu.lt
Received 5 Apr 2004; accepted 31 May 2004
Abstract. Investigations of acoustics of halls of different types and volumes have disclosed good or acceptable values
of the music sound clarity index C
80
, while the reverberation times of these halls were found to be either very short or
very long. Subjectively, the acoustics of these halls are considered to be unsatisfactory. .or example, the reverberation
time measured in the hall of the Vilnius Opera and Ballet Theatre is about 1 s, while the values of the C
80
index are
either good or acceptable. A church the volume of which is 22650 m
3
has acceptable C
80
values both in the performerss
places and in the front rows. Similar results were obtained in other halls too.
Keywords: Clarity index C
80
, reverberation time, absorption, early energy, late energy, sound reflections.
1. Introduction
In order to evaluate the acoustics of existing halls
or halls being designed, objective acoustic indicators and
subjective energy indicators are to be measured. The
voice clarity index C
50
and the music sound clarity in-
dex C
80
are widely used. It is possible to calculate these
indices in the designing stage, though, much more exact
results are offered by measured indices when a pulsed
sound source is used. The music sound clarity index is a
ratio between the early energy and the late energy The
values of the index are different at varied points of the
hall [18]. Its values are determined by the early sound
reflections reaching the listener within the first 80 ms.
The structure of these reflections strongly depends on
the positions of the sound source and the listener. If the
sound source is located at the front of the stage and the
hall itself is high and wide at this point, the first reflec-
tion will reach listeners relatively late, whereas listeners
in the back rows of the hall will receive the first reflec-
tions much earlier [912]. .urthermore, the quantity of
reflections received there within 80 ms will be much
larger than that received in the first rows of the hall.
The intensity of reflections which depends on the type
of materials used for ceiling and wall planes is also im-
portant. After multiple reflection from all surfaces of the
hall, energy will be lost and the energy losses will be
directly proportional to the sound energy absorption by
these surfaces. These will be late reflections reaching the
listener after 80 ms. The hall reverberation time, an im-
portant acoustic indicator, will also depend on these re-
flections.
Listeners are sensitive to the changes both in the
intensity of early reflections and in the time during which
the reflections arrive. Due to these reflections the lis-
tener may evaluate the dimensions and proportions of
the hall and the work of music itself. Different structure
and intensity of reflections determine different sounding
of music at the front, middle and back rows of the hall.
In an attempt to evaluate the different reflection
structures, the impact upon the listener, Reichardt et al
[14] offered optimum values of the music sound clarity
index C
80
at the front and back rows of a hall. The au-
thors of the study considered that, with the recommended
values of C
80
, the hall acoustics will receive good sub-
jective evaluation. The recommended values were ac-
cepted all over the world and for many years already
have been used for evaluating the acoustic properties of
halls. [13]. The question arises, however, as to whether
optimum values of C
80
are tantamount to good acoustics
of a hall. It is understandable that the index alone is not
sufficient for the evaluation. However, a good correla-
tion between the C
80
index and certain objective acous-
tic indicators has been found.
The purpose of this work is to determine the values
of the music clarity index C
80
at various points of actual
halls with very different acoustic characteristics, to com-
pare them with the recommended values, and to link them
with the acoustic quality of the halls as evaluated by
music specialists and the audience.
236
236
2. Subject of investigation
Investigations into acoustics of halls of different
types and volumes have shown that the values of the
music sound clarity index are good or acceptable, while
those of reverberation time, time centre of gravity, etc
are completely unacceptable. Subjectively, acoustics of
these halls are considered to be unsatisfactory.
In all cases the C
80
values correspond to an unfil-
tered signal. At the same time, reverberation times, cen-
tre times of gravity and other indicators were calculated.
The clarity index C
80
was calculated from experi-
mental data by the formula [1, 2].
dt t p
dt t p
C
) (
) (
lg 10
08
2
80
0
2
80
.
In order to establish the relationship between the change
in the index C
80
in halls of various types and at various
points of the halls, six halls were selected. Their geo-
metrical characteristics are presented in Table.
No Name of hall L
max
B
max
H
max
V, m
3

1 St. Johns
Church
58 25,5 20,5 2 7000
2 Cathedral 56 24 19,8 22 650
3 Cinema 25 15 9,5 3 306
A calibre No 9 sound pistol was used as a sound
source. It was located in the places usually occupied by
performers. The microphone was placed in different rows
along the hall. The sound signal was transmitted from
the microphone to a 16-bit analog-to-digital converter
having a wide dynamic range and stored into a computer
memory. Afterwards the signals were analysed by means
of a special computer program.
The reverberation times, time centres of gravity and
other indicators of these halls were measured according
to the requirements of the standard [1416].
The investigations were conducted in unoccupied
halls. When a hall is filled with listeners the overall hall
absorption will increase and the reverberation time will
be reduced. However, this reduction also depends on the
chairs in the hall: the softer the chairs, the lesser the
reduction attributed to the listeners. If a hall contains
wooden benches, the reduction of the reverberation time
will be much more marked. This will also produce ef-
fect on both early and late energy reaching the listeners,
which, in its turn, will affect the clarity index C
80
. In
order to establish the extent of change in the index with
the presence of listeners and to determine whether such
change may be examined in an unoccupied hall, investi-
gations were conducted in the physical model of the hall
(scale 1:25) and in two real halls.
3. Dependence of the clarity index C
80
on changes in
the halls absorption
.or this purpose investigations were conducted in a
physical model of the hall of scale 1:25. Placing sound-
absorbing materials on various planes in the hall varied
the overall absorption of the hall. The results of the in-
vestigation are shown in .ig 1.

3
2
4
1
0
20
40
60
80
100
120
140
160
125 250 500 1000 2000 4000
.requency, Hz
A
b
s
o
r
p
t
i
o
n
,

m

.ig 1. Dependence of sound absorption of the hall on
location of sound-absorbing materials: 1 hall without
special absorbing materials; 2 absorbing materials only
on the ceiling; 3 only on the floor; 4 on the ceiling
and the floor
When the absorbing materials are placed only on
the ceiling, the maximum hall absorption is reached at
500 Hz, and at high frequencies it is almost equal to the
absorption of an empty hall. When such materials are
placed only on the floor, the character of absorption
changes entirely. Allocation of the materials both on the
floor and the ceiling results in the maximum absorption
only at 1000 Hz, with almost no increase at low and
high frequencies.
In practice, ceilings are only partially covered with
absorbing materials. The audience fills the floor plane
either by 100% or by 50%. Their sitting positions may
vary. The results of this investigation show that absorp-
tion changes, consequently, C
80
index is subject to change
as well. The results of the investigation for unfiltered
signal are presented in .ig 2.
The graph clearly shows that the values of C
80
in-
crease with the increase in sound absorption. A 1,5-fold
increase in absorption results in a ~ 3 dB increase in
C
80
. This applies both to positive and negative values of
CC
80
.
These results allow to draw a conclusion that lis-
teners sitting on semi-upholstered or upholstered chairs
in the hall cause no significant changes in the overall
hall absorption. Therefore one may assert that the differ-
ence in the clarity index of the hall without listeners and
the hall filled with listeners up to 70-80% will make up
only 1-3 dB. The presence of listeners/viewers will al-
ways increase and never reduce the clarity index. This
V. J. Stauskis / JOURNAL O. CIVIL ENGINEERING AND MANAGEMENT 2004, Vol X, No 3, 235239
237
237
.ig 2. Dependence of Index C
80
on changes in the hall
absorption. .igures denote sound absorption and letters
denote location of absorbing materials: A1 absorbing
materials only on the floor; A2 only on the ceiling;
A3 on the ceiling and the floor; A4 no absorbing
materials
-8
-7
-6
-5
-4
-3
-2
-1
0
A1-59 A2-38 A3-68 A4-29
Absorption, m
"
C
l
a
r
i
t
y
"

i
n
d
e
x

C
8
0
,

d
B
means that changes in the clarity index may also be in-
vestigated in unoccupied halls.
4. The change in the values of clarity index C
80
in
churches
Let us take an extreme example. In churches of St
John and the Cathedral there is a very long reverberation
time. The results of investigations are presented in Fig 3.
In the Cathedral church the values of C
80
are good
near the sound source and acceptable at the distance of
10 m from it. In St John church this distance is shorter
and is equal to 5 m. When listeners are present, the C
80
values will undergo a considerable increase of about 3-4
dB. Therefore it may be expected that they will be good
and acceptable at a greater distance from the sound
source. In such a case, even in the churches characterised
by a long reverberation time, the index values are good
and acceptable quite far from the sound source.

