Professional Documents
Culture Documents
Pgina 1 de 26
page_229
Page 229
9
Modelling of Structures Subjected to Seismic Loading
T.P.Tassios
Laboratory of Reinforced Concrete, National Technical University of Athens, Greece
1 INTRODUCTION
Although considerable advances have been made in the use of analytical methods to predict the seismic performance of structures,
increasing use is being made of scaled reinforced concrete models in research and design.
Seismic behaviour is complex and the techniques of the associated modelling are continuously developing. This chapter, therefore,
draws upon recent research developments.
It is worth considering some details of seismic design. During a seismic event, all structural components are subjected to
deformations due to a sequence of horizontal and vertical displacements acting on the foundation of the structure. Critical sections
may yield, while the stiffness and hysteretic damping can be changed dramatically during the excitation. Thus, the characteristics of
motion (such as natural period and induced acceleration) can be continuously altered. Also, considerable response degradation due
to cycling may occur, thus reducing the resistance of several critical regions. Yet, the overall seismic behaviour of a structure may be
considered as satisfactory, provided that the structure does not collapse under the design earthquake. In this respect, the provision of
sufficient ductility is a fundamental requirement in seismic engineering. Seismic design forces are calculated on the basis of the
expected overall ductility of the structure and a simplified expression of these forces acting
page_229
page_230
Page 230
on an elastoplastic structure is given by
(1)
where W denotes the total weight of the superstructure and c, the base shear coefficient. This coefficient may be written as
(2)
where s(T)=normalised acceleration, from an elastic response spectrum, corresponding to the natural period T of the structure; and
q=the behaviour factor, reflecting mainly the overall ductility of the structure. In Table 1 the conditions affecting this behaviour
factor are listed.
Table 1
Conditions affecting the seismic behaviour factor of RC structures
No.
Conditions
1 Ductility of critical
regions
Parameters
1, 1 =percentages of longitudinal reinforcement
1:1 =their ratio
w =mechanical volumetric percentage of confining reinforcement
v =normalised axial compression force (N:Acfc)
s =M:Vd=shear ratio
VR,V:VR,M =ratio of ultimate shear force resistance values, corresponding to shear or flexural
failure modes
file:///C:/Users/hp%20home/AppData/Local/Temp/~hhAD49.htm
20/07/2016
page_229
3 Input accelerogram
Pgina 2 de 26
Redundancy
Regularity of mass, stiffness, and overstrengths distribution
Availability of secondary resistance
Extent of imposed post-yield displacements (d:dy)
Number of large amplitude full reversals (n)
A reliable simulation of seismic behaviour of real structures should take account of these factors as well as the usual factors, such as
the value of the elastic modulus, etc. Ductility similitude must be retained as an additional requirement in model design (see section
4.2).
It is obvious that the complexity of dynamic behaviour reduces the ability of analytical methods to describe this behaviour.
Analytical
page_230
page_231
Page 231
modelling is not an exact science.1 It is possible, however, to improve analytical modelling of the behaviour of structures subjected
to static loading by direct comparison with the behaviour of real structures.
This is not the case for structures subjected to seismic effects in which a large number of factors influence response and full-size
testing is not realistic. Model testing, therefore, offers a solution in the confirmation, modification or calibration of analytical
solutions. It seems, however, that in structural engineering a step-by-step procedure has been adopted consisting of the calibration of
a less costly method of model analysis against a more costly one, for example:
Analytical methods versus small-scale dynamic models.
Small-scale dynamic models versus large-scale (or full-scale) models, tested under pseudo-dynamic conditions.
All models versus fully instrumented real structures, subjected to real earthquakes.
It is believed that in the above framework, small-scale dynamic modelling will be much more important than in the past.
page_231
page_232
Page 232
tions, whereas models in the second category (mainly small-scale) are more frequently tested under dynamic conditions. There is not
an exclusive connection, however, between models scale and testing method. Technical and economical restraints dictate each time
file:///C:/Users/hp%20home/AppData/Local/Temp/~hhAD49.htm
20/07/2016
page_229
Pgina 3 de 26
page_232
page_233
Page 233
A lumped mass model of the structure is assumed with a finite number of degrees of freedom. The seismic motion of a planar
structure may be described by the following equation:
(3)
where M=the mass matrix
=the nodal acceleration vector
C=the damping matrix
=the nodal velocity vector
=the vector of the response-forces of the structure (restoring forces)
g=horizontal ground acceleration, parallel to the plan of the structure given by the input accelerogram to be considered.
If a stiffness matrix K can be formulated by means of a standard finite element procedure, then
(4)
where
denotes the displacement vector, and the equations are solved by means of a step-by-step integration. Inaccuracies in the
prediction of the performance of the structure caused by non-linear behaviour may be overcome by means of a hybrid
(analytical/experimental) method. An elastic solution is adopted for the initial (small amplitude) part of the motion and the
corresponding displacements are computed on the basis of the input accelerogram. These displacements are quasi-statically applied
file:///C:/Users/hp%20home/AppData/Local/Temp/~hhAD49.htm
20/07/2016
page_229
Pgina 4 de 26
to the model at appropriate points by electrohydraulic actuators linked to the computer. Corresponding forces are simultaneously
measured, thus providing the
vector. A new analytical solution is found for the next time interval, and hence new displacements
are imposed on the model. The method is called the pseudo-dynamic testing method (PD method). The method has been used by,
among others, Mahin et al.4 and a schematic arrangement of the implementation scheme is given in Fig. 1.
For structures with a predominant first mode, e.g. wall systems with a moderate number of floors, the technique may be simplified to
a single degree of freedom pseudo-dynamic (SPD) method.
The advantages of this method are evident. In fact, the absence of any really dynamic phenomenon is of a great importance from a
mechanical and a functional safety point of view. Conventional equipment, such as that used in quasi-static testing, is suitable. In
addition, the low rate or even intermittent testing procedure allows for a direct observation of gradual damage. The most interesting
feature of the pseudo-dynamic
page_233
page_234
Page 234
page_234
page_235
Page 235
3.3.1 Free vibration tests (pull-back and sudden release)
The model performs free vibrations about its static equilibrium position. The natural period can be directly determined from the
recorded cycles. Damping is also calculated from the ratios of successive decayed vibrational amplitudes.
3.3.2 Forced vibrations
Counter rotating eccentric weight exciters may be mounted on top of large scale models, in order to produce steady state sinusoidal
excitations with a constant period. The corresponding response amplitudes are measured. The period is subsequently adjusted to
another value and the measurements are repeated. Thus, a resonance curve is plotted depicting the measured amplitudes for the
file:///C:/Users/hp%20home/AppData/Local/Temp/~hhAD49.htm
20/07/2016
page_229
Pgina 5 de 26
whole range of imposed periods. From such a curve the natural period and the viscous damping can be obtained. By increasing the
magnitude of the exciting force, some non-linear characteristics may also be studied.8
Mode shapes and accelerations at several levels may also be investigated, especially using the forced vibration method.
