Você está na página 1de 13

Soil Dynamics and Earthquake Engineering 25 (2005) 341353 www.elsevier.

com/locate/soildyn

Inuence of seismic cyclic loading history on small strain shear modulus of saturated sands
Yan-guo Zhou, Yun-min Chen*
Department of Civil Engineering, Zhejiang University, Zheda Road 38, Hangzhou 310027, Peoples Republic of China Accepted 2 March 2005

Abstract Small strain shear modulus Gmax is a key parameter together with the state of stress and shear strain amplitude for predicting the dynamic behavior of soils. Although the seismic cyclic loading on saturated soil deposits induces a decrease in effective stress and a rearrangement of the soil skeleton which may both lead to a degradation in undrained stiffness and strength of soils, only the contribution of effective stress reduction to Gmax degradation is considered in the Hardin and Richart equation which is widely used in seismic response analysis nowadays, and that of soil fabric change is neglected. In this paper, undrained cyclic triaxial tests were conducted on normally and isotropically consolidated saturated sands with the shear wave velocity measured intermittently by bender element during cyclic liquefaction, to study the inuences of seismic cyclic loading history on small strain shear modulus Gmax during earthquake. And the Gmax values of samples without such inuences were investigated for comparison. The tests results indicate that Gmax of sand under high amplitude seismic cyclic loading history inuences is moderately lower than the corresponding value of non-cyclic loading effects at the same effective stress. Hence it is necessary to reinvestigate the determination of Gmax in seismic response analysis carefully to predict the ground responses more reasonably. q 2005 Elsevier Ltd. All rights reserved.
Keywords: Cyclic loading; Seismic response analysis; Undrained cyclic triaxial test; Sands; Small strain shear modulus; Effective stress; Soil fabric; Bender element

1. Introduction Adequate information on dynamic soil properties, especially dynamic shear modulus and damping ratio, is essential for accurate computations of ground response and soilstructure interaction problems. Many experimental investigations carried out on sandy soils through resonant column test or improved cyclic triaxial test in early studies (Hardin and Richart [1]; Hardin and Black [2]; Drnevich and Richart [3]; Seed and Idriss [4]; Kokusho [5]) showed that the small strain shear modulus Gmax (g!10K5) was basically related to the mean effective principal stress s 0 m and void ratio e of the soil, and even overconsolidation ratio OCR for cohesive soil, expressed by the well known Hardin
* Corresponding author. Tel.: C86 571 8795 1340; fax: C86 571 8795 2165. E-mail address: cym@civil.zju.edu.cn (Y.-m. Chen).

and Richart equations taking the general form:


0 n OCRk Gmax Z AF esm

(1)

In which A is empirical constant reecting soil fabric formed through various stress and strain histories; n is empirically determined exponent, approximately equal to 1/2; s 0 m is mean effective conning pressure, s 0 mZ(s 0 vC 2s 0 h)/3, where s 0 v is vertical effective consolidation stress and s 0 h is horizontal effective consolidation stress; k is exponent of OCR that depends on plasticity index, which means that for cohesionless soils OCR has no effect on Gmax; e is void ratio and F(e) is void ratio function, which is usually given by (for angular grain sand) Fe Z 2:973 K e2 =1 C e (2)

0267-7261/$ - see front matter q 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.soildyn.2005.03.001

Although the Hardin and Richart equations have since undergone several adjustments (Hardin and Drnevich [6,7]; Iwasaki and Tatsuoka [8]; Hardin [9]; Hardin and Blandford [10]; Hryciw and Thomann [11]), for cohesionless soils these various expressions could be readily reduced to almost

342

Y.-g. Zhou, Y.-m. Chen / Soil Dynamics and Earthquake Engineering 25 (2005) 341353

Nomenclature A e F( e ) G Gmax Gmax0 GI max GII max K0 L n N N1 empirical constant reecting soil sample fabric; void ratio of the soil; void ratio function; secant shear modulus; small strain shear modulus; initial small strain shear modulus before the pore water pressure generation; small strain shear modulus without effects of seismic cyclic loading history; small strain shear modulus under seismic cyclic loading history inuences; coefcient of earth pressure at rest; distance between the bender elements; empirically determined exponent, approximately equal to 1/2; cycles in undrained cyclic triaxial test corresponding to T; accumulative cycle number of liquefaction failure; OCR t T Vs DT g gc gr gtv r s0m s 0 m0 s0v s0h td overconsolidation ratio; shear wave travel time in soil specimen; accumulative time of seismic loading on soil deposits during earthquake; shear wave velocity of soil specimen; time interval of subsequent motion of in situ soil deposits during earthquake; shear strain amplitude; cyclic shear strain amplitude; residual strain during cyclic loading; cyclic threshold shear strain amplitude; mass density of soil; mean effective stress; initial mean effective stress due to the connement; vertical effective consolidation stress; horizontal effective consolidation stress; and cyclic shear stress amplitude.

the same form, with the most widely used one for Gmax determination in the past decades being Eq. (1), especially in site-specic dynamic response analysis with total stress method (Idriss and Seed [12]) or effective stress method (Finn et al. [13,14]; Martin and Seed [15]; Papadimitriou and Bouckovalas [16]). It is generally known that in total stress analysis, Eq. (1) is used to estimate the small strain shear modulus variation in different layers with depth of soil proles as the initial modulus for nonlinear analysis. Besides, in effective stress analysis Eq. (1) will also account for the stiffness degradation caused by the gradual reduction of effective stresses in a given layer during seismic shaking, by updating the initial modulus Gmax at various time intervals so as to be consistent with the s 0 m during the same interval. Thus the small strain shear modulus, Gmax, is the key parameter for predicting the dynamic response and behavior of soils both in total stress and effective stress analyses, since it degrades with the reduction of s 0 m and acts as the basis of secant shear modulus G evaluation corresponding to the dynamic shear strain amplitude at the existing constant connement (see Fig. 1). However, it should be noted that Eq. (1) is summarized from the small strain (generally less than about 10K5) measurement with static connement in laboratory or in situ, while during seismic shaking the soils behave dynamically and undergo relatively large cyclic shearing strain (about 10K4w10K3 in general), when the effective stress will decrease with the pore water pressures build up due to plastic deformations in the soil skeleton, and therefore the loading effects on soils during earthquake are not the same as those reected in Eq. (1).

