Escolar Documentos
Profissional Documentos
Cultura Documentos
[~"~U T T ~" R VV 0 R T M
0141-0296(95)00053-4
17, No. 7, pp. 505-511, 1995 Copyright 1995 Elsevier Science Ltd Printed in Great Britain. All rights reserved 0141-0296/95 $10.00 + 0.00
Shaking table tests on twelve reinforced concrete model columns are described. The system identification method is used to process the test data and to find the restoring force for the scaled columns at different deformation stages. The analysis shows that a semidegrading trilinear model can be used for the cracking and yielding stages of the reinforced concrete columns, and a semidegrading trilinear model with a descending branch can be used for the failure stage.
Keywords: reinforced concrete, model column, shaking table test, system identification, restoring force model
Numerical analysis and structural model testing are the two main approaches to the study of the structural behaviour of building structures, and the two disciplines have developed rapidly with the use of advanced computers and modem test equipment. Numeri~cal methods, especially the finite element method, can be used, with a supercomputer, to solve most engineering problems. However, for complex and innovative structures with complex boundaries and various loading conditions, it is impossible to simulate-the whole structural behaviour taking all the factors into account. Therefore structural model tests, are often used to study the structural behaviour and for the verification of the analysis method. As to research on the seismic resistance of reinforced concrete structures, a number of static tests on structural members and structural models have been carded out by many investigators. However, earthquake motion is a kind of dynamic action applied to the structure, and obviously static tests alone cannot simulate seismic forces properly. It is therefore necessary to carry out shaking table tests to study the dynamic behaviour of structures, especially reinforced concrete structures, under simulated strong earthquake loading. Since the capacity and size of shaking tables are limited, it is very common to test models on a shaking table instead of prototypes. In this paper the results of shaking table tests on a group of three RC model columns are described and the study of the failure mechanism and hysteretic behaviour of reinforced concrete model columns of different sizes using shaking table test data is emphasized. The results of the analysis can be incorporated into the time history analysis of the nonlinear earthquake responses of RC structures.
Notation
C J
K~,Kz, }
K3, K,,
KI2, KI3
coefficient of viscous damping error function stiffness mass cracking force yielding force ultimate force duration weighting factors displacement from analysis velocity from analysis acceleration from analysis ground (table) acceleration displacement from test acceleration from test vector cracking deflection yielding deflection deflection at ultimate load
M Pc P, T x Jc ~g y
Py Wa, Wb
/3 6~ 6,
6y
Testdesign
Twelve RC columns were designed and constructed with different scaling factors. The cross-section of the basic column, which was on the 1/2 scale of prototype, numbered Z2-1, was 200 mm x 300 mm with a height of 1800 mm; and the cross-section of the model columns, numbered Z4-1 and Z4-
505
506 (too)
D y n a m i c test a n d a n a l y s i s o f RC c o l u m n s : X i l i n Lu
L!
l I j
M
ST 100.0 39.86 39.86 0.0112 0.0112 0.0100
_1
2, was 100 mm x 150 mm with a height of 900 mm. The details of the specimens are shown in Figure 1. The aggregates and the reinforcement were equally scaled during the manufacture of the specimens, and an additional load was applied to the specimens to simulate the vertical load with different scale factors in order to study the effect of the gravity load on the dynamic behaviour. The similitude relationships are listed in Table 1, the details of which can be found in Reference 1, and the applications of the theory can be found in References 2 and 3. According to the properties of the specimens and the similitude requirements, the compressed El Centro (N-S, 1940) earthquake record was used as input acceleration of
Table I Relationships between specimens
Scaling parameters* Similitude requirements Bending strain Bending stress Young's modulus Normal strain Normal stress Mass density Length Displacement Sb, 1 S~ Sb~= Sb, Designed value
the shaking table with 1000 sampling points. The peak value of the input acceleration was gradually increased to study the behaviour of the reinforced concrete at different stages of deformation. The acceleration and displacement of the specimens were monitored during excitations, and the strains of the reinforcements and the surface strains of the concrete were also measured. The test set-up is shown in Figure 2. All the data were recorded and stored in a PDP 11/34 computer for further analysis. The model parameters, sampling time intervals and durations are listed in Table 2. T e s t results All the specimens underwent four deformation stages from small displacement to failure during the tests.
