Four centrifuge experiments were conducted at Rensselaer Polytechnic Institute to explore the impact of stratum thickness on the behavior of silty sand soil under seismic conditions, specifically focusing on pore water pressure dynamics during and after shaking. This research also aimed to calculate the overburden pressure correction factor (K sigma) and to assess how variations in stratum thickness affect this parameter under field-simulated single drainage conditions. The experimental design includes tests on 2.5 and 5 m soil strata subjected to low (1 atm) and high (6 atm) overburden pressures. All tests maintained a consistent relative density of 45% and were conducted under a centrifugal acceleration of 45 g, with each test subjected to a 10-cycle sinusoidal wave base shaking. Instrumentation captured data that enabled detailed analysis of acceleration time histories, excess pore pressure ratio time histories, pore pressure profiles, pore pressure dissipation, settlement, shear stress ratios, and shear strains. Notably, despite the presence of a free boundary at the top of the silty sand layer, the maximum excess pore pressure ratio (ru) consistently occurred in the upper half of the stratum across all tests. The dissipation of excess pore pressure was observed to be faster in the 2.5 m tests as compared to the 5 m counterparts. A key finding from this study is the inverse relationship between stratum thickness and the field K sigma correction factor, with higher K sigma values observed for the 2.5 m stratum compared to the 5 m stratum. This outcome suggests that partial drainage may influence this factor, potentially elevating it to values above 1, contrary to the values typically predicted by current state of practice (SoP) that rely heavily on undrained triaxial tests. The SoP neglects the critical effects of drainage and stratum thickness. This study underscores the need for a reevaluation of current methodologies to incorporate these influential factors more accurately in the assessment of soil liquefaction potential under seismic loading.

The effect of high overburden pressure above 1 atm on sand liquefaction potential is typically evaluated based on cyclic undrained testing, with the overburden pressure correction factor, K-sigma < 1.0 and K-sigma decreasing as the pressure increases. Recent centrifuge experiments of a prototype 5 m-thick clean sand layer having a permeability 1.2 x 10(-4) m/s, with free drainage at the top and subjected to 1 and 6 atm overburden pressures, show that a high overburden pressure may increase partial drainage. As a result, the measured field overburden pressure factor, (K-sigma)(field) was estimated to be >1.0 instead of <1.0 in these centrifuge tests. A parametric study is presented here that extends the centrifuge results for a relative density, D-r = 45% and free top drainage, utilizing a high-fidelity, calibrated numerical model (P2Psand in FLAC 3D). A stepped acceleration base input is used that ensures a uniform cyclic shear stress amplitude at the elevation of maximum pore pressure ratio. The main parameters varied in the numerical study are overburden pressure, sigma'(v0) (1-12 atm); sand permeability, k (10(-6) to 10(-3) m/s); and sand layer thickness, H (2-10 m). A new drainage factor, K-dr >= 1.0, is proposed to separate the usual undrained K-sigma from the effect of partial drainage. The recommendation is to evaluate the overburden pressure factor to be used in liquefaction charts, (K-sigma)field, as the product of two factors, (K-sigma)(field) = K-sigma x K-dr. The study shows that for constant sigma'(v0) = 6 atm, K-dr approximate to 1.4-1.7 when k = 10(-4) to 10(-5 )m/s, even for a very thick sand layer of H = 10 m. Still for a constant sigma'(v0) = 6 atm, K-dr decreases considerably to values close to 1.0 for a low k approximate to 10(-5) m/s, when the layer thickness is H = 7 or 8m or greater. And for constant k approximate to 10(-4) m/s and sigma'(v0) = 12 atm, K-dr approximate to 1.4 if H = 5(-10) m, increasing to K-dr approximate to 1.8 if the layer thickness decreases to H = 2 m. The parametric study provides insight on when the current State-of-Practice of using only undrained K sigma becomes too conservative and should be supplemented with consideration of partial drainage through use of factor K-dr. Ready-to-use charts are provided to evaluate K-dr and (K-sigma)(field) for a D-r = 45% sand layer with free top drainage.

Two centrifuge experiments conducted at Rensselaer Polytechnic Institute (RPI) aimed to evaluate the effectiveness of the used centrifuge scaling laws and validate results obtained within the project, related to the liquefaction behavior of silty sand soils under simulated field drainage conditions. These experiments, replicating a 5-m thick silty sand layer under 1 atm overburden pressure and featuring a double drainage condition, were performed at two different centrifugal accelerations. Due to the difficulties encountered in saturating silty sand, this paper presents a bottom-up saturation method, verified through assessments of remaining air volume after saturation. Despite differing g-levels, the tests consistently demonstrated similar behaviors in terms of acceleration, pore pressure buildup and dissipation, and stress-strain responses, validating the saturation and modeling technique for silty sand.