Repeated dynamic stress effects are of a different nature compared with that of static stress, and repeated straining at relatively high amplitude will cause the change of soil dynamic properties in subsequent low amplitude vibrations (Drnevich et al. [17]). Vucetic [18] found empirically that irreversible strains become increasingly signicant for cyclic shear strain amplitude gc larger than the cyclic threshold shear strain g tv (between 6.5! 10K5 and 2.5!10K4 for nonplastic sands and silts). And large accumulative strains could be expected to develop a nonuniform structure in the sand leading to nonuniform deformations in the sample during a cyclic loading test [19]. Wichtmann and Triantafyllidis [20,21] also pointed out that due to its cyclic loading history the fabric of a soil

1.0

50 40

G max
1 1 G

Shear modulus reduction, G/Gmax

0.6
0

30

0.4 0.2

20 10
G/Gmax

W = 4 W

0.0

1E-4

1E-3 0.01 0.1 Shear strain amplitude, (%)

0 10

Fig. 1. Nonlinear characteristics of soils (after Martin and Seed, 1979).

Damping ratio, (%)

0.8

Y.-g. Zhou, Y.-m. Chen / Soil Dynamics and Earthquake Engineering 25 (2005) 341353

343

(i.e. number, orientation and shape of particle contacts) as well as the distribution of interparticle forces are changed, and the small strain shear modulus Gmax of dry sand was reduced somewhat at high amplitude low-number of cyclic prestraining. Evidently sand fabric evolves during undrained cyclic shearing, and the saturated sand may acquire a cyclic loading history as the result of such fabric evolution besides the effective stress reduction. And this loading history is strongly characterized by its instantaneity because of the continuity and short duration of seismic shaking on soil deposits permitting no rest time for soil to regain its original structure. Hence directly assuming the small strain shear modulus Gmax using Eq. (1) equal to that of sands during earthquake is not appropriate and rigorous enough, and there is a necessity to examine more carefully the Gmax of sands under high amplitude cyclic loading. To study the problems mentioned above, a piezoelectric ceramic bender element measuring system was established using conventional cyclic triaxial apparatus, and a series of stress-controlled undrained cyclic triaxial tests were carried out on saturated sands with measured of shear wave velocity during cyclic liquefaction, to simulate the inuences of seismic cyclic loading history on soil stiffness, and investigate the mechanism of small strain shear modulus variation under this inuence. For comparison, another series of tests on Gmax without cyclic loading history effects were conducted. These test results are presented in this paper, which show that Gmax of sand under the inuences of relatively high amplitude and low-number cyclic loading is moderately lower than that under no such inuences at the same effective pressures.

(a) 1.0 Acceleration (g) 0.5 0 -0.5 -1.0 -1.5 0 (b) 2 Time (s)

T
4 6

T
8 10

Fig. 2. Equivalent cycles simulation of seismic loading. (a) Acceleration record during earthquake (Tangshan earthquake, MwZ7.8, 1976); (b) Schematic of equivalent cycles simulation.

including the small strain shear modulus, residual pore water pressure and residual strain. For it is hard to measure the shear wave velocity of specimens by bender element during continuous cyclic loading test, intermittences were adopted only long enough to take reading during the whole cyclic liquefaction process in following tests.

3. Test apparatus 2. Test fundamentals The behavior of soil deposits subjected to seismic loading, is closely associated with the undrained cyclic shear deformation. Thus the examination of seismic cyclic loading history during earthquake can be simulated through undrained cyclic triaxial tests in laboratory according to the procedure suggested in Ref. [4], which allows the irregular stress sequences produced by an earthquake to be replaced by an equivalent series of uniform cyclic stress cycles. Fig. 2 (a) and (b) show how the process will be simulated: The seismic cyclic loading on soil deposits for a period of loading time T will induce both the excess pore water pressure generation and the soil fabric evolution, and their effects on Gmax can be simulated by evaluating the small strain stiffness of corresponding soil specimen subjected to N cycles in undrained cyclic triaxial test; then the loading history effects on the subsequent motion of in situ soil deposits during following time interval DT may also be understood through investigating the specimen state during the (N C1)th cycle in tiraxial test, wherein the vital thing is to monitor correctly the development of the state variables To measure Gmax of soils during undrained cyclic shearing, a piezoelectricceramic bender element system was established based on Model HX-100 multi-use triaxial apparatus (Zhou et al. [22]). The test system frame is shown in Fig. 3.

Fig. 3. Schematic of triaxial test system with bender elements.

344

Y.-g. Zhou, Y.-m. Chen / Soil Dynamics and Earthquake Engineering 25 (2005) 341353

Fig. 4. Triaxial apparatus equipped with bender elements.

3.1. Brief introduction of bender element method The bender element method was originally developed by Shirley and Hampton [23] to obtain the very small strain shear modulus of a soil Gmax by measuring the velocity of shear wave propagating through a sample, which have attracted intensive study since then (De Alba et al. [24]; Dyvik and Madshus [25]; Thomann and Hryciw [26]). A bender element is a piece of piezoelectric ceramic plate which bends if a voltage across it is changed or, if bent by an external force, the voltage across it changes. Bender elements are set into the top and bottom platens of a triaxial cell and penetrate about 3 mm into the sample (see Fig. 4). One element is vibrated by changing the voltage across it, and shear waves propagate through the sample and vibrate the other element. The input and output voltages are continuously recorded and the travel time determined. The shear wave velocity Vs in the specimen can be calculated from the wave travel time t and the known separation L between the bender elements as Vs Z L=t (3) For a known material density r, according elastic theory, the small strain shear modulus can thus be calculated as Gmax Z rVs2 (4)

