Induced liquefaction experiment in relatively dense, clay ... et al JGR 2009.pdf · Induced liquefaction experiment in relatively dense, ... In this paper we report results from a

Induced liquefaction experiment in relatively dense, clay-rich

sand deposits

Yossef H. Hatzor,1 Haim Gvirtzman,2 Ilia Wainshtein,1 and Itay Orian1

Received 21 July 2008; revised 10 November 2008; accepted 3 December 2008; published 20 February 2009.

[1] In this paper we report results from a controlled blast-induced liquefaction experimentat the field scale. The physical and mechanical properties of the materials at the subsurfaceare characterized by a suite of in situ and laboratory tests, including the StandardPenetration Test (SPT); downhole and cross-hole seismic velocity tests; density, porosity,and gradation tests; and direct shear tests. Since the blast experiment was performedabove groundwater table, the subsurface was saturated by a sequence of controlledinfiltration tests. A 50-kg TNT charge was detonated at a depth of 10 m, and seismicground motions were recorded in a vertical geophone array positioned at a horizontaldistance of 30 m from the blast borehole. Obtained liquefaction features include a waterfountain that erupted from the blast borehole, prolonged bubbling of the water surfaceinside the infiltration trench (a process equivalent to ‘‘sand boils’’ typically observed atsites which have experienced liquefaction), lateral spreading, and surface settlement. Weargue that in contrast to conventional predictions, liquefaction may be induced inrelatively dense silty and clayey sands (shear wave velocity >300 m s�1; relative density= 63–89%) relatively rich in clays (fines content >30%) and that the driving mechanismshould not necessarily be restricted to cyclic shear stress loading.

Citation: Hatzor, Y. H., H. Gvirtzman, I. Wainshtein, and I. Orian (2009), Induced liquefaction experiment in relatively dense, clay-rich

sand deposits, J. Geophys. Res., 114, B02311, doi:10.1029/2008JB005943.

1. Introduction

[2] Soil liquefaction, perhaps the single most importantsource of earthquake damage around the world, has beenassumed traditionally to be generated by the vertical prop-agation of shear waves through a saturated soil columnfollowing strong earthquakes [Seed and Idriss, 1970].Liquefaction may be defined as the transformation of agranular material from a solid to a liquefied state as aconsequence of increased pore water pressure and reducedeffective stress [Marcuson, 1978]. It is commonly assumedthat cyclic shear stresses which develop within the soilcolumn during strong earthquakes induce shear modulusdegradation [Seed, 1979, 1980], leading ultimately to struc-tural collapse of the soil skeleton and rapid pore pressureincrease typically manifesting in ‘‘sand boils’’ on thesurface and irreversible settlement [Seed et al., 1981]. Itwas found experimentally that in relatively soft sedimentsthat contain a certain amount of fines, and where the porespace is completely filled with water, the collapse of the soilskeleton will induce a positive increase in pore pressure,which will in turn reduce the effective stress, followingconventional effective stress and pore pressure concepts

[e.g., Skempton, 1954; Terzaghi, 1943]. When the effectivestress is reduced to zero the soil is said to liquefy.[3] Because of the assumed shear-based failure mecha-

nism, shear modulus degradation and void ratio changesduring cyclic shearing in saturated soils of various relativedensities and clay contents have been the subject of expe-rimental research during the past three decades [Boulangerand Idriss, 2007; Evans et al., 1992; Martin et al., 1978;Mulilis et al., 1977; Vucetic and Dobry, 1991]. Numericalcodes have also been developed to compute one-dimensionalshear wave propagation from an input motion at the base ofthe analyzed column and upward through the strata [e.g.,Schnabel et al., 1972]. Such codes can be modified toincorporate experimentally obtained shear modulus degra-dation and material damping curves, and are used exten-sively throughout the world for one dimensional siteresponse analysis and for liquefaction prediction. Whiletheoretical work is constantly being published [Cristescu,2000; de Groot et al., 2006; Sawicki and Swidzinski, 2007]liquefaction prediction is still largely empirical, and is basedon international documentation of liquefaction events onone hand and on the material properties of the sediments atsite on the other [e.g., Bray et al., 2004; Cetin et al., 2004a,2004b; Chu et al., 2004].[4] The relative density of the material in situ (Dr) is

considered to be of prime importance for liquefactionprediction and is best estimated in the field using correla-tions with the in situ standard penetration test (SPT). Sincethe pioneering work of H. B. Seed on the correlationbetween liquefaction potential and SPT values [Seed and

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 114, B02311, doi:10.1029/2008JB005943, 2009ClickHere

for

FullArticle

1Department of Geological and Environmental Sciences, Ben-GurionUniversity of the Negev, Beer-Sheva, Israel.

2Institute of Earth Sciences, Hebrew University, Jerusalem, Israel.

Copyright 2009 by the American Geophysical Union.0148-0227/09/2008JB005943$09.00

B02311 1 of 22

http://dx.doi.org/10.1029/2008JB005943

Idriss, 1971] a substantial amount of data has been accu-mulated and synthesized from liquefaction events from allover the world [Cetin et al., 2004a; Youd et al., 2001].These correlations suggest that with increasing relativedensity, as scaled by the corrected SPT blow count value(i.e., the number of hammer strikes to advance 0.3 m in thesoil, for strict definition of terms, see for example Idriss andBoulanger [2006]), the liquefaction potential of the materialis decreased; a corrected SPT value of N1,60 = 30 seems tobe the upper bound for liquefaction.[5] The shear wave velocity of the material (Vs) is

another measure of the relative density but is much betterdefined mechanically. A comprehensive compilation of casehistories performed by Andrus and Stokoe [2000] suggeststhat liquefaction potential decreases with increasing shearwave velocity. An overburden stress-corrected shear wavevelocity of Vs1 = 300 m s�1 seems to be an upper bound forliquefaction; Vs,1 is given by Vs,1 = Vs(Pa/s

0o)0.25, where Pa

is atmospheric pressure and s0o is effective stress (forcomprehensive definition of terms see Andrus and Stokoe[2000]).[6] Finally, the fines content (FC), defined as the percent

of total weight of grains smaller than 0.075 mm, plays animportant role because it controls the permeability of thematerial and consequently its potential to develop undrainedconditions in the short term, necessary for the build up ofpore water pressure as a result of shaking. From a recentcompilation [Cetin et al., 2004a] it is evident that liquefac-tion potential increases with increasing fines content; finescontent of 5% seems to be the lower bound for thedevelopment of undrained conditions during and aftershaking. The upper bound of FC is poorly defined althoughmost documented failures are found in sediments with FCbetween 5% and 35%. Silts and sands with fines contentgreater than 35% are therefore considered less prone toliquefaction although liquefaction events in cohesive sedi-ments with up to 48% [Miura et al., 1995], 70% [Kishida,1969] and even 90% [Tohno and Yasuda, 1981] fines havebeen reported from documented case studies in Japan. InNorway liquefaction in clays has been studied experimen-tally [Andersen et al., 1980; VanEekelen and Potts, 1978].[7] While an assumed shear based failure mechanism is

currently the norm in liquefaction prediction, there isnevertheless ample evidence for pressure-induced liquefac-tion in granular materials. Florin and Ivanov [1961] showedexperimentally that sands may transform into a liquid stateby subjecting the soil column to different sources ofpressure waves. In a series of papers, Charlie et al. [1985,1995, 1996] and Charlie and Doehring [2007] discussedliquefaction in granular deposits owing to explosions ofburied charges. They show that explosive-induced groundmotions can alter well water levels and induce liquefactionin water-saturated cohesionless geological profiles. Theyfurther show that the relationship between blast-inducedpore pressure rise and scaled distance to the source is linearover 11 orders of magnitude of energy (Joules) or trinitro-toluene (TNT) equivalent mass (kilograms). Regardingearthquake-induced liquefaction owing to P waves, Lin[1997] demonstrated convincingly that during the magni-tude 7.2, 1995 southern Hyogo prefecture earthquake inJapan liquefaction features in the field were induced by

P wave arrival because they formed during the first 3 to 4 safter the main shock; before S wave arrivals were recordedat the site. Finally, Bachrach et al. [2001] discussedpressure-induced liquefaction using concepts of dynamicporoelasticity and demonstrated that critical conditions forliquefaction may be achieved owing to a resonant mode ofBiot’s type II wave, a mechanism which is compatible withP wave loading.[8] In this paper we present results from a carefully laid

out field experiment where a 50-kg trinitrotoluene (TNT)charge was detonated at a depth of 10 m below groundsurface, in a saturated sequence of silty sands, clayey sands,and clays. Since the experiment was conducted within thevadose zone some 50 m above the groundwater table,saturation was achieved by controlled seepage from a trenchthat was excavated on the surface. Particle velocities weremeasured by a vertical array of geophones beneath thetrench at a horizontal distance of 30 m from the blast.Following the blast, liquefaction features such as sand boils,lateral spreading, and irreversible settlement were observed.[9] On the basis of the results of in situ seismic velocity

and SPT tests conducted in the field along with finescontent analysis conducted at the laboratory, we concludethat blast-induced liquefaction may develop in relativelydense sediments (Dr = 63–89%) with shear wave velocitygreater than 300 m s�1, and with fines content between 30and 60%. Such sediments are not considered ‘‘prone toliquefaction’’ in standard prediction procedures whichassume a shear-based failure mechanism.