5
2
3
1
4
-15
-12
-9
-6
-3
0
3
6
9
12
1 2,5 5 7,5 10 20 25
Distance from source, m
"
C
l
a
r
i
t
y
"

I
n
d
e
x

C
8
0
,

d
B

.ig 3. Changes in the clarity index in churches depending
on the distance to the sound source: 1 good values for
front and back rows (lower and upper limits); 2, 3 ac-
ceptable values for front and back rows (lower and upper
limits); 4 measured values church No 2; 5 mea-
sured values church No 1
.ig 4 depicts the frequency characteristics of rever-
beration time. The volumes of these churches are very
large and all surfaces are made of sound-reflecting ma-
terials. Both churches have benches made of hardwood.

1
2
0
2
4
6
8
10
125 250 500 1000 2000 4000
.requency, Hz
R
e
v
e
r
b
e
r
a
t
i
o
n

t
i
m
e
,

s

.ig 4. .requency dependence on reverberation time: 1
measured in an unoccupied St John; 2 recommended
The reverberation times in both churches are very
long, particularly at low and medium frequencies. With
600 listeners present, the reverberation time is reduced
by about 1-1,8 s but still remains very long and unac-
ceptable from the point of view of good music sound.
Fig 5 shows the results of investigation into the time
centre of gravity.

1
3
2
0
100
200
300
400
500
600
1 2,5 5 7,5 10 20 25
C
e
n
t
r
e

t
i
m
e

o
f

g
r
a
v
i
t
y

T
S
,

m
s

.ig 5. The dependence of the time of centre of gravity on
the distance to the sound source: 1 St. John church; 2
Cathedral church; 3 maximal permitted values
.igure shows that the values of the time centre of
gravity are very large when the distance from the sound
source exceeds 2,5 m. Large values of this indicator
demonstrate the presence of echoes and long reverbera-
tion time of the hall.
5. The change in the values of clarity index C
80
in the
cinema hall
This particular hall was selected for investigations
because its acoustics are absolutely unsuitable for a natu-
ral sounding of music. Therefore it was interesting to
observe the changes in the values of the clarity index
C
80
before and after the reconstruction of the hall. The
238
238
Dolby Surround Digital sound recording system was in-
stalled in the hall.
.or the installation the Dolby Surround Digital sound
recording system, the reverberation time and the air vol-
ume resonances had to be strongly reduced. Therefore
all walls and the ceiling of the hall were covered with
sound-absorbing materials and the overall sound absorp-
tion was markedly increased after reconstruction. The
results of investigations are presented in .ig 6.

2
1
0
100
200
300
400
500
600
700
800
900
125 250 500 1000 2000 4000
.requency, Hz
A
b
s
o
r
p
t
i
o
n
,

m

.ig 6. .requency dependence of change in the overall
sound absorption in the cinema hall: 1 before reconstruc-
tion; 2 after reconstruction
The graph shows that after the reconstruction the
absorption has increased by about 230-250 m
2
through-
out the frequency range. The absolute values of absorp-
tion are large for a hall of such small volume.
Such marked changes in the sound absorption undoubt-
edly influence reverberation time, the results of investiga-
tion of which are shown in Fig 7.

1
2
0
0,2
0,4
0,6
0,8
1
1,2
1,4
125 250 500 1000 2000 4000
.requency, Hz
R
e
v
e
r
b
e
r
a
t
i
o
n

t
i
m
e
,

s

.ig 7. .requency dependence of reverberation time in a
cinema hall: 1 measured values before reconstruction;
2 after reconstruction
Before the reconstruction the reverberation time
values were 1,2-1,3 s at the low and medium frequen-
cies and only 0,8-0,9 s at high frequencies. Such rever-
beration time is too low for a good sounding of music.
After the reconstruction it was further reduced and is
0,8-0,5 s throughout the frequency range.
The values of the clarity index C
80
are shown in
.ig 8.

4
5
3
2
1
-6
-4
-2
0
2
4
6
8
10
12
11 14 17 21 18(b) 22(b)
'
'
C
l
a
r
i
t
y
'
'

i
n
d
e
x
,

C
8
0
,

d
B

.ig 8. The change in the clarity index in the cinema hall
depending on the distance to the sound source: 1 good
values for front and back rows (lower and upper limits);
2, 3 acceptable values for front and back rows (lower
and upper limits); 4 measured values before reconstruc-
tion; 5 after reconstruction
The index C
80
values were good in all rows of the
stalls and the balcony before reconstruction. After the
reconstruction the reverberation time became much
shorter but in spite of this the clarity index values are
acceptable at the front rows of the stalls. It is obvious
that no orchestra may sound well in such a hall. The
values of the clarity index C
80
, however, were good at
all points of the hall prior to reconstruction and remained
good in the front rows even after the reconstruction.
.ig 9 shows the results of investigation of the time
centre of gravity.

3
2
1
0
20
40
60
80
100
120
140
160
125 250 500 1000 2000 4000
.requency, Hz
C
e
n
t
r
e