denition (S-A or S-B in Fig. 5) were compared with those obtained from resonant column tests (see Table 1), improving the condence in the data. Resonant column tests were conducted by using Drnevich Long-Tor Apparatus. As shown in Fig. 5 (a) and (b), the travel time values remained almost the same despite the exciting frequencies in (a) where the travel time is dened from S to A; but were relatively scattered in (b) where the travel time is dened as S-B. Further comparison in Table 1 shows that the shear modulus measured by bender element was a little higher than that obtained from resonant column tests for S-A travel time, but much lower for S-B travel time. Since the vibration shear strain amplitude in resonant column test was about 10K6w10K5 and a bit higher than that induced in bender element (generally about 10K6 or less), it is reasonable to accept the S-A travel time as the correct travel time for about 10 kHz sine wave signal. So in the subsequent tests sine pulse was selected as the transmitted signal and the rst major reversal point A at the received trace was considered as the rst arrival of the shear wave at the receiver bend element. Exciting frequencies selected were mainly about 1015 kHz, but adjusted to lower values for low connement such as s 0 mZ50 kPa or higher values for higher connement such as s 0 mZ400 kPa to obtain the most explicit traces of receiver bender element. And the effective length L through which the shear wave travel time dened was taken as the distance between the tips of the elements.

4. Test procedures and contents The mechanism of Gmax of saturated sand under the inuences of seismic cyclic loading was investigated, and that of Gmax without such inuences was investigated for comparison. Comparison between the two series of tests will show the differences between them, and yield information on the inuences of cyclic loading on small strain shear modulus. 4.1. Tested sands and specimen preparation Saturated specimens of 39.1 mm in diameter and 80 mm in height were sampled with the spooning method. Two types of sand (medium sand and ne sand, angular silica grains) were tested here with physical properties as given in Table 2, and the grain size distribution curves are shown in Fig. 6. The relative densities of the specimens were around 60%. After a specimen was formed in a split mold, a 15 kPa negative pressure was applied to support the specimen. After the mold was removed and the pressure chamber and other components were installed, a 20 kPa cell pressure was applied and the negative pressure was

3.2. Preliminary test for determining the correct travel time It is generally recognized that the principal thing with bender element method has always been the determination of the travel time t used to calculate Vs (Viggiani and Atkinson [27]; Jovicic et al. [28]; Arulnathan et al. [29]). The travel time of an impulse wave between two points in specimen may be taken as the time between the rst direct arrival of shear wave at each point. To reduce the degree of subjectivity in the interpretation of t, several types of input signal and different frequencies for each signal were tried in preliminary tests, and the results with different travel time

Y.-g. Zhou, Y.-m. Chen / Soil Dynamics and Earthquake Engineering 25 (2005) 341353

345

Fig. 5. Typical bender element traces with different input frequencies (s 0 mZ100 kPa). (a) Selected points (SP) for travel time: S-A (b) Selected points (SP) for travel time: S-B.

removed. Then the required full conning pressure was applied. As the minimum conning pressure in the tests was 50 kPa, this procedure prevented overconsolidation prior to testing.

4.2. Examination of Gmax without cyclic loading effects To obtain the dependence of small strain shear modulus Gmax on void ratio and conning pressure, a test series

Table 1 Comparison between bender element tests and resonant column tests (medium sand) s 0 m0 (kPa) Bender element tests Dr (%) 100 57.4 SP S-A S-A S-A S-B S-B S-B S-A S-A F (kHz) 5 10 20 5 10 20 10 10 Dt (ms) 354 352 352 400 402 394 296 264 Vs (m/s) 198.5 199.6 199.6 175.4 174.5 178.1 236.1 262.1 Gmax/ F(e)(MPa) 21.9 22.1 22.1 17.1 16.9 17.6 30.3 37.5 Resonant column tests Dr (%) 58.0 g (!10K6) 5.1 Vs (m/s) 197.6 Gmax/F(e) (MPa) 21.6

200 300

62.8 59.6

61.2 60.9

7.4 1.0

234.6 259.9

30.1 36.9

346

Y.-g. Zhou, Y.-m. Chen / Soil Dynamics and Earthquake Engineering 25 (2005) 341353

Table 2 Fundamental properties of tested sands Sand Type Specic gravity (g/cm3) 2.65 2.69 Void ratio and Relative density emax Medium sand Fine sand 0.788 1.096 emin 0.432 0.593 Dr (%) 52.3w65.0 59.6w73.7 Grain size distribution D10 (mm) 0.13 0.10 D50 (mm) 0.34 0.14 D90 (mm) 0.80 0.22 Cu 3.1 1.6

without cyclic loading history effects was conducted. Two ways were adopted to change the effective conning pressure of specimens with different initial void ratios e0 and relative densities Dr0: (1) Two specimens were consolidated isotropically with effective conning pressure increasing gradually in steps of 50, 100, 200, 300, 400 kPa for medium sand (Specimen-M1) and ne sand (SpecimenF1) respectively; (2) The other two specimens were consolidated isotropically under high conning pressure rst, then subjected to multi-stage backpressure with the effective stress decreasing gradually to zero. The backpressure tests were conducted in steps of 100, 200, 300, 350, 400 kPa for medium sand (Specimen-M2) and ne sand (Specimen-F2) respectively. Shear wave velocity was measured 2 h after each of the loading stages being applied, which duration is usually taken as the required time period of primary consolidation completion for sands. Note that in way (1) specimens were under normal consolidation while in (2) they were in the state of overconsolidation after back pressure being applied. Thus it is convenient to check out whether OCR will affect the Gmax variation of sand indeed or not by comparing the two ways of examination. Since the tests performed in the studies herein involve hours of connement duration, and the so called ageing effect were found more pronounced for clay than for sand (A and Richart [30], Vucetic and Dobry [31]), it is necessary to evaluate the connement duration effect on Gmax. Thus two medium sand specimens were prepared and conned totally for about 30 h under constant conning
120 Gravel 100