Free vibration tests are currently used in order to study the condition of small-scale models before and after their testing on
earthquake simulators. The gradual increase of their natural period after consecutive excitations of increasing intensity, is very easily
followed by means of free vibration tests. Figure 2 shows the simple test set-up used on the 1:10
Fig. 2. A 1:10 RC frame-wall model is tested by means of free vibrations before and after testing on an earthquake simulator.9
page_235
page_236
Page 236
page_236
page_237
Page 237
An extremely large variety of facilities of this kind are in use, starting from very small one-degree-of-freedom tables, up to
sophisticated very heavy six-degrees-of-freedom simulators.
The following main components of large simulators can be distinguished (Fig. 4):
(a) The platform. Normally a steel structure on which models will be fixed (Fig. 5). Considerable stiffness is needed. Natural
frequencies as remote as possible from the operating frequency range are desired.
Fig. 4. General layout of a large earthquake simulator (3-DOF system, Julich, FRG).
file:///C:/Users/hp%20home/AppData/Local/Temp/~hhAD49.htm
20/07/2016
page_229
Pgina 6 de 26
(b) The static support system. When motionless, the platform may be supported on vertical actuators or on air-cushions or on a liquid
mass (oil or water).
(c) The driving system. Several actuators (depending on the degrees of freedom of the simulator) are appropriately attached and are
usually actuated by hydraulic power (although mechanical systems are sometimes used). Pressure accumulators are needed in order
to maintain the required pressure when peak flow is demanded.
(d) The reaction system. A monolithic massive reinforced concrete structure surrounds the whole system. Its mass is many times
larger than the total table and model mass. Springs and dampers may
page_237
page_238
Page 238
Fig. 5. Earthquake simulator of the Nat. Tech. University of Athens (6 degrees of freedom, design MTS/Carydis).
further improve the conditions of the systems isolation from the environment.
(e) The control system. The motion of the platform along, or around the axes of available freedom may be controlled by mechanical
or, normally, by electrical servo-systems. Displacement, velocity and acceleration are continuously controlled. Analogue facilities
are normally used to receive the input signals and a feedback system is
page_238
page_239
Page 239
used. Modern installations also include control systems to correct possible motion errors.
(f) The data acquisition system. Appropriate multichannel systems are used.
The following features of an earthquake simulator are of interest:
(i) Geometrical data: table size and maximum height of centre of gravity of the specimen.
(ii) Maximum weight of specimen (without substantial modification of the performances of the simulator).
(iii) Degrees of freedom (total and simultaneously controlled).
(iv) Operating frequency range.
(v) Maximum values of acceleration, velocity and displacement. For a given operating frequency and a given specimens weight,
however, the available motion characteristics are modified according to a performance spectrum (Fig. 6).
Fig. 6. Performance spectrum of the SAMSON Earthquake simulator (3-DOF system, HRB Julich, FRG). 1=empty platform; 2=150
KN payload; 3=250 KN payload.
A world-wide list of earthquake simulators with a model capacity greater than 50 KN is given in the Appendices.
4.2 Basic similitude requirements
It is beyond the scope of this section to repeat the fundamentals of structural similitude. Chapter 2 of this book constitutes the best
reference
file:///C:/Users/hp%20home/AppData/Local/Temp/~hhAD49.htm
20/07/2016
page_229
Pgina 7 de 26
page_239
page_240
Page 240
in this respect for static loading. Similitude requirements for the specific applications in earthquake engineering, however, are
extensively described in Moncarz & Krawinkler.11
The essential data needed for the design of small-scale dynamic models, however, will be reproduced here, as an introduction to
section 4.4.
The dynamic behaviour of a model is fully described by means of the following three basic quantities: length (L), mass (M) and time
(T).
Only three independent scales can be selected when designing a model. Nevertheless, depending on practical necessities, other
quantities such as stress or acceleration, etc., may be selected for scaling, provided that the total number of independently chosen
scales will again be equal to three. Subsequently, the scales of all the other parameters are expressed in terms of the basic scale
factors chosen, by means of the equations governing the phenomenon, e.g. Table 2, column 3.
Such a complete observation of similitude requirements (termed a true replica model) is almost impossible in earthquake
engineering. Post-yield phenomena of concrete models are governed by the descending part of the stress-strain curve of concrete in
compression, by its cyclic response degradation and by the steel concrete bond characteristics. Consequently, the use of
microconcrete as a model material becomes imperative. The stress and stiffness scale-factors are taken equal to unity (S=SE=1) and
the similitude requirement for density
(5)
cannot be satisfied: since microconcrete has been selected as the model material, the S should be equal to unity.
Reinforced concrete small-scale models in seismic analysis cannot be true replica models. Instead, in order to reproduce intertia
effects, additional masses are provided at appropriate points on the model and are loosely attached so that the stiffness and strength
of members are not altered. These dynamically effective masses are decoupled from the density similitude requirement, but of
course they will be appropriately calculated (see section 4.4).
These additional masses are also acting vertically on the model (gravity forces). In many cases, the level of action-effects due to
gravity forces is of paramount importance for the seismic behaviour of the structure: The available ductility of RC columns depends
very much on their axial load, Table 1. Besides, the non-seismic loads acting on beam-ends on both
page_240
page_241
Page 241
Table 2
Similitude relationships11
Scaled entities
True replica
Model type
Artificial mass simulation
(3)
SL
(4)
SL
(1)
Length displacement
(2)
SL
Time
ST
SL
Vibrational period
ST
SL
Velocity
Sv
Gravitational acceleration
Sg
Acceleration
file:///C:/Users/hp%20home/AppData/Local/Temp/~hhAD49.htm
(neglected)
20/07/2016
page_229
Pgina 8 de 26
SA
Density
SE:SL
Strain
Stress
SE
SE
Modulus of elasticity
SE
SE
SE
Force
SF
Energy
aWhen
SW
SE
SE
SE
SE
sides of a beam column joint, predetermine the amplitude of local shear force reversals. The seismic behaviour of these beam-ends
and joints is adversely affected by eventual full reversals. Consequently, these gravity force action-effects should also be present in
the model. Thus, an appropriate distribution of the additional masses should be achieved, (section 4.4).