2. Site Investigations

2.1. Geological Setting

[10] The study was performed in the southwestern coastalplain of Israel where sands, silts, and clays are abundant.Massive deposition of sands along the shoreline of Israelstarted at the upper Pliocene [Menashe, 2003] and contin-ued until late Holocene. This was associated with loessdeposition, consisting largely of silty-clayey deposits[Crouvi et al., 2008; Enzel et al., 2008; Yaalon and Dan,1974; Zilberman, 1992]. The soils in the study site areaclassified as sandy Regosols, arid Brown soils and sanddunes [Dan et al., 1976]. Shallow (0.5–1.5 m) young sandydeposits cover the entire landscape. Arid calcic Brown soilsare found underneath the sand. In the northern part of thestudy site loessial calcic alluvial soils are either covered byor mixed with the sandy calcic Regosols. This association istypical to the southwestern part of the coastal plain of Israel,as well as of alluvial valleys in between the dunes ofnorthern Negev and northern Sinai deserts [Zilberman etal., 2007].

2.2. In Situ Test Methods

[11] The test site was restricted to a representative area of4 km2, in which six 14-cm-diameter boreholes were drilledto a depth of 20 m, to provide information on physical andmechanical properties of the material. The material wassampled every 1 m in all boreholes for soil classificationtests and determination of index properties such as density,porosity, and fines content. The general layout of the fieldtest program and site investigation campaign is shown inFigure 1.

B02311 HATZOR ET AL.: INDUCED LIQUEFACTION IN DENSE SANDS

2 of 22

B02311

[12] In order to determine the liquefaction potential of thesediments at the site according to standard predictionmethods, SPT and seismic velocity measurements wereperformed in situ, and density, porosity, and fines contentfor the different layers were determined at the laboratory.[13] In situ SPT tests were conducted at the test site when

the material was in its natural water content condition, amoisture condition which will be referred to as ‘‘dry’’throughout this paper. Some samples were dried at the labto actual zero water content for porosity and dry densitydetermination (Table 1), but the term ‘‘dry’’ will be used todiscuss the natural moisture of the material at the site beforecontrolled infiltration was conducted, unless stated otherwise.[14] In situ seismic velocity tests were performed below the

infiltration trench (Figure 1) twice, before and after the firstcontrolled infiltration test. The obtained velocities and elastic

parameters are referred to as ‘‘dry’’ and ‘‘wet’’ parameters, todescribe material moisture conditions before and after the firstinfiltration test, respectively. As will be discussed later, webelieve 100% saturation was only reached after the secondinfiltration test, when the ‘‘wet’’ blast test was conducted.[15] Direct shear tests at the laboratory were performed

on undisturbed samples at their natural water content andtherefore the obtained shear strength parameters represent‘‘dry’’ material.[16] SPT tests were performed following ASTM standard

D1586 in boreholes 1, 2, 3, 4, and 6 (Figure 1), andcorrected N1,60 values [Seed et al., 1985] were calculatedfor every 2-m interval down to a depth of 20 m, in allboreholes. The SPT test involves driving a standard cylin-drical sampler into the bottom of a borehole. The totalnumber of hammer blows required to drive the sampler over

Figure 1. Field test layout: (a) location of exploratory boreholes in test site, the wet blast experimentsite with the infiltration trench and the dry blast experiment site and (b) vertical cross section through theinfiltration trench used for the wet blast experiment. Boreholes B6, B6a, and B6b were used for dry andwet seismic velocity tests conducted before and after the first infiltration phase and prior to the ‘‘wet’’blast experiment. (c) Plan view of infiltration trench.

Table 1. Physical Properties of the Sediments Obtained at the Laboratory and in Situa

Depth (m) USCS FC (%) rdry (g/cm3) wo (%) n ro (g/cm

3) rsat (g/cm3) N N1,60 Dr (%)

2 SM 30.5 1.55 3.8 0.420 1.61 1.97 16 15 80.34 CL 56.7 1.74 8.1 0.356 1.88 2.09 40 31 —6 CL 61.9 1.72 10.5 0.364 1.90 2.08 27 19 —8 CL 59.0 1.70 8.7 0.369 1.85 2.07 25 15 —10 SM 16.3 1.67 3.5 0.373 1.73 2.05 24 14 63.212 CL 67.7 1.87 12.9 0.306 2.11 2.18 35 18 —14 SC 35.8 1.73 8.9 0.353 1.88 2.08 31 15 62.616 SM 31.6 1.66 8.2 0.378 1.80 2.04 57 25 80.518 SM 27.6 1.64 6.1 0.387 1.74 2.02 57 24 76.920 SC 35.4 1.76 6.5 0.340 1.88 2.10 84 33 89.3

aAbbreviations are as follows: USCS, Unified Soils Classification System; FC, fines content; rdry, dry density; wo, initialwater content; n, bulk porosity; ro, initial density at natural water content; rsat, density at saturation; N, standard penetration test(SPT) blow count value; N1,60, modified SPT blow count value; Dr, relative density, empirical correlation based on Gibbs andHoltz [1957].


3 of 22

B02311

the interval 150 to 450 mm gives the blow count N value,defined as the number of blows per foot. The N value isused universally as the primary index for liquefactionresistance of sandy deposits.[17] Downhole and cross-hole seismic velocity tests were

performed by Ezersky [2004] following ASTM and ISRMstandards [Takahashi et al., 2006] to determine P and Swavevelocities both in dry and wet subsurface conditions. Thesequence of events was as follows. Seismic velocity testswere first performed below the infiltration trench before itwas flooded to obtain the ‘‘dry’’ velocity data set represent-ing the material at natural moisture conditions. At that stageall preparations for the ‘‘wet’’ velocity tests were completedby extending the inner PVC casings in the three boreholesabove the anticipated water level in the trench whenflooded. Drilling was performed using a wireline techniqueto ensure verticality with an accuracy of up to 2�. Aftercompletion of the ‘‘dry’’ velocity tests a 17-day-long floodingphase commenced while water content levels in the subsur-face were continuously monitored using TDR probes. Afterthe completion of the first flooding phase the seismicvelocity tests were repeated to obtain the set of ‘‘wet’’velocity values. As will be discussed below, it seems that atend of the first flooding phase some sections of the stratadid not reach full saturation, and therefore the ‘‘wet’’seismic velocity data set represents a material which isnearly, but possibly not 100%, saturated.

[18] Boreholes B6, B6A, and B6B were used to performseismic velocity tests (Figure 1). The three boreholes weredrilled inside the infiltration trench before it was flooded,down do a depth of 20 m at 3-m spacing, with individualborehole diameter of 76 mm.[19] The essential concept of the two methods is illus-

trated in Figure 2. For the downhole method (Figure 2a) aseismic source is set up at the ground surface and receiversare set in the borehole to measure seismic waves generatedby the surface source. The cross-hole method (Figure 2b)uses three boreholes for setting up the source and tworeceivers to measure the travel time of seismic wavesbetween these boreholes. While cross-hole tests, performedbetween boreholes B6 and B6a, provided a profile of S wavevelocities only, both S and P wave velocities were obtainedfrom two separate downhole tests performed inside theseboreholes.[20] The cross-hole test was performed in adherence to

ASTM standard D4428. The energy source, designed togenerate shear waves (SV type, manufactured by SoilEngineering and Geophysics Inc., USA.) was attached toborehole B6B using an air inflated, flexible flat jack. Acable was used to operate a hammer, the application ofwhich triggered a seismic recording unit. The seismic waveswere recorded in boreholes B6 and B6a using three com-ponent geophones (BHG-2 type, manufactured by GEO-STUFF, United States) which were attached to the walls of

Figure 2. Schematic illustration of the in situ seismic velocity tests which were performed inside theinfiltration test using boreholes 6, 6A, and 6B. (a) Downhole method. (b) Cross-hole method.


4 of 22

B02311

the boreholes using springs. One horizontal (longitudinal)sensor was aligned in parallel and the other horizontalsensor was aligned in transverse direction to the trenchaxis. Geophone orientation was achieved by means of afluxgate compass connected to these sensors. The shearwave source was applied in transverse direction such thatthe transverse horizontal geophone could detect SH typeshear waves. Two BHG-2 geophones were placed in bore-holes B6 and B6a. The measurements were performed in 1-mintervals from the surface to a depth of 20 m such that ateach measurement the source and the two geophones werepositioned at exactly the same level. The recording unitmeasured the arrival times to each geophone, thus enablingseismic velocity calculations. This process was repeatedcontinuously resulting in an SV shear wave velocity profilebetween boreholes B6 and B6a.[21] The downhole test used a single borehole. The

seismic source was placed at the ground surface at adistance of 1.5 m from the top of the tested borehole, andseveral receivers were positioned successively at differentdepths inside the borehole. The energy source was a 9-kghammer which generated P and S waves by vertical orhorizontal blows, respectively, delivered to a metal platewhich was placed on the ground. After each measurementthe geophone was lowered in 1-m depth intervals. Thedownhole log represents material properties at the vicinityof the borehole; for a characteristic shear wave velocity of400 m s�1 and a frequency range of 100–500 Hz, the widthof the zone explored by this method is between 0.2 and 1.0 m.[22] All seismic data were acquired on a seismic record-

ing unit. The length of each seismic recording was 48 msand 100 ms for P and S wave velocities, respectively. Theaccuracy of the travel time picks is estimated at 2–4%.Arrival times were measured during the survey, data wasprocessed using dedicated software [Shtivelman, 1996],and velocities were obtained using the available distancesbetween boreholes (in the cross-hole method) or geophonelevels (in downhole method).[23] The borehole geophysical surveys provided the vari-

ation of P and S wave velocities with depth for dry and wetsubsurface conditions, and enabled calculation of in situ,dynamic, elastic parameters for the sediments at the twomoisture conditions.