pressure 100 and 300 kPa, respectively. The shear wave velocities of the specimens were measured in denite time intervals during the whole conning duration, and the corresponding Gmax variation were assessed. 4.3. Examination of Gmax under seismic cyclic loading inuences In cyclic loading tests, samples were consolidated isotropically for 12 h at a given connement to obtain a sufciently stable soil fabric, and the shear wave velocity was measured for Gmax, before applying cyclic loading. Then they were subjected to a few number of uniform high amplitude cyclic stresses at frequency of 1 Hz under undrained conditions; and when the cyclic loading ended, the shear velocity for Gmax, residual pore water pressure and residual strain were recorded at the pause intervals only long enough to take reading (within 1 min). This process was repeated throughout one test series until the cyclic failure occurred (see Fig. 7). The ve conning pressure levels selected for the tests were 50, 100, 200, 300, 400 kPa, and spanned the range of soil pressure in situ for seismic response analyses under normal circumstances. Meanwhile, different stress amplitudes were tried at a given conning pressure to check out the diversity of test results between different tdwN1 combinations, in which td is cyclic shear stress amplitude, Nl is accumulative cycle number of liquefaction failure. For simulating seismic cyclic loading more reasonably, Nl was controlled within several hundred cycles in tests.
Consolidation for 12 h

Course

Sand Medium

Fine

Silt
Shear wave velocity measurement for Gmax0

Finer by weight, %

80 60 40 20 0 10 Medium sand Fine sand 1 0.1 0.01

Undrained cyclic loading

Fail or not? NO

YES

Cyclic liquefaction failure

Diameter, mm
Fig. 6. Grain distribution curves of tested sands.

Shear wave velocity measurement for Gmax


Fig. 7. Cyclic test procedure ow chart.

Y.-g. Zhou, Y.-m. Chen / Soil Dynamics and Earthquake Engineering 25 (2005) 341353

347

Small strain shear modulus Gmax/F(e), MPa

In consideration of the inuence of pause interval length on reading, some other observations were carried out to examine the change of Gmax during the pause interval under undrained conditions. The test procedure was similar to the abovementioned one, and the whole cyclic liquefaction process interrupted three times to study the Gmax variation during each pause interval. Each pause interval was held for about 120 min, and the shear velocities and residual pore water pressures of the specimen were measured after 0.5, 1, 5, 10, 30, 60, 120 min from the beginning of the pause. These tests were performed on two medium sand specimens under constant conning pressure 100 and 300 kPa, respectively. 5. Test results and analysis 5.1. Tests results without cyclic loading effects The original test data of Gmax versus the effective conning pressure s 0 m of the both sands are plotted in Fig. 8. To eliminate the effect of void ratio nonuniformity, Gmax was further divided by F(e) and the results are plotted in Fig. 9, where e is the void ratio corresponding to the Gmax at a given loading stage. As shown in Fig. 9, if divided by F(e), the data points of the two ways of tests are almost identical, and the small strain shear modulus is well correlated with the effective stress s 0 m regardless of the void ratio. And no evident OCR effect on Gmax was observed in back pressures applied tests of way (2). Then the following approximation for Gmax can be tted:
0 n Gmax Z AF esm

60

M1: Dr0= 49.8%, e0= 0.592 M2: Dr0= 51.8%, e0= 0.585
45

F1: Dr0= 48.5%, e0= 0.818 F2: Dr0= 54.0%, e0= 0.791

30

15

Fit curve equation : Gmax/F(e) = A*( 'm)

(medium sand) A = 2.121, n = 0.505 (fine sand) A = 2.227, n = 0.510


0 0 50 100 150 200 250 300 350 400 450

Effective confining pressure m ' , kPa


Fig. 9. Dependence of Gmax/F(e) on conning pressure.

(5)

presented in Fig. 10. A very slight increase of Gmax with time could be determined about 3w4% after 30 h of connement, especially for the modulus values divided by F(e) this increase lay only between 1w2%. This results tally with those presented in previous study (Petrakis and Dobry [32]) showing that for cohesionless soils, Gmax is ultimately a function of the number of intergranular contact points per soil grain and the effective conning pressure, and F(e) in Eq. (5) is a surrogate for the particle contacts. In other words, in tests on sands under static connement there was no signicant effect of previous stress history and no evident effect of loading time was observed, and the small strain stiffness variations were mainly related to effective stress s 0 m and void ratio e changes. 5.2. Undrained cyclic loading test results As shown in Fig. 11 (ac), throughout the whole cyclic shearing process, interrupted frequently to measure
160

Where AZ2.121, nZ0.505 for medium sand and AZ2.227, nZ0.510 for ne sand. In Eq. (5), Gmax and s 0 m are expressed in MPa and kPa, respectively. The test results of further study on the dependence of small strain shear modulus on connement duration are
225

Small strain shear modulus Gmax, MPa

200 175 150 125 100 75 50 25 0 0

M2: Dr0= 51.8%, e0= 0.585 F1: Dr0= 48.5%, e0= 0.818 F2: Dr0= 54.0%, e0= 0.791

Small strain shear modulus Gmax, MPa

M1: Dr0= 49.8%, e0= 0.592

120

Gmax Gmax

Gmax/F(e), m ' =100 kPa Gmax/F(e), m ' =300 kPa

80

40

Data in tests without cyclic loading effects 50 100 150 200 250 300 350 400 450

0 1

10

100

1000

5000

Effective confining pressure m ' , kPa


Fig. 8. Dependence of Gmax on void ratio and conning pressure.

Time since confinement application, min


Fig. 10. Effects of connement duration on Gmax.