1 1
(1) Elastic stage When the peak value of the input acceleration was very low, the deformation of the specimens was very small, and cracking did not occur. The peak values of the table input and largest response of the specimens are listed in Table 3. (2) Cracking stage When the values of the input accelerations were increased to 0.2764 g for Z4-1, 0.3450 g for ZA-2, and 0.4414 g for
Table 3 Test results for elastic stage
Specimen no. Z2-1 Z4-1 Z4-2 Input acceleration (g) Response acceleration (g) 0.4883 0.3692 0.2762
1
0.6272 0.6272 0.3136 2
Sx Sx= SL
Sm Sm = SpS~L
2
2.509
Mass
Stiffness Time or period Damping Acceleration
2
1.120 2.240 1.594
Scaling parameters refer to ratio of units between prototype and model for a physical quantity, i.e. SL = LJLm (where L,, Lm are lengths of prototype and model, respectively)
Dynamic test and analysis o f RC columns: Xilin Lu Table4 Test results for cracking stage
Specimen
no.
507
Response acceleration
(g)
history of acceleration for specimen 24-1 is shown in Figure 4. (4) Failure stage When models 24-1 and Z4-2 were subjected to peak acceleration of values 1.5801 g and 1.3319 g, respectively, the concrete in the cracked zone crushed and a plastic hinge formed in each specimen. As a result, the specimens swung strongly, and the test was finally stopped. The results for this stage are listed in Table 6; the acceleration response of specimen Z2-1 is shown in Figure 5; and the displacement response of 24-1 is shown in Figure 6. From Figure 6 we can see that residual deformation occurred 4 s after the excitation began and the specimen began to vibrate on one side of the balance position, which resulted in the termination of the test.
Response acceleration
(g) 1.6191 0.9375 0.9174
This section aims to use the data from the shaking table tests to formulate an analytical model that can predict the nonlinear response of an RC structure subjected to an earthquake input. The system identification method is used in this formulation. Analytical model According to the test set-up shown in Figure 2, the RC structure considered in this study is simplified as a singledegree-of-freedom system with the following equation M Y ( t ) + C 2(t) + P(x) = - M 'g(t) x(O) = ~(0) = 0 (1)
H
(120
0.00
-Q20
-O.4O
0.(X) 1.20 2AO seconds 3.60
~8o
600
508
1.20
-0.60
-1.20
AA [ ,l
0.00
20o
z40o
seconds
6.0o
80o
Figure 4 Acceleration time history of Z4-1 Table 6 Test results for failure stage
KI3 =
P./8.
(5)
Equations (1)-(5), together with the rules for using the hysteretic model, constitute the analytical model. This model contains eight independent parameters: Kt, K2, K3, K4, Pc, Py, P. and C which can be represented by a vector /3={K,, K2, K3,
1(4, P~,
Py,
P.,
C} r
(6)
relative to the table; M is the mass; C is the coefficient of viscous damping; ~g(t) is the ground (table) acceleration; and P(x) is the restoring force of the structure. In the formulation of the restoring force, the following equation
K1 x
The purpose of this study is to find the best values for the vector/3.
x~ < 8c
IPy+ K3(x -
By)
8c<X<Sy 8y<X<8.
8. < x
(2)
Error function In shaking table tests, the data available are the time histories of acceleration and displacement. Therefore the error function can be written as
J(/3,T) = I / T
is used for the skeletal curve, and
{Wa[x(/3,t) - y(t)] 2 +
0
Wb['(/3t) - y(t)]2}dt
(7)
P(x) =]Ktz x
!