[24] Representative, undisturbed, samples were extractedfrom the field and tested at the laboratory to determine theshear strength of the materials at natural water contentconditions.

2.3. In Situ Test Results

[25] The sediments at the test site can be classified intothree major soil types according to ASTM unified soilclassification system (ASTM D2487–00): (1) Nonplasticsilty sand with fine content less than 20% (SM); (2) low-plasticity clayey sand with fine content greater than 33%(SC); and (3) low-plasticity clay with some amount of sand(CL). The clayey sand and the clays contain 3–5% carbonateconcretions which are randomly distributed in the section.The variations in soil type, density, fines content, porosity,and in situ SPT values obtained at dry conditions in borehole6 before the first infiltration phase are listed in Table 1.Seismic velocities obtained before and after the first infiltra-tion phase are listed in Table 2 and plotted in Figure 3 for dryand wet subsurface conditions.[26] The density of the material in its natural water content

(ro) increases with depth, or with vertical overburden stress,from ro = 1.61 g cm

�3 close to the surface to ro = 1.88 g cm�3

at a depth of 20 m (Table 1). Similarly, SPT values increasewith overburden stress fromN = 16 close to the surface toN =84 at a depth of 20 m. This increase in N value must reflectincrease in the relative density of the material with increasingdepth. Modified SPT values naturally do not reflect thistrend because N1,60 values are corrected for, among otherparameters, the level of overburden stress [see Idriss andBoulanger, 2006; Seed et al., 1985]. N1,60 values should,however, reflect material property variations. Indeed, N1,60

values seem to be higher for the clayey sands and clays.[27] The seismic velocities obtained clearly increase with

depth, supporting the previous conclusion that the relativedensity of the material increases with depth. This is bestshown in Figure 3e where the shear wave velocitiesobtained in the Cross Hole survey are plotted versus depthfor dry and wet subsurface conditions. Close to the surfacethe shear wave velocity for dry subsurface is Vs = 402 m s�1

and it increases to Vs = 547 m s�1 at a depth of 20 m(Figure 3e). P wave velocities also increase with depth fromVp = 690 m s�1 close to the surface to Vp = 810 m s�1 at 20-m

Table 2. Measured Mechanical Properties of the Sediments Before (Initial) and After the First Infiltration Phase Obtained From in Situ

Seismic Velocity Testsa

Depth (m)

Initial Moisture Conditions (Dry) After First Infiltration Phase (Wet) At 100% Saturation

Vs (m/s) Vs1 (m/s) Vp (m/s) Gd (MPa) Kd (MPa) Vs (m/s) Vs1 (m/s) Vp (m/s) Gd (MPa) Kd (MPa) Kd,sat (GPa) Vp,sat (m/s)

2 402 536 690 250 404 364 593 480 261 106 6.18 18194 418 464 710 304 472 360 430 645 271 508 7.34 19256 445 442 710 340 412 367 373 645 280 492 7.16 19138 425 391 710 308 449 382 349 645 302 458 7.08 190310 442 384 710 327 408 405 349 645 336 405 6.89 189012 433 359 710 351 476 382 312 645 318 483 8.50 202814 472 375 810 384 620 459 358 800 438 747 7.46 195716 515 395 810 440 502 530 401 800 573 542 6.90 191618 540 403 810 477 438 561 412 800 636 445 6.70 190520 547 398 810 527 453 512 366 800 551 610 7.59 1987

aThe values for 100% saturation were calculated. Abbreviations are as follows: Vs, shear wave velocity; Vs1, over burden stress-corrected shear wavevelocity; Vp, P wave velocity; Gd, dynamic shear modulus; Kd, dynamic bulk modulus.


5 of 22

B02311

Figure 3. Results of downhole and cross-hole seismic velocity tests before (solid line) and after (dashedline) first infiltration test: (a) columnar section below the infiltration trench; (b) P wave velocity fromdownhole test in borehole 6; (c) P wave velocity from downhole test in borehole 6a; (d) S wave velocityfrom downhole test in borehole 6; (e) S wave velocity from cross-hole tests performed between boreholes6 and 6a; (f) S wave velocity from downhole test in borehole 6a.


6 of 22

B02311

depth in dry subsurface conditions (Figures 3b and 3c).Overburden stress-corrected shear wave velocities [Andrusand Stokoe, 2000], the geophysical equivalent to the N1,60

value, are also shown in Table 2.[28] The dynamic elastic parameters for each layer are

obtained from the following relationships:

nd ¼

Vp

Vs

� �2

�2

2Vp

Vs

� �2

�2

ð1Þ

Gd ¼ rV 2s ð2Þ

Ed ¼ 2Gd 1þ ndð Þ ð3Þ

Kd ¼Ed

3� 6ndð Þ ; ð4Þ

where the subscript d denotes ‘‘dynamic’’ and n, G, E, andK are, respectively, the Poisson’s ratio, shear, Young’s, andbulk moduli. The calculated elastic moduli for dry and wetsubsurface conditions are shown in Table 2 using therelevant velocities and densities for each layer.[29] The elastic constants for wet subsurface conditions

after the first infiltration phase (Table 2) were obtainedusing saturated densities (rsat), namely the density of thematerial when the pore space is completely filled withwater. The saturated density for each layer was obtainedanalytically using the measured values of dry density andporosity at the lab (Table 1). We use the saturated densitiesfor each layer in our calculation of elastic parameters forwet conditions because the difference in material densitiesafter the first infiltration phase and at 100% saturation isbelieved to be negligible.

2.4. Effect of Wetting on Seismic Velocities

[30] Interestingly, both measured S and P wave velocitiesdecreased following the first, 17-day-long, infiltrationphase. While theoretically the shear modulus should notbe affected by saturation, the shear wave velocities shoulddecrease upon wetting owing to the increased density ofthe material upon wetting. The expected reduction inshear wave velocity is readily given by the density correc-tion factor (see equation (2)) if we assume that the shear

modulus remains a constant: Vwet /Vdry =ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffirdry=rwet

q.

Indeed, the average density correction factor for the10 measured layers when using initial and saturated densi-ties (Table 1) is 0.942 and the average ratio between the wetand dry shear wave velocities (Table 2) is 0.926. Thesevalues are very close suggesting that the change in totaldensity owing to saturation is mainly responsible for themeasured decrease in shear wave velocity. Note that the wetshear wave velocities obtained after the first infiltrationphase (Table 2) although lower than in dry state, are stillhigher than the accepted upper bound for liquefaction thatfrom Andrus and Stokoe’s [2000] compilation appears to beVs1 = 300 m s�1.

[31] The decrease in P wave velocity following the firstinfiltration phase clearly suggests that the material did notreach 100% saturation at the end of that phase, because infully saturated subsurface conditions the P wave velocityshould be greater than about 1500 m s�1, the velocity infresh water at 25 C [Clark, 1966]. The presence of trappedair bubbles in the sequence, amounting to as low as 1–2%of the void space, would be sufficient to keep the P wavevelocities low, as shown for example by Bachrach and Nur[1998] using poroelasticity. The reduction of the wet P wavevelocities below the dry P wave values can be explainedagain by the increase in total density upon increased degreeof saturation in the voids. This will hold true until the voidspace is completely filled with water, at which point theP wave velocity will sharply increase to values greater than1500 m s�1, as illustrated by Bachrach and Nur [1998].[32] It is also possible that in cohesive, weakly cemented,

granular materials stiffness degradation may develop withincreasing degree of saturation [e.g., Assallay et al., 1996;Dijkstra, 2001; Fener et al., 2005; Jefferson et al., 2004]thus reducing the values of both the bulk and shear moduli.This process, if encountered in the subsurface, may affectboth S and P wave velocities.

2.5. Theoretical Determination of P Wave Velocity for100% Saturation

[33] For estimating the P wave velocities of the testedmaterials for fully saturated conditions poroelasticity maybe invoked. Let the effective P wave velocity of the materialbe

Vp ¼

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiK þ 4

3G

r

vuut: ð5Þ

To find the P wave velocity for saturated sediments thevalues of Ksat, Gsat, and rsat must be obtained. Sincetheoretically Gsat = Gdrained [Gassmann, 1951] only thevalue of Ksat must be resolved. Using the Biot-Gassmannequation [Bachrach and Nur, 1998] Ksat may be obtainedfrom

Ksat

Ko � Ksat

¼ Kdry

Ko � Kdry

þ Kfl

n Ko � Kfl

� ; ð6Þ

where Ko is the mineral bulk modulus, Kdry is the dry bulkmodulus obtained from the results of the dry seismicvelocity tests, Kfl is the bulk modulus for the pore fluid, inour case water where Kwater = 2.29 GPa, and n is theporosity, the value of which is known for each layer (Table 1).[34] The mineral bulk modulus may be estimated using

weighted average of the quartz and clay fractions in thesediments given that Ko,Qz = 38 GPa and Ko,clay = 20.9 GPa,by assuming that the ‘‘fines’’ fraction determined from sieveanalyses represents clays and the coarser fractions representquartz; a reasonable assumption for loess deposits. Thecomputed Ksat and Vp values for each layer are shown inright column of Table 2. We used the shear modulusobtained for wet conditions for these calculations as it is


7 of 22

B02311

assumed to better represent the elastic response of thematerial when fully saturated.

