ejge paper 2007 -0677

Seismic Behavior of Gravel Drains and Compacted Sand Piles using Physical and Numerical Models

Department. of Civil & Environmental Engineering,
University of Illinois at Urbana-Champaign, Urbana, USA
asadrek2@uiuc.edu

ABSTRACT

During recent earthquakes, it was observed that liquefaction can cause severe damages to buildings in the form of significant subsidence and shear failure of foundation soil. This damages is known to be due to the build up of pore water pressure and hence a reduction of soil strength. A number of remediation methods exist which reduce the excess pore pressure, enhance the shear deformability of the soil and fortify the soil. Two well known methods which are gravel drains and compacted sand piles are discussed and compared in this paper. Some precisely prepared 1-g shaking table tests were performed regarding these methods. The results show that compacted sand piles were more efficient than gravel drains in the case of liquefaction resistance and settlement of the subsoil during the shaking period. On the other hand after shaking the efficiency of the gravel drains was improved by the means of excess pore pressure dissipation. Also the cases were modeled with a numerical finite element program which could only handle the excess pore water pressure. By evaluating the numerical results with the experimental ones it is concluded that replicating liquefaction merely by an excess pore pressure generation/dissipation model would underestimated the excess pore pressure as well as the settlement.

Keywords: Gravel Drains, Compacted Sand Pile, Liquefaction, Shaking Table, Excess Pore Water Pressure, Numerical Modeling.

INTRODUCTION

Liquefaction is the most important geotechnical factor which has caused damages to buildings and coastal structures during earthquakes (Shibata et al., 1996; Inagaki et al., 1996; Kamon et al., 1996; Hamada et al., 1996; EERC, 1995). There are a large number of new projects involving the construction or utilization of reclaimed areas worldwide. Conventional reclamation work uses hydraulically placed fills resulting in loose deposits of essentially cohesionless soils highly prone to liquefaction. As an example, in Japan a large artificial island was constructed using 180 million cubic meters of dredged material for the extension of Tokyo International Airport with an estimated cost of 12 billion dollars. The continuing development of waterfront properties and the construction of offshore artificial islands impose increasing pressure on the development and implementation of remediation measures for liquefaction prone sites. Several mitigating actions can be taken such as removal or replacement of undesirable soil, densification of the insitu material, insitu soil improvement by grouting and chemical stabilization and using of relief wells such as gravel or rock drains for the control of undesirable pore water pressure. Although these types of mitigation techniques are developed, the effectiveness of these methods are not well defined and understood (Das, 1983).

All mitigation techniques which are frequently employed to reduce large deformations and subsidence of buildings are based on the following philosophies:

Reducing the build up of pore water pressure by means of quick drainage of water during and immediately after the earthquake.

Improving shear deformability of the soil skeleton to prevent large cyclic deformation during the earthquake.

Reinforcing the soil skeleton, which in turn can reduce both shear strain and generation of excess pore water pressure and increases the soil strength.

One of the widely used mitigation methods is using gravel drains. The possible benefits of gravel drains are densification of surrounding non-cohesive soil, dissipation of excess pore water pressure and re-distribution of earthquake-induced or pre-existing stresses (due to introduction of the stiffer columns). When dealing with non-plastic silty soils, only the third benefit can be expected primarily to mitigate liquefaction (Baez, 1995). The gravel drain technique is ideally suited for improving soft silts and clays, and loose silty sands. The level of improvement depends on the soil type, installation technique, relative spacing of the drains, and drain diameter. Crushed stones made of recycled concrete from torn-down apartment buildings and complexes are suitable alternatives to be used as drain materials (Orense et al., 2003). Gravel drains operate by providing preferential drainage paths which enable accumulated pore pressures to dissipate ideally before the surrounding soil reaches a state of initial liquefaction (Brennan and Madabhushi, 2002). One of the first studies on gravel drains as liquefaction remediation is that done by Seed and Booker (1977). Since this was published, drains have been subjected to real earthquakes, such as Koshiro-Oki (Sonu et al., 1993), Northridge (Boulanger et al., 1998) and Kobe (Yasuda et al., 1996). Several shortcomings of gravel drains have been reported in the literature. A collected experience suggests that while drains can certainly provide a solution, settlement can still occur to an unsatisfactory degree (Brennan and Madabhushi, 2002).

