Escolar Documentos
Profissional Documentos
Cultura Documentos
SUMMARY
To enable acceptable seismic displacement demands to be speci5ed, the displacement capacity of a
structural system needs to be known. This is controlled by selected performance criteria. It is postulated
that the assignment of fractions of the required strength of the system to its components may be arbitrary.
This then enables the displacement capacity of buildings to be evaluated without the knowledge of
its seismic strength. A rede5nition of some traditionally used structural properties is a prerequisite to
applications. The study of a reinforced concrete wallframe dual system illustrates rationale and extreme
simplicity of application. Copyright ? 2001 John Wiley & Sons, Ltd.
KEY WORDS:
1. INTRODUCTION
Numerous recent publications, addressing issues of performance-based earthquake engineering,
described invaluable approaches to the estimation of displacement demands that may be imposed on building systems by earthquakes. However, relatively few studies considered ways
and means by which displacement capacities of structural systems, particularly reinforced
concrete buildings, especially those comprising components with very di@erent behavioural
characteristics, could be estimated. A rational evaluation of structural displacement capacities,
corresponding to speci5c performance criteria, should enable acceptable seismic displacement
demands, associated with appropriate earthquake records, to be established. This in turn should
facilitate the corresponding seismic strength of systems to be quanti5ed. It should ensure that
expected displacement demands are not likely to exceed capacities, which can be reliably
provided.
To allow displacement capacities to be quanti5ed, some traditionally used structural properties needed to be rede5ned. It is considered that several assumptions, particularly those related
Correspondence to: Tom Paulay, Department of Civil Engineering, University of Canterbury, Private Bag 4800,
Christchurch, New Zealand.
E-mail: t.paulay@civil.canterbury.ac.nz
584
T. PAULAY
to elastic structures using homogeneous materials, do not meet realistic criteria, for example those relevant to ductile reinforced concrete systems. Some important 5ndings of recent
studies, which critically examined fallacious, yet widely used, concepts [1], were adopted in
this work. This enabled displacement capacities of components and elements of structural
systems to be more realistically evaluated. Moreover, they may be simply related to each
other. In this context an element comprises components, which are subjected to identical
lateral displacements. On the other hand, due to torsional phenomena, elements of a system, generally placed parallel to each other, may be subjected to di@erent earthquake-induced
displacements [2].
A behaviour-based strategy is outlined, which enables the designer to estimate limiting displacements of systems about to be designed. This can be undertaken without the knowledge
of the strength to be eventually assigned to the system. A relatively complex structure, comprising reinforced concrete walls and frames, was chosen to demonstrate the rationale and
simplicity of implementation.
The procedure avoids the use of engineering judgement-based global displacement ductility
ratios, extensively used in modern building codes. Instead, it addresses displacement capacities
of components. This then allows the displacement capacities of elements to be readily established. A similar approach is used to derive the ductility capacity of an entire system, as a
function of the displacement capacities of its critical components or elements.
2. PERFORMANCE-BASED EARTHQUAKE-ENGINEERING
One of the most important aspects of the speci5cation for seismic performance criteria is
that addressing structural displacements. By necessity the estimates of earthquake-imposed
displacement demands must be based on a probabilistic approach. However, the meeting
of these criteria in the process, for example of structural design, may be better guided by a
deterministic approach. In simple terms, with con5dence the designer should tell the structure
to be constructed, what it should do, rather than ask, by way on analysis, what it might do [3].
585
rather than analysis-oriented approach is to identify the major sources of component, element
and system displacements.
3.1. Preliminary considerations
Traditionally, the structural design process starts with experience-based estimates of dimensions, particularly component sizes, which are likely to satisfy functional requirements of the
building. Once this information, based on both architectural and engineering perspectives, is
available, displacement estimates can be readily made.
3.2. Curvature estimates
A fundamental property of a component is the nominal yield curvature at its critical section or
sections. This will de5ne its response in the elastic and post-elastic domain of behaviour. In
reinforced concrete members, it is more reliable to base this estimate on quanti5able section
properties, rather than on an assumed or recommended [4] values of Mexural rigidity, Ec Ie ,
where Ec is the modulus of elasticity of the concrete and Ie is the second moment of the
e@ective sectional area of the cracked component.