3. Laboratory Direct Shear Tests

3.1. Test Systems

[35] Direct shear tests were performed at the laboratoryon undisturbed samples from the two dominant materialtypes, namely the silty sands (SM) and the clayey sands(SC) at their natural water content conditions, to obtain thecharacteristic shear strength of the material.[36] Two test systems were employed, a high-capacity

hydraulic load frame with a maximum shear force of 300 kNoriginally designed for testing rock joints, and a low-capacity mechanical load frame with a maximum shearforce of 5 kN, originally designed for testing soils. Cubicblock samples were quarried out of an exploratory trench inthe test site for testing in the high-capacity load frame;cylindrical samples were extracted from the sampling tubeused when drilling boreholes 6, for testing in the low-capacity load frame.[37] The high-capacity load frame is a hydraulic, closed-

loop servo-controlled system (TerraTek Systems modelDS-4250) with a particularly large shear box with dimen-sions of 170 mm � 170 mm � 300 mm allowing for a largesample cross-sectional area of up to 225 cm2 (Figure 4).Both normal and axial cylinders are operated hydraulicallyin this system under two possible control modes: load anddisplacement. The normal and horizontal load capacities are1000 kN and 300 kN respectively, with load-cell linearity of0.25% full scale. Two 50-mm-range LVDTs were used tomeasure horizontal shear displacement and four 50-mm-range LVDTs were used to record vertical motions duringshear; the LVDTs were manufactured by Macro Sensors2

(model GHSA-750-1000) with 0.25% linearity full scale. Thelow-capacity load frame (Wykeham Farrance 31-WF25421)is a commercially available microprocessor controlled drivesystem operated with a high-precision step motor.

3.2. Direct Shear Test Results

[38] Shear displacement versus shear load curvesobtained with the servo-controlled system are plotted inFigures 5a and 5b for the silty sands and clayey sands,

respectively for different levels of normal stress. All testswere performed under an imposed constant normal stressboundary condition and under a constant shear displacementrate of 0.025 mm s�1. Since the normal piston maintainedconstant normal stress under load control the material wasfree to dilate or contract during shear. The correspondingaverage vertical displacements recorded by the four verticalLVDTs during horizontal shear displacement are plotted inthe lower panels in Figure 5 where as a convention dilationis positive.[39] The tendency of the silty sands to dilate during shear

is suppressed with increasing normal stress (Figure 5a). Theclayey sands exhibit a more brittle deformation (Figure 5b)and the material dilates during shear under all levels ofnormal stress. The direct shear test results therefore suggestthat the clayey sands are stiffer than the silty sands. Thisfinding is supported by the obtained in situ N1,60 values(Table 1). The seismic velocities measured in situ do notseem to be sensitive enough to detect such differences inmaterial stiffness, probably because the difference betweenthe wavelength of the in situ tests and the sample size testedat the laboratory.[40] The resulting failure envelopes for the two materials

in t�s space are shown in Figure 6 including test resultsobtained with the sensitive 5-kN system. Both materialsseem to obey the linear Coulomb-Mohr failure criterion.The silty sands exhibit negligible cohesion and an effectivefriction angle of 32�, or a friction coefficient of m = 0.62(Figure 6a). The clayey sands exhibit cohesion of approx-imately 50 kPa and an effective friction angle of 41.5�, or afriction coefficient of m = 0.88 (Figure 6b).

4. Controlled Infiltration

[41] Two controlled infiltration experiments were per-formed. The first infiltration test lasted 17 days, then aftera period of 3 months, the second 25-day-long infiltrationtest was performed. In both tests the excavated trench wasflooded allowing downward water infiltration down to atleast 20 m. The trench sidewalls were sealed so thatinfiltration took place only through its base. At the firstexperiment, after which the wet seismic velocity tests wereperformed, a constant head of 1.0 m above the trench basewas maintained. During the first test the dimensions of thebase of the trench were 17 m by 3 m, and the total amountof water used was 1400 m3. An important objective of thefirst experiment was to track the infiltration rate and itstemporal and spatial patterns. Therefore, the original trenchdimensions were relatively small. For the second infiltrationexperiment the trench was elongated by additional 31 m toprovide infiltration area of 48 m by 3 m. Infiltrationcommenced for a period of 25 days during which a constanthead of 1.25 m above the base of the trench was maintained.The total amount of water used during the second infiltra-tion period was 10,300 m3. The propagation of the wettingfront during that stage was faster, as expected.[42] The temporal changes of water content were moni-

tored at 28 points beneath the trench, using time domainreflectometry (TDR) probes placed in 4 boreholes, provid-ing a two-dimensional vertical image below and across thetrench. Detailed records of a 17-day period of wetting,followed by a 3-month period of drying, and finally,

Figure 4. The high-capacity, hydraulic, servo-controlleddirect-shear apparatus used for obtaining shear strengthparameters for the sandy units.


8 of 22

B02311

followed by a second 25-day period of re-wetting, wereobtained [Gvirtzman et al., 2008].[43] Time domain reflectometry (TDR) is a common

method for measuring water content in unsaturated soilsin the field. The TDR technique involves measurements ofthe propagation velocity of a high-frequency signal trans-mitted along a waveguide. The dielectric constant of water(�80) is significantly higher than that of most minerals (2–7);therefore, propagation velocity measurements can be used toestimate the volumetric soil water content. Lately, monitoringdeep vadose zone horizons (down to 25 m) has been madepossible using flexible waveguides that are pressed againstthe borehole walls using a sleeve filled with liquid resin[Dahan et al., 2003].[44] In this study, we followed Dahan’s TDR setup and

monitoring procedure. Four flexible polyethylene sleeves,each with 10-cm diameter and 22-m long, were prepared forthe four boreholes. Eight TDR probes (waveguides) wereglued to each sleeve, at a distance of about 2 m from eachother. The probes were made of two stainless steel foilstrips, each 30-cm long and 2-cm wide, glued parallel toeach other, at a spacing of 5 cm. The sleeves were perfectlysealed to prevent future liquid leakage. Shielded coaxial

cables were connected to the probes to allow in situ, real-time readings. All individual waveguides were connected toa digital cable tester (model 1502C, Tektronix, Beaverton,OR) through a multiplexer (model SDM50, CampbellScientific, Inc. Logan, UT), and the entire system wasconnected to a data logger (model CR10X, CampbellScientific, Inc. Logan, UT) for data collection. The probeswere checked for TDR readings before the trench wasflooded with water. Out of the 32 pairs of probes, 28 werefunctional after installation.[45] During the experiment, numerous TDR readings

were collected, stored on a laptop, and the volumetric watercontent was calculated by the standard technique of wave-form analysis. However, because of temporal and spatialchanges in moisture salinity (owing to flooding and saltwashing), the relationship between TDR signal and mois-ture content was not unique, and therefore the results havesome inherent degree of uncertainty. Nevertheless, thewetting front propagation was reconstructed very precisely[see Gvirtzman et al., 2008].[46] Traced by the TDR probes, the wetting front was

onion-shaped, and slowly propagated vertically and hori-zontally. Detailed description of the water flow in the

Figure 5. Shear stress (top panels) and dilation (bottom panels) versus shear displacement during servo-controlled direct-shear tests; dilation is positive. (a) Silty sands. (b) Clayey sands.


9 of 22

B02311

unsaturated zone is provided by Gvirtzman et al. [2008].The spatial and temporal variations during the wetting,drying and re-wetting processes were reconstructed usinga numerical code that solves the flow equation, under bothsaturated and unsaturated conditions. We divided thesediment column into a sequence of two fundamentalsediment types, each of which was treated as an equivalenthomogeneous medium. This approach transfers the hetero-geneity problem within each of the units to a problem ofidentification of the effective parameters [see Wierenga etal., 1991; Wildenschild and Jensen, 1999]. The 2-D simu-lation domain was 35 m by 30 m in horizontal and verticaldirections respectively, reflecting the actual size of the fieldexperiment. This domain was discretized into 153 by79 pairs of triangle elements (24,174 elements) with meansize of 0.2 m by 0.4 m. The bottom of the domain wasassumed to be under a unit gradient condition; the left andright hand sides and the top of the domain, except for thecenter zone corresponding to the trench, were modeled asno-flow boundaries. A constant head boundary wasassumed for the base of the trapezoid-shaped trench at thetop boundary, reflecting the water head in the trench. Theinitial capillary tension distribution was assumed uniform at�3.0 m, reflecting the natural moisture content.

[47] The retention curve, as well as the hydraulic con-ductivity versus moisture content curve for each unit, wereplotted using the parameters of the Mualem–Van Genuchtenmodel [Van Genuchten, 1980]. These curves were firstestimated using field data and then adjusted by the numericalmodeling results [see Gvirtzman et al., 2008]. In Figure 7 thetemporal and spatial changes in moisture content during thesecond infiltration experiment, and the resulting pressurebuildup, are shown graphically. The numerical simulationreconstructed the changes in the position of the onion-shaped wetting front detected in the field by the TDRprobes. It is important to emphasize that sediment stratifi-cation, and variations in hydraulic conductivity under sat-urated conditions, result in differential flow rates throughoutthe profile and consequently pressure buildup inside theonion-shaped saturated zone. The results of the numericalsimulations indicate that a head of up to 3.1 m developed inthe saturated zone and in particular within the central siltysand unit, bounded in the field by the less-permeable sandyclay units form above and below.