Kl x
x<~&t
8y<X~<8, 8. < x (3)
[KI 3 x
for branch curves (Figure 7), where 6c, 6y and 6, are the cracking deflection, yielding deflection and the deflection at ultimate load, respectively. Pc, Py and Pu are the cracking force, yielding force and ultimate force, respectively. K~ is the initial stiffness; and Kl2 and K13 are the equivalent stiffnesses given by
K,= = Py/6y (4)
where x(/3,t) and 2(13,t) are the responses calculated using the model with parameters/3 and excitation )tg(t); y(t) and 3~(t) are the real responses of the structure tested on a shaking table with the same excitation ;g(t); and W~ and Wb are positive scalar weighting factors. The value of T can be all or part of the duration of the table acceleration.
Parameter optimization algorithm In this step an algorithm is selected to systematically adjust the parameters in the model until the error function is minimized. The optimization technique used in this paper is a random search method called the adaptive step size random
509
1.00
Qoo k /
/%
jr
O.90
1.80
seconds
-1.00
-200 00
Figure 5 Acceleration time history of Z2-1
27O
36O
5DO
C
M
2.50
o.oo I V
-2.50
AA
JJ, A
'Ill
2.00 4.00
seco nds
"U" v vV
6.0 0 8DO 10.00
-5.00 0.00
Identification results
The identification result:~ for the 12 RC columns are summarized as follows. In the elastic stage., the values of elastic stiffness obtained from free vibrations can be used to predict the linear seismic response,; of RC structures. The identified and measured responses of Z4-1 are shown in Figure 8.
In the cracking and yielding stages, the 'semidegrading trilinear model', which is similar to the model obtained from the static test but with different parameters, can be used to predict the nonlinear seismic response of RC structures. The structural parameters of specimen Z2-1, contained in vector/3, are shown in Table 7. Since specimen Z2-1 did not reach its ultimate state, there is no ultimate load P, available in this model. The identified and measured acceleration time histories are shown in Figure 9. In the failure stages, the 'semidegrading tfilinear model
510
with descending branch' can be used to predict the nonlinear seismic response of RC structures as shown in Figure 10 which is the acceleration time history of specimen Z41. The structural parameters identified from the test data for Z4-1 and Z4-2 are also shown in Table 7.
Conclusions
Based on the above test results and analysis, the following conclusions can be drawn. Firstly, the failure mechanism and hysteretic behaviour of the RC model columns are very similar to those of the prototype columns under seismic excitations. Secondly, the restoring force models of RC columns under seismic excitations up to yielding stage are similar to those obtained from static tests but with different parameters, which must be considered in the dynamic analysis in which the static models are used. Finally, when plastic hinges have formed and concrete has crushed, the 'semidegrading trilinear model with descending branches' can be used to predict the seismic response of RC structures, and the structural parameters for this model are different from those based on static tests. The determination of these parameters needs to be investigated further.
Acknowledgment
The financial support of the National Natural Science Foundation of China is gratefully acknowledged.
5209260 3 240160 1 133 568 1 133 568 138 190 176343 113067
1 319020 867 823 184753 -23 464 7448 12869 14396 11 131
References
1 Sabnis,G. M. and Harris, H. G. Structural modeling and experimental techniques Prentice-Hall, Englewood Cliffs, NJ, 1983 2 Harris, H. G. Dynamic modeling of concrete structures, Publication SP-73, ACI, Detroit, USA, 1982 3 Lu, X. L. and Liao, G. M. 'Study on similitude relationships of reinforced concrete columns based on shaking table tests', Proc. US-
Identified
0.40
G
----Measured
Q20
~1
tJ
Q00
/IIAtltJ
'l
0.00 1J)0 21)0 Seconds 3D0
Ivvw
AA,J
-020
511
0.80
(I00
OIIll
r!
-0.80
.,, i
tA vyl
3.00 &O0 500
Measured
-1.60
0.00
1.00
ZOO Seconds
2.50 G
1.25
0.00
iA
0.00 1.00
Iv"
2.00 3.00 Seconds
-1.25
4.00
5.O0
4 5