Figure 1 shows curvature relationships relevant to a typical structural wall section. With
some design experience, the neutral axis depth can be estimated. If necessary, this estimate
can be subsequently veri5ed once details of the Mexural reinforcement are known. For a
given tensile yield strain, y , and the location of the neutral axis, D, the curvature at the
onset of yielding is established as y = y =D, where D is the overall depth of the section. When details of the reinforcement are known, the associated yield moment, My , may
also be evaluated. However, the need for this will seldom arise. The designer will primarily
Copyright ? 2001 John Wiley & Sons, Ltd.
586
T. PAULAY
address nominal strengths, Mn , at sections, once strength requirements for components are
known.
For seismic design purposes a bi-linear simulation of the non-linear moment curvature
relationship, as shown in Figure 1, is adequate. With linear extrapolation, this then leads to
the nominal yield curvature
y = (Mn =My )y = (Mn =My ) ( y =D) =
y =D
(1)
587
curvature, y , of the component. Hence, the nominal strength-independent yield displacement, Qy , may be based on the nominal yield curvature, y , at the critical section, given by
Equation (1). Subsequent examples will illustrate applications. An important conclusion, drawn
from the bi-linear relationship seen in Figure 3, is that component sti@ness is de5ned as
k = Vy =Qy = Vn =Qy
(2)
(3)
where Qim is the component displacement associated with a stipulated performance limit state.
4. THE FREEDOM IN ASSIGNING ELEMENT STRENGTHS
As stated previously, the nominal yield curvature in a component with given dimensions,
and hence the relevant nominal yield displacement, may be considered for seismic design
purposes to be independent of the nominal strength of the component. Therefore, strength
to components and elements, comprising components, may be assigned arbitrarily [2; 7]. The
implications are as follows:
(1) Irrespective of the strength assigned to components, they will commence yielding when
the imposed system displacement attains the nominal yield displacement of the relevant
component.
Copyright ? 2001 John Wiley & Sons, Ltd.
588
T. PAULAY
(2) Simultaneous onset of yielding of components of elements with di@erent overall depths,
and hence di@erent aspect ratios, is not possible.
(3) Lateral force-resisting components of a system with di@erent nominal yield displacements, when subjected to identical translational displacements, will be exposed to
di@erent displacement ductility demands.
(4) A structural system, comprising elements having di@erent nominal yield displacements,
does not exhibit a distinct yield displacement. The de5nition of the nominal yield
displacement of a system may become ambiguous and hence it needs to be rede5ned.
(5) To ensure that all components of a system will perform satisfactorily, the displacement
capacity of the system is to be restricted to that of the component with the smallest
displacement capacity.
Arbitrariness in strength assignments imparts to the astute designer the ability to choose
from a number of possible, viable and appealing solutions. To illustrate the relevance of the
structural features listed above, a somewhat idealized system, deliberately made simple, is
studied here brieMy.
Figure 4(a) shows the plan of a structure in which lateral forces in the y direction are resisted by 5ve rectangular reinforced concrete cantilever walls of identical heights but with different lengths, Di . Components transmitting some gravity loads are not shown in Figure 4(a).
A traditional design approach would assign nominal wall strength, i.e. base shear, in proportions of Di3 . Under uniform element translations this would require reinforcement ratios in
Walls (1) and (3), 43 per cent larger than for Walls (4) and (5). However, if wall nominal
strengths are made proportional to Di2 , all walls would require the same ratio of vertical reinforcement. This would enable identical arrangement of bars to be used in all walls of the
structure.
Wall strengths so provided would, however, result in this case in a strength eccentricity
with respect to the centre of mass (CM) of 0:095a. With some 6 per cent reduction of the
strength of Wall (1) and a corresponding increase of the strengths of Walls (4) and (5), even
this relatively small eccentricity would be eliminated. Thereby torsional phenomena, a@ecting
element displacement demands primarily in the post-yield domain of response, would become
Copyright ? 2001 John Wiley & Sons, Ltd.
589
negligible [2; 8], and would thus not need to be addressed. It may be shown [2] that, based
on Equation (2), the sti@ness eccentricity, a@ecting elastic response of the model structure,
would also be rather small (0:21a, i.e. approximately 4 per cent of the length of the plan).