5. Blast-Induced Liquefaction Experiment

[48] To investigate induced liquefaction wet and dry blasttests were conducted in the field. Recall that the two

Figure 5. (continued)


10 of 22

B02311

infiltration periods were followed by two different sets oftests. Before and after the first infiltration period seismicvelocity tests were performed below the trench to obtainseismic velocities at dry and wet conditions. The infiltrationtrench was then let to drain without additional water inflowfor a period of three months, during which it was elongatedand preparations were made for the wet blast test. After thesecond 25-day-long infiltration period, the wet blast test wasperformed below the trench and after several hours the dryblast test was performed 300 m away from the trench on dryand leveled ground, with exactly the same subsurfacecomposition as found below the infiltration trench.[49] Since the wet blast test was performed after the

completion of the second infiltration phase we assume, onthe basis of our hydrological model and TDR measure-ments, that the subsurface below the trench was 100%saturated at that time down to a depth of at least 14 m,before the charge was detonated.

5.1. Blast Test Setup

[50] The layout of the blast tests is shown in Figure 1. A50-kg trinitrotoluene (TNT) charge was placed and buriedin a 40-cm diameter borehole that was drilled below one end

of the infiltration trench to a depth of 10 m, the depth ofdetonation. After placing the charge the boreholes wereback filled tightly with sand. At a horizontal distance of30 m from the blast borehole along the trench axis a 20-cm-diameter borehole with 15.2-cm-diameter casing was drilledto a depth of 15 m for the placement of a vertical geophonearray. The geophones (BMIII type three-component sensors,Instantel) measured particle velocities in longitudinal, trans-verse, and vertical directions. The triaxial geophones weretriggered by the blast wave and data were stored 0.25 sbefore the trigger for a total acquisition time of 1.5 s at asampling rate of 1024 samples per second. The range ofthe geophones was 254 mm s�1 with a resolution of0.127 mm s�1. The natural frequency of the geophoneswas 10 Hz and the frequency range 2–250 Hz.[51] Two blast tests were performed: the first below the

infiltration trench after the material in the subsurfaceattained full saturation, and the second 3 h later and 300 maway from the infiltration trench at leveled ground in similarmaterial but under natural moisture conditions. The samecharge, distance, and vertical array configuration were usedin both tests. Similarly, the same four geophone packageswere placed in the vertical arrays in the wet and dry blastexperiments: sensor BA8850 was placed on the surface;sensors BA8109, BA8835, and BA8836 were placed in thevertical borehole 6 m, 10 m, and 14 m below the surface, inboth tests.[52] A single electric piezometer (model LV610, DINA

electronics) was placed immediately below the verticalarray (Figure 1) to enable direct measurements of excesspore water pressures during and after the wet blast exper-iment, should they develop at that location.

5.2. Blast Test Results

[53] The acquired wave forms in the vertical arrays for thedry and wet blast tests are presented in Figure 8. Peakparticle velocities measured in each geophone are tabulatedin Table 3. The longitudinal (radial–horizontal) particlevelocity output at depth of 6 m for the dry test wascorrupted owing to an electrical problem and unfortunatelycannot be used. Nevertheless, the obtained trend is quiteclear. First, particle velocities at saturated subsurface con-ditions are much higher than particle velocities for drysubsurface, down to a depth of 14 m where the particlevelocity outputs seem to be the same. Second, the verticalcomponent is higher than the two horizontal componentsclose to the surface but diminishes to almost zero at 14 mdepth, for both subsurface conditions. Third, the longitudi-nal component is much higher than the transverse compo-nent in both subsurface conditions, as would be expected.[54] To compare between velocity readings at different

depths and moisture levels the magnitude of the resultantpeak particle velocity (PPV) is first discussed. In the wetblast test the surface geophone measured resultant peakparticle velocity of PPV(h=0 m) = 92 mm s�1. At 6 m and10 m the results were identical and similar to the surfacereading: PPV(h=6,10 m) = 90 mm s�1. At 14 m depththe resultant peak particle velocity was much lower:PPV(h=14 m) = 35 mm s�1. For comparison, in the dry blasttest the surface geophone measured PPV(h=0 m) = 42 mm s�1,at 10 m PPV(h=10 m) = 28 mm s�1, and at 14 m PPV(h=14 m) =18 mm s�1. The results of the blast tests therefore clearly

Figure 6. Failure envelopes obtained from direct sheartests. (a) Silty sands. (b) Clayey sands. Open symbols areoutliers not included in linear regression.


11 of 22

B02311

Figure 7. Two-dimensional numerical solution of saturation degree (left panels) and pressure head(right panels) below the infiltration trench from day 0 to day 21 during the second infiltration test,demonstrating the process of water penetration and spreading beneath the flooded trench. The onion-shaped, solid black lines represent the measured wetting front.


12 of 22

B02311

Figure

8.

Resultsofverticalgeophonearraymeasurementsin

wetanddry

blasttests.Theverticalarraywas

positioned

atahorizontaldistance

of30m

from

thesource.


13 of 22

B02311

show that peak particle velocities decreased with decreasingdegree of saturation, and with depth.[55] The acquired wave forms (Figure 8) do not permit

reliable differentiation between the wave types that reachedeach geophone in the vertical array. While P waves wouldbe expected to be dominant owing to the nature of theenergy source, possibly S waves as well as surface wavescould have been generated by the blast. The presence ofsurface waves for example is suggested by the observationthat the vertical component was higher than the twohorizontal components close to the surface but diminishedto almost zero at 14-m depth, for both subsurface conditions.[56] Seven minutes after the wet blast test a water

fountain erupted from the blast borehole to a height ofseveral meters. After the ‘‘geyser’’ subsided (Figure 9a) thewater that remained at the base of the trench continued tobubble (Figure 9b) for a period of about 30 min, to ahorizontal distance of 15 m from the blast borehole alongthe trench axis. Approximately 20 min after the blast lateralspreading features were observed at opposite banks acrossfrom the blast borehole (Figure 9c). An accurate topograph-ic survey performed immediately after the blast indicatedthat the ground subsided by about 85 mm close to theborehole and by up to 15 cm within a radius of 15 m fromthe borehole (Figure 10). An ultimate settlement of 150 cmwas measured around the blast borehole 70 days after theexperiment. The radius of the disturbed zone remained 15 mafter 70 days. In contract, the ground immediately aboveand all around the dry blast test site remained completelyleveled after the charge was detonated. Moreover, a waterfountain did not erupt from the dry blast borehole, no sandboils were observed on the ground surface, and no lateralspreading was detected.[57] Finally, no excess pore water pressure was recorded

in the electric piezometer that was placed immediatelybelow the vertical array in the wet blast experiment site(Figure 1), because the wetting front did not reach thatdepth at the time of the wet blast experiment.

6. Discussion

6.1. Failure Modes Triggered by the Wet BlastExperiment

[58] Four different deformation patterns were observed atthe test site following the wet blast experiment, all indicateexcess pore pressure generation and all exhibit time depen-dency: (1) a water fountain that erupted from the blastborehole immediately after the blast with continued ejectionof a water jet for a duration of �8 min (Figure 9a),(2) bubbling of the water surface inside the infiltrationtrench for a distance of 15 m from the blast borehole that

continued for �45 min after the blast (Figure 9b), (3) lateralspreading on two opposite banks of the infiltration trenchtriggered �20 min after the blast (Figures 9c, 10b, and 10c),and (4) settlement of the ground surface immediately afterthe blast for a period of at least 70 days (Figure 10). None ofthese features was observed following the dry blast exper-iment that was performed in dry subsurface conditions andleveled ground, 300 m away from the wet blast experiment.[59] All deformation patterns suggest rapid development

of excess pore water pressure, but not all necessarily implyliquefaction. The eruption of a water fountain from the blastborehole for example does not necessarily imply thatliquefaction was triggered in the adjacent saturated domain,as the immediate blast region is characterized by singular-ities in the generated pressure field typically resulting inmud fountains.[60] The bubbling of the water table inside the infiltration

trench however (Figure 9b) does prove that liquefaction wastriggered, as this is a common feature in sites undergoinginstantaneous liquefaction following strong earthquakes,typically marked by ‘‘sand boils’’ that remain on the surfaceafter the event [e.g., Fiegel and Kutter, 1994; Moretti et al.,1999; Rossetti, 1999; Yuan et al., 2004].[61] The breakage of the surface on two opposite banks of

the infiltration trench (Figures 9c, 10b, and 10c) resembleslateral spreading features documented at sites that haveexperienced earthquake-induced liquefaction [e.g., Bardetet al., 2002; Boulanger et al., 1997; Cetin et al., 2004b;Rajendran et al., 2001]. The observed breakage of theground surface may reflect brittle response of the dry upperlayers to the liquefaction of underlying, saturated, layers(Figure 9c), as documented in many case studies reportingearthquake-induced lateral spreading phenomena [Rauch,1997]. Volume changes leading to a more compact config-uration in the underlying layers undergoing liquefactioncould be followed by brittle breakage of the dry upperlayers resulting in lateral spreading at the surface, asobserved on both slopes (Figure 9c).[62] Finally, the measured ground settlement (Figure 10),

while often reported from sites that have experiencedearthquake-induced liquefaction [e.g., Cetin et al., 2002],is classic evidence for blast-induced liquefaction [Charlie etal., 2001; Gandhi et al., 1999]. The measured cone shapedsettlement of the ground surface reflects a transformation ofthe deeper liquefied sand layers into a denser configuration.This process, although evidently time-dependent, should notbe confused with consolidation, a very slow process the rateof which depends upon the hydraulic conductivity and thecompressibility of the soils. The observed failure modesowing to the blast are further discussed below.