Regarding aforesaid remarks, in the current study some aspects of the effectiveness of gravel drains and compacted sand piles in mitigating excess pore water pressure and reducing the subsidence of buildings has been studied using 1g shaking table tests and following that the experimental results are compared with a numerical method which incorporates an improved procedure of that used by Seed and Booker (1977).

PHYSICAL MODELING

A series of shaking table tests were conducted on model gravel drains and compacted sand piles. Figure 1 shows a three dimensional view of the model. Models were constructed in a transparent plexiglass container of 180cm long, 45cm wide and 70cm high. The bottom of the container was covered by a fine screen mesh so that the saturation process could be performed by percolating water gradually and uniformly from the bottom of the soil box.

Figure 1. Three-dimensional view of the model apparatus

Firuzkooh sand was used as the subsoil. The characteristics of this sand are Gs = 2.658, emax = 0.943, emin = 0.603, D50 = 0.3mm, Cu = 2.58, Cc = 0.97 and permeability of 0.0125 cm/sec. The model foundation had dimensions of 20cmx30cm and was applying an overburden pressure of 3 kPa on the sand. A geometrical scaling factor of 1:25 can be assumed throughout these tests to model a prototype with a width of 5m. Different types of transducers were employed to measure acceleration, pore water pressure and displacement at different positions as shown in Figure 2. The pore pressure transducers were fixed in place to record the pore water pressures at the exact locations; however the acceleration transducers were free to move with the adjacent soil.

Moist tamping method, in which the Firuzkooh sand was mixed with 5% moisture, was used to prepare a uniform soil profile. Wet Firuzkooh sand was poured inside the container and carefully tamped to a total unit weight of 14.41 kN/m3, thus a target void ratio of 0.9 was gained for the liquefiable soil through the tests. Dyed grid lines were created to make the behavior of model ground visible. The soil models were percolated with carbon dioxide to help dissolve the air in the void space, in order to facilitate full saturation by water. Afterwards the model was saturated from bottom with a very slow steady flow of water in order to sustain the controlled density of the tamped sand. Input shaking in all tests was a harmonic wave. The frequency of shaking and amplitude of base acceleration were 3 Hz and 0.28g respectively.

Figure 2. Schematic view of the model and transducers (A1~A4: Accelerometers, P1~P5: Pore water pressure transducers and D1:Displacement transducer)

Test O was performed without any improvements; in test G gravel drains were placed in the model ground. These gravel drains had a diameter of 5cm and were sandwiched by geo-textile filters. The longitudinal and transverse center-to-center spacing of these drains was 25cm and 30cm respectively. Dynamic compaction in test C was applied by dropping a 2.0 kg weight with a circular bottom area of 19.6cm2 from a height of 30cm, ten times. The ground could be improved to a depth of almost 30cm in model scale using this method. The resulting compacted sand piles had a relative density of about 65-70%. Relative densities as high as 90% can be achieved in field with crushed stones (Adalier et al., 2003). The center-to-center distance of the piles was 5cm. These piles covered an abscissa of 1B. Figure 3 schematically shows different types of models as described above. A dynamic data acquisition system was utilized to record the behavior of the model during the test. During all tests, data were recorded at a sampling rate of 1000 samples per second.

Figure 3. Schematic views of test arrangements.

NUMERICAL MODELING

The conducted experiments were simulated with a two-dimensional finite element code (FEQDrain) programmed by Pestana et al. (1998). This program allows for generation and dissipation of the pore water pressure during dynamic loading, and the basic differential equation is the one used by Seed and Booker (1977), and Onoue (1988) with some modifications in the treatment of boundary conditions and drain elements, which are as below:

(1)

Where u is the pore water pressure, t is time, mv is the coefficient of volumetric compressibility, ?w is the unit weight of water, z is depth within the soil, and k is the coefficient of permeability. The last term, ?ug/?t, is the undrained rate of pore pressure buildup which is calculated through empirical findings of development of pore water pressure in granular soils under cyclic loading conditions.

(2)

(3)

(4)

The above relation is an empirical relationship and ? is the empirical constant, which depends on the soil type. N is the number of uniform shear stress cycles undergone by the soil at the given depth during the earthquake loading and N1 is the number of cycles at the same stress level required to cause liquefaction under undrained conditions (EERC, 1975).