In other cases these gravity forces may be neglected without substantial error. An example of this situation occurs in the case of a
building whose lateral resistance is primarily secured by means of reinforced concrete walls. It is also the case for nuclear reactor
containments where the effects
page_241
page_242
Page 242
of gravity are rather small compared to internal pressure and severe seismic action. The acceleration scale factor is relaxed and the
condition SA:Sg=1 is not fulfilled.
4.3 Modelling materials
The use of microconcrete in small-scale seismic models constitutes the only available practical solution. The reinforcement is
simulated by ribbed small diameter steel wires.
This direct limitation of the composition of prototype reinforced concrete is dictated by the complicated post-yield cyclic
degradation phenomena which govern the seismic behaviour of RC structures.
The basic information on these modelling materials is given in Chapter 3 and Chapter 4 of this book. A short account is also given
below, however, together with a discussion on the effectiveness of simulation of several mechanical properties of the prototype
materials.
4.3.1 Microconcrete
Production. Cement content should be kept as low as possible (for the same design compressive strength) in order to reduce the
tendency of the model to exhibit higher shrinkage than prototype. In this respect it is useful to recall a practical rule. For each 1-mm
decrease of the maximum aggregate size between 25 and 5 mm, a 2 kg/m cement increase should be provided, for constant
compressive strength under constant slump.
On the other hand, the maximum aggregate size (2560 mm) follows, approximately, the overall geometrical scaling in order to
match possible variation in size of aggregates across the wall sections and clearances.
Slightly higher values of maximum size are suggested, however, in order to decrease the discrepancies observed in tensile strength
and shrinkage between model and prototype concrete. Usually, three aggregate classes are mixed to produce the desired gradation.
Full similitude in grading curves should be avoided, however, in the area of very fine materials: It is suggested that the amount of
aggregate passing the 015 mm sieve is limited to less than 10%.
Additives may also be used, but their effect on bond and shrinkage should be carefully studied.
file:///C:/Users/hp%20home/AppData/Local/Temp/~hhAD49.htm
20/07/2016
page_229
Pgina 9 de 26
Extensive trial mixes are suggested, followed by various measurements of mechanical properties, such as compression, tension,
bond, friction and shrinkage tests, after realistic curing conditions. Since a complete repro-
page_242
page_243
Page 243
duction of all prototype concrete properties is not feasible, an appropriate optimisation has to be made. It is worth noting that ageeffects should be carefully taken into account, since microconcrete seems to gain strength with age more rapidly than ordinary
concrete.
Finally, the curing of the model has an important role in determining the mechanical properties. It is suggested that the casting
sequence of the prototype frame is reproduced (e.g. storey by storey). Furthermore, thin microconcrete model elements could be
subject to a higher shrinkage and this should be taken into account during the curing process.
Compression simulation. In order to secure equal modulus of elasticity, equal strength and equal post-elastic behaviour, identical
cc curves are needed for model and prototype concrete. Experience shows that such a similitude is possible. Some eventual
shifting of the peak of the stressstrain curve, however, is to be expected. The problem of size-effect should also be considered: the
target-strength of microconcrete should be measured on control specimens which are tested under machine head-blocks of
appropriate stiffness. Finally, possible differential rate-effects should be considered. Figure 7 provides a summary of results of
several studies both on normal and small-aggregate concretes. Since the experimental condi-
Fig. 7. Strain rate effects on compressive strength fc and on strain at peakstress of concrete, co.
page_243
page_244
Page 244
tions might not be comparable, there is considerable doubt in the amount of the differences in sensitivity shown in the figure.
Strain rate effects could help to explain the differences in response of models and prototypes. For instance, the magnitude of the
strain co at peak stress of concrete under compression depends upon strain rate (Fig. 7). Little experimental evidence seems to be
available which might be related to microconcrete.
Finally, Poissons ratio at 40% strength-level does not seem to be substantially influenced by rate-effects. A 15% decrease is
expected from 103 to 101 s1 strain rate.12
Tension simulation. Normally, a microconcrete has higher tension strength than its prototype, for equal compressive strength. This is
due to the higher cement content needed to counterbalance the higher amount of water used to achieve equal workability. Thus,
every effort should be made to decrease this difference in tensile strength which has an effect upon bond properties. Cracking strain
at failure in tension seems to increase by say 50% for a mean size of aggregate decreasing from 8 to 2mm.13
Regarding size-effects, the comments made for compressive control specimens are also applicable to the tensile specimens.
Differential rate-effects for the tensile strength between microconcrete and normal concrete are not known. It should be noted that
these rate-effects for normal concrete are more pronounced in tension than in compression. A 30% increase of tensile strength is
expected from 103 to 101 s1 strain-rate.14 Similarly, a 30% increase of Poissons ratio has been found in tension for the above
strain-rate change.12
Bond simulation. Bond action is of considerable importance for post-elastic phenomena governing the seismic behaviour. Transverse
cracking phenomena are related to stiffness, thus influencing the natural period of the model and its spectral response. Longitudinal
splitting and pull-out resistance are related to anchorage behaviour under cyclic loading.
Bond simulation poses a major problem, and Chapters 3 and 4 of this book offer considerable assistance in this respect. Local bond
stress/slip tests need to be performed, on the smallest possible bonded length, e.g. equal to two maximum aggregate sizes. On the
file:///C:/Users/hp%20home/AppData/Local/Temp/~hhAD49.htm
20/07/2016
page_229
Pgina 10 de 26
basis of such local bond stress/slip curves, computerised predictions may be found for all bond-controlled phenomena, including
cyclic behaviour.15
Comparative rate-effects on bond strength are not known. Even if
page_244
page_245
Page 245
identical model/prototype behaviour is assumed, however, the time-compression frequently needed in seismic modelling (see ST,
Table 2) would lead to an additional bond increase, since, according to Vos et al.16
(6)
where n=07(125s)
,o=local bond stresses corresponding to a local slip, s (mm)
,
Details of the characteristics and the fabrication of deformed steel wires are given in Chapter 4 of this book.
Cold-rolling operations of steel wires can be used in order to increase the yield strength. The ductility lost during this process is
subsequently restored by means of appropriate heat treatment.
Since the most important factor in model reinforcement is bond
page_245
page_246
Page 246
simulation, steel wires need to be deformed by means of appropriate cold-rolling. Large radius knurling wheels are used in order to
produce mechanical deformation of the wires surface. An accurate ratio of rib height to wires effective diameter is sought, together
with appropriate ribs spacing.24,25 Such cold-rolling causes simultaneous yield-strength increase, accompanied by an elimination of
the sharp yield-point and a significant decrease in ductility. Subsequent annealing can bring down the yield strength if needed and
control the adverse change in properties. Acceptable yield-plateaus and strain-hardening forms may be obtained by a combination of
file:///C:/Users/hp%20home/AppData/Local/Temp/~hhAD49.htm
20/07/2016
page_229
Pgina 11 de 26
=104 s1.
page_246
page_247
Page 247
or
(9)
This is the so-called time-compression factor.