t
i
m
e

o
f

g
r
a
v
i
t
y

T
S
,

m
s

.ig 9. Dependence of the time centre of gravity on the
distance to the sound source: 1 maximal permitted val-
ues; 2 measured values before reconstruction; 3 after
reconstruction
It can be seen from .ig 17 that the values of the
time centre of gravity are very small both before and
after the reconstruction. This demonstrates that the hall
is very suppressed and all reflection energy is concen-
trated in the initial stage of the energy decrease.
Investigation results are presented for the halls of
various type and volume, the acoustics of which have
been subjectively recognised by music specialists as un-
suitable for a natural sounding of music. This is con-
firmed by the investigations of objective indicators. In
239
239
the churches, acceptable values were obtained at the
performers places and in the front rows of the hall. Even
in the cinema hall, which is absolutely unsuitable for good
sounding of music, the C
80
values are good in all rows
of the stalls. This means that the clarity index C
80
may
not be applied as an indicator for proper characterisation
of the acoustics of a hall. Its recommended values should
be adjusted. .urthermore, much acoustic information
could be provided by the data on frequency dependence
of this indicator.
The music sound clarity index is a ratio between the
early energy and the late energy. It is determined by the
structure of sound reflections which is different at various
points of the same hall. .or the front rows, good values
are within the range of 3 to 8 dB. This means that the
early energy reaching the listener within the first 80 m
must exceed the remaining pulse energy by 3-8 dB. One
may hardly agree with this. In order for the pulse energy
from 80 ms to oo be lower by 3-8 dB than the early en-
ergy up to 80 ms, the final process of energy decrease
must be sufficiently quick. However, in such a case the
hall reverberation will be short and this is undesirable in
the halls used for music purposes. At the back rows, good
values of C
80
are within the range of 0 to 5 dB. The ratio
0 shows a balance between the early energy and the
late energy. Such balance is obtained in the Opera and
Ballet Theatre (.ig 2) and the Small Hall of the Philhar-
monic Society (.ig 11). Due to higher intensity of reflec-
tions and longer reverberation time both the early and the
late energy are larger in the concert hall than in the the-
atre, though the energy ratio is equal for both halls and
the quality of music sound is quite different.
.or instance, when an orchestra plays in the Opera
Theatre, the first intensive reflection reaches the front
rows of listeners after 23-25 ms and its amplitude is lower
by 7,3 10,7 dB than that of the direct sound. No inten-
sive reflections will be present in the interval between
the direct sound and the first reflection. This interval
will be filled by diffracted reflections with the intensity
lower by 20-30 dB than that of the direct sound. At the
back rows, the first reflections will reach the listener after
4 and 5 ms and their intensity will be lower by 9,1 and
12,8 dB respectively than that of the direct sound.
6. Conclusions
1) The presence of audience always increases the
clarity index C
80
, the increase amounting to 1-3 dB.
2) When the reverberation time is short and abso-
lutely unacceptable in different halls, the values of the
clarity index C
80
are good or acceptable.
3) The values of the clarity index C
80
are very dif-
ferent at different places of the same hall and they
strongly depend on the positions of the sound source and
the listener.
4) Recommended values of C
80
are also found in
the halls whose acoustics is assessed as bad or unac-
ceptable by music specialists.
References
1. Reichadt, W.; Abdel Alim O, Schmidt, W. Definition and
measurement basies of objective mass to determine limits
between usable and unusable transparance in music.
(Difinition und Messgrundlage eines objectiven Masses zur
Ermittlung der Grenze zwischen brauchbarer und
unbrauchbarer Durchsichtigkeit bei Musikdarbietungen).
Acustica, 1975, Vol 32, No 3, p. 128137 (in German).
2. Reichadt W. und Lehmann P. Room volume and
transparance optimising in music by impulsive noise test
(Optimierung von Raumeindruck und Durchsichtigkeit von
Musikdarbietungen durch Auswertung von Impulsschalltest.
Acustica, 1981, Vol 48, No 4, p. 174185 (in German).
3. Lehmann, P.; Wilkens, H. Relation between subjective es-
timate and room acoustics in concert halls. (Zusammenhang
der subjectiver Beurteilungen von Konzertsle mit
raumakustischen Kriterien). Acustica, 1980, Vol 45, No 4,
p. 256268 (in German).
4. Beranek, L. L. Audience and seat absorption in large halls.
The Journal Acoustic Society of America, Vol 45, No 5,
1969, p. 1319.
5. Bradley, J. S. Auditorium acoustic measures grom pistol
Shots. The Journal Acoustic Society of America, Vol 80 (1),
No 2, July 1986, p. 199205.
6. Bradley, J. S. Experience width new auditorium acoustic
Measurements. The Journal Acoustic Society of America,
73, 1983, No 4, p. 20512058.
7. Bradley, J. S. A comparison of three classical concert halls.
The Journal Acoustic Society of America, 69 (3), No 2,
1991, March, p. 11761192.
8. Bradley, J. S. Acoustical comparison of three theaters. The
Journal Acoustic Society of America, 79 (6), No 3, 1986,
June, p. 18271832.
9. Beranek, L. L. Audience and seat Absorption in large halls.
The Journal Acoustic Society of America, 45, No 4, 1969,
p. 661670.
10. Kuttruff, H. Objective measurements in room acoustics.
In: Reports of the 6th International Congress on acoustics,
Vol 1, GP. P. Tokyo, 1968.
11. Kuttruf, H. Room Acoustics. New-York: Springer, 1991.
380 p.
12. Kuttruf, H. Reverberation and effective absorption in rooms
with diffuze wall reflexions. Acustica, Vol 35, No 1, 1976,
p. 141153.
13. Voelker, J. S. Home cinema surround sound - acoustics
and neighbourhood. The 100th convention AES.
Copenhagen, 1996. 7 p.
14. Lewers, T. H., Anderson, J. S. Some acoustical properties
of St Pauls Catedral, London. Journal of Sound and Vi-
bration, 1984, 92 (2), p. 285297.
15. Thiele, R. About measuring of side noise in different halls
in rooms (Richtungsverteilung und Zeitfolge der schallruck-
wurfe in Raumen). Acustica, 1953, Vol 3, No 4, p. 291
302 (in German).
16. ICO 3382:1997(E). Acoustics. Measurement of the rever-
beration time of rooms with reference to other acoustical
parameters.
241
241
2004, Vol X, No 3, 241245
ISSN 13923730
COMPOSITE AND INTERACTION E..ECTS IN STEEL-CONCRETE STRUCTURES
.OR HIGHER .IRE RESISTANCE
Antanas apalas
Dept of Steel and Timber Structures, Vilnius Gediminas Technical University, Saultekio al. 11,
LT-10223 Vilnius-40, Lithuania. E-mail: Antanas.Shapalas@st.vtu.lt
Received 21 June 2004; accepted 30 Aug 2004
Abstract. Wide application of steel structures for office and industrial buildings is one of the main features of Lithuanian
and world building industry. This fact is related to an essentially wider use of new structural solutions, new structural
members and materials. Usage of different types of materials in one structural system for better utilisation of positive
properties of each component is very popular in current building industry. Composite beams, columns, beams-columns
and composite floors can be mentioned. Structures that were designed and erected twenty or more years ago are mostly
a collection of different types of structural elements. In general, interaction or composite effect is not taken into account,
when designing such structures. Though a considerable number of roofs and ceilings made up of reinforced concrete
slabs and steel beams or steel trusses are still in maintenance. Structural solution of the slabs, welding of embedded
steel parts of slabs to steel structures, filling with concrete the joints between slabs create the conditions for interaction
of slabs and steel structures. The result of this effect is that slabs take up a part of the internal forces of the steel
structures and consequently diminish it in steel beams and trusses. It is obvious that this fact leads to the increase of
load bearing capacity and fire resistance. The paper deals with the problem of evaluation of composite and structures
interaction effect on fire resistance.
Keywords: composite structures, steel beams, reinforced concrete slabs, interaction of structures, fire resistance.
1. Introduction
Wide application of steel structures for office and
industrial buildings is one of the main features of
Lithuanian and world building industry. Steel structures
offer exciting architectural possibilities, quick fabrication
and erection, possibility to use steel in combination with
other structural materials of high effectiveness. All these
advantages lead to cost effective construction.
Type of buildings, their geometrical parameters and
installed industrial technologies determine the level of
fire resistance requirements of load bearing structures [1,
2]. .ire resistance of structural elements depends on a
wide range of different parameters. These parameters can
be divided into two main groups.
One group affects the bearing capacity of structural
element under fire conditions and related critical tem-
perature [2, 3]. Main mechanical materials properties
belong to this group. Modification law for these proper-
ties under high-temperature conditions is defined in de-
sign codes [3, 4]. A lot of investigations are performed
in this field to derive properties-temperature relations.
So material behaviour in a high-temperature environment
is more or less known. This is true on material level, but
not true on structural element or global structure level.
On a higher-level of interaction between materials and
between it separate structural elements become very im-
portant and can have substantial influence on bearing
capacity under fire conditions too. In this article we do
not discuss the problem of global analysis under fire
conditions. We will try to investigate how interaction of
different materials and separate structural elements af-