348

Y.-g. Zhou, Y.-m. Chen / Soil Dynamics and Earthquake Engineering 25 (2005) 341353

(a) 100
75 Axial cyclic loading d, kPa 50 25 0 25 50 75 0 20 40 60 80 Number of cycles, N 100 120

(b) 1
'm0= 300 kPa, Dr= 62.1%
Axial cyclic strain d, %

d= 49.8 kPa, Nl= 115

0 1 2 3 4 5 0

'm0= 300 kPa, Dr= 62.1% d= 49.8 kPa, Nl= 115


25 50 75 Number of cycles, N 100 125

100

(c)
Pore water pressure u, kPa

300 250 200 150 100 50 0 0

'm0= 300 kPa, Dr= 62.1% d= 49.8 kPa, Nl= 115

25

50 75 Number of cycles, N

100

125

Fig. 11. Time histories of undrained tiraxial cyclic loading tests (s 0 m0Z300 kPa). (a) Cyclic loading versus number of cycles (b) Cyclic strain versus number of cycles (c) Pore water pressure versus number of cycles.

the shear wave velocity, the strain and pore water pressure behaviors were similar to those in conventional cyclic triaxial test (Seed and Lee [33]; Kokusho [5]), which indicated that the test procedure adopted here was reasonable enough for simulating the seismic cyclic loading. The test conditions of all the cyclic tests are listed in Table 3. And the small strain shear modulus Gmax obtained
Table 3 Test conditions of cyclic triaxial tests Sand type Medium sand Specimen label MS-50 MS-400 MS-100A MS-100B MS-200A MS-200B MS-300A MS-300B FS-50 FS-400 FS-100A FS-100B FS-200A FS-200B FS-300A FS-300B Conning pressure s 0 m0 (kPa) 50 400 100 100 200 200 300 300 50 400 100 100 200 200 300 300

at ve different conning pressures are plotted against effective stress s 0 m in Fig. 12(a) and (b) for medium sand and ne sand, respectively. There was similar trend of Gmax variations with reduction of effective stress regardless of the initial connement. As shown in Fig. 13(a) and (b) with Gmax divided by F(e), besides the rst point of each data group hitting almost the tting curve of test data without

Relative density Dr (%) 58.8 64.5 54.9 57.8 59.9 62.8 62.1 59.6 61.0 73.7 59.6 61.6 63.0 65.1 68.5 72.0

Cyclic stress amplitude td (kPa) 5.4 27.2 9.5 6.3 16.8 13.2 24.9 18.0 8.3 48.9 14.1 9.6 29.7 21.3 37.2 26.3

Cyclic strain amplitude gd 1.3!10K3 0.8!10K4 2.1!10K3 7.6!10K4 1.1!10K3 8.8!10K4 1.0!10K3 6.2!10K4 1.7!10K3 1.3!10K3 1.4!10K3 5.9!10K4 1.2!10K3 6.5!10K4 1.8!10K3 5.6!10K4

Failure cycle number Nl 66 188 15 575 43 479 115 713 39 95 34 612 47 394 71 857

Fine sand

Y.-g. Zhou, Y.-m. Chen / Soil Dynamics and Earthquake Engineering 25 (2005) 341353

349

(a) Small strain shear modulus Gmax, MPa 180


Cyclic tests data of medium sand

(b) Small strain shear modulus Gmax, MPa 160 140 120 100 80 60 40 20
FS-50 FS-100A FS-200A FS-300A FS-100A FS-100B FS-200B FS-300B Cyclic tests data of fine sand

150 120 90 60 30
MS-50 MS-100A MS-200A MS-300A MS-400 MS-100B MS-200B MS-300B

0 450 400 350 300 250 200 150 100 Effective stress 'm, kPa

50

0 450 400 350 300 250 200 150 100 Effective stress 'm, kPa

50

Fig. 12. Variation of Gmax against s 0 m during cyclic loading tests. (a) Original data of Gmax versus s 0 m (medium sand) (b) Original data of Gmax versus s 0 m (ne sand).

cyclic loading effects, the other points are lower than the curve at the same effective stress: the former shows the variation of Gmax without cyclic loading effects while the latter reveals the characteristic of Gmax inuenced by seismic cyclic loading history. For explicitness, the small strain shear modulus and effective stress are normalized by initial values Gmax0 and s 0 m0 respectively, and the results are plotted in Fig. 14(a) and (b) for the two tested sands, where s 0 m0 is the initial mean effective stress due to the connement, and Gmax0 is the initial small strain shear modulus before the pore water pressure generation. If the small strain shear modulus without effects of cyclic loading is denoted as GI max , and the one under the inuence of seismic cyclic loading as GII max , the modulus reduction I I K G = G under the same effective stress can be GII max max max computed from the data in Fig. 14(a) and (b), with the results being plotted in Fig. 15(a) and (b). As shown in I Fig. 15, the modulus reduction GII max K Gmax appears at the beginning of cyclic shearing, increases with the effective
(a) Small strain shear modulus Gmax/F(e), MPa 50 40 30 20 10
Medium sand data and fitting curve in tests without cyclic loading effects

stress degradation and reaches to a relatively stable value round the midpoint of normalized effective stress; then stabilizes or decreases slightly ultimately. In the case of the amplitude geZ5.6!10K4w2.1!10K3 investigated herein, this stable value of modulus reduction was approximately 6w9% of GI maxfor medium sand and 3w5% for ne sand respectively, which demonstrates that the further modulus reduction was larger for medium sand than for ne sand. In the case of the dynamic prestraining amplitude gc Z 5 ! 10K3 for dry medium sand presented in Ref. [20], similar modulus reduction amounted 7% in average was observed. It should be noted that in Fig. 15 a tendency can be gured out, that for a given sample at the same conning pressure, the higher the cyclic shear strain were applied, the larger modulus reduction could be obtained. Results of the two conducted tests on the inuence of pause interval length on data reading are shown in Fig. 16 (a)(d). During pause interval, the pore water pressure decreases a little, and small strain shear modulus will
(b) Small strain shear modulus Gmax/F(e), MPa 50 40 30 20 10
Fine sand data and fitting curve in tests without cyclic loading effects

Cyclic tests data of medium sand MS-50 MS-100A MS-200A MS-300A MS-400 MS-100B MS-200B MS-300B

Cyclic tests data of fine sand FS-50 FS-100A FS-200A FS-300A FS-400 FS-100B FS-200B FS-300B

0 450 400 350 300 250 200 150 100 Effective stress 'm, kPa

50

0 450 400 350 300 250 200 150 100 Effective stress 'm, kPa

50

Fig. 13. Variation of Gmax/F(e) against s 0 m during cyclic loading tests. (a) Data of Gmax/F(e) versus s 0 m (medium sand) (b) Data of Gmax/F(e) versus s 0 m (ne sand).