(b) Now assume a unit stress scale-factor S=1, because of the fundamental importance of using a model material that exhibits the
same strength and same modulus of elasticity as the full-scale material, section 4.2. Therefore, since =ma:l2,
(10)
and since SA=1,
(11)
This relationship applies to the entire mass of each area of the model and not necessarily to the density of the model material alone.
(c) Select an initial value of geometrical scale-factor SL that is based on approximate criteria such as the bearing capacity of the
earthquake simulator.
All the above findings have been listed in Table 2. Now that the three independent scale-factors have been selected, the dynamic
problem can be fully described. The important consequences of this basic selection now follow:
(i) It is observed that another similitude condition is satisfied. Equal resonance sensitivity is also achieved both for the prototype and
file:///C:/Users/hp%20home/AppData/Local/Temp/~hhAD49.htm
20/07/2016
page_229
Pgina 12 de 26
the model. The ratio of the natural period Tp of the prototype and the predominant period Tpred., p of the input prototype
accelerogram is equal to the relevant ratio for the model. In fact, since stiffness k=(l2):l,
(12)
Therefore
(13)
and
(14)
page_247
page_248
Page 248
This is the time compression factor found in eqn (9). Combining eqns (13) and (14), it is found that
(15)
Therefore, the observance of equal resonance sensitivity is confirmed.
(ii) Another consequence of the correct selection of the independent scale-factors is that we are now able to calculate the additional
masses required on the appropriate areas of the model in order to simulate both the overall mass present and the gravity actioneffects. From eqn (11), the ratio of all masses present on prototype and model may be written as follows:
(16)
where mp, mm denote the masses of the building elements themselves (in prototype and in model, respectively), QE is the vertical
live load which is probabilistically present on the prototype at the moment of the design earthquake, and G is the additional
(artificial) weight that must be applied to the model.
The ratio of dead weights of the structures alone is
(17)
and from eqn (16)
(18)
This expression allows the calculation of the additional weight that must be applied on every area of the model, accounting for the
dead weight G and the vertical live load QE acting on the appropriate area of the prototype.
4.4.2 Models (partly or totally) neglecting vertical force action effects
Two compatibility checks are normally made in the previous case and
page_248
page_249
Page 249
these are:
file:///C:/Users/hp%20home/AppData/Local/Temp/~hhAD49.htm
20/07/2016
page_229
Pgina 13 de 26
Is the total model weight (mmg+G) compatible with the capacity of the available earthquake simulator (see section 4.1)?
Is there any space within the model to properly attach the additional masses to the appropriate areas, (see section 4.4.1)?
If the first check is not satisfied a possible decrease of the geometrical scale-factor could be considered. This is provided that a
second check can be satisfied. Similarly, if the second check cannot be satisfied then an increase in the geometrical scale-factor
could be considered provided that the first check may then be satisfied.
If all these are not satisfied, other possibilities could be explored:
I Gravity forces are totally neglected. This is acceptable in only a few cases and then only as a first approximation (see section 4.2).
For this case the following similitude conditions should be satisfied:
(a) Select the highest possible value of SL, but in doing so remember the difficulties in trying to satisfy the conditions of section
4.4.1.
(b) Adopt a unit stress scale-factor S=1, since materials of equal strength and modulus of elasticity for both the prototype and the
model are used. This unit stress scale-factor cannot reproduce inertia stresses.
(c) Since no additional mass will be applied and identical materials will be used, a density scale-factor defined by
(19)
will be used.
Having selected these three scale-factors, some practical consequences will be discussed.
(i) Since (fully simulated) inertia stresses are expressed as
(20)
or
(21)
it is found that
(22)
page_249
page_250
Page 250
(ii) On the other hand, since
(23)
and using eqn (22), we find
(24)
(iii) It is very important now to note that the foregoing simplification which disregards the gravity force action-effect causes some
difficulty. The time compression factor now (ST=SL) is much larger than in Section 4.4.1. Therefore, in addition to some possible
distortions in test results, strain-rate effects are greatly accentuated (both in the models behaviour and the high frequency
reproduction of the simulator). The acceleration scale-factor has also changed. Instead of SA=1, the value SA=
file:///C:/Users/hp%20home/AppData/Local/Temp/~hhAD49.htm
is required.
20/07/2016
page_229
Pgina 14 de 26
page_250
page_251
Page 251
This value lies between the time scale-factors given in eqns (9) and (21).
At the same time, since
(29)
it is concluded that
(30)
Here again, this value lies between the values found for the cases discussed in sections 4.4.1 and 4.4.2, respectively.
Consequently, a considerable reduction has been made possible on the values of both the necessary time compression factor and the
acceleration amplitude increase needed in model testing. Of course, simulation is still not complete since gravity force action effects
are only partly taken into account.
4.4.3 Model design steps
In Fig. 8 a summarised presentation of design steps is given, together with the necessary checks for the case of gravity forces being
taken into account. Similar steps are followed for the design of models neglecting gravity forces, G=0.
4.4.4 Model dimensioning
In principle, a bar-for-bar substitution of prototype reinforcements is needed for the following reasons:
Bond similitude will be better followed.
Eccentricities mainly in wall sections will be avoided.
Confinement will be better simulated.
file:///C:/Users/hp%20home/AppData/Local/Temp/~hhAD49.htm
20/07/2016
page_229
Pgina 15 de 26
The relationship between action effects leading to flexural and shear failure modes will be better observed.
It has to be admitted that such a proportioning policy is, however, not practical for low scale-factors, possibly for SL=6. If this is the
case, the basic moment and shear ductility simulation is not fully observed.
Particular problems are related to splicing and anchorage. Splicing of longitudinal model reinforcement is not needed and can, in
fact, be avoided unless this is a parameter to be investigated.
Whenever anchorage is expected to be a decisive characteristic of large amplitude cyclic behaviour, a direct scaling down of
prototypes anchorage length may not be the best decision. The improved bond characteristics in the model should be somehow
accounted for.
page_251
page_252
Page 252
Fig. 8. An algorithm for seismic models design, accounting for gravity forces.
A particular problem is posed by the confinement of critical regions of the model elements. Local ductility depends on the available
ultimate strain cu of concrete, which in turn is a function of the mechanical volumetric percentage w of the confining
reinforcement.28 The ultimate strain is given by
(31)
where n denotes the number of intermediate longitudinal bars, other than those at the corners, on each side of a rectangular crosssection which are restrained by a link (Fig. 9).