fects the bearing capacity under fire conditions.
The other group of parameters determines the time
in which critical temperature will be reached. This time
is simply called fire resistance. Thermal properties of
materials included in structural element can be assigned
to this group. Values of these properties also depend on
temperature conditions. Such relations are included in
design code [3] and values of properties can be evalu-
ated in the necessary temperature situation. Because steel
is very homogeneous material the design code provides
simple solutions for determination of section tempera-
ture in dependence on section shape and heating situa-
tion. .or sections made of different materials and with
complex shape finite elements method can be used to
create section temperature fields for each moment of
simulated fire situation. These investigations will be per-
formed in future.
242
242
In most cases unprotected steel structures do not fit
for fire resistance requirements and effective fire protec-
tion materials are still expensive enough. One of the ways
to make fire protection cheaper is to choose special types
of composite structural elements, which, in some cases,
can be used without additional fire protection measures
[5, 6]. The other way the use of more precise evalua-
tion of behaviour of structural elements in case of fire.
Effectiveness of composite structures under fire con-
ditions and the method giving possibility to take into
account the interaction of load bearing structures and re-
inforced concrete slabs under normal and fire conditions,
is discussed in the paper.
A great deal of substantial changes in the construc-
tion sphere of various types and of different destination
buildings in Lithuania took place in recent years. Espe-
cially this fact is related to a wider use of new structural
solutions, new structural members and materials. Though
a considerable number of roofs and ceilings made up of
reinforced concrete slabs and steel beams or steel trusses
are still in maintenance.
Discussing the problem in question, first of all analy-
sis of the behaviour of roof and ceiling structures com-
posed by reinforced concrete slabs and steel beams or
steel trusses has to be made. And then, on the basis of
these results, fire resistance of steel structures may be
cleared up. The result of the above-mentioned analysis
will be a more precise determination of internal forces
in the steel structures and consequently a more reliable
evaluation of their fire resistance.
2. Composite structures and fire situation
It has already been mentioned that composite sec-
tion cannot only increase level of strength utilisation of
materials from which section is composed, but also pro-
vides a higher fire resistance of structural element [5,
6]. Proper decision in the case of composite structure
can allow to design high performance structural elements
for fire conditions. Some of such solutions can exclude
usage of expensive fire protection materials. Structural
fire design experience shows that in most cases for build-
ings with unprotected steel columns and beams fire re-
sistance can be about 15 min. The investigations [6] show
that it is possible to reach 60 min fire resistance limit by
using composite structural elements without fire protec-
tion.
In Eurocode as in almost all national codes for fire
design temperature increasing rate depends on section
factor:
V
A
tor SectionFac
m
, (1)
where
m
A is the exposed surface area of the member
per unit length; V is the volume of the member per unit
length. A higher value of fire resistance can be obtained
in case of high V value and low
m
A value. Utilisation
of bearing capacity (load level) is also important for fire
resistance calculations. Main goal for fire resistance in
case of using composite floor system is reduction of
exposed surface area
m
A .
Typical composite beams with shear connectors and
concrete floor on corrugated sheet is not a very advanced
system for fire situation. In such structural solu-tion only
top edge of upper flange of beam can be prote-cted from
fire. Unfilled voids between corrugated sheet waves and
upper flanges must be protected as well. Otherwise it
will reduce floor system fire resistance. Much better ef-
fect can be reached by using Slim .loor system (.ig 1,
2). In such systems, special modified I beams with addi-
tional bottom plate extending beam flange for support-
ing deep steel decking are used. ASB or I.B beams are
also successfully applied for Slim .loor system.
Another floor type ensuring the same reduced sec-
tion factor can be arranged from the mentioned types of
special beams and prefabricated reinforced hollow con-
crete plates. In such a system corrugated steel sheets are
not used (.ig 3).
Prefabricated plates are placed on bottom flange of
beams and top concrete layer with additional reinforce-
ment is used.
.ire resistance of a building depends on fire resis-
tance of all bearing structures. The second important
structural element in building skeleton is column. .illing
with con-crete space between flanges can increase fire
.ig 3. Cross-section of composite floor system with pre-
fabricated plates
.ig 1. Cross-section of Slim .loor beam with extended
bottom flange
.ig 2. Cross-section of Slim .loor beam with extended
bottom flange
A. apalas / JOURNAL O. CIVIL ENGINEERING AND MANAGEMENT 2004, Vol X, No 3, 241245
243
243
resistance of steel columns. Concrete must be joined to
column body by shear connectors. Reinforced concrete
may be used for achieving higher fire resistance (.ig 4).
In some cases the same method for increasing the fire
resistance of steel beams can be effective. .or such beams
longitudinal reinforcement bars must be joined with trans-
verse bars welded to beam web.
.ire resistance of effective composite structures with
sections from steel CHS or RHS sections can also be in-
creased with simple adding of longitudinal reinforcement
(.ig 5). .or such sections transverse reinforcement is only
necessary to fix bars on proper position.
All described methods for increasing the fire resis-
tance of steel members can guarantee up to 60 min fire
resistance depending on the quantity of reinforcement.
The required fire resistance level with such measures can
be reached much more cheaper in comparison with ordi-
narily applied special passive fire protection materials.
Additional quantity of steel and concrete is necessary
when fire resistance of elements must be 30 min or
higher.
As it has been mentioned, there exists another pos-
sibility to increase the fire resistance. It does not require
additional financial resources, but in some situations can
ensure necessary fire resistance or reduce the quantity of
fire protection materials. Such results can be obtained
by using more precise methods for evaluating the inter-
nal forces of bearing structures. It may be made taking
into account interaction between different structural mem-
bers.
3. Interaction between slabs and beams
In accordance with the existing design practice re-
inforced concrete slabs are considered as exterior load
.ig 6. Structural and computational schemes for investi-
gation of beamslab interaction: 1 steel beam, 2 con-
crete plate, 3 rigid links
only. On the other hand, really the structural solution of
the slabs, welding of embedded steel parts of slabs to
the steel structures, filling with concrete the joints be-
tween slabs create the conditions for interaction between
slabs and steel structures. The result of this effect is that
slabs take up a part of the internal forces of the steel
structures and consequently diminish them in steel beams
and trusses. Interaction effect was proved in a number
of theoretical and experimental investigations. Some of
them were carried out in Lithuania [79].
Interaction effect between structures may be exami-
ned in the following way.
Computational scheme (.ig 6) of roof or ceiling
structures composed by reinforced concrete slabs and
steel beams instead of traditionally used in such cases is
substituted by the presented scheme in which slabs are
Pshaped members and steel beam as a beam with can-
tilevers of the length equal to a half of the beam depth.
It should be mentioned that the rigidity problem of
Pshaped members substituting the roof slabs was ex-
amined very accurately in a great number of investiga-
tions. The ceiling slabs rigidity was determined in a rather
exact way.
Quantitative changes of bending moments, as inter-
nal forces of the greatest interest in the roof and ceiling
beams for various spans, have been analysed. The val-
ues of bending moments using two computational
schemes were determined in two cases: in one case -
taking into account the interaction effect and without it.
By the way, peculiar magnitudes of roof loads for a cer-
tain time in the past were used in calculations. Ceiling
structures were calculated for a load equal to 10kPa.
The calculation results are presented in .igs 7, 8.
In .ig 7 curves 1 and 3 represent the relationship
between bending moments and beam span when the in-
teraction of structures is taken into account and curves
2, 4 the same relationship when the interaction of struc-
tures is neglected.
Analysis of the results presented in .igs 7 and 8
shows the increase of the load bearing capacity of steel
beams.
.ig 4. Cross-section of partially encased column
.ig 5. Cross-section of composite columns with additional
reinforcement
244
244
4. .ire resistance of structures
.ire resistance of structures in Lithuania must be
determined according to Construction Code .ire Safety,
Basic Requirements [2]. The code suggests very conser-
vative and simple solution in comparison with Eurocode
3 part 12 [3].
The method suggested in this paper can be useful
only when using the Eurocode 3. The current Lithuanian
Code does not take into account the degree of utilisation
of load bearing capacity of a structure. .ire resistance of
steel structures according to [2] depends only on cross-
section factor:
i
A
t
red