350

Y.-g. Zhou, Y.-m. Chen / Soil Dynamics and Earthquake Engineering 25 (2005) 341353

(a) Normalized shear modulus Gmax/Gmax0 1.0 0.8 0.6 0.4 0.2 0.0 1.0
Fitting curve of GImax/Gmax0~( 'm/ 'm0)

(b) Normalized shear modulus Gmax/Gmax0

1.0 0.8 0.6 0.4 0.2 0.0 1.0


Fitting curve of GImax/Gmax0~( 'm/ 'm0)

Cyclic tests data of medium sand MS-50 MS-100A MS-200A MS-300A MS-400 MS-100B MS-200B MS-300B

Cyclic tests data of fine sand FS-50 FS-400 FS-100A FS-100B FS-200A FS-200B FS-300A FS-300B

0.8

0.6

0.4

0.2

0.0

0.8

0.6

0.4

0.2

0.0

Normalized effective stress 'm / 'm0

Normalized effective stress 'm/ 'm0

Fig. 14. Variation of Gmax/ Gmax0 against s 0 m/s 0 m0 during cyclic loading tests. (a) Data of Gmax/ Gmax0 versus s 0 m/s 0 m0 (medium sand) (b) Data of Gmax/ Gmax0 versus s 0 m/s 0 m0 (ne sand).

rapidly increase to a stable value within 5w10 min, but this value is still moderately lower than GI max at the corresponding effective stress. Fig. 16 shows that further reduction of shear modulus compared to GI max is most probably attributed to the rearrangement of soilparticle structure (including interparticle contact slippage and grains connect crushing) under the cyclic shearing, and this reduction could not recover at relatively short rest time. Fig. 16 (a) and (c) also implies that if the shear wave velocity could be measured at the exact beginning of the pause interval or even during continuous cyclic shaking, the small strain shear modulus calculated consequently would be smaller than that measured at pause time of 1 min adopted herein, since the unstable state of soil fabric requires some time (5w10 min in tests) to stabilize primarily. 5.3. Reasons for further reduction of small strain stiffness The variation of GII max with effective stress reduction could be understood based on GI max , with further consideration of high amplitude dynamic stresses induced drastic
(a) Shear modulus difference , % 0
Cyclic tests data of medium sand

rearrangement of soilparticle structure effects (macroscopically shown as the residual strain gr accumulation in Fig. 17(a) and (b)). Finn et al. [19] studied the strain history effect on liquefaction of sand and concluded that a given number of cycles of a shear strain greater than 0.5% had a weakening effect on the resistance of sand samples to liquefaction under cyclic loads. And the reason for this loss of liquefaction resistance was postulated that the effect of any shear strain beyond a threshold value was to create a nonuniform structure in the sand sample, which was more sensitive to liquefaction than the structure created by consolidation. These statements are helpful to gain insight into the seismic cyclic loading history effect on small strain modulus of sand investigated herein, and indicate that although the reduction of liquefaction resistance and Gmax of sand under this loading history effects are phenomenally different aspects of cyclic liquefaction process, they are somewhat due to the same substantial mechanism: the soil fabric evolution at cyclic strain levels above the threshold value. Cascante and Santamarina [34] and Santamarina [35] also concluded that the small strain stiffness of particulate materials was strongly determined by the behavior of
(b) Shear modulus difference , % 0
Cyclic tests data of fine sand

10

10

15

MS50 MS100A MS200A MS300A

MS400 MS100B MS200B MS300B

15

FS50 FS100A FS200A FS300A

FS400 FS100B FS200B FS300B

20 1.0

0.8 0.6 0.4 0.2 Normalized effective stress 'm/ 'm0

0.0

20 1.0

0.8

0.6

0.4

0.2

0.0

Normalized effective stress 'm/ 'm0

I I II I I 0 0 Fig. 15. Dependence of shear modulus reductionGII max K Gmax =Gmax on effective stress. (a) Data of Gmax K Gmax =Gmax versus s m/s m0 (medium sand) (b) I I 0 0 Data of GII K G = G versus s / s (ne sand). m m0 max max max

Y.-g. Zhou, Y.-m. Chen / Soil Dynamics and Earthquake Engineering 25 (2005) 341353

351

(a) Small strain shear modulus Gmax, MPa 80

(b) 1.0 Normalized shear modulus Gmax/Gmax0


Fitting curve GImax/Gmax0= ( 'm/ 'm0)0.505

0.8

60

0.6

0.4

40

'm0= 100 kPa, Dr= 54.4% Gmax0= 76.2 MPa


Group1, u= 8.9~8.83 kPa Group2, u= 26.8~26.5 kPa Group3, u= 53.0~54.95kPa

'm0=100 kPa, Dr= 54.4%, Gmax0=76.2 MPa


Measured at puase time 1 min Group1, u= 8.9~8.83 kPa Group2, u= 26.8~26.5 kPa Group3, u= 53.0~54.95 kPa

0.2

20 0 30 60 90 120 150 Pause time, min

0.0 1.0

0.8 0.6 0.4 0.2 Normalized effective stress 'm/ 'm0

0.0

(c) Small strain shear modulus Gmax, MPa 140

(d) 1.0 Normalized shear modulus Gmax/Gmax0


Fitting curve GImax/Gmax0= ( 'm/ 'm0)0.505

0.8

120

0.6

100

'm0= 300kPa, Dr= 67.2%


Gmax0= 144.9 MPa Group1, u=52.4~49.2 kPa Group2, u=94.8~93.2 kPa Group3, u=201.3~199.1 kPa