4.5 Instrumentation
4.5.1 Scope
page_252
page_253
Page 253
page_253
page_254
Page 254
(b) The horizontal and vertical displacements and the rotations of model elements relative to the base.
(c) Accelerations around the model.
(d) Internal angular deformations, strains and crack widths along critical regions of the model.
file:///C:/Users/hp%20home/AppData/Local/Temp/~hhAD49.htm
20/07/2016
page_229
Pgina 16 de 26
The following instrumentation may be used to obtain the desired experimental objectives.
An external frame outside the shaking platform may be needed to record the displacement history of the platform. Accelerometers
are also attached on the platform in order to check the accuracy of reproduction of the input accelerogram.
A rigid reference frame is fixed on the platform, close to the model, (Fig. 10). A horizontal reference framework is also needed if
uplift displacements are also to be measured. A large number of displacement monitoring instruments, e.g. linear-variabledifferential-transformers or linear-potentiometers are used. Several horizontal displacements are measured both in the direction of
motion and perpendicular to it. Similarly, some vertical displacements are also recorded, mainly on upper regions, in order to
measure possible uplifts.
Appropriate accelerometers are attached to several levels of the model, in order to measure horizontal response accelerations,
parallel and transverse to the input motion. Vertical accelerations may also be needed to be measured, even when the input motion
does not contain vertical components. The measurement of localised deformations of models may pose problems because of their
small scale. Internal angular distortions may be measured by arragements like those shown in Fig. 11. Normal strain gauges are used
for strain measurements on wires and microconcrete. Finally, the time history of the width of some selected cracks may be recorded
by means of small type strain gauges. It is very advisable to insert appropriate load-cells (force transducers), when the scale of
the model offers such a possibility, in order to follow the time history of internal forces generated during the seismic motion at some
particular cross-sections, e.g. the columns of an RC building (Fig. 12).
4.5.3 Data acquisition system
It is beyond the scope of this chapter to describe in detail the data acquisition system needed for earthquake simulator tests.
Essentially, all
page_254
page_255
Page 255
Fig. 10. Instrumentation, location and orientation on a 1:10 RC seismic model of a seven-storey building.9
page_255
page_256
Page 256
Fig. 11. Column-footing interface details and diagonal distortion measurement in a 1:5 scale seismic model of a seven-storey
building.7
analogue signals from instruments connected to the model and the platform have to be amplified, recoded, digitised and stored. The
earthquake simulator motion, simultaneously recorded, is used as a time-reference source in order to synchronise records which
might be taken on different recorders. Graphic display and printer facilities are also part of the system.
Motion-picture cameras are used to record all phases of tests.
One of the main characteristics of these data acquisition facilities is their precision and resolution for high speed events.
4.6 Testing procedure
A check-list of steps for testing a seismic model is now given.
file:///C:/Users/hp%20home/AppData/Local/Temp/~hhAD49.htm
20/07/2016
page_229
Pgina 17 de 26
Model visual inspection. Identification of possible initial cracks due to shrinkage, thermal stresses or accidental shocks during
handling.
Check again the stability of fixings.
Check the serviceability of all instruments (small amplitude arbitrary motions).
Horizontal static loadings are applied at several levels of the model
page_256
page_257
Page 257
Fig. 12. Internal force transducer installation in mid-height of a ground-floor RC column of a 1:5 seismic model.7
by means of light jacks, in order to measure the corresponding displacements and check the initial stiffness of the model. A
comparison between calculated and experimentally found initial natural frequency values should be made.
Free-vibration tests (small amplitude) carried out before and after attachment of the additional masses. The platform should be
immovable during these tests. Initial natural periods and viscous damping factors of the model are found.
Check the accuracy of reproduction of the input seismic record.
page_257
page_258
Page 258
After the selection of the appropriate time-compression factor, a given seismic record is used to control the motion of the shaking
table in the form of a displacement-time history. The displacements may be obtained by integrating the accelerograms.
An initial run is performed with properly scaled down acceleration amplitudes, so that the linear behaviour of the model is
observed. During this run, the elastic response of the model is recorded and checked against analytical finding. A good comparison
is expected as proof of satisfactory overall experimental conditions before investigations into the non-linear domain are attempted.
The acceleration-time history of the platform is checked against the input accelerogram. Some discrepancies are expected in peak
acceleration values due mainly to the simulator-model interaction (Fig. 6).
Experiments planned for the specific investigation are performed. Besides instrumental measurements, systematic visual
inspection of the model is carried-out after a seismic test. Crack formation, crack widths, possible microconcrete spalling and the
condition of steel wire at critical regions are recorded and described.
It is worth noting, that strictly speaking, only one non-linear dynamic test can be considered as accurate. Every additional test,
e.g. a second identical run, a scaled-up acceleration amplitude or a run with another input accelerogram is influenced by the lowcycling fatigue induced in the model during its first non-linear excursion. This is particularly so in the case of seismic loading close
to a partial collapse condition.
After each of these planned runs, additional free-vibration tests are carried out in order to assess the general condition of the
model. This may be expressed by the modified natural period and damping.
It is customary to perform a final destructive test of the model by means of disproportionate scaling-up of the input accelerogram.
Sometimes repair of the model and retesting is envisaged (mainly for medium scale models).
4.7 Applications and reliability of small-scale models
file:///C:/Users/hp%20home/AppData/Local/Temp/~hhAD49.htm
20/07/2016
page_229
Pgina 18 de 26
Comprehensive details of the model testing of buildings are to be found in Refs 7, 29, 30. In these papers the dynamic behaviour of a
wall-frame multi-storey building, a five-storey pre-cast panel building and a seven-storey three-framed monolithic building are
discussed.
page_258
page_259
Page 259
A short discussion is now presented on the possible inconsistencies of small-scale RC models tested on earthquake simulators.
4.7.1 Initial conditions
Assuming that material modelling has been achieved as accurately as possible, the following eventual sources of error have to be
accounted for. The importance of the accurate simulation of strain-hardening of prototype and model reinforcement must be
emphasised. The lack of such hardening in some model wires may cause more rapid local deterioration and a lower global ductility
(inability for redistribution of action-effects).
Geometrical errors. Concrete cover of steel wires should be a matter of concern. Standard deviations of more than 1 mm have been
measured,9 for model elements of 8 mm nominal thickness. Post-testing systematic checks are recommended before the final
evaluation of experimental findings.
Output accelerations. The base accelerations induced to the model (i.e. the accelerations measured on the platform) differ from the
input signals, due to simulation errors and to modelsimulator interaction. Local differences of the order of 20% at peak values are
not rare. The overall damage potential of the two accelerograms may be considered as sufficiently equal if, however, the intensity
coefficient
is practically identical.