The bi-linear simulation of the translational behaviour of the system and its 5 components
is shown in Figure 4(b). Displacements plotted are normalized in terms of the nominal yield
displacement of Walls (1) and (3), taken as unity. The modelling of the displacements of the
example structure, comprising of components with di@erent strength and sti@ness properties,
demonstrates:
(i) the sequence of component yielding, from that of Wall (1) to Wall (5), during monotonic displacements of the system. Nominal yield displacements are inversely proportional to wall lengths, Di , and, as stated, are independent of strength assigned to
them.
(ii) the superposition of bi-linear component responses results in non-linear system response.
(iii) evidence that assuming bi-linear simulation also of system behaviour would be justi5ed.
This assumption allows the nominal yield displacement of the system to be expressed
as
Qy =
Vin =
ki
(4)
In this example Qy = 1:00=0:879 = 1:14 displacement units. It is to be used as a reference to enable the system displacement ductility, given by Equation (5), to be quanti5ed. At this displacement some elements, such as Walls (1) and (3), will have entered
the inelastic domain, while others are still elastic.
(iv) that, contrary to the characteristics of nominal yield displacements of components,
Qiy , that of the system, Qy , will be a@ected to a small extent by the distribution of
component strengths, Vin .
(v) that, as stated previously, the translational displacement capacity should be controlled
by that of the critical element, i.e. Walls (1) and (3). In this example, it is assumed
that appropriate detailing of the walls [3; 4] allows a component displacement ductility
capacity of iQ = 4 to be relied on. Therefore, Qm = 4 1 = 4 units. When the ductility
capacity of the system is considered, as in strength-based seismic design procedures,
the system displacement ductility capacity to be assumed should, therefore, be restricted
to
Q 6(Qiy =Qy )iQ = (1:0=1:14) 4 = 3:5
(5)
(vi) that, to a relatively small degree, the choice of strength distribution within a system
will a@ect its sti@ness.
This simple example was presented in support of previously made claims, that displacement
limits applicable to a system and its components can be established before required strengths
are determined. In terms of the required strength of the total system, the strengths of its
components may be freely chosen.
Copyright ? 2001 John Wiley & Sons, Ltd.
590
T. PAULAY
591
Figure 6. Design actions and displacement limits relevant to the 12-storey dual system.
Copyright ? 2001 John Wiley & Sons, Ltd.
592
T. PAULAY
Selected shear resistances over the height of the building also determine element contributions to the resistance of the total overturning moment, M . The condensed frame, being
subjected to a single lateral force at level 13, will sustain a linearly increasing moment with
a maximum value of 0:35hVb at the base. This happens to be close to 50 per cent of the total
overturning moment at the base, 0:693hVb . The share of the overturning moment assigned to
the walls is shown shaded in Figure 6(b).
The walls also need to support gravity loads. In accordance with the principles of capacity
design, the nominal Mexural strength of the walls at all levels above the base, will need to
be well in excess of those indicated in Figure 6(b). This is to ensure that a plastic hinge,
requiring special detailing, will not develop during the dynamic response at any region of
a wall above its base. These regions, although thoroughly cracked, are expected to remain
elastic.
With a known moment pattern applicable to the walls, their deformations may be readily
estimated. It is evident that deformations of this dual system, when all components develop
at least their nominal strengths, will be controlled by those evolving in the walls.
5.3. Nominal yield deformations
Because wall deformations control target displacements of the system, their estimation is of
great importance. Using Equation (1) and the data provided in column 6 of Figure 2, i.e.
= 1:8, for a rectangular wall under moderate axial compression load, the nominal yield
curvature at the base section is
wy = 1:8 y =Dw = 1:8 0:002=0:14h = 0:0257=h
(6)
The moment pattern, applicable to the walls and shown in Figure 6(b), exhibits near-linear
variation over the height he , equivalent to that of a cantilever wall with a concentrated lateral force at that level. This approximation is shown by a thin dashed line. Therefore, the
displacement in the vicinity of the eighth storey will be close to
Qwy = wy h2e =3 = (
he =3) y Arw
(7)
(8)
Two important terms a@ecting wall yield displacements, the chosen yield strain y and the wall
aspect (slenderness) ratio, Arw , are worth noting. Using these data for the example structure,
it is found that
Qwy = (1:8 0:63h=3)0:002 4:5 = 0:0034h
(9)
Displacement ductilities and sti@ness of walls can be subsequently expressed in terms of its
nominal yield displacement, Qwy .