, (2)
where A and i cross-section area and perimeter respec-
tively. The same method of calculation is used for all
structural elements types (beams, columns).
It has already been shown above that interaction
between bearing and enclosure structures can increase
the load bearing capacity of structure. Possibility to use
advantages of interaction between different structural el-
ements for fire engineering calculations was discussed
in [5, 6]. The possibility to use benefits of interaction
between steel beam and reinforced concrete floor and
roof slabs for achieving a higher critical temperature of
steel beam were investigated.
Determination of critical temperatures according
Eurocode was carried out for the same cases as it was
performed for internal forces:
the first case - for beam when interaction between
structures is neglected;
the second case - for beam when interaction effect
with slabs is taken into account.
Critical temperature calculations were performed for
the floor and ceiling beams which load bearing capacity
has already been examined (.igs 7, 8). Simple calcula-
tion model was chosen for determination of critical tem-
perature of steel beams.
Because the simple supported beams sections fit the
Class 1 requirements, the design moment resistance was
calculated by the formula:
Rd fi M M y Rd fi
M k M ] / [
, 1 , , , ,

, (3)
where
, y
k reduction factor for yield strength of steel
at temperature ;
1 , M
and
fi M,
partial safety fac-

tors for normal and fire situation;
Rd
M plastic mo-
ment resistance of gross cross-section. Calculation of
degree of utilisation
0
is based on design moment re-
sistance for time t = 0.
Critical temperature is calculated by the formula of
Eurocode:
482 1
9674 , 0
1
ln 19 , 39
8333 , 3
0
,
+
,
_
cr a
. (4)
These calculations results are presented in .igs 9, 10.
.ig 7. Bending moments of beams: 1, 2 roof beams, 3,
4 ceiling beams
.ig 8. Relationship between beams, bending moments:
M
int
evaluating interaction of structures, M interac-
tion of structures is neglected, 1 roof beams, 2 ceiling
beams
.ig 9. Critical temperatures for floor beams with and with-
out beam-slab interaction
M
int
M
%
6 9 12
span of beam, m
620
600
580
560
540
520
500
480
c
r
i
t
i
c
a
l

t
e
m
p
e
r
a
t
u
r
e
with interaction without interaction
245
245
They illustrate the possibility to allow higher criti-
cal temperature for steel beams when beam-slabs inter-
action is taken into account. The interaction effect of
critical temperature was higher for the structure composed
of steel beams and ceiling slabs. Of course, it is related
to character of bending moments reduction in the beams.
5. Conclusions
1. The necessary fire resistance can be achieved by
using composite floor systems and composite columns
elements without applying fire protection.
2. Evaluation of steel beam concrete slabs inter-
action in fire engineering calculation leads to higher criti-
cal temperatures and higher fire resistance of steel beams.
On the other hand, it permits to evaluate more precisely
the real state of structures.
2. .ormulas for critical temperature of Eurocode [3]
take into account the degree of utilisation what is not
included in National code of Lithuania [2]. In such a
situation calculations of fire resistance for elements with
a low degree of utilisation according to Eurocode is much
more cost effective.
.ig 10. Critical temperatures for ceiling beams with and
without beam-slab interaction
References
1. RSN 134-92 Office Buidings. .ire protection requirements.
State building code (Visuomeniniai pastatai ir statiniai.
Priegaisriniai reikalavimai). Construction and Urban Plan-
ning Ministry, Vilnius, 1992. 34 p. (in Lithuanian).
2. RSN 133-91. .ire safety. General requirements. State build-
ing code (Priegaisrin sauga. Bendrieji reikalavimai).
Construction and Urban Planning Ministry, Vilnius, 1992.
78 p. (in Lithuanian).
3. Eurocode 3 Design of steel structures Part 12: Gen-
eral rules Structural fire design. Sept 1995. 64 p.
4. Lawson R. M., Newman G. M. Structural fire design to
EC3 & EC4, and comparison with BS 5950. Ascot, Berk-
shire: New Steel Construction Institute, 1996. 191 p.
5. Ma Z., Makelainen P. Behaviour of composite Slim floor
structures in fire. Journal of Structural Engineering,
Vol 126, No 7, 2000, p. 830837.
6. Bailey C. G., Newman G. M., Simms W. I. Design of
steel framed buildings without fire protection. Ascot, Berk-
shire: New Steel Construction Institute, 1999. 83 p.
7. Juozaitis J. Relationship between stiffness of rafters and
P-frames substituting the slabs of single-storey industrial
buildings. In: Transactions of the Lithuanian Univs, Build-
ing structures (Lietuvos auktj mokykl mokslo darbai,
Statybins konstrukcijos). Vilnius: Mokslas, 1983, p. 27
32 (in Lithuanian).
8. Juozaitis J., Vaitkeviius V. The effect of large-size rein-
forced concrete roof slabs on the behaviour of steel trusses.
In: Transactions of the Lithuanian Univs, Building struc-
tures (Lietuvos auktj mokykl mokslo darbai, Statybins
konstrukcijos). Vilnius: Mokslas, 1985, p. 35 (in
Lithuanian).
9. Juozaitis J., Praneviius J., Vaitkeviius V. Interaction be-
tween steel trusses and reinforced concrete roof slabs. In:
Transactions of Vilnius Technical University, Building struc-
tures (Vilniaus technikos universiteto mokslo darbai,
Statybins konstrukcijos), Vol 20. Vilnius: Technika, 1994,
p. 1823 (in Lithuanian).
c
r
i
t
i
c
a
l