This article studies the effect of overburden pressure on liquefaction behavior of saturated Ottawa F65 sand using centrifuge experiments as well as numerical simulations. A series of centrifuge tests were conducted simulating a 5 m layer of Ottawa sand having different relative densities and subjected to overburden effective pressures of 1 atm and 6 atm. The objective was to study the pore pressure response of sand to base acceleration under low and high overburden pressure. The sand layer had a bottom impervious boundary and a top pervious boundary, approximating an idealized field condition. The analysis shows that sand during liquefaction is not fully undrained but rather partially drained. It was also found that partial drainage was more significant at 6 atm than at 1 atm due to the increased coefficient of consolidation, cv at higher overburden pressure.

Practical application of the state of practice field liquefaction charts at low and high overburden is assessed with centrifuge testing. Available results of four centrifuge experiments of a 5 m-thick clean sand layer having free drainage at the top are used as case histories. The centrifuge tests include relative densities of 45% and 80% and overburden pressures of about 1 atm and 6 atm. The shear wave velocity (Vs) was measured in the centrifuge models using bender elements. The cone penetration testing (CPT) tip resistance was estimated with correlations from the literature. Vs-based and CPT-based liquefaction charts proposed by Andrus and Stokoe (2000) and Idriss and Boulanger (2008) were used. Both liquefaction charts worked well for the centrifuge tests having an effective overburden, σ'v0, of about 1 atm. This was expected as the charts were originally calibrated with field earthquake case histories having σ'v0 < 2 atm. Both liquefaction charts are too conservative for the centrifuge tests having an effective overburden, σ'v0 of about 6 atm. The agreement at 6 atm deteriorates even further when state-of-practice overburden correction factors, Kσ < 1 are applied, including a prediction of liquefaction when the measured maximum pore pressure ratio in the test was only 0.6. Values of Kσ > 1 were found to provide good agreement with the charts for the 6 atm tests.

Liquefaction strength curves (LSCs) are commonly generated based on undrained stress-controlled cyclic testing to define the liquefaction potential of sands. However, these LSCs do not consider the effect of partial drainage on liquefaction potential. This is mainly because there are very limited experimental techniques to run stress-controlled tests for partially drained sands. This manuscript proposes a procedure to numerically simulate a saturated sand column with a free drainage boundary at the top (or top and bottom), accelerated horizontally at the bottom with constant acceleration cycles of duration designed to generate constant cyclic shear stress histories in the column. This is done for columns subjected to low and high overburden pressures. The paper starts with numerical simulations of a typical undrained cyclic direct simple shear, CDSS tests by applying a stepped velocity wave to a single soil element, an already established approach. This approach only works for a single element in undrained condition which is fixed at the base. Then, the proposed procedure is implemented by applying stepped acceleration time histories at the base of soil columns having one or two drainage boundaries. In these stepped acceleration runs, the durations of the acceleration cycles are controlled to achieve a partially drained stress-controlled CDSS kind-of-loading at specific elevations within the soil columns. The technique is used to show the effect of partial drainage on sand liquefaction behavior at both low and high overburden pressures. The results show that the effect of drainage is much more significant at higher overburden. The existence of both top and bottom drainage boundaries resulted in less liquefaction vulnerability, as compared to having only one drainage boundary at the top of the sand column. The results show that liquefiable sand layers in the field may be less prone to liquefaction under high overburden pressure than predicted by the current state-of-practice, which relies mainly on undrained small-scale cyclic tests.

An extensive instrumented quay wall model embedded in the liquefiable soils of various permeabilities subjected to a base shaking in 120 g can duplicate the field damage patterns and capture the key mechanisms that may not be disclosed in the field investigation after earthquakes. The test results show that the backfill in the soil-structure interaction zone and the foundation soils beneath the quay wall still contained the lower excess pore water pressures during shaking even if liquefaction occurred in the far field. The vibration modes of rotational and translation are in phase each other for caissons in the soil of lower permeability, while those in the higher permeability soils are 90 degrees out of the phase and that in the dry sand are 180 degrees out of phase during 1 Hz base excitation. The wall will produce the larger seaward horizontal displacement and tilt provided that the combinations of two vibration modes make the foundation soil weaker due to high pore water pressure during the wall is seaward accelerating.