6.2. Field Measurements of Excess Pore WaterPressure

[63] The magnitude of excess pore pressure required forblast-induced liquefaction can be determined analytically, ifthe change in total stress owing to the blast wave (Dp) canbe estimated. Let Po, P

0o, and uo represent, respectively,

initial mean total stress, initial mean effective stress, andinitial pore water pressure, at a given point below thegroundwater table that for the sake of this discussion, willbe assumed to coincide with the leveled ground surface. Theinitial mean effective stress is then given by [Terzaghi,

Table 3. Longitudinal, Transverse, Vertical, and Resultant Peak

Particle Velocities As Measured in the Vertical Arraya

Depth (m)

Longitudinal Transverse Vertical Resultant

Initial Sat. Initial Sat. Initial Sat. Initial Sat.

0 32.8 51.9 4.06 31.4 34.3 90.8 42.5 92.36 . . . 89.0 24.6 18.3 45.0 30.7 . . . 90.210 27.7 89.5 3.05 24.3 16.1 10.5 27.8 90.414 16.8 33.5 15.6 19.0 4.44 5.71 18.2 35.0

aInitial refers to moisture conditions at natural water content.


14 of 22

B02311

Figure 9. Failure modes associated with induced liquefaction as observed in the field following the‘‘wet’’ blast experiments: (a) termination of water geyser eruption from the blast borehole, (b) bubbling ofwater surface inside the infiltration trench (sand boils), and (c) lateral spreading observed on the dryslopes of the infiltration trench.


15 of 22

B02311

1943]: P0o = Po�uo. During transient loading owing to the

blast the mean effective stress becomes P0 = P�u = (Po +Dp)�(uo + Du) where Dp is the transient change in meantotal stress owing to the blast and Du is the resulting excesspore water pressure, both of which are time-dependent. Thelimiting condition for liquefaction can be stated as the pointin time when the mean effective stress diminishes:

P0 ¼ 0: ð7Þ

[64] The excess pore pressure required for blast-inducedliquefaction Duc at that point in time is therefore

Duc ¼ P0o þDp: ð8Þ

Figure 10. Time-dependent ground settlement owing to liquefaction. (a) Topographic elevation contourmap of the infiltration trench (hatched) and vicinity prior to the ‘‘wet’’ blast experiment. Topographicdifference maps indicating surface deformation depict (b) immediately after the ‘‘wet’’ blast, (c) one dayafter the ‘‘wet’’ blast, and (d) 70 days after the blast. (e) Three vertical cross sections normal to trenchaxis and through the ‘‘wet’’ blast borehole performed immediately after, one day after, and 70 days afterthe blast.


16 of 22

B02311

FindingDuc by equation (8) requires accurate determinationof Dp, a task which is beyond the scope of this paper.[65] The limiting condition for liquefaction (equation (7)),

if and where reached, will only be satisfied for a very short

period of time. After the dynamic loading owing to thehigh-amplitude blast waves diminish, a process which maylast 101–102 ms (e.g., Figure 8), excess pore pressuredissipation will ensue at rate and form similar to experi-

Figure 10. (continued)


17 of 22

B02311

mental results as obtained for example by Ashford et al.[2004] and reproduced in Figure 11. The exact form of thepore pressure dissipation curve will depend upon the typeand amplitude of the blast-induced waves, the compress-ibility of the soil skeleton, the hydraulic conductivity of thematerial, and the availability of drainage paths at thesubsurface.[66] The excess pore pressure at any given time and

location in the saturated domain is typically expressed bythe excess pore pressure ratio (Ru), Ru = Du/s0v, where s

0v is

the initial vertical effective stress. A pore pressure ratio ofRu = 1.0 is generally defined as liquefaction in sandy soils,regardless of the loading mechanism. For blast-inducedliquefaction existing reports suggest that it can be triggeredat lower excess pore pressure ratios, with a lower bound atRu = 0.8 [Kok and Trense, 1979]. The bubbling of the watersurface inside the trench, as well as the water fountain, aresurface manifestations of areas in the subsurface where thedeveloped excess pore pressure ratio must have been withinthe range of 0.8 < Ru < 1.0.[67] Since analytical determination of Du owing to blast-

ing requires a determination of the transient load Dp, amagnitude of which is very difficult to obtain eitherexperimentally or numerically, an empirical relationshipbetween the scaled distance of the charge mass; the Hop-kinson’s number (

ffiffiffiffiffiM3

p/R;

ffiffiffiffiffikg3

p/m) and the resulting excess

pore pressure has been proposed by several authors on thebasis of field tests [e.g., Kok and Trense, 1979; Studer andKok, 1980] and confirmed in recent studies [Ashford et al.,2004]:

Du

s0v

¼0 for

ffiffiffiffiffiW3

p

R< 0:12

1:65þ 0:64 lnffiffiffiffiffiW3

p

R

� �for

ffiffiffiffiffiW3

p

R� 0:12

averageð Þ

8>>><>>>:

;

ð9Þ

where M and R are the charge mass and the radial distancebetween the charge and the target, respectively. Equation (9)can be used to delineate the boundaries for blast-inducedliquefaction at the subsurface on the basis of existingexperience, given the charge mass and radial distance fromthe source point. This was done in Figure 12 for the scaleand outline of the wet blast experiment. Indeed, the extent

of the region where water bubbling (sand boils) wasobserved inside the infiltration trench coincides with theempirical Ru = 0.8 boundary.[68] From the computed Ru boundaries plotted in

Figure 12 it appears that the vertical array was drilled atthe edge of the region where excess pore water pressurewould be expected. This explains the lack of excess porepressure readings in the piezometer during and after the wetblast test.

6.3. Ground Settlement

[69] Finally, it would be instructive to confirm that ourmeasurements of ground settlement following the wet blastexperiment (Figure 10) are consistent with reports fromother sites that have experience liquefaction under similarloading conditions. In our test site two liquefiable sandylayers are separated by an intermediate, nonliquefiableclayey layer: an upper 350-cm-thick layer of liquefiablesand is underlain by intermediate 490-cm-thick nonliquefi-able clays, which overlay a lower 290-cm-thick liquefiablesand layer (Figure 12). In the preceding sections of thisdiscussion we have argued that liquefaction must havedeveloped in both liquefiable layers by virtue of excesspore pressure. Yet, the intermediate clay layer may absorbthe deformations associated with liquefaction of the lowerliquefiable layers, such that its failure will not have anysurface manifestation. In such a case the amount of surfacesettlement would only reflect the thickness of the uppermostliquefiable layer.[70] The thickness of an intermediate, nonliquefiable

layer, required to absorb deformations associated withliquefaction of underlying liquefiable layers was investigatedby Ishihara [1985] who compiled reports of ground failurestriggered by earthquake-induced liquefaction at sites whichhave experienced strong ground motions (up to 0.2 g).Ishihara [1985] found that a nonliquefiable layer thicknessof 3 m would be sufficient to absorb all deformationsassociated with liquefaction of underlying liquefiablelayers. Ishihara’s study confirmed that with sufficient thick-ness of the intermediate, nonliquefiable layer, the liquefac-tion of underlying liquefiable layers will not have anysurface manifestation, regardless of their thickness.[71] In our test site the thickness of the intermediate

nonliquefiable layer is 490 cm and therefore, on the basisof Ishihara’s study, it may be assumed that the liquefactionof the underlying liquefiable layers did not produce anysurface manifestation. Consequently, all measured surfacesettlements should be associated with the uppermost lique-fiable sandy layer.[72] Ground settlement is typically discussed in terms of the

ratio between themeasured amount of surface settlement (DH)and the thickness of the liquefied layer (H). For blast-inducedliquefaction Kok [1981] proposed an empirical relationshipbetween the observed ground settlement (DH/H) andHopiknson’s number for tests performed in saturated sands:

DH

H%ð Þ ¼

0 forDu

s0v

< 0:8

2:73þ 0:9 lnffiffiffiffiffiW3

p

R

� �for

Du

s0v

� 0:8

8>>><>>>:

; ð10Þ

Figure 11. Typical example of time-dependent excesspore pressure buildup and dissipation at a point in thesubsurface owing to a blast wave expressed in terms ofexcess pore pressure ratio Ru (modified after Ashford et al.[2004]).


18 of 22

B02311

where DH is the surface settlement as a result of blasting, His the initial thickness of the liquefiable layer, W is the massof the charge in equivalent kg of TNT (kg), and R is theradial distance between the charge and the target (m).Application of equation (10) to this study (Figure 13a)suggests that settlement should be expected up to ahorizontal distance of 11.20 m from the source, a predictionthat is confirmed by our test results (see Figure 10e).Equation (10) is plotted graphically in Figure 13b alongwith our settlement measurements performed one day afterthe wet blast test. Our results are within the range of averagesettlements predicted by equation (10), but show better theobtained settlement cone around the blast borehole.