4.7.2 Model prototype basic dissimilarities
Some of these dissimilarities seem to be inevitable but they have to be considered when evaluating test results.
Cracking pattern. In general, models exhibit fewer cracks than prototypes tested pseudo-dynamically. In areas of low steel ratio, the
higher tensile strength of microconcrete may result in a considerable delay in the formation of first cracking. Sometimes, cracking
occurs together with yielding. Thus, non-yielded regions may remain uncracked. A longer yield plateau of the models
reinforcement may also be the reason for fewer cracks in model areas further from critical cross-sections.
Strain-rate effects may further accentuate bond dissimilarities between
page_259
page_260
Page 260
models tested on earthquake simulators and real full-scale structures or full-scale models tested pseudo-dynamically. A small bond
increase in models has been recognised in section 4.3.1 due to strain-rate effects. In addition, bond-stress relaxation,31 can hardly
occur under these very high rate conditions. Thus, a further increase of bond resistance occurs. It is worth noting here that such a
model-to-prototype dissimilarity is observed even in the case of equal-scale structures made of identical materials.32 Thus,
explanations other than strain-rate effects may be less important.
Now, whatever the reasons for such cracking dissimilarity may be, the consequences should be taken into consideration. Both
flexural and shear stiffness of models tested on earthquake simulators are higher than in prototypes. However, this difference is not
clear during the quasi-elastic loading conditions (less pronounced cracking due to loading, counter-effects of initial shrinkage cracks,
etc). Besides, more concentrated rotations at a models critical cross-sections, may lead to more rapid steel-hardening.
Axial force simulation. For several reasons, vertical RC building elements under large amplitude seismic reversals are very sensitive
to the actual axial forces. An increase of compressive axial forces leads to higher stiffness and thus to higher flexural and shear
action-effects. It is not always certain whether the corresponding increased flexural or shear strengths are high enough to cover the
file:///C:/Users/hp%20home/AppData/Local/Temp/~hhAD49.htm
20/07/2016
page_229
Pgina 19 de 26
higher action effects. Higher axial compression leads to disproportionately lower ductilities. It is, therefore, very important to secure
the highest possible accuracy of axial force reproduction. To this end, gravity forces should in general be fully reproduced.
Similarly, all possible membrane and three-dimensional effects influencing axial forces, should also be simulated. Planar frames
detached from a real building can hardly fulfill this requirement.
Strength increase due to strain-rate effects. It has been observed in Section 4.3 that model concrete and less clearly, steel wire,
exhibit a differential sensitivity in strain-rate effects. In any event, small-scale seismic models under time-compression factors which
are considerably higher than unity are expected to be stronger, except perhaps for possible localised failures.
Localised deterioration events. Localised failure phenomena which are so critical for the final level of available ductility of a
structure can hardly be simulated. The basic dissimilarities already discussed and the over-
page_260
page_261
Page 261
sensitivity of these phenomena to local detailing, may explain this difficulty. Therefore, a detailed observation of the damage to the
model is needed at every step in order to attribute a specific result to normal global response or to localised deterioration. In the
latter case, parallel seismic experiments of isolated critical building elements, tested under larger scale, may offer a better insight.
For such tests, the response history of the element obtained from an earthquake simulator test of model structure, can serve as the
loading history applied to the isolated element.11.
Other seismic components. Under some conditions, the response of a structure to a given horizontal seismic excitation is
considerably influenced by a simultaneous transverse and vertical excitation. Unless a more complete input excitation is feasible the
disadvantage of one-direction time histories should be recognised.
4.8 The potential of small-scale models
In spite of the limitations described in the previous section, small-scale RC models tested on earthquake simulators remain a
powerful tool for the realistic prediction of the seismic behaviour of given structures. The use of small-scale models together with
pseudo-dynamic tests should be continued. Many areas not considered by seismic codes may be investigated using experimental
methods. Examples of these are:
Assessment of the limits of mass and stiffness irregularities and
Fig. 13. Force-displacement relationship for three seismic model frames tested under several input accelerograms. An overall
ductility factor equal to 35 seems to be suggested.9
page_261
page_262
Page 262
overstrengths, if equivalent static or even linear dynamic analysis is to be acceptable from a public safety standpoint.
The contradictory role of infill walls in RC frame structures can be parametrically investigated.
Realistic values of behaviour factors can be sought (see Section 1). In this respect small-scale models seem to be quite reliable
(Fig. 13).
Hybrid design methods can be implemented for unusually complex structures or for structures at high risk.
The consequences of foundation behaviour (rocking and sliding) on the overall seismic response can be studied. Few
investigations have been carried out in this area.
ACKNOWLEDGEMENT
file:///C:/Users/hp%20home/AppData/Local/Temp/~hhAD49.htm
20/07/2016
page_229
Pgina 20 de 26
The Author of this chapter acknowledges the bibliographical assistance and the fruitful discussion offered by Professor P.Carydis.
page_262
page_263
Page 263
APPENDICES
Earthquake simulators (with model weight more than 50 KN) around the world (courtesy of P. Carydis)
Appendix 1. Tables in the European Community
No.
1
2
Institution
Construction Table Specimen Controlled Frequency Acceleration Maximum Table motion Country
year
size (m) weight degrees of range (Hz)
(g)
velocity displacement
(KN)
freedom
(cm/s.)
(cm)
CEN VESUVE
3131
200
1
0200 H 18
1000
100 France
Saclay
CEN
2020
100
2
0200 H
2200
125 France
TOURNESOL,
Saclay
V
1200
100
SOPEMEA,
3030
100
2
050
H 30
600
110 France
VelizyV 30
600
110
Villacoublay
HRB GmbH
5050
250
3
05100 H 15
1000
200 Germany
SAMSON,
Julich
H 15
1000
200
V 15
1000
200
National
1983
4040
100
6
0160 H 15
890
100 Greece
Technical
University,
H 15
890
100
Athens
V 29
850
100
Ansaldo
1980
3535
70
2
0160 H 13
860
70 Italy
Impianti,
Geneva
H 063
550
70
ISMES,
1984
4040
300
6
0120 H 30
550
100 Italy
Bergamo
H 30
550
100
V 20
450
100
ENEA,
1985
4040
100
3
0550 H 30
500
125 Italy
Rome
H 30
500
125
V 3.0
500
125
page_263
page_264
Page 264
No.
Institution
9 ELCOM
SYSTEM,
Perugia
Construction Table size Specimen Controlled Frequency Acceleration Maximum Table motion Country
year
(m)
weight degrees of range (Hz)
(g)
velocity displacement
(KN)
freedom
(cm/s.)