0.4

'm0=300 kPa, Dr= 67.2%, Gmax0=144.9 MPa


Measured at puase time 1 min Group1, u = 52.4~49.2 kPa Group2, u = 94.8~93.2 kPa Group3, u = 201.3~199.1 kPa

80

0.2

60 0

30

60

90

120

150

0.0 1.0

Pause time, min

0.8 0.6 0.4 0.2 Normalized effective stress 'm/ 'm0

0.0

Fig. 16. Time-dependent regain in small strain shear modulus during pause interval. (a) Gmax versus pause time (s 0 m0Z100 kPa) (b) Gmax/ Gmax0 versus s 0 m/s 0 m0 (s 0 m0Z100 kPa) (c) Gmax versus pause time (s 0 m0Z300 kPa) (d) Gmax/ Gmax0 versus s 0 m/s 0 m0 (s 0 m0Z300 kPa).

contacts, and that compared with elastic contact, grains contact crushing will result in more rapid change in sand stiffness. It seems reasonable to conclude, therefore, that the possible cause of further small strain stiffness reduction is the change of shape of particle contacts. Under a number of high amplitude cyclic shearing, the relatively stable contacts
(a) 1 Residual shear strain r, % 0

formed under long time consolidation are replaced by softer and more unstable contacts due to particle re-orientation. And the observation on Gmax variation during pause intervals is an additional evidence of this conclusion. Objectively, there was also a little fabric change in soil samples drained under static connement which was mainly
(b) 1 0 Residual shear strain r, % 1 2 3 4 5 1.0
FS-50 FS-100A FS-200A FS-300A FS-400 FS-100B FS-200B FS-300B

1 2 3 4 5 1.0
MS-50 MS-100A MS-200A MS-300A MS-400 MS-100B MS-200B MS-300B

0.8 0.6 0.4 0.2 Normalized effective stress 'm/ 'm0

0.0

0.8

0.6

0.4

0.2

0.0

Normalized effective stress 'm/ 'm0

Fig. 17. Residual shear strain development with effective stresses. (a) Data of gr versus s 0 m/s 0 m0 (medium sand) (b) Data of gr versus s 0 m/s 0 m0 (ne sand).

352

Y.-g. Zhou, Y.-m. Chen / Soil Dynamics and Earthquake Engineering 25 (2005) 341353

associated with the minimal void ratios reduction, which was reported by Richart [36] and also observed in tests herein, but differs from the relatively more drastic interparticle contact slippage and grains crushing induced by high amplitude undrained cyclic shearing, and the latter causes further stiffness degradation with constant void ratio. It is interesting that similar degradation was observed on the immediate postcyclic undrained secant modulus of cohesive soils (Yasuhara and Hyde [37]), which may imply some correlation between the small strain shear modulus and secant modulus under undrained cyclic loading. It should be noted that, although the tests conducted herein were just under isotropic stress states, the test results evidently revealed the particular phenomena of Gmax inuenced by seismic cyclic loading history. In further study, other factors such as K0 conditions, grain characteristics, and cyclic frequencies will be taken into consideration to gain more general and profound understanding of the Gmax variation mechanism under the inuence of seismic cyclic loading.

During earthquake, the ground deposit shakes continuously, so the determination of small strain shear modulus in seismic response analyses should take into account the cyclic loading history effects, including both the induced effective stress reduction and rearrangement of the soil particle structure. The test results presented herein indicate that present seismic response analyses may overestimate the value of small strain shear modulus, which overestimation will increase with the development of earthquake process, thus the variety of response analysis results will be inuenced to some extent. Therefore, it is necessary to carefully reinvestigate the determination of small strain shear modulus in present seismic response analyses and even to make some modication based on it.

Acknowledgements This study was supported by the National Natural Science Foundation of China (No.10372089). The writers would like to acknowledge Prof. Charles W. W. Ng and Dr Bo Huang for their valuable suggestions on this study, and express the gratitude to Yan-Hong Ma, Wei-An Lin and Zhi Yang for their kind help in conducting part of the laboratory tests. The writers also wish to thank Prof. W. D. Liam Finn and the SDEE anonymous reviewers whose comments led to substantial improvement of this paper.

6. Summary and conclusions In the present study, the small strain shear moduli of saturated sands consolidated normally under isotropic connement were investigated throughout the undrained cyclic triaxial tests where cyclic shear strain amplitude of about 5.6!10K4w2.1!10K3, and further reduction of Gmax different from Eq. (1) were observed at almost I constant void ratio. The modulus difference GII max K Gmax appeared at the beginning of cyclic shearing, increased with the effective stress reduction and reached to a relatively stable value around the midpoint of cyclic liquefaction process; then stabilized or decreased slightly ultimately. This stable value of modulus reduction amounted approximately 6w9% of GI maxfor medium sand and 3w5% for ne sand, which indicates that this reduction is soil type dependent. And for a given sample at the same conning pressure, the higher the cyclic shear strain was applied, the larger modulus reduction could be obtained. These observations are consistent with part of the ndings reported in those literatures, which were based on intensive investigation on the dry sand dynamic properties subjected to high amplitude and low-number previbration. The Gmax difference of sands between the two types of test (namely, with or without cyclic loading effects) at the same effective stress is due to the inuence of relatively high amplitude cyclic loading, which results in not only reduction of effective stress, but also the change of shape of particle contacts, namely, the interparticle contact slippage and grains connect crushing. Under high amplitude cyclic shearing, the relatively stable contacts formed under long connement duration are replaced by softer and unstable contacts due to particle re-orientation, which contributes to further reduction of Gmax.