7. Summary and Conclusions

[73] Results from a controlled blast-induced liquefactionexperiment are reported in this paper. The experiment wasperformed above the groundwater table and therefore con-trolled infiltration tests, executed in two phases, precededthe blast test. The sediments at the site consist of alternatinglayers of silty sands, clayey sands, and clays. The mechan-ical behavior of the material was determined using in situstandard penetration (SPT) tests, in situ seismic velocity

tests performed in both dry and wet subsurface conditionsbefore and after the first infiltration phase, and laboratorydirect shear tests. The physical properties of the materialsincluding density, porosity, and fines content were alsodetermined at the laboratory, for every 2-m interval.[74] The density of the material in its natural water

content (ro) increased from ro = 1.61 g cm�3 close to thesurface to ro = 1.88 g cm�3 at a depth of 20 m reflectingmaterial densification with increased overburden. Similarly,SPT values increased from N = 16 close to the surface toN = 84 at a depth of 20 m reflecting an increase in therelative density with depth. The fines content varied from16 to 32% in the silty sands, 35% in the clayey sands, andfrom 56 to 68% in the clays.[75] Shear wave velocity for dry subsurface increased

from Vs = 402 m s�1 close to the surface to Vs = 547 m s�1

at a depth of 20 m. P wave velocities also increased fromVp = 690 m s�1 close to the surface to Vp = 810 m s�1 at20-m depth in dry subsurface conditions. Interestingly, bothmeasured S and P wave velocities decreased as a result ofwetting, following the first infiltration phase. The reducedseismic velocities are explained by degraded material stiff-ness owing to wetting, a typical response of clay rich,

Figure 12. Boundaries for excess pore pressure generation owing to the blast on the basis of empiricalrelationships [Kok and Trense, 1979; Studer and Kok, 1980]. The 100% saturation boundary is plotted bythe dashed line for reference.


19 of 22

B02311

weakly cemented, granular soils to the presence of water inthe voids. The decrease in P wave velocity following thefirst infiltration phase suggests that the material did notreach 100% saturation at that time. P wave velocities for100% saturation are therefore determined analytically. Ournumerical modeling of the groundwater flow below theinfiltration trench clearly indicates that the material at thesubsurface did reach 100% saturation down to a depth of14 m after the second infiltration phase and before the wetblast test was executed.[76] Direct shear tests were performed at natural water

content; the clays were not tested owing to samplingdifficulties. It was found that the sandy materials at the siteobey the Coulomb-Mohr failure criterion. The silty sandsexhibit negligible cohesion and an effective friction angleof 32� (m = 0.62); the clayey sands exhibit cohesion ofapproximately 50 kPa and an effective friction angle of41.5� (m = 0.88).[77] Blast tests were performed at both saturated and dry

subsurface conditions, below the infiltration trench and300 m away from it, respectively. Particle velocities owingto the blast experiments were measured using a verticalgeophone array positioned 30 m away from the source. Wefound that particle velocities at saturated subsurface con-

ditions were much higher than particle velocities at drysubsurface, down to a depth of 14 m where the particlevelocity outputs of the two tests were the same. Thelongitudinal component was much higher than the trans-verse component in both subsurface conditions. The verticalcomponent was higher than the two horizontal componentsclose to the surface and diminished to almost zero at 14-mdepth, for both subsurface conditions.[78] Four different types of liquefaction features were

observed at the wet blast experiment site, whereas nonewere detected at the dry blast test site. Seven minutes afterthe wet blast test a water fountain (or a ‘‘geyser’’) eruptedfrom the blast borehole to a height of several meters. Afterthe ‘‘geyser’’ subsided the water surface at the base of thetrench continued to bubble for a period of about 30 min, to ahorizontal distance of 15 m from the blast borehole alongthe trench axis. This procedure is assumed to be equivalentto ‘‘sand boils’’ typically observed on the surface of sitesthat have experienced liquefaction. Approximately 20 minafter the blast lateral spreading features were observed atopposite banks of the trench across from the blast borehole.An accurate topographic survey performed immediatelyafter the blast indicated that the ground subsided by about85 mm close to the borehole and by up to 15 cm within a

Figure 13. Empirical determination of relative settlement [Kok, 1981] versus field measurements (thisstudy) after wet blast test. (a) Geometrical layout. (b) Comparison between predicted (solid circles) andmeasured (solid diamonds) settlement owing to blast-induced liquefaction.


20 of 22

B02311

radius of 15 m from the borehole. An ultimate settlement of150 cm was measured around the blast borehole 70 daysafter the experiment. The radius of the disturbed zoneremained 15 m after 70 days.[79] The water fountain and bubbling of the water surface

inside the infiltration trench indicate excess pore pressuregeneration at the subsurface. The lateral spreading observedon opposite banks of the infiltration trench could be asurface manifestation of material failure at the saturatedsubsurface, owing to diminished effective normal stress as aresult of excess pore pressure build up. It may also reflectthe brittle response of the dry material forming the banks ofthe infiltration trench to the settlement of underlying layersthat have experienced liquefaction in the saturated zone.The measured settlement represents re arrangement of thematerial at the subsurface into a more compact configura-tion, after excess pore pressure has been dissipated. Theextent of the area that has been subjected to excess porepressures and the amount of surface settlement are in goodagreement with published empirical relationships.[80] The major contribution of this paper is the detailed

and comprehensive characterization of the materials at thesite using a wide range of testing methods that cross theboundaries between established scientific and engineeringdisciplines, and the careful documentation of the controlledfield experiments as well as the obtained failure modes inthe field. We show that liquefaction can be induced whensubjecting the saturated sediments to a variety of high-amplitude waves that in the particular case study reportedhere are emitted from a blast. We conclude that liquefactioncan be induced in relatively dense sediments (relativedensity 63–89%) with shear wave velocities greater than300 m s�1, and with relatively high fines content. Suchsediments are not considered prone to liquefaction instandard liquefaction prediction procedures.

[81] Acknowledgments. We thank Michael Ezersky for performingthe seismic velocity tests and for stimulating discussions. Rivka Amit isthanked for discussions on the loess deposits in the studied area. DougSchmitt and two anonymous reviewers are thanked for critical readings ofearlier versions of this manuscript and for their thoughtful comments, whichhelped improve this paper significantly.

ReferencesAndersen, K. H., J. H. Pool, S. F. Brown, and W. F. Rosenbrand (1980),Cyclic and static laboratory tests on Drammen clay, J. Geotech. Eng. Div.Am. Soc. Civ. Eng., 106(GT5), 499–529.

Andrus, R. D., and K. H. Stokoe (2000), Liquefaction resistance of soilsfrom shear-wave velocity, J. Geotech. Geoenviron., 126(11), 1015–1025,doi:10.1061/(ASCE)1090-0241(2000)126:11(1015).

Ashford, S. A., K. M. Rollins, and J. D. Lane (2004), Blast-induced lique-faction for full-scale foundation testing, J. Geotech. Geoenviron., 130(8),798–806, doi:10.1061/(ASCE)1090-0241(2004)130:8(798).

Assallay, A. M., C. D. F. Rogers, and I. J. Smalley (1996), Engineeringproperties of loess in Libya, J. Arid Environ., 32(4), 373 – 386,doi:10.1006/jare.1996.0031.

Bachrach, R., and A. Nur (1998), High-resolution shallow-seismic experi-ments in sand, part I: Water table, fluid flow, and saturation, Geophysics,63(4), 1225–1233, doi:10.1190/1.1444423.

Bachrach, R., A. Nur, and A. Agnon (2001), Liquefaction and dynamicporoelasticity in soft sediments, J. Geophys. Res., 106(B7), 13,515–13,526, doi:10.1029/2000JB900474.

Bardet, J. P., T. Tobita, N. Mace, and J. P. Hu (2002), Regional modeling ofliquefaction-induced ground deformation, Earthquake Spectra, 18(1),19–46, doi:10.1193/1.1463409.

Boulanger, R. W., and I. M. Idriss (2007), Evaluation of cyclic softening insilts and clays, J. Geotech. Geoenviron., 133(6), 641–652, doi:10.1061/(ASCE)1090-0241(2007)133:6(641).

Boulanger, R. W., L. H. Mejia, and I. M. Idriss (1997), Liquefaction atMoss Landing during Loma Prieta earthquake, J. Geotech. Geoenviron.,123(5), 453–467, doi:10.1061/(ASCE)1090-0241(1997)123:5(453).

Bray, J. D., et al. (2004), Subsurface characterization at ground failure sitesin Adapazari, Turkey, J. Geotech. Geoenviron., 130(7), 673 – 685,doi:10.1061/(ASCE)1090-0241(2004)130:7(673).

Cetin, K. O., T. L. Youd, R. B. Seed, J. D. Bray, R. Sancio, W. Lettis, M. T.Yimaz, and H. T. Durgunoglu (2002), Liquefaction-induced grounddeformations at Hotel Sapanca during Kocaeli (Izmit), Turkey, earthquake,Soil Dyn. Earthquake Eng., 22(9–12), 1083–1092, doi:10.1016/S0267-7261(02)00134-3.