t
e
m
p
e
r
a
t
u
r
e
6 9 12 15 18
span of beam, m
620
600
580
560
540
520
500
480
with interaction without interaction
247
247
Life counts years quickly a nice jubilee is cel-
ebrated by Romanas Karkauskas, Professor Doctor of
Department of Structural Mechanics of Vilnius
Gediminas Technical University. He is the University's
veteran one of a few lecturers employed from the first
day in the first higher technical educational institution
in Vilnius Civil Engineering Institute (VCEI), which
was founded in 1969. Vilnius was the place where his
research, pedagogical, engineering talents have been
successfully realised. Here he has started his family, has
built his own house
In the village Kaltinnai (former Varniai region),
where Professor was born on September 11, 1944, now
a memorial stone is erected The reason is that a lot
of the native village land currently has returned to him
as owner, is interseced by a highway Vilnius-Klaipda
with its highest point (approximately 20 km from
Krykalnis on the way to Klaipda), marked by the
above memorial. Here, not far from the point, is the
birth place of R. Karkauskas.
R. Karkauskas finished secondary school in Taurag
and in 1962 entered the Civil Engineering faculty of
Kaunas Politechnical Institute (KPI) aiming to obtain
the civil engineering qualification. When taking full-time
studies during the first two years he worked as a simple
worker in process of building construction (following
the idea of the contemporary Soviet Union leader N.
Chruschiov) and in evening attended the lectures in the
big auditorium of the 5th building. Nowadays it is the
wonderful city hall of Kaunas On the third year he
started activities in the student research group of De-
partment of Structural Mechanics of KPI. In 1967 he
graduated with honour from the KPI obtaining the civil
engineer qualification and became a theoretically quali-
fied young specialist.
The R. Karkauskas' dreams to enter the famous
Structural Mechanics school of Academician A. yras
became true in 1967. The graduate was directed as a
probator to Department of Mechanics of the Vilnius
branch of KPI. Working in this position the young spe-
cialist proved himself as creative and talented investi-
gator; on 1968 he entered direct Doctoral studies of the
Department. The research topic was Design of rigid-
plastic shallow spherical shell applying the mathemati-
cal programming. At that time there was not many sci-
entists in the world applying the non-linear mathemati-
cal programming for optimization of plastic structures.
R. Karkauskas was one among the first investigators who
created the sufficiently universal computer program for
non-linear problem solution, which was employed for
limit equilibrium problem solution of shell of revolu-
tion. The latter numerical result was among the first
attempts in this field of investigation in the Soviet Union.
In 1972 R. Karkauskas was honoured the candidate of
technical sciences (now PhD) scientific degree. In 1974
PRO.ESSOR ROMANAS KARKAUSKAS CELEBRATES HIS 60
CHRONICLE
together with Academician A. yras and A. Borkowski
he prepared and published in Russian the scientific mono-
graph Theory and methods of optimization of elastic-
plastic systems (in 2004 the second extended edition of
the monograph in English Theory and methods of opti-
mization of rigid-plastic systems was published). The
book in concert with the successive A.yras' monographs
was highly appreciated by many scientists.
In 1980 R. Karkauskas was honoured the title of As-
sociate Professor. He is the co-author of the manual
Structural Mechanics. Problems solution applying com-
puters (in Russian and Lithuanian). In 1993 for a thor-
oughly compiled set of computerised manuals
R. Karkauskas and a group of investigators of the De-
partment of Structural Mechanics (A. Krutinis, S. Kalanta,
J. Nageviius, J. Akoinas) was awarded Lithuanian Na-
tional Prize. Now R. Karkauskas lectures the main courses
on structural mechanics and optimization for civil engi-
neering students. Many Lithuanian civil engineers VCEI
graduates remember R. Karkauskas, their lecturer of
structural mechanics.
His deep theoretical knowledge R. Karkauskas con-
veys not only to Bachelor, Master, Doctoral students but
also provides consultations for designers he has per-
formed statical calculations of Vilnius stadium, of many
buildings in Vilnius and Kaunas. He is certified (certifi-
cation No 9378) specialist in the following fields: build-
ing design, building structural part, author's and techni-
cal supervision, building project expertise and building
expertise supervision.
The field of scientific interests optimization of
elastic and elastic-plastic structures of buildings, origi-
nal results of scientific investigations allowed to
R. Karkauskas to probate in Polish Academy of Sciences,
scientific institutes of Kiev and Moscow, to present his
own works at structural optimization international con-
ferences, held in Lithuania and abroad. He is the author
of over 80 manuscripts printed in local and foreign sci-
entific journals. In beginning of the summer 2004
R. Karkauskas presented a review of his scientific inves-
tigations titled Theory and methods of optimization
rigid-plastic systems: geometrical and physical non-
linearities, nominated for Habilitation procedure and
successfully maintained it receiving the Professor's title.
On September 11 this year Romanas Karkauskas cel-
ebrates his 60th birthday and 35th annivesary of scien-
tific research. The Department of Structural Mechanics
of Vilnius Gediminas Technical University congratulates
Professor Romanas Karkauskas on these festive occasions
and wishes him good health, happiness, long and fruitful
creative work.
Professor Dr Habil Juozas Atkoinas,
Head of Department of Structural Mechanics
of Vilnius Gediminas Technical University
Ia
SANTRAUKOS
A. Alejev, M. Grikeviius, R. ukys, I. .rith, A. Norta. Ignalinos atomins elektrins gaisrins rizikos analizs
apvalga // Journal of Civil Engineering and Management. Vilnius: Technika, 2004, t. X, Nr. 3, p. 169175.
Gaisrin sauga svarbi atomins elektrins ilgaamikumui, j projektuojant ir statant, inspektuojant bei eksploatuojant.
Pateikta VATESI dokument, reikaling licencijai gauti, apvalga bei informacija apie tai, kaip Ignalinos atomin elektrin
atitinka respublikinius reikalavimus bei tarptautines rekomendacijas utikrinant gaisrin saug. Veikianio RBMK-1500
tipo reaktoriaus gaisrins saugos analiz atlikta pasaulyje pirm kart.
Raktaodiai: sauga, gaisrin sauga, atomin elektrin, gaisrins rizikos analiz, apsauga, licencija, apvalga, saugos
gerinimas.
J. Brauns, K. Rocens. Mediagos stiprio taka betonerdi plien element elgsenai // Journal of Civil
Engineering and Management. Vilnius: Technika, 2004, t. X, Nr. 3, p. 177182.
Sudtins kolonos ir sijos yra betonini ir plienini element kombinacija, jungianti abiej tip mediag pranaumus.
Betonerdi kolon normose skerspjvio plastinis atsparis pateikiamas kaip komponent suma, kartu atsivelgiant apkabos
efekt apskrit skerspjvi atveju. Tyrime sudtini kolon tempi bvis yra nustatytas, atsivelgiant tamprumo moduli
ir Puasono koeficient netiesinius sryius su betono erdies tempi lygiu. Nustatyta, kad apkabos efektas atsiranda esant
auktam tempi lygiui, kai statybinis plienas yra tempiamas, o betonas gniudomas i ono. Lenkiamo pjvio tempi
bvis ir atsparis nustatytas, atsivelgiant netiesin neutraliosios aies padties priklausomyb. Vis dali mediagos
saugos ribinis bvis gaunamas ne tuo paiu laiku. Sudtinio elemento elgsenos slygoms pagerinti ir norint ivengti
suirties dl mao statybinio plieno storio, turi bti naudojamas atitinkam stiprumo klasi betonas ir plienas.
Raktaodiai: sudtins kolonos, sijos, tempi bvis, tamprumo modulis, Puasono koeficientas, netiesin deformacija.
B. .aggiano, G. De Matteis, R. Landolfo, .. M. Mazzolani. Aliuminio lydini konstrukcij elgsena gaisro
metu // Journal of Civil Engineering and Management. Vilnius: Technika, 2004, t. X, Nr. 3, p. 183190.
Daug dmesio skiriama Euronormose 9 pateikt statybini aliuminio lydini mechanini savybi priklausomybi nuo
aukt temperatr analizei. Atlikus techninje literatroje pateikt duomen analiz, gautos tamprumo modulio, slygins
takumo ribos, ribinio stiprio, kietjimo koeficiento ir ribins deformacijos priklausomybs nuo temperatros. Silomas
mechaninis modelis paremtas gerai inoma Ramberg-Osgood formuluote, vertinania ios mediagos elgsenos auktoje
temperatroje ypatybes. Darbe vertintas ir ianalizuotas bendras sustiprjimo veiksnio ir auktos temperatros poveikis