The article presents simulations of the seismic liquefaction response of dense and loose clean Ottawa sand under low and high overburden in the centrifuge, using Program FLAC3D and the P2Psand constitutive model. P2Psand was initially calibrated in the paper with cyclic stress-controlled triaxial tests and then modified with information from two centrifuge experiments. The calibrated model was used to simulate four centrifuge tests covering relative densities from 45% to 80% and overburden pressures from about 100 kPa (similar to 1 atm) to 600 kPa (similar to 6 atm). The four numerical computations were fully coupled effective stress simulations that allowed for pore water pressure buildup and dissipation at every time step. The calibration yielded very good matches between numerical and experimental results in all four centrifuge experiments, and calibrated P2Psand input parameters are suggested for practitioners in similar clean sands. The simulations confirmed the increased diffusivity of the sand layer under high overburden obtained before from the centrifuge results. The reason is that P2Psand assumes that the sand bulk modulus is proportional to the square root of the mean effective stress, consistent with the similar conclusion derived from the centrifuge data by the authors. The calibrated P2Psand model was also used to perform "no flow" simulations of the same four centrifuge experiments, in which fluid flow was not allowed during or after shaking. No flow simulations are used sometimes in practice to reduce numerical effort, on the assumption that liquefaction in the field is mostly undrained. It was found that this assumption may produce useful engineering results for a low overburden of 1 atm, but it may become increasingly incorrect and too conservative at higher overburden. The reason is that for certain field conditions, fluid flow becomes more significant during shaking under high overburden due to increased sand diffusivity.

This paper studies the effect of a high effective overburden pressure [sigma'(v0) = similar to 600 kPa (6 atm)] under two drainage conditions on the field liquefaction behavior of saturated Ottawa sand. A series of eight centrifuge experiments with relative densities D-r = 45% and 80% and base shaking are considered that include a 5-m saturated sand layer under a pressure of either sigma'(v0) = similar to 100 kPa (1 atm) or similar to 600 kPa (6 atm). Four of the tests had single drainage at the top of the layer (SD), whereas the other four tests had double drainage (DD) at top and bottom. The four SD test results had been reported before, whereas the four DD tests are new. A novel centrifuge technique was developed to achieve the double-drainage boundary condition of two pervious boundaries at the top and bottom of the sand layer, using a geocomposite at the bottom. Measured responses are compared at the same sigma'(v0) between SD and DD tests having the same input acceleration, as well as between SD and DD tests where the shaking induced a similar maximum excess pore pressure ratio (r(u))max approximate to 0.8. These comparisons include acceleration time histories, excess pore pressure time histories and profiles during and after shaking, and stress ratio and shear strain time histories. Comparisons between corresponding tests at similar to 100 and similar to 600 kPa (1 and 6 atm) revealed significantly more partial drainage at similar to 600 kPa (6 atm) than at similar to 100 kPa (1 atm), with even more significant variation in excess pore pressures in the DD than in the SD tests. Best estimates of field overburden pressure correction factors at -600 kPa (6 atm), K-sigma were obtained, were obtained from the centrifuge results with two independent methods for a failure criterion of (r(u))(max) = 0.8. Those K-sigma = 1.2-1.3 > 1.0 for both SD and DD drainage conditions due to the significantly lower compressibility of the sand at similar to 600 kPa (6 atm). The results further emphasize the important role partial drainage may play in the field during shaking at high sigma'(v0) on the excess pore pressures and values of K-sigma. (C) 2021 American Society of Civil Engineers.

This article is the second of two companion papers studying the effect of a high overburden pressure on the liquefaction behavior of saturated Ottawa sand. A series of four centrifuge experiments were conducted simulating a 5-m prototype layer of this sand in the field, having two relative densities and subjected to overburden effective pressures, sigma(v0)', of similar to 100 and 600 kPa (1 and 6 atm). The layer was on a rigid impervious base and could drain freely at the top. This was supplemented by undrained stress-controlled and strain-controlled cyclic triaxial tests on the same sand consolidated at 1 and 6 atm. The laboratory undrained overburden pressure factor at 6 atm obtained from the triaxial tests on loose sand, K-sigma=0.85, is consistent with the state of practice (SoP), which assumes that K(sigma)1.0 and K-sigma decreases with sigma(v)0 '. However, in the centrifuge experiments and sigma(v0)'=6 atm, the field K-sigma=1.28 for loose sand and K(sigma)1.15 for dense sand. The discrepancy is due to more significant partial drainage during shaking in the 6-atm centrifuge models. Although the excess pore pressures at the bottom of the sand layer seem to have been close to undrained in the four experiments, they were much smaller at shallower elevations in the 6-atm tests compared with the 1-atm tests. Further analysis is conducted by evaluation in the four centrifuge experiments of the coefficient of consolidation, cv, during the dissipation phase. This is done using the recorded pore pressure and settlement data. It is concluded that c(v) was two to four times greater during dissipation in the 6-atm centrifuge tests. The reason is the increase-also by a factor of two to three-of the drained constrained volumetric stiffness of the sand, M '=1/mv, when going from 1 to 6 atm. This finding plus other data from the literature suggest that for a range of sands, layer thicknesses, field conditions, earthquake shaking, and values of sigma(v0)', both M ' and c(v) may increase proportionally to root sigma(v0)', with the field K-sigma > 1.0, and with K-sigma increasing instead of decreasing with sigma(v0)'. (c) 2020 American Society of Civil Engineers.