Drains can be modeled in four different ways. In the first case, there is no drain, thus allowing the site to be analyzed prior to remediation. The second method uses a “perfect” drain, similar to the LARF (EERC, 1976) code, in which excess pore pressures below the water table are assumed uniform. Thus, if the water level in the drain starts out at the ground surface, the excess pore pressures in the drain will always be zero. If however, the water level is below the ground surface, water can accumulate within the drain, leading to a uniform rise in the excess pore pressures within the drain, thus retarding subsequent entry of water. The third approach follows an Onoue-type analysis (Onoue, 1988) in which the drain is represented by a soil element with both horizontal and vertical hydraulic conductivities which can be set independently of the soil outside the drain. Thus, a very high permeability channel can be created. As in the “perfect” drain method, the code has the additional capability for allowing water to accumulate within the drain itself. The boundary conditions are as Figure 4.

The input parameters were applied as following. Only one layer in the soil profile was defined due to the relatively uniform sand deposit prepared throughout the tests. The depth to static ground water table was set to zero since water level was at the ground surface in the model tests. An effective vertical stress of 3 kPa was used to model the foundation overburden pressure and a hydraulic conductivity of 1.25x10-4 m/s was assigned. Also coefficient of volumetric compressibility of 5x10-5 m2/kN was used, provided by some consolidation experiments carried out on Firuzkooh sand.

The number of cycles to cause liquefaction in each test was extracted from the acceleration time history response. Equivalent number of cycles due to earthquake loading was selected from the recorded input acceleration time history (acceleration transducer A4) of each test and 10 second duration was used. According to tests specifications relative density (Dr), total layer thickness and total unit weight of 12.65%, 0.6 m and 18.56 KN/m3 were used respectively. An axisymmetric analysis with a variable compressibility was performed. The gravel drains were modeled with a constant hydraulic conductivity of 0.25 m/s, an outside radius of 2.5 cm and a tributary area radius of 20.34 cm according to the experiments. Besides, the compacted sand piles were modeled as uniform soils with higher relative densities and less hydraulic conductivities.

Figure 4. Boundary conditions used in the numerical modeling (EERC, 1997)

Experimental results

Acceleration time histories

The acceleration responses of the models are shown in Figures 5-7. A very clear reduction of acceleration occurred after the second cycle in test O (Figure 6) which indicates severe liquefaction and softening of the soil particularly at the positions of A1 and A2. This kind of softening also happened in test G (Figure 6) after the seventh cycle. The presence of gravel drains delayed the softening of the soil; however it didn’t mitigate it completely. The larger amplitudes of acceleration response in test G implies that the seismic shaking was transferred, with some amplification from the base of the deposit up to the footing by the composite ground of sand and gravel drains. The loss of strength in test G was larger and faster than that in test C (Figure 7). At corresponding locations, attenuations were smaller and delayed in test C comparing to tests G and O. This can be attributed to the reinforcing effect of the compacted sand piles. In general, throughout shaking, model test C behaved in a stiffer manner and the cyclic mobility induced softening occurred gradually after eight cycles of shaking in shallower depths of A1 and A2. Compacted sand piles due to their dilative characteristic appeared to be stronger and degradation of their strength was not very significant. The spikier and an overall stiffer response of the compacted sand piles exhibit a more pronounced cyclic-mobility behavior of the stratum. This cyclic mobility behavior explains why the accelerations reduce at a later time than those of the unremediated ground.

Figure 5. Time histories of accelerations recorded in Test O

Figure 6. Time histories of accelerations recorded in Test G

Figure 7. Time histories of accelerations recorded in Test C

Excess pore water pressure

Time histories of excess pore pressure ratio, ru, recorded at depths of 15cm and 35cm below the center of the foundation are shown in Figures 8-10. Comparing these figures with the acceleration time histories indicates that acceleration amplitudes attenuated due to excess pore pressure buildup since the chronological agreement between the maximum excess pore pressure and the attenuation of acceleration amplitude is clearly seen from these figures. Maximum excess pore pressure ratio (ru) was achieved during some initial cycles and remained almost unchanged within the shaking period. The maximum excess pore pressure ratios in all tests, were almost the same however, the number of cycles causing this maximum ru was different. The gravel drains increased the resistance against liquefaction and ru reached its maximum within a larger number of cycles. A similar behavior was observed in test C with the compacted subsoil. Compaction was able to increase liquefaction resistance more than gravel drains. During shaking, gravel drains were not able to reduce the excess pore pressure considerably; and changes in the behavior of the remediated ground was primarily a result of the stiffening effect of the gravel drains.