mediagos tempi ir deformacij priklausomybei. Silomas modelis pritaikytas baigtini element programai, skirtai
konstrukcij analizei gaisro slygomis. Pateikti rezultatai, gauti tiriant P formos rm, sudaryt i skirting aliuminio
lydini. Gauti rezultatai patvirtina mediagos savybi vertinimo svarb modeliuojant gaisro veikiam aliuminio konstrukcij
elgsen.
Raktaodiai: aliuminio lydiniai, mechanins savybes gaisro metu, sustiprjimo efektas, tempi ir deformacij priklau-
somyb auktoje temperatroje, gaisro veikiam konstrukcij analiz.
A. Garstecki, A. Knitter-Pitkowska, Z. Pozorski ir K. Ziopaja. Paeidim nustatymas parametriniais
dinaminiais eksperimentais ir taikant bangeli transformacij // Journal of Civil Engineering and Management.
Vilnius: Technika, 2004, t. X, Nr. 3, p. 191197.
Nagrinjamos lokali paeidim nustatymo taikant statinius ir dinaminius bandymus problemos. Konstrukcijos elgsena
analizuojama taikant bangeli transformacij kaip signal apdorojimo bd. Lokalus sijini konstrukcij paeidimas
modeliuojamas lenkimo standumo sumajimu. Pateikta statini ir dinamini bandym efektyvumo analiz. Aptariami
minimals eksperimentini matavim skaiiaus ir matavimo tikslumo reikalavimai, kad paeidimas skmingai bt
lokalizuotas. Pateikti keli skaitiniai pavyzdiai.
Raktaodiai: konstrukcijos identifikavimas, paeidim nustatymas, bangeli transformacija.
M. A. Gizejowski, C. J. Branicki, A. M. Barszcz, P. Krol. Plienini rm patikslinta analiz vertinant sandr
deformacijas bei j dalin stipr // Journal of Civil Engineering and Management. Vilnius: Technika, 2004, t. X,
Nr. 3, p. 199208.
Apibendrinti rm su pusiau standiomis jungtimis iuolaikiniai patobulinti skaiiavimo metodai. Darbe pateikti metodai
priklauso bendrajai antrojo laipsnio patobulint plastini lankst metod grupei, kuriais vertinta sandr standio sumajimo
bei iilgai elemento ir jo skerspjvyje pasireikiani plastini deformacij sumin taka. Autori pasilytas santvar
skaiiavimo metodas pagrstas spyruokli sistemos modeliu. Jungi dalinis standis bei stipris vertinti pasirinkus tam
tikras jungt atitinkani spyruokli pradinio standio, ribinio momento bei momento-poskio charakteristikos formos
faktoriaus reikmes. Gniudomj konstrukcini element elgsenos modeliavimo netobulumas vertinamas taikant
supaprastint tangentinio modulio koncepcij bei sumainant pradin tamprumo modul. Liekamj tempi taka vertinama
imant tam tikras spyruokls momento-poskio charakteristikos formos parametro reikmes, kurios reprezentuoja palaipsn
elemento tekjim. ios reikms priklauso nuo skerspjvio tipo ir gamybos bdo (t. y. nuo valcavimo ir suvirinimo metu
................................................................................................................................................................................... SANTRAUKOS
Ib
atsiradusi liekamj tempi pasiskirstymo). Pateikta skaitinio eksperimento analiz. Darbe, remiantis minto eksperimento
rezultatais, suformuluotos ivados, susijusios su pateiktos analizs pritaikymu projektavimui.
Raktaodiai: plieninis rmas, pusiau standi jungtis, patikslinta analiz, spyruokli sistemos modelis.
H. Y. Leung. Plieno ir anglies pluoto polimero strypais armuot lenkiamj betono sij laikomoji galia //
Journal of Civil Engineering and Management. Vilnius: Technika, 2004, t. X, Nr. 3, p. 209215.
Nepaisant to, kad pastaraisiais metais atlikta daug tyrim, susijusi su plieno ir anglies pluoto polimero strypais armuotomis
betono sijomis, daugelis ininieri projektuotoj j naudojim iri labai atsargiai ir pirmiausia dl to, kad anglies
pluoto polimeras yra trapi mediaga, todl iuo polimeru armuot betono sij laikomosios galios netekimas gali bti
katastrofinis. Dl ios prieasties racionalu projektuoti sijas, amuotas tiek anglies pluoto polimero, tiek plieniniais strypais.
Dabartins projektavimo normos reglamentuoja tik plieno arba tik anglies pluoto polimero strypais armuot betonini
element projektavim. Pateiktos plieno ir anglies pluoto polimero strypais armuot betonini sij projektavimo gairs
bei priklausomybs. Gauti teorini tyrim rezultatai gerai sutapo su eksperimentini bandym duomenimis.
Raktaodiai: betonas, sija, lenkimo laikomoji galia, anglies pluoto polimeras, plienas.
A. Norkus, R. Karkauskas. Santvaros optimizacija vertinant kompleksinius apribojimus ir atsitiktin apkrov
// Journal of Civil Engineering and Management. Vilnius: Technika, 2004, t. X, Nr. 3, p. 217226.
Realaus lengvo svorio konstrukcij projektavimo proceso metu turi bti vertinti stiprumo, standumo ir stabilumo reikalavimai
bei apkrovimo proceso pobdis. Suprojektuota konstrukcija turi atitikti optimalumo ir patikimumo kriterijus per vis jos
eksploatacijos laikotarp. traukus bent kelet ivardyt reikalavim optimizavimo udavinio formuluot susiduriama su
inomais optimizavimo udavinio skaitmenins realizacijos sunkumais. Pateikiamas metodas, kaip formuluoti strypins
konstrukcijos optimizavimo udavin vertinant visus mintus reikalavimus, udavinio matematinis modelis ir jo skaitins
realizacijos algoritmas vertinant ir apkrovimo proceso stochastin prigimt. Pateikiamas skaitinis pavyzdys.
Raktaodiai: tampriai plastin konstrukcija, optimizacija, standumo ir stabilumo apribojimai, stochastinis apkrovimo
procesas, baigtini element diskretinis modelis.
G. Skripkinas, M. Daukys. Cemento tel su cheminiais priedais dilatantikumas // Journal of Civil Engineering
and Management. Vilnius: Technika, 2004, t. X, Nr. 3, p. 227233.
Cemento telos ir betono miiniai danai priskiriami prie Binhamo kn (sistem), kurioms bdingos dvi reologins
charakteristikos ribiniai lyties tempiai
0
ir struktrin klampa . Tyrimai rodo, kad Binhamo modeliu netiksliai
apraomos cemento tel tekjimo kreivs. i tel tekjimo kreivs nra tiess kaip Binhamo modelyje, didjant lyties
tempiams jos gauna kreivs pavidal. Tai rodo, kad cemento teloms bdingas dilatantikumas, t. y. sistemos klampos
didjimas didjant lyties tempiams. Atlikti eksperimentiniai tyrimai, kuriais nustatyta plastifikuojani, or traukiani ir
stabilizuojani pried taka cemento tel reologinms savybms ir dilatantikumui. Cemento tel reologins savybs
nustatytos naudojant rotacin viskozimetr su koaksialiniais cilindrais. Apie cemento tel dilatantikum buvo sprendiama
i tekjimo kreivi pobdio ir dilatantikumo rodiklio D, skaiiuojamo pagal pasilyt metodik. Nustatyta, kad
plastifikuojantys priedai nepriklausomai nuo chemins sudties padidina cemento tel dilatantikum, palyginti su cemento
tela be priedo. Or traukiantys ir stabilizuojantys priedai neymiai sumaina cemento tel dilatantikum, palyginti su
cemento tela be priedo. Tyrim rezultatai parod, kad cemento telos dilatantikumas maai priklauso nuo skystos dispersins
terps savybi.
Raktaodiai: cemento tela, cheminiai priedai, reologija, lyties tempiai, klampa, dilatancija, viskozimetras.
V. J. Stauskis. Subjektyvi akustikos rodikli kitimas salse, kuriose yra didel ir maa aidjimo trukm //
Journal of Civil Engineering and Management. Vilnius: Technika, 2004, t. X, Nr. 3, p. 235239.
Tiriant vairios paskirties ir labai skirting tri sali akustik pastebta, kad muzikos skambjimo aikumo indekso C
80
reikms yra geros arba priimtinos, o i sali aidjimo trukm yra arba labai maa, arba labai didel, todl i sali
akustika subjektyviai yra vertinama nepatenkinamai. Vilniaus operos ir baleto teatre aidjimo trukm yra apie 1 s, o C
80
indekso reikms visose sals vietose yra arba geros, arba priimtinos. Banyioje, kurios tris 22 650 m
3
, yra priimtinos
C
80
reikms atlikj vietose ir priekinse eilse. Panas rezultatai gauti ir kitokiose salse.
Raktaodiai: aikumo indeksas C
80
, aidjimo trukm, absorbcija, ankstyvoji energija, vlyvoji energija, garso atspindiai.
A. apalas. Sudtini mediag taikymas plieno ir betono konstrukcijose siekiant didesnio j atsparumo ugniai
// Journal of Civil Engineering and Management. Vilnius: Technika, 2004, t. X, Nr. 3, p. 241245.
Viena i pagrindini biur ir pramonins statybos Lietuvoje ir pasaulyje ypatybi yra platus plienini konstrukcij taikymas.
Tai yra susij su nauj konstrukcini sprendim ir konstrukcini element bei mediag taikymu. Skirting savybi mediag
sujungimas vienoje konstrukcinje sistemoje siekiant geresnio teigiam j savybi panaudojimo yra gana populiarus.
Aptariama mediag ir konstrukcini element sveikos taka konstrukcijos atsparumui ugniai.
Raktaodiai: sudtins konstrukcijos, plienins sijos, gelbetonins plokts, konstrukcij sveika, atsparumas ugniai.
. , . , . , . , . .
// Journal of Civil Engineering and Management. : , 2004, X .,
3, c. 169175.
, ,
. VATESI
,
.
RBMK-1500 .
: , , , , ,
, , .
. , . . //
Journal of Civil Engineering and Management. : , 2004, X ., 3, c. 177182,