The maximum slope of the wall, at the onset of yielding at its base, being a good measure
of storey drift at level he ; wy , is
wy = wy = wy he =2 = 0:0257 0:63=2 = 0:0081 rad
Copyright ? 2001 John Wiley & Sons, Ltd.
(10)
593
Maximum storey drifts developed in the post-yield range of behaviour of the dual system,
will be in or close to the eighth storey.
Similarly, storey yield displacements generated in the frames can be realistically estimated.
When the development of plastic hinges is restricted to the ends of beams, so that, with the
exception at level 1, columns over the full height of the building will remain elastic, the
storey drift associated with nominal yield curvatures developed in the beams, was shown [13]
to be of the order of
fy = 0:5 y Arb
(11)
where Arb is the aspect ratio of the beams. Details of the derivation [13] of Equation (11)
are not given here. In this approximation, allowance was made also for the deformations of
elastic columns and those due to shear, and particularly for beamcolumn joint distortions.
Sensitivity of storey drift in frames to y and Arb , may again be noted. For the frames in this
example
fy = 0:5 0:002 (0:21h=0:018h) = 0:0117 rad
(12)
(13)
with the assumed centre of plastic hinge rotation at 0:028h above the base. The post-yield
rotation of the wall should thus not exceed
wp = 0:0250 0:0081 = 0:0169 rad
(14)
(15)
(16)
With appropriate detailing of the walls, this displacement, Qwm = 4 0:0034h = 0:0136h, can
be readily attained several times.
The displacement pro5les associated with the drift limits of wy = 0:8 per cent and m = 2:5
per cent in the eighth storey, respectively, are plotted in Figure 6(d). The hypothetical displacement of the frames only at the attainment of the uniform storey yield drift of fy = 0.0117,
Copyright ? 2001 John Wiley & Sons, Ltd.
594
T. PAULAY
i.e. approximately 1.2 per cent in every storey over the height of the building, is shown in
Figure 6(d) by the straight dashed line. This indicates that at the development on the maximum storey drift, 2.5 per cent, all frames would have entered the inelastic domain in every
storey. The maximum storey displacement ductility demand in the eighth storey of the frames
would be of the order of f Q = 0:025=0117 = 2:14, a very moderate quantity.
Another performance criterion may require that, irrespective of drift limits, the displacement ductility demand on a wall should not exceed its ductility capacity, say wQ = 5,
i.e. Qwmax = 5 0:0034h = 0:017h. The inelastic rotation of the wall at this stage would be
wpmax = 4 0:0034h=0:602h = 0:0226 rad , leading to a maximum drift in the eighth storey of
max = 0:0081 + 0:0226 = 0:0307 rad .
5.5. Sti+ness and ductility relationships
By similarity to the example structure presented in Figure 4, the idealized bi-linear force
displacement responses of the wall and frame components of the dual system, can be superimposed. The nominal yield displacement of the walls (0:0034h) was de5ned in Section 5.3.
For the estimation of the nominal yield displacement of the condensed frame element, some
judgement is required. Figure 6(d) suggests that a displacement in the vicinity of the eighth
storey of 0:0075h, is a reasonable estimate for this. In terms of system response, errors involved in this estimate are negligible. These quantities lead then to the bi-linear simulation of
the system response. The results are presented in Figure 7. In accordance with the de5nition
of component sti@ness by Equation (2), the following values are obtained for the condensed
elements:
k walls = 0:65Vb =(0:0034h) = 191Vb =h
k frames = 0:35Vb =(0:0075h) = 47Vb =h
Copyright ? 2001 John Wiley & Sons, Ltd.
595
This leads to the total system sti@ness K s = 238Vb =h, which allows the nominal yield displacement of this dual system to be de5ned as
(17)
Qy = Vin = ki = Vb =Ks = (1:0=238)h = 0:0042h
The idealized responses of the elements and the system are seen in Figure 7.
Probable non-linear monotonic forcedisplacement responses are also indicated in Figure
7 by the dashed line curves. These relationships are often obtained by pushover analysis
techniques. It is claimed, that bi-linear modelling used in this study is equally capable to
predict displacement limits associated with any performance criterion speci5ed for a building
to be designed. Because no allowance in the modelling used was made for strain hardening
e@ects, and hence for post-yield sti@ness, corresponding procedures to estimate the required
system strength, are likely to be conservative. It appears that pushover analysis are unnecessary
for the purpose of verifying the ductile response of dual systems which are being designed
using the postulated features.