After the shaking excess pore water pressures at deeper locations started to dissipate, however they increased in shallower deposits due to the upward movement of water from deeper strata and flows draining from the surrounding far field soils; such a phenomenon was also observed by Liu and Dobry (1997) as well as during the 1995 Kobe earthquake, where upward seepage was observed in Rokko Island an hour after the main event (Shibata et al., 1996). This migration of water may reduce the strength of surface soils and generate "secondary" (or seepage induced) liquefaction, causing large deformations or loss of bearing capacity (EERC, 1975; Yoshimi and Kuwabara, 1973).

Furthermore the excess pore water pressure ratios show that at any specific depth there was a moment after which, the excess pore pressures started to dissipate faster. This is the initial period, where vertical dissipation had not had a chance to get hold on the soil at that corresponding depth and only radial drainage was experienced at that depth.

After shaking the differences in dissipation rates of various tests were remarkable which indicates that gravel drains accelerated the excess pore pressure dissipation after shaking, showing their effectiveness in non-dynamic cases i.e. effectively mitigating secondary liquefaction due to the upward flowing water after earthquake. The deeper pore water pressure used the full drain capacity and overlying deposits waited for the way to be clear. At shallower sections water left through surface rather than the drain itself. Such phenomenon was also observed by Brennan and Madabhushi (2002).

Figure 8. Time histories of excess pore pressure ratio under the foundation centerline in Test O

Figure 9. Time histories of excess pore pressure ratio under the foundation centerline in Test G

Figure 2. Figure 10: Time histories of excess pore pressure ratio under the foundation centerline in Test C

Figure 11. Excess pore water pressure ratio isopiestic lines in Test O

Figure 12. Excess pore water pressure ratio isopiestic lines in Test G

Figure 13. Excess pore water pressure ratio isopiestic lines in Test C

The isopiestic lines for excess pore water pressure ratio, five seconds after the shaking had started, are depicted in Figures 11-13. It can be observed that the excess pore water pressures right under the foundation never reached zero effective stress conditions in any of the tests and the corresponding excess pore pressure ratios never gained a value of 100%. This is due to the presence of foundation, other wise the shaking intensity was enough to develop complete liquefaction. Looking at the locations farther from the effect of the foundation, shows that achieving higher excess pore water pressure ratios was possible. In other words, ru values were lowest immediately below the foundation, revealing a significantly less contractive soil response within the foundation soil. If there was no foundation placed on the soil the soils at shallower depths would be more susceptible to liquefy than soils at deeper depths. Centrifuge and other 1g shaking table tests on foundations supported by sandy deposits have shown that excess pore water pressure was generally smaller under the foundation than the free field (Laak et al., 1994; Whitman and Lambe, 1988; Liu and Dobry, 1997; Adalier et al., 1998; Adalier et al., 2002; Koga and Matsuo, 1990) i.e. the superposed footing loads caused a beneficial reduction of liquefaction potential. This is similar to sloping ground conditions, in which the maximum achievable pore water pressures are suppressed by the static driving shear stress and may not reach full liquefaction, no matter how many additional loading cycles are applied. Koga and Matsuo (1990) attributed it to the inability of the earlier liquefied free-field soil to provide lateral stress more than its initial vertical effective stress to the foundation soil.

Settlement

Earthquake induced settlement frequently causes damages to structures supported on shallow foundations, damage to utilities that serve pile-supported structures, and damage to lifelines that are commonly buried at shallow depths. Failure is observed in the form of considerable subsidence due to the following reasons:

Softening of the subsoil resulting in lateral deformations of the soil which can be indicated by the curved shapes of the dyed sand under the footing in Figure 14.

Loss of shear strength, which causes a punching settlement of the model foundation.

General settlement of subsoil following liquefaction, which is caused by excess pore pressure dissipation during and after the earthquake.