This paper introduces a series of centrifuge tests-under low and high confining pressures-designed to study the effect of effective overburden pressure on liquefaction potential of clean sand. All the centrifuge tests simulate about 5 m saturated clean sand deposit under effective overburden pressure of 1 and 6 atm with relative density ranging from 45% to 80%. To achieve the targeted overburden pressures, a dry layer of lead shots with different thickness was deposited on top of the clean sand. Viscous fluid was used for saturation to keep constant prototype permeability. All the centrifuge tests were subjected to 10-cycle sinusoidal seismic motions with different prototype peak acceleration to achieve the targeted maximum excess pore pressure buildup. Acceleration, pore pressure build-up and dissipation, and shear wave velocity were monitored, recorded, and analyzed during or after shaking. Acceleration amplification was observed in experiments conducted under low confining pressures, while de-amplification was found in experiments conducted under high confining pressures.

This article is the first of two companion papers studying the effect of overburden pressure on the liquefaction behavior of saturated Ottawa sand. A series of four centrifuge tests were conducted simulating a 5-m layer of this sand having two different relative densities, and subjected to overburden effective pressures of similar to 100 and 600 kPa (1 and 6 atm). The objective was to study the pore pressure response of the soil to base acceleration under low and high pressure, including evaluation of the overburden pressure factor K-sigma for idealized field drainage conditions. The sand layer had a bottom impervious and a top pervious boundary, approximating a common field situation. A novel experimental technique was developed using a dry lead shot layer to provide the necessary high level of pressure. The performances of the sand layer under low and high confining pressure were compared in terms of times histories and profiles of excess pore pressures, cyclic stress ratios (CSR), and cyclic shear strains gamma c, with some of the parameters determined using system identification. It was found that pore pressure dissipation started earlier at shallower depths, and that partial drainage was more significant in the 6-atm than in the 1-atm tests. Field overburden pressure correction factors at 6 atm, K-sigma, obtained from the centrifuge tests for (r(u))(max)=0.8 in 10 cycles of shaking and including the partial drainage effect, were found to be higher than 1.0 for both D-r=45% and 80% This is different from the usual laboratory undrained K-sigma < 1 based on cyclic triaxial and simple shear laboratory tests and reflected in the current state of practice. The discrepancy is related to the more significant effect of partial drainage and deviation from the undrained assumption at the higher confining pressure for the field drainage and other conditions of these centrifuge tests. (c) 2020 American Society of Civil Engineers.

AbstractThis article presents numerical simulations investigating pore pressure buildup of a sand layer with a free drainage boundary at the top under both low and high overburden pressures and sub...

Preshaking by previous seismic activity can significantly affect the liquefaction resistance of saturated sands. Field evidence, cyclic laboratory testing and centrifuge modeling show that: (i) earthquake events that build up excess pore pressures short of liquefaction strengthen the soil; and (ii) liquefying earthquakes may weaken the soil. The paper presents the results of a centrifuge test (Experiment 2), where a 6 m saturated loose silty sand deposit was subjected to 52 successive base shakings. Three types of shakings were used: preshaking Events A, with a shaking duration of 5 cycles, and stronger Events B and C having a duration of 15 cycles, with Events C having the largest input accelerations. A total of 35 Events A, 9 Events B, and 8 Events C were applied at the base of the model in an alternating sequence. All Events C liquefied the deposit, with the Events B inducing liquefaction at the beginning but not at the end of Experiment 2. Events A generally induced excess pore pressures but not liquefaction. In addition to the pore pressures, horizontal accelerations, settlement and densification, and the soil shear wave velocity, V-s, were also monitored during the test. The results were compared with those of Experiment 1, reported in a previous publication, where a similar 6 m deposit of the same silty sand was subjected to a different sequence of shakings A and B which did not include any Event C. Conclusions are drawn on the effect of the strong Events C on the deposit's response. Events A, B and C of Experiment 2 are plotted on existing V-s-based liquefaction charts. The charts predict well the liquefaction response of the deposit at the beginning of Experiment 2. On the other hand, for shakings near the end of the test, the charts predict liquefaction for the Events B, which by this time have stopped liquefying the deposit due to the previous history of shakings. This result is consistent with the field evidence including the presence of a number of "false positives" in the charts for silty sand sites in the Imperial Valley of California, an area of intense seismic activity.