References
[1] Hardin BO, Richart Jr FE. Elastic wave velocities in granular soils. J Soil Mech Foundations Div, ASCE 1963;89(1):3365. [2] Hardin BO, Black WL. Vibration modulus of normally consolidated clay. J Soil Mech Foundations Div, ASCE 1968;94(2):35369. [3] Drnevich VP, Richart Jr FE. Dynamic prestraining of dry sand. J Soil Mech Foundations Div, ASCE 1970;96(2):45369. [4] Seed HB, Idriss IM. Simplied procedure for evaluating soil liquefaction potential. J Soil Mech Foundation Div, ASCE 1971; 97(9):124973. [5] Kokusho T. Cyclic triaxial test of dynamic soil properties for wide strain range. Soils Foundations 1980;20(2):4560. [6] Hardin BO, Drnervich VP. Shear modulus and damping in soils: measurement and parameter effects. J Soil Mech Foundations Div, ASCE 1972;98(6):60324. [7] Hardin BO, Drnervich VP. Shear modulus and damping in soils: design equations and curves. J Soil Mech Foundations Div, ASCE 1972;98(7):66792. [8] Iwasaki T, Tatsuoka F. Effects of grain size and grading on dynamic shear moduli of sands. Soils Foundations 1977;17(3):1935. [9] Hardin BO. The nature of stress-strain behavior for soils. In: Conference on Earthquake Engineering and Soil Dynamics, ASCE 1978: 390. [10] Hardin BO, Blandford GE. Elasticity of particulate materials. J Geotech Eng, ASCE 1989;115(6):788805. [11] Hryciw RD, Thomann TG. Stress-history-based model for Ge of cohesionless soils. J Geotech Eng, ASCE 1993;119(7):107393. [12] Idriss IM, Seed HB. Seismic response of horizontal soil layers. Soil Mech Foundations Div, ASCE 1968;94(4):100331.

Y.-g. Zhou, Y.-m. Chen / Soil Dynamics and Earthquake Engineering 25 (2005) 341353 [13] Finn WDL, Byrne PM, Martin GR. Seismic response and liquefaction of sands. J Geotech Eng Div, ASCE 1976;102(8):84156. [14] Finn WDL, Lee KW, Martin GR. An effectivestress model for liquefaction. J Geotech Eng Div, ASCE 1977;103(6):51733. [15] Martin PP, Seed HB. Simplied procedure for effective stress analysis of ground response. J Geotech Eng Div, ASCE 1979;105(6):73958. [16] Papadimitriou AG, Bouckovalas GD. Plasticity model for sand under small and large cyclic strains: a multiaxial formulation. Soil Dyn Earthquake Eng 2002;22:191204. [17] Drnevich VP, Hall Jr JR, Richart Jr FE. Effects of amplitude of vibration of the shear modulus of sand. In: Proceedings of international symposium on wave propagation and dynamic properties of earth materials, Albuquerque, NM, 1967. [18] Vucetic M. Cyclic threshold shear strains in soils. J Geotech Eng ASCE 1994;120(12):220828. [19] Finn WDL, Bransby PL, Pickering DJ. Effect of strain history on liquefaction of sand. J Soil Mech Foundations Div, ASCE 1970;96(6): 191734. [20] Wichtmann T, Triantafyllidis Th. Inuence of a cyclic and dynamic loading history on dynamic properties of dry sand, part I: cyclic and dynamic torsional prestraining. Soil Dyn Earthquake Eng 2004;24: 12747. [21] Wichtmann T, Triantafyllidis Th. Inuence of a cyclic and dynamic loading history on dynamic properties of dry sand, part II: cyclic axial preloading. Soil Dyn Earthquake Eng 2004;24:789803. [22] Zhou YG, Chen YM, Ke H. Improvement on simplied procedure for liquefaction potential evaluation of sands by shear wave velocity (in Chinese). Chinese J Rock Mech Eng 2005;24(13). [23] Shirley DJ, Hampton LD. Shear wave measurements in laboratory sediments. J Acoustical Soc Am 1978;63(2):60713. [24] De Alba P, Baldwin K, Janoo V, Roe G, Celikkol B. Elastic-wave velocities and liquefaction potential. Geotech Testing J ASTM 1984; 7(2):7787.

353

[25] Dyvik R, Madshus C. Lab measurement of Gmax using bender elements. In: Advances in the art of testing soils under cyclic conditions. Detroit, MI. Geotechnical Engineering Division, ASCE; 1985, 18696. [26] Thomann TG, Hryciw RD. Laboratory measurement of small strain shear modulus under K0 conditions. Geotech Testing J ASTM 1990; 13(2):97105. [27] Viggiani G, Atkinson JH. The interpretation of the bender element tests. Geotechnique 1995;45(1):14954. [28] Jovicic V, Coop MR, Simic M. Objective criteria for determing Gmax from bender element. Geotechnique 1996;46(2):35762. [29] Arulnathan R, Boulanger RW, Riemer MF. Analysis of bender element tests. Geotech Testing J ASTM 1998;21(2):12031. [30] A SS, Richart Jr FE. Stress-history effects on shear modulus of soils. Soils Foundations 1973;13(1):7795. [31] Vucetic M, Dobry R. Effect of soil plasticity on cyclic response. J Geotech Eng, ASCE 1991;117(1):89107. [32] Petrakis E, Dobry R. A self consistent estimated of the elastic constants of a random array of equal spheres with application to granular soil under isotropic conditions. NY: Rensselaer Polytechnic Inst., Troy; 1986 (Rep. No. CE-86-04). [33] Seed HB, Lee KL. Liquefaction of saturated sands during cyclic loading. J Soil Mech Foundation Div, ASCE 1966;92(6):10534. [34] Cascante G, Santamarina JC. Interparticle contact behavior and wave propagation. J Geotech Eng 1996;122(10):8319. [35] Santamarina JC. Soil behavior at the microscale: particle forces. Geotechnical Special Publication, ASCE 2003;119:2556. [36] Richart Jr FE. Some effects of dynamic soil properties on soil structure interaction. J Geotech Eng Division, ASCE 1975;101(12): 447. [37] Yasuhara K, Hyde AFL. Method for estimating postcyclic undrained modulus of clays. J Geotech Geoenviron Eng, ASCE 1997;204(3): 20411.

Você também pode gostar