Earthquake shaking causes excess pore pressure to build up under undrained conditions, thereby reducing the effective stress. The excess pore pressure produces a hydraulic gradient that drives the pore water out of the voids. The flow of water reduces the hydraulic gradient until the excess pore pressure completely dissipates. As the water flows from the voids, the volume of the soil decreases. The magnitude of the volume change increases with the magnitude of the seismically induced excess pore pressure. Even small excess pore pressures which may not be sufficient to produce flow liquefaction or cyclic mobility, can produce some post-earthquake settlements. The time required for this settlement to occur depends on the permeability and compressibility of the soil, and on the length of the drainage path (Kramer, 1996).

Among the mentioned reasons, the effects of the first and second mechanisms were more remarkable.

Figure 14. Failure patterns observed after shaking of (a) Test O, (b) Test G

Figure 15 shows the recorded time histories of foundation settlements. The initial settlement rate of tests O and G was larger than that of test C; however after this initial fast settlement the settlement rates in all of the tests become very similar. This implies a similar initial settlement mechanism in tests O and G which because of its high rate can be attributed to a loss of shear strength and punching type of settlement. This observation is analogical to the behavior of the response acceleration and excess pore pressure ratio which were discussed earlier. Due to the larger voids in gravel drains this type of settlement was larger in gravel drains. Afterwards the rate and mechanisms of settlement becomes a softening type in all of the tests which was manifested by lateral deformations in the subsoil as shown in Figure 14 for tests O and G. Furthermore the amount of settlement due to softening and lateral deformation in all of the tests was almost the same and equal to 50mm. This can be justified by the same amount of the maximum pore water pressure ratio developed in all of the tests.

Settlement in test C seems to be reasonably controlled. Compaction could reduce the rate and the maximum amount of settlement as well as delaying the settlement initiation time. Figure 10 shows an initial negative excess pore water pressure developed at 15 cm below the foundation centerline in test C and correspondingly the ground surface was observed to have a smaller subsidence. Settlement in test C was due to the migration of underlying foundation soil towards the free field and was partially masked by heave.

Figure 15. Foundation settlements observed in different tests

In all of the tests regarding the ten second shaking period, most of the foundation settlements occurred during shaking, and a smaller portion of the total settlement was caused by post-shaking soil reconsolidation due to excess pore water pressure dissipation. The less efficiency of gravel drains can be realized by comparing these two phenomena that excess pore water pressure was reduced mostly after shaking and most of the settlement occurred during shaking. However, more recently, gravel drains have been used to reduce post-earthquake settlements resulting from soil consolidation due to excess pore pressure dissipation and secondary liquefaction (EERC, 1997).

NUMERICAL RESULTS

The numerical analysis presented in figures 16-18 illustrate that the numerical method was only able to predict the excess pore pressure ratio correctly in test O. In the other tests using gravel drains and compacted sand piles, the computed excess pore pressure ratios were much smaller than the measured values. However the trends in each test were predicted correctly i.e. the excess pore pressure ratios in deeper layers were larger than those in shallower layers.

Figure 16. Time histories of excess pore pressure ratio under the center of foundation by the numerical analyzing of Test O

Figure 17. Time histories of excess pore pressure ratio under the center of foundation by the numerical analyzing of Test G

Figure 18. Time histories of excess pore pressure ratio under the center of foundation by the numerical analyzing of Test C

By not considering the softening and loss of strength of the subsoil, which had a significant contribution in foundation settlements, the computed settlements were much smaller than the recorded values in Figure 19. Ignoring the loss of shear strength mechanism has made the settlement of the gravel drains to be much smaller than the non-remediated case. However the efficiency of the compacted piles in reducing the rate and amount of settlement is demonstrated in the numerical simulations too.

Figure 19. Foundation settlement in different tests obtained by numerical analysis.

Figure 20. Excess pore water pressure ratio isopiestic lines from numerical simulation of Test O.

Figure 21. Excess pore water pressure ratio isopiestic lines from numerical simulation of Test G.

Figure 22. Excess pore water pressure ratio isopiestic lines from numerical simulation of Test C.

The isopiestic lines of the excess pore water pressure ratios, obtained from the numerical simulations, are depicted in Figures 20-22, and can be compared with their counterparts in Figures 11-13. Higher excess pore pressure ratios in test O and lower ones in tests G and C are the effects of the ground remediation. The numerical code applies the overburden pressure uniformly all over the surface, that's why the contours are not the same shape as the actual ones in Figures 11-13, especially in tests O and C. However, in corresponding depths the computed excess pore water pressure ratios agreed very well with the actual ones in test O but in tests G and C they were much less than the real values. In addition the numerical code was not able to simulate the dilative response and the consequent negative excess pore pressure beneath the foundation in test C. Besides the pore pressure reduction capability of the gravel drains was expressed by the numerical simulation.