.
.

. ,
, ,
.
. ,
,
.
.
: , , ,
.
. , . , . , . . .
// Journal of Civil Engineering and Management. : ,
2004, X ., 3, c. 183190.
,
, 9- Eurocode.
, , ,
, ,
. -
, , .
- .
, .
.
,
.
: , , ,
- , ,
.
. , . -, . . .
// Journal of Civil
Engineering and Management. : , 2004, X ., 3, c. 191197.
,
.
.
. .
,
. .
: , , .
. , . . , . . , . .
// Journal of Civil Engineering and Management.
: , 2004, X ., 3, c. 199208.
.
, , .

, .
, .
.
. , ,
.

. ,
, .
IIa
..................................................................................................................................................................................... IIb
.
.
, .
: , , , .
. . . ,
// Journal of Civil Engineering and Management. : , 2004, X ., 3, c. 209215.
, ,
, ,
. ,
.
.
.

, .
.
: , , , , .
. , . .
// Journal of Civil Engineering and Management. : , 2004, X ., 3, c. 217226.
, ,
, .
.
, .
, ,

. .
: , , ,
, - .
. , M. . // Journal
of Civil Engineering and Management. : , 2004, X ., 3, c. 227233.
() ,

0

.
, .
, .
. , , . .
. ,
,
.
.
D,
. , ,
, .

. ,
.
: , , , , , ,
.
.-. .
// Journal of Civil Engineering and Management. : , 2004, X ., 3,
c. 235239.
, , ,

80
,
, . . ,
1 ,

80
, . ,
22650 m
3
,
80

. 5-6 .
.
: , , , , , .
. .
// Journal of Civil Engineering and Management. :
, 2004, X ., 3, c. 241245.

.
.
.
: , , ,
, .
International Journal of Civil Engineering and Management
Order form
ISSN 1392-3730 n Volume 10 (2004) n Frequency:
4 times a year (1 volume of 4 issues)
oYes, I want to subscribe to
for 2004
Institutional Rate
oEurope and rest of World: 140 EUR
Personal Rate
oEurope and rest of World: 140 EUR
oEurope and rest of World for members of CIB, ARCOM, CIOB,
RICS, 70 EUR
oEurope and rest of World for students: 40 EUR
Personal details (Please use BLOCK CAPITALS):
Name: ....................................................................................................
Job Title: ................................................................................................
Address: .................................................................................................
...............................................................................................................
...............................................................................................................
Postcode/Zip: .........................................................................................
Country: .................................................................................................
Telephone: .............................................................................................
Fax: ........................................................................................................
E-mail: ....................................................................................................
Sample Copy Request Form
o Please send me a FREE sample copy of International Journal of Civil Engineering and Management
Name: ................................................................................................... Job Title: ............................................................................
Address: .............................................................................................................................................................................................
...........................................................................................................................................................................................................
Postcode/Zip: .....................................................................................................................................................................................
Country: .............................................................................................................................................................................................
Telephone: ............................................................................................ Fax: ....................................................................................
E-mail: ................................................................................................................................................................................................
Library Recommendation Form
To: Periodicals Librarian
Name: .................................................................. Department: ..................................................................................
Signature: ............................................................ Date: ..............................................................................................
I recommend that the Library subscribe to the International Journal of Civil Engineering and Management
(ISSN 1392-3730) published by Technika.
Methods of payment
o Payment enclosed. Cheques or bank drafts should be made payable to Vilnius Gediminas Technical
University and sent to: VGTU Publishing House Technika, Sauletekio al. 11, LT-10223 Vilnius, Lithuania.
o Payment has been made by Bank Transfer to:
Vilniaus Bankas AB, Gedimino pr. 12, LT-01103 Vilnius, Lithuania.
Account No. LT767044060000317763; Swift Code: CBVI LT 2X
All subscriptions are payable in advance. Subscribers are requested to send payment with their order whenever
possible. Issues will be sent on receipt of payment.
Technika contact details:
VGTU Publishing House Technika, Sauletekio al. 11, LT-10223 Vilnius, Lithuania.
Telephone: + 370 5 274 50 38 n Fax: + 370 5 2700112
E-mail: leidykla@adm.vtu.lt n Internet: http://www.vtu.lt/english/editions/
To order your FREE sample copy, complete and return this order card
By post:
n VGTU Publishing House Technika, Sauletekio al. 11, LT-10223 Vilnius, Lithuania.
or
n Fax: + 370 5 2700112
To request a sample copy please visit the Technika website at http://www.vtu.lt/english/editions/
2004 Institutional Subscription Rate
ISSN 1392-3730 n Volume 10 (2004) n Frequency: 4 times a year (1 volume of 4 issues)
o Europe and rest of World: 140 EUR
IJSPM provides a forum for discussion and debate relating to all areas of strategic property management.
Technika contact details:
VGTU Publishing House Technika, Sauletekio al. 11, LT-10223 Vilnius, Lithuania.
Telephone: + 370 5 274 50 38 n Fax: + 370 5 2700112
E-mail: leidykla@adm.vtu.lt n Internet: http://www.vtu.lt/english/editions/

Civil - Eng 2004 3

Enviado por

Dados do documento

Título original

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

Civil - Eng 2004 3

Enviado por

Direitos autorais:

Formatos disponíveis

Vilnius Gediminas Technical University Lithuanian Academy of Sciences

Journal of Civil Engineering

is rate of combustion, m is mass of combus-

, placed with the given step. It is taken into account

characteristic of the stress diagram form. The cross-sec-

ultimate moment of the member section (if necessary

ultimate strain of .RP rods. The linear strain pro-

, the limiting ratios of

are material yield and critical stresses for tensile and

is certain bar reduction coefficient, depending

The second load ) (

The third load ) (

The loads processes are described by the following

for the different rotation rate of the

partial safety fac-

Você também pode gostar