(cm)
160160
1500
1
0520 H
15 Italy
file:///C:/Users/hp%20home/AppData/Local/Temp/~hhAD49.htm
20/07/2016
page_229
Pgina 21 de 26
10 Laborat.
National
de Engen.
Civil,
Lisbon
11 NEL,
Glasgow
245145
50
3030
200
12 GEC,
Whetstone
13 GEC Power
Engineering,
Stafford
14 University
of Bristol
38532
250
4343
1000
3030
150
1987
100 Portugal
033
H
H
V
125 UK
125
125
UK
31000
75 UK
100
V
H 10
H 10
V 10
500
500
500
75
150 UK
page_264
page_265
Page 265
Appendix 2. Tables in the North and South America
No.
4
5
Institution
Construction Table size Specimen Controlled Frequency Acceleration Maximum Table motion Country
year
(m)
weight degrees of range (Hz)
(g)
velocity displacement
(KN)
freedom
(cm/s.)
(cm)
University of
1968
365365
45
1
050
H 50
381
51 USA
Illinois,
Urbana
University of
1971
6161
450
5
0550
H 067
635
152 USA
California,
Berkeley
V 022
254
51
Construction
1973
365365
60
5
01200
H 15
813
73 USA
Engineering
Research Lab.
V 30
686
35
Champaign,
IL
Westinghouse
1978
4949
178
3
0100
H 35
762
305 USA
Pittsburg, PA
V
762
305
University of
1978
3030
68
1
050
H 10
635
75 Canada
British
Columbia
Union
1980
183183
340
2
0120
H 025
305
193 USA
Carbide Corp.
Oak Ridge,
V 025
305
193
TN
E.G. & G.
1981
3030
100
5
030
H 10
635
152 USA
Idaho Falls,
V 05
318
76
10
State of
1983
37x37
180
2
0.160
H 10
760
150 USA
University
of New York,
V 10
500
75
Buffalo
file:///C:/Users/hp%20home/AppData/Local/Temp/~hhAD49.htm
20/07/2016
page_229
Pgina 22 de 26
9 Wyle
Laboratories,
Huntsville, AI
10 Wyle
Laboratories,
Huntsville, AI
11 CERL US
Army
1972
06x06
890
0100
H 150
2000
610 USA
1973
0606
890
0100
H 100
1520
280 USA
1973
37x37
53
0200
H 200
890
152 USA
1978
2418
89
0100
V 40.0
H 06
890
12 University of
Washington
13 University of
Mexico,
Mexico City
76
76 USA
1975
4525
200
0150
H 12
381
51 Mexico
page_265
page_266
Page 266
Appendix 3. Tables in Japan
No.
Institution
1 Kajima
Institute of
Construction
Technol,
Chofu
2 Ministry of
Construction,
Tsukuba
3 Toshiba
Electric Co.,
Kawasaki
4 Ministry of
Construction,
Tsukuba
5 NUPEC,
Tadotsu,
Takamatsu
6 IshikawajimaHarima
Heavy
Industries,
Yokohama
7 Nippon
Telephone
and
Telegraph
Tokyo
8 Instit. of
Construction Table size Specimen Controlled Frequency Acceleration Maximum Table motion Country
year
(m)
weight degrees of range (Hz)
(g)
velocity displacement
(KN)
freedom
(cm/s.)
(cm)
1975
4040
2000
2
0150
H 20
1140
150 Japan
V 10
445
75
1979
6080
1000
0130
H 07
600
75 Japan
1980
5050
200
(2, 1983)
2
0130
H 10
400
75 Japan
250
600
38
75 Japan
1981
2030
1982
(4 tables)
150150
1983
4545
1985
3030
10020
250
0150
V 07
H 07
10000
030
H 18
750
200 Japan
350
0150
V 09
H 15
375
750
200
100 Japan
H 15
750
100
V 10
H 30
500
65
67
100 Japan
H 30
65
100
V 30
H 04
65
120
100 Japan
100
1700
1100
file:///C:/Users/hp%20home/AppData/Local/Temp/~hhAD49.htm
20/07/2016
page_229
Industr.
Sc., Univ. of
Tokyo, Chiba
9 Railway
Techn. Res.
Inst. JNR,
Tokyo
Pgina 23 de 26
10020
1000
H 04
600
50 Japan
page_266
page_267
Page 267
10 CRIEPI, Abiko,
Chiba
11 Natl. Res. Cent. for
Dis. Prevention, Tsukuba
12 Takasago Res. Inst.
Mitsub. Ind., Himeji
13 Natl. Res. Inst. Agric.
Eng., Tsukuba
14 Port & Harb. Res. Inst.,
Yokosuka
15 Port and Harb. Res. Inst.,
Yokosuka
16 Shimizu Const. Comp.,
Tokyo
17 Build. Res. Institute,
Tsukuba
18 Ohbayashi Gumi,
Tokyo
19 Div. Prev. Res. Inst.,
Kyoto
20 Div. Prev. Res. Inst.,
Kyoto
21 Natl. Res. Center for Dis. Prevention, Tsukuba
6560
1250
6060
750
6060
500
6032
300
5520
170
4035
H 12 600
100
75
420
420
320
200 Japan
100
50 Japan
50
50 Japan
H 05 150
50 Japan
300
H 045 250
50 Japan
5040
120
H 10 720
100 Japan
4030
200
H 10 600
75 Japan
4030
100
H 10
100 Japan
3030
120
H 05
100 Japan
2525
80
150150
050
050
050
H 12
V 10
H 12
V 12
H 04
50 Japan
H 5000 1 (H or V) 050
H 05 500
V 05 500
H 055 370
50 Japan
50
30 Japan
V 2000
V 100 370
30
page_267
page_268
Page 268
Appendix 4. Tables in other countries
No.
Institution
Construction Table Specimen Controlled Frequency Acceleratio Maximum Table motion Country
year
size (m) weight degrees of range (Hz)
(g)
velocity displacement
(KN)
freedom
(cm/s.)
(cm)
1 University of
1980
5050
400
2
0130 H 067
635
125 Yugoslavia
Kiril
and Metodij,
V 040
380
50
file:///C:/Users/hp%20home/AppData/Local/Temp/~hhAD49.htm
20/07/2016
page_229
Skopje
2 Tong Ji
University,
Shanghai
3 East China
Technical
University of
Water
Resources,
Nanjing
4 Dalian
Institute of
Technology,
Dalian
5 Hydroproject
Res.