CONCLUSIONS

A series of 1g shaking table tests were carried out to evaluate the performance of two common ground improvement techniques, gravel drains and sand compacted piles. For comparison a test with no improvement was also performed to compare the behaviors. Following the experimental work, these techniques were modeled with a finite element code.

The experiments presented that the improvement provided by the dynamic compaction method not only reduces the excess pore pressure, but also the stiffer compacted sand piles provided higher overall foundation shear strength and bearing capacity, preventing settlements better than gravel drains. Settlement was mainly due to migration of underlying foundation soil towards the free field (lateral spreading) in the tests with sand compacted piles. However with the gravel drains, the settlement was considerably raised due to the loss of shear strength and punching attained by the shaking process. Since gravel is a frictional material possessing negligible cohesion, confining pressure applied by the soil is of paramount importance. Sufficient vertical stress or confining pressure might be required to engage the full reinforcing effect of the gravel drains. This confinement can be obtained with the weight of the structure and method of installation. The installation process should embed the drains tightly within the soil matrix, while preventing mixing of the in-situ soft soil with the drain material. Such contamination not only compromises the strength of the columns, but also reduces their drainage capacity. Furthermore it was observed that the intensity of shaking was enough to produce liquefaction, however the excess pore pressure ratio never reached 100% under the center and edge of the foundation due to the static driving shear stress.

In general, the test results suggest that compacted sand columns as stiffer elements are likely to be more viable solutions in mitigating liquefaction where the only possible mitigation benefit is from the stiffening stress concentration criterion. The real advantages of drains may lie not in preventing liquefaction but in reducing the time that deposits spend in a liquefied state. This should prevent problems caused by prolonged post-earthquake excess pore pressures and secondary liquefaction, such as rotation of bridge piers or high backfill pressures behind quay walls and this implies that the efficiency of gravel drains would be improved in small duration earthquakes.

The numerical analysis illustrated that simulating the liquefaction phenomenon by only considering pore pressure generation and dissipation, would be far from the margins of safety, especially in the remediated cases. Finally it should be emphasized that the effectiveness of the improvement methods depends not only on the mechanism of their behavior but also on the quality and quantity of the employed techniques.

REFERENCES

Adalier, K., A. Elgamal, and G.R. Martin ((1998) “Foundation liquefaction countermeasures for earth embankments.” Journal of Geotechnical & Geoenvironmental Engineering, ASCE, Vol. 124, No.6, 500-517.

Adalier, K., A. Elgamal, J. Meneses, and J.I. Baez ((2003) “Stone columns as liquefaction countermeasure in non-plastic silty soils.” Soil Dynamics and Earthquake Engineering, Vol. 23, 571-584.

Adalier, K., T.F. Zimmie, and A. Pamuk ((2002) “Seismic behavior of rubble-mound moles on sandy marine deposits.” Proceedings of the 21st International Conference on Offshore Mechanics and Arctic Engineering, Oslo, 281-289.

Baez, J.I. (1995) “A design model for the reduction of soil liquefaction by vibro-stone columns.” University of Southern California, PhD Thesis.

Boulanger, R.W., I.M. Idriss, D.P. Stewart, Y. Hashash and B. Schmidt (1998) “Drainage capacity of stone columns or gravel drains for mitigating liquefaction.” ASCE Geotechnical Special Publication, Vol. 75, No.1, 678-690.

Brennan, A.J., and S.P.G. Madabhushi (2002) “Effectiveness of vertical drains in mitigation of liquefaction.” Soil Dynamics and Earthquake Engineering, Vol. 22, 1059-1065.

Das, B.M. (1983) “Fundamentals of Soil Dynamics.” Elsevier Science, New York.

Earthquake Engineering Research Center (1997) “FEQDrain: A finite element computer program for the analysis of the earthquake generation and dissipation of pore water pressure in layered sand deposits with vertical drains.” EERC, University of California, Berkeley, CA, Report No. EERC 97-15.