An examination of dynamic responses of clean sand under high effective confining pressure (σv′) is conducted with two centrifuge tests using laminar container. The centrifuge experiments simulate a 5 m saturated clean sand layer under high effective overburden pressure of 6 atm with the relative density of 40%. To achieve the targeted overburden pressure, a dry layer of lead shot with an appropriate thickness was deposited on the top of the clean sand. Viscous fluid was used for saturation to keep a constant prototype permeability. The centrifuge tests were subjected to 10 cycles sinusoidal seismic motions with a different prototype peak acceleration to study the effects of pre-shaking history. Acceleration, pore pressure build-up and dissipation, and shear wave velocity were recorded during and after shaking. The recorded accelerations were further analyzed using an established system identification (SI) technique to evaluate the cyclic stresses and strains induced in the soil deposit. Furthermore, the effects of pre-shaking history were analyzed in the aspects of excess pore pressure build-up, shear wave velocity changes, and soil deposit densification.

The effect of preshaking and repeated liquefaction on liquefaction resistance was studied in a large-scale shaking table experiment, in which a sequence of 51 shakings was applied to the base of a 5-m uniform deposit of saturated clean Ottawa sand. Three event types were used in a very intense repeated pattern: mild preshaking Events A, stronger preshaking Events B, and extensive liquefaction Events C, with each Event C typically liquefying most or all of the deposit. Relative density, cone penetration test (CPT) tip resistance, and liquefaction resistance to Events A and B were found to increase significantly throughout the 51-shaking sequence, with the shear wave velocity (Vs) increasing slightly. However, the CPT tip resistance and liquefaction resistance decreased temporarily after each Event C, recovering rapidly with additional preshaking-presumably due to a decrease and subsequent increase in the soil lateral stresses. The results for the different shakings were compared with available CPT- and Vs-based field liquefaction charts, with and without accounting for the fact that the soil deposit was much younger than the case histories covered by the charts (age factor). The liquefaction response for Events A, B, and C was reasonably well predicted by the CPT chart when the age factor was considered, including Events A immediately after liquefaction by an Event C. The implications of the research were discussed for the geologic age, preshaking and liquefaction effects observed in the field, including reliquefaction response of the same site by milder aftershocks after the main earthquake shock.

Brief details of a 1-g full scale laminar box system capable of simulating the liquefaction phenomenon of saturated loose soil deposit up to 6 m deep are presented. The internal dimensions of the box are 5 m in length and 2.74 m in width. The system consists of a two dimensional laminar box made of 40 rectangular laminates stacked on top of each other supported by high capacity and very low friction bearings, a base shaker sitting on the strong floor, two high speed actuators mounted on the rigid blocks, a dense instrumentation array and a close loop hydraulic filling system for building loose sand deposit. Intricate details of the system are presented and its capabilities are illustrated using a recent shake test leading up to liquefaction.

The paper discusses the effect of seismic preshaking history on liquefaction of silty sand soils using case histories in California, as well as experimental data. The case histories are: (a) the response of the Wildlife site in the Imperial Valley to the 2010 El-Mayor Cucapah earthquake (M-w = 7.2, a(max) = 0.15g); and (b) the response of the Treasure Island Fire Station (F. S.) site in the San Francisco Bay area to the 1989 Loma Prieta earthquake (M-w = 6.9, a(max) = 0.16g). Both sites are similar in almost every respect except for their seismic history. The magnitude and intensity of the corresponding earthquakes were also very similar at both locations. While Treasure Island F. S. did liquefy during the shaking, Wildlife did not and was far from liquefaction as indicated by piezometers at the site. The experiments conducted in this research were a crude simulation of the seismic history of the sites. From the results of the experimental simulation as well as the field case histories, it is concluded that preshaking by previous earthquakes is the most probable explanation of the higher liquefaction resistance exhibited by the Wildlife site and other sites in the Imperial Valley of Southern California.