As Figure 7 illustrates, a drift limit of 2.5 per cent in the eighth storey, limits the lateral
displacement at that level to 0:0136h units. The ductility demand on the frames is of the
order of f Q 13:6=7:5 1:8: The system displacement ductility, used in force-based seismic
design procedures, should, therefore, be limited to Q = 13:6=4:2 3:2. Increased displacement ductility capacities, associated with the assumed (maximum) displacement capacity of
Qwm = 0:017h (at level he ) of the walls, are also shown in Figure 7.
5.6. The analysis of dual systems
Once the the required base shear intensity has been established, analysis of the structure may
be undertaken, if necessary. In compliance with Equation (2), strength must be known before
sti@ness can be quanti5ed. To this end the nominal Mexural strength, Min , of each component,
as detailed for construction, needs to be evaluated. For use in standard software, the Mexural
rigidity of each component can then be de5ned as
Ec Ie = Min =iy
(18)
596
T. PAULAY
other hand, are protected against the development of storey shear forces in excess of those
evaluated from capacity design considerations.
When frames comprise beams with long spans, their strengths are likely to be dominated
by gravity load demands rather than those resulting from seismic actions. In such cases, it
may be diPcult to enforce a strong columnweak beam frame system. However, as storey
displacements are controlled by walls, so-called soft storeys should not develop. Accordingly,
storey mechanisms and strength may utilize plastic hinges in columns only. Corresponding
nominal storey yield drifts in such frames can be readily determined by a similar approach
as that used in establishing Equation (11).
There are possibilities in the assignment of seismic strength to the two types of elements,
other than that chosen in this example and shown in Figure 6. For example, a designer may
follow traditional techniques [4; 11] related to elastic systems. These are based on strength
being proportional to predetermined sti@ness. Thereby the analysis designates larger storey
shear forces to the frames in the vicinity of the mid-height of the building. Such a design
solution may produce storey shears for the condensed frame, as indicated by the dotted stepped
vertical lines in Figure 6(c). Accordingly, the condensed frames could be designed to resist
30 per cent of the total base shear in the lower and upper 4 storeys and 45 per cent between
levels 5 and 9. As Figure 6(b) shows, this hardly a@ects the moment pattern applicable to
the walls. Therefore, e@ects on displacement limitations, which are controlled by the walls,
would be negligible.
Special attention in the design of dual systems needs to be paid to diaphragm actions. Contributions of inertia forces to bending moments and shear forces generated in Moor diaphragm
change at every Moor. Particularly, severe demands will arise at roof level where forces introduced to the walls and frames act in opposite directions. The magnitudes of diaphragm
moments and shear forces are controlled by the positions of the walls in relation to those of
the frames. Moreover, owing to Moor accelerations, diaphragm actions, derived from equilibrium considerations of static forces only (Figure 6(c)), must be expected to be signi5cantly
ampli5ed at di@erent levels. Slender diaphragms and those using prefabricated elements, with
possibly inferior connections to the lateral force resisting elements, may not satisfy the initially made assumption of in5nite rigidity. This may then result in frame displacements larger
than those, based in this study to be identical to those of the walls. As a result, performancedictated drift limits for frames may be violated. However, because of the inevitably reduced
displacement ductility demands on frame elements, as seen in Figure 6(d), excess frame deformations due to diaphragm Mexibility are not likely to jeopardize the integrity and nominal
strength of frames.
6. CONCLUDING REMARKS
To complement research e@orts, addressing earthquake-imposed displacement demands, there
is also a need to be able to evaluate displacement capacities. These are limited by speci5c
performance criteria, applicable to structural systems .
Bi-linear modelling of forcedisplacement relationships, enables displacement ductilities of
elements of ductile reinforced concrete systems, as well as the system, to be estimated and
related to each other.
Copyright ? 2001 John Wiley & Sons, Ltd.
597
Yield deformations are primarily functions of the geometry of components and the yield
strain of the reinforcing steel used. Contrary to traditionally used concepts, such deformations
are properties that are essentially independent of the nominal strengths of components.