Institute,
Tbilisy
6 Arya Mehz
University,
Teheran
7 University of
Pahlavi,
Shiraz
8 INCERC,
IASI
Pgina 24 de 26
1983
4040
100
0150
H 15
890
100 China
625
625
500
100
100
50 China
500
75 China
1984
2028
60
0180
H 15
H27
H 12
1984
3030
100
0150
H 10
1980
6060
500
01100
H 12
100 USSR
100
75
50 Iran
50 Iran
5050
500
050
H 12
V 10
H 06
4040
200
050
H 11
7070
6000
H 08
130 Romania
page_268
page_269
Page 269
REFERENCES
1. Actan, A.E. & Bertero V.V., Analytical and physical modelling of R.C. frame-wall/coupled wall structure. In Design for Dynamic
Loading, The Use of Model Analysis, ed. G.Armer, F.K.Garas. Construction Press, London, 1982.
2. Eto, H. & Takeda, T., Simulated earthquake tests of R.C. frame structures with columns subjected to high compressive stress. In
Proc. 8th WCEE, Vol. VI, San Francisco, CA, 1984.
3. Tassios, T.P., Discussion paper. In Proc. Interassociation Symp. Concrete Structures under Impact Loading, Final Vol. p. 43,
BAM, Berlin, 1982.
4. Mahin, S.A. & Shing, P.B., Pseudodynamic method for seismic testing. ASCE, Struct. Div, July 1985.
5. Shing, P.B. & Mahin S.A., Pseudodynamic test method for seismic performance evaluation: theory and implementation.
UCB/EERC, 84/01, 1984.
6. Xu Peifu, Seismic resistance of shear wall structure with large space ground floor. Building Science (quarterly), China Academy
of Building Research, 1 (1985).
7. Bertero, V.V., Actan, A.E., Charney, F.A. & Sause, R., US-Japan co-operative earthquake research program: Earthquake
simulation tests and associated studies of a 1/5th. scale model of a 7-storey R.C. test structure. UCB/EERC-84/05, June 1984.
8. Wiegel, R.L. (Ed.), Earthquake Engineering. Prentice-Hall, London, 1970.
9. Wolfgram, C.E., Experimental modelling and analysis of three one-tenth-scale R.C. frame-wall structures. PhD thesis, Graduate
file:///C:/Users/hp%20home/AppData/Local/Temp/~hhAD49.htm
20/07/2016
page_229
Pgina 25 de 26
page_269
page_270
Page 270
18. Taylor, H.P.J., Investigation of the dowel shear forces carried by the tensile steel in R.C. beams. Tech. Report, CCA, London,
November 1969.
19. Tassios, T.P. & Vintzeleou, E., Concrete-to-concrete friction. ASCE, Struct. Div. April 1987.
20. Daschner, F., Schubkraftubertragung in Rissen von Normal-und Leichtbeton. Inst. f. Massivbau, T.U.Muenchen, FRG, March
1980.
21. Loeber, P.J., Shear transfer by aggregate interlock. Master thesis, University of Canterbury, Christchurch, New Zealand, 1970.
22. White, R.N. & Holley, M.J., Experimental studies on membrane shear transfer. ASCE, Struct. Div., August 1972.
23. Walraven, J.C., Aggregate interlock: A theoretical and experimental analysis. PhD thesis, Delft University of Technology, The
Netherlands, 1980.
24. Maisel, E., Mikrobeton fur modelstatische Untersuchungen. Inst. f. Modelstatik, Univ. Stuttgart, FRG, 1979.
25. Noor, F.A. & Khalid, M., Deformed wire reinforcement for microconcrete models. In Reinforced and Prestressed Microconcrete
Models. Construction Press, London, 1980.
26. Ammann, W., Mhlematter, M. & Bachmann, H., Stress-strain behaviour of non-prestressed and prestressed reinforcing steel at
high strain rates. In Proc. Interassociation Symp. Concrete Structures under Impact Loading. BAM, Berlin, 1982.
27. Staffier, S.R. & Sozen, M.A., Effect of strain rate on yield stress of model reinforcement. Civ. Eng. Studies, Struct. Res. Series
No. 145, University of Illinois, February 1975.
28. Tassios T.P., Fundamental mechanisms of force transfer across RC critical surfaces. CEB Bulletin No. 178/179, 1986.
29. Shimazu, T. & Araki, H., Fundamental study on dynamic behaviour of multi-storey shear walls. In Proc. 8th WCEE, Vol. VI,
San Francisco, CA, 1984.
30. Caccesse, V. & Harris, H.G., Seismic behaviour of precast concrete large panel building using a small shaking table. Report 3,
Dept of Civil Engineering, Drexel University, PA, June 1985.
file:///C:/Users/hp%20home/AppData/Local/Temp/~hhAD49.htm
20/07/2016
page_229
Pgina 26 de 26
31. Plaines, P., Tassios, T. & Vintzeleou, E., Bond relaxation and bond-slip creep under monotonic and cyclic actions. In Proc. Int.
Conf. on Bond in Concrete. Paisley, UK, 1982.
32. Kitagawa, Y., Kubota, T., Kaminosono, T. & Kubo, T., Correlation study on shaking table tests and pseudo-dynamic tests by
R.C.models. In Proceedings 8th WCEE, Vol. VI. San Francisco, CA, 1984.
BIBLIOGRAPHY
ACI Committee 444, Models of concrete structures, a state-of-the-art report. 1979.
Harris, H.G., Schwindt, D.C., Taher, I. & Werner, S.D., Techniques and materials in the modelling of R.C. structures under dynamic
loads. Research Report 6324, Dept Civ. Eng., MIT, MA, 1963.
page_270
page_271
Page 271
Knig, G. & Dargel, H.J., A constitutive law for R.C. with consideration to the effect of high strain rates. In Proc. Interassoc. Symp.
Concrete Structures under Impact Loading. BAM, Berlin, 1982.
Kowalzyk, R. & Dilger, W., Deformability of reinforced and unreinforced concrete (in Polish). Arch. Inz. Lad., 18 (1972) 287.
Morgan, B.J., Hiraishi, H. & Corley, W.G., U.S.-Japan co-operative research program, Tests of 1/3-scale planar wall assemblies. In
Proc. 8th WCEE, San Francisco, CA, 1984.
Tassios, T.P., Fundamental mechanisms of force-transfer across RC critical interfaces. In CEB Bulletin No. 174, 1987.
Tassios, T.P. & Tsoukantas, S., Behaviour of large panel connections. Journal of CIB, JulyAugust (1983) 22632.
page_271
page_272
page_272
Page 272
This page intentionally left blank.
file:///C:/Users/hp%20home/AppData/Local/Temp/~hhAD49.htm
20/07/2016