Earthquake Engineering Research Center (1995) “Geotechnical reconnaissance of the effects of the January 17, 1995 Hyogoken-Nanbu earthquake, Japan.” EERC, University of California, Berkeley, CA, Report No. EERC 95-01.

Earthquake Engineering Research Center (1976) “Stabilization of Potentially Liquefiable Deposits Using Gravel Drain Systems” EERC, University of California, Berkeley, CA, Report No. EERC 76-10.

Earthquake Engineering Research Center (1975) “The generation and dissipation of pore water pressure during soil liquefaction” EERC, University of California, Berkeley, CA, Report No. EERC 75-26.

Hamada, M., R. Isoyama, and K. Wakamatsu (1996) “Liquefaction-induced ground displacement and its related damage to lifeline facilities” Soils and Foundations, Special issue on the geotechnical aspects of the January, 17 1995 Hyogoken-Nanbu earthquake, 81-97.

Inagaki, H., S. Iai, T. Sugano, H. Yamazaki, and T. Inatomi (1996) “Performance of Caisson Type Quay Walls at Kobe Port” Soils and Foundations, Special issue on the geotechnical aspects of the January, 17 1995 Hyogoken-Nanbu earthquake, 119-136.

Kamon, M., T. Wako, K. Isemura, K. Sawa, M. Mimura, K. Tateyama, and S. Kobayashi (1996) “Geotechnical Disasters on the Waterfront.” Soils and Foundations, Special issue on the geotechnical aspects of the January, 17 1995 Hyogoken-Nanbu earthquake, 137-147.

Koga, Y. and O. Matsuo (1990) “Shaking table tests of embankments resting on liquefiable sandy ground.” Soils & Foundations, Vol. 30, No.4, 162-174.

Kramer, S.L. (1996) “Geotechnical Earthquake Engineering” Prentice-Hall, New Jersey.

Laak, V.P., A.W. Elgamal and R. Dobry (1994) “Design and performance of an electro-hydraulic shaker for the RPI centrifuge.” Proceedings of the Centrifuge' 94, 139-144.

Liu, L., and R.Dobry (1997) “Seismic response of shallow foundation on liquefiable sand.” Journal of Geotechnical & Geoenvironmental Engineering, ASCE, Vol. 123, No.6, 557-567.

Onoue, A. (1988) “Diagrams Considering Well Resistance for Designing Spacing Ratio of Gravel Drains.” Soils and Foundations, Vol. 28, No.3, 160-168.

Orense, R.P., I. Morimoto, T. Yamamoto, T. Yumiyama, H. Yamamoto, and K. Sugawara (2003) “Study on wall type gravel drains as liquefaction countermeasure for underground structures.” Soil Dynamics and Earthquake Engineering, Vol. 23, 19-39.

Pestana, J.M., C.E. Hunt, A.M. Kammerer, and R.R. Goughnour (1998) “Use of Prefabricated Drains with Storage Capacity for Reduction of Liquefaction Potential.” Geotechnical Engineering Report, University of California at Berkeley, CA, UCB/GT/98.

Seed, H.B., and J.R. Booker (1977) “Stabilization of potentially liquefiable sand deposits.” Journal of Geotechnical Division, ASCE, Vol. 103, No.7, 757-768.

Shibata, T., F. Oka and Y. Ozawa (1996) “Characteristics of ground deformation due to liquefaction.” Soils and Foundations, Special issue on the geotechnical aspects of the January, 17 1995 Hyogoken-Nanbu earthquake, 65-79.

Sonu, C.J., K. Ito, H. Oishi, and H.Seed (1993) “Liquefaction and the gravel drain.” Civil Engineering, Vol. 63, No.12, 58-60.

Whitman, R.V., and P.C. Lambe (1988) “Earthquake like shaking of a structure founded on saturated sand.” Proceedings of the International Conference on Geotechnical Centrifuge Modeling, Paris, 529-538.

Yasuda, S., K. Ishihara, K. Harada, and N. Sinkawa (1996) “Effect of soil improvement on ground subsidence due to liquefaction.” Soils and Foundations, Special issue on the geotechnical aspects of the January 17 1995 Hyogoken-Nanbu earthquake, 99-108.

Yoshimi, Y., and F. Kuwabara (1973) “Effects of subsurface liquefaction on the strength of surface soil.” Soils and Foundations, Vol. 13, No.2, 67-81.