The 5ndings lead to the precept that, within rational limits, fractions of the required seismic
system strength to lateral force-resisting elements may be assigned arbitrarily. The astute
designer can advantageously exploit this freedom of choice.
Nominal strengths are developed when strength-independent nominal yield displacements
of components are attained or exceeded.
Instead of being an a priori quantity, the sti@ness of a component is proportional to the
nominal strength eventually assigned to it.
As nominal yield displacements are independent of strength, displacement limits for ductile
systems, dictated by performance criteria, can be estimated with a high degree of precision
before the design proper of the system commences. To enable adverse seismic behavioural
features to be avoided, this feature permits designers to introduce changes in the geometry
or position of elements during the preliminary study of the proposed structural system for its
suitability.
A step-by-step study of a framewall dual system was presented to facilitate the appreciation
and exploitation of postulated concepts, the use of which should enable displacement and
ductility limits to be very simply established.
The procedure does not address estimations of probable earthquake-imposed displacement
demands.
Apart from performance-dictated drift limits, the displacement capacity of critical elements,
i.e. those with the smallest yield displacements, should control the maximum acceptable displacement demand to be imposed on the system.
With a more realistic estimate of system sti@ness, the procedure can be readily incorporated in currently used force-based design approaches. However, with the ability to predict
strength-independent displacement limits, the postulated approach should particularly appeal
to displacement-based design strategies [14].
APPENDIX: NOMENCLATURE
Arb
Arw
CM
D
Db
Dc
Dw
Ec
Ie
h
he
k
Ks
p
598
M
My
Mn
r
Vb
Vn
Vs
Vy
T. PAULAY
Subscripts
f
i
w
a frame element
component i
a wall component
Greek symbols
m
fy
wy
Qm
Qwm
Qwp
Qwy
Qy
Qy
y
Q
iQ
wp
wy
y
y
1. Priestley MJN. Myth and fallacies in earthquake engineeringconMicts between design and reality. Recent
Developments in Lateral Force Transfer in Buildings, American Concrete Institute (SP-157), 1995; 231 254.
2. Paulay T. Some design principles relevant to torsional phenomena in ductile buildings. Journal of Earthquake
Engineering 2000; 5(3):273 308.
3. Paulay T. Simplicity and con5dence in seismic design. The Fourth MalletMilne Lecture. Wiley: Chichester,
1993.
4. Paulay T, Priestley MJN. Seismic Design of Reinforced Concrete and Masonry Buildings. Wiley: New York,
1992.
5. Priestley MJN, Kowalsky MJ. Aspects of drift and ductility capacity of rectangular structural walls. Bulletin of
the New Zealand Society for Earthquake Engineering 1998; 31(2):73 85.
Copyright ? 2001 John Wiley & Sons, Ltd.
599
6. Paulay T. A re-de5nition of sti@ness of reinforced concrete elements and its implications in seismic design.
Structural Engineering International 2001; 11(1):36 41.
7. Paulay T. A simple displacement compatibility-based design strategy for reinforced concrete buildings. 12th
World Conference on Earthquake Engineering (12WCEE), Auckland, New Zealand, 2000; Paper No. 0062.
8. Paulay T. Torsional mechanisms in ductile building systems. Earthquake Engineering and Structural Dynamics
1998; 27:11011121.
9. Goodsir WJ, Paulay T, Carr AJ. A design procedure for interacting wallframe structures under seismic actions.
Eighth World Conference on Earthquake Engineering, San Francisco, vol. V. 1984; 621 628.
10. Paulay T, Goodsir WJ. The capacity design of reinforced concrete hybrid structures for multistorey buildings.
Bulletin of the New Zealand Society for Earthquake Engineering 1998; 19(1):117.
11. Paulay T. A seismic design strategy for Hybrid structures. Fifth Canadian Conference on Earthquake
Engineering, Ottawa. A.A. Balkema: Rotterdam, 1986; 3 25.
12. Park R, Paulay T. Reinforced Concrete Structures. Wiley: New York, 1975.
13. Priestley MJN. Brief comments on elastic Mexibility of reinforced concrete frames and signi5cance in seismic
design. Bulletin of the New Zealand Society for Earthquake Engineering 1998; 31(4):246 259.
14. Priestley MJN. Performance based seismic design. 12th World Conference on Earthquake Engineering
(12WCEE), Auckland, New Zealand, 2000; Paper No. 2831.