Cetin, K. O., R. B. Seed, A. Der Kiureghian, K. Tokimatsu, L. F. Harder,R. E. Kayen, and R. E. S. Moss (2004a), Standard penetration test-basedprobabilistic and deterministic assessment of seismic soil liquefactionpotential, J. Geotech. Geoenviron., 130(12), 1314–1340, doi:10.1061/(ASCE)1090-0241(2004)130:12(1314).

Cetin, K. O., T. L. Youd, R. B. Seed, J. D. Bray, J. P. Stewart, H. T.Durgunoglu, W. Lettis, and M. T. Yilmaz (2004b), Liquefaction-inducedlateral spreading at Izmit bay during the Kocaeli (Izmit), Turkey,earthquake, J. Geotech. Geoenviron., 130(12), 1300–1313, doi:10.1061/(ASCE)1090-0241(2004)130:12(1300).

Charlie, W. A., and D. O. Doehring (2007), Groundwater table mounding,pore pressure, and liquefaction induced by explosions: Energy-distancerelations, Rev. Geophys., 45, RG4006, doi:10.1029/2006RG000205.

Charlie, W. A., G. E. Veyera, and S. R. Abt (1985), Predicting blast inducedpore water pressure increases in soil-A review, Civ. Eng. Pract. Des.,4(4), 311–328.

Charlie, W. A., C. E. Scott, T. J. Siller, L. W. Butler, and D. O. Doehring(1995), Estimating liquefaction potential of sand using the Piezovane,Geotechnique, 45(1), 55–67.

Charlie, W. A., G. E. Veyera, D. S. Durnford, and D. O. Doehring (1996),Pore water pressure increases in soil and rock from underground chemicaland nuclear explosions, Eng. Geol., 43(4), 225–236, doi:10.1016/S0013-7952(96)00032-4.

Charlie, W. A., W. A. Lewis, and D. O. Doehring (2001), Explosiveinduced pore pressure in a sandfill dam,Geotech. Test. J., 24(4), 391–400.

Chu, D. B., J. P. Stewart, S. Lee, J. S. Tsai, P. S. Lin, B. L. Chu, R. B. Seed,S. C. Hsu, M. S. Yu, and M. C. H. Wang (2004), Documentation of soilconditions at liquefaction and non-liquefaction sites from 1999 Chi-Chi(Taiwan) earthquake, Soil Dyn. Earthquake Eng., 24(9–10), 647–657,doi:10.1016/j.soildyn.2004.06.005.

Clark, P. S. (1966), Handbook of Physical Constants, Geol. Soc. of Am.,Boulder, Colo.

Cristescu, N. D. (2000), Theoretical approach to sand liquefaction, in SoilDynamics and Liquefaction 2000, Geotech. Spec. Publ., 107, edited byR. Y. S. Pak and J. Yamamura, p. 208, Am. Soc. Civ. Eng., Reston, Va.

Crouvi, O., R. Amit, Y. Enzel, N. Porat, and A. Sandler (2008), Sand dunesas a major proximal dust source for late Pleistocene loess in the NegevDesert, Israel, Quat. Res., 70(2), 275–282.

Dahan, O., E. V. McDonald, and M. H. Young (2003), Flexible timedomain reflectometry probe for deep vadose zone monitoring, VadoseZone J., 2, 270–275.

Dan, J., D. H. Yaalon, H. Koyumdisky, and Z. Raz (1976), The Soil Map ofIsrael, Div. of Sci. Publ., Volcani Cent., Bet Dagan, Israel.

deGroot,M. B.,M.D. Bolton, P. Foray, P.Meijers, A. C. Palmer, R. Sandven,A. Sawicki, and T. C. Teh (2006), Physics of liquefaction phenomenaaround marine structures, J. Waterw. Port Coastal Ocean Eng., 132(4),227–243, doi:10.1061/(ASCE)0733-950X(2006)132:4(227).

Dijkstra, T. A. (2001), Geotechnical thresholds in the Lanzhou loess ofChina, Quat. Int., 76–77, 21–28, doi:10.1016/S1040-6182(00)00086-0.

Enzel, Y., R. Amit, U. Dayan, O. Crouvi, R. Kahana, B. Ziv, and D. Sharon(2008), The climatic and physiographic controls of the eastern Mediter-ranean over the late Pleistocene climates in the southern Levant and itsneighboring deserts, Global Planet. Change, 60, 165–192, doi:10.1016/j.gloplacha.2007.02.003.

Evans, M. D., H. B. Seed, and R. B. Seed (1992), Membrane complianceand liquefaction of sluiced gravel specimens, J. Geotech. Eng., 118(6),856–872, doi:10.1061/(ASCE)0733-9410(1992)118:6(856).

Ezersky, M. (2004), Seismic survey for determination of shear and pres-sure wave velocities in Amitai area, Geophys. Inst. of Israel, Holon,Israel.

Fener, M., S. Kahraman, Y. Bay, and O. Gunaydin (2005), Correlationsbetween P wave velocity and Atterberg limits of cohesive soils, Can.Geotech. J., 42(2), 673–677, doi:10.1139/t04-102.

Fiegel, G. L., and B. L. Kutter (1994), Liquefaction mechanism for layeredsoils, J. Geotech. Eng., 120(4), 737–757, doi:10.1061/(ASCE)0733-9410(1994)120:4(737).

Florin, V. A., and P. L. Ivanov (1961), Liquefaction of saturated sandy soils,paper presented at 5th Annual Conference of the ICSMFE, Paris.


21 of 22

B02311

Gandhi, S. R., A. K. Dey, and S. Selvam (1999), Densification of pond ashby blasting, J. Geotech. Geoenviron., 125(10), 889–899, doi:10.1061/(ASCE)1090-0241(1999)125:10(889).

Gassmann, F. (1951), Uber die elastizitat poroser medien, Vierteljahrsschr.Nat. Geschr. Zurich, 96, 1–23.

Gibbs, H. J., and W. G. Holtz (1957), Research on determining the densityof sands by spoon penetration testing, paper presented at 4th AnnualConference of ICSMFE, London.

Gvirtzman, H., E. Shalev, O. Dahan, and Y. H. Hatzor (2008), Large-scaleinfiltration experiments into unsaturated stratified loess sediments:Monitoring and modeling, J. Hydrol. Amsterdam, 349(1–2), 214–229,doi:10.1016/j.jhydrol.2007.11.002.

Idriss, I. M., and R. W. Boulanger (2006), Semi-empirical procedures forevaluating liquefaction potential during earthquakes, Soil. Dyn. Earth-quake Eng., 26(2–4), 115–130, doi:10.1016/j.soildyn.2004.11.023.

Ishihara, K. (1985), Stability of natural deposits during earthquakes, inProceedings of the 11th International Conference on Soil Mechanicsand Foundations Engineering, pp. 321–376, Soil Mech. and Found.Eng., San Francisco, Calif.

Jefferson, I. F., N. Mavlyanova, K. O’Hara-Dhand, and I. J. Smalley (2004),The engineering geology of loess ground: 15 tasks for investigators-TheMavlyanov programme of loess research, Eng. Geol., 74(1–2), 33–37,doi:10.1016/j.enggeo.2004.01.012.

Kishida, H. (1969), Characteristics of liquefied sands during Mino-Owari,Tohankai and Fukui earthquakes, Soil Found., 9(1), 75–92.

Kok, L. (1981), Settlements due to contained explosions in water-saturatedsands, paper presented at 7th International Symposium on MilitaryApplications of Blast Simulation, Medicine Hat, Alberta, Canada.

Kok, L., and R. W. Trense (1979), Blast-induced soil liquefaction: State ofthe art, paper presented at 6th International Symposium on MilitaryApplications of Blast Simulation, Cahors, France, 25–29 June.

Lin, A. M. (1997), Instantaneous-shaking liquefaction induced by the M7.21995 Southern Hyogo Prefecture earthquake, Japan, Geology, 25(5),435–438, doi:10.1130/0091-7613(1997)025<0435:ISLIBT>2.3.CO;2.

Marcuson, W. F. I. (1978), Definition of terms related to liquefaction,J. Geotech. Eng. Div. Am. Soc. Civ. Eng., 104(9), 1197–1200.

Martin, G. R., W. D. L. Finn, and H. B. Seed (1978), Effects of systemcompliance on liquefaction tests, J. Geotech. Eng. Div. Am. Soc. Civ.Eng., 104(4), 463–479.

Menashe, R. (2003), The stratigraphy and paleo-geography of Tel Sharu-hen Section, Northern Negev, Israel, 95 pp., Hebrew Univ. of Jerusalem,Jerusalem.

Miura, S., S. Kawamura, and K. Yagi (1995), Liquefaction damage ofsandy and volcanic grounds in the 1993 Hokkaido Nansei-Oki earth-quake, paper presented at 3rd International Conference on RecentAdvances in Geotechnical Earthquake Engineering and Soil Dynamics,St. Louis, Miss., 2–7 Apr.

Moretti, M., P. Alfaro, O. Caselles, and J. A. Canas (1999), Modellingseismites with a digital shaking table, Tectonophysics, 304(4), 369–383, doi:10.1016/S0040-1951(98)00289-3.

Mulilis, J. P., H. B. Seed, C. K. Chan, J. K. Mitchell, and K. Arulanandan(1977), Effects of sample preparation on sand liquefaction, J. Geotech.Eng. Div. Am. Soc. Civ. Eng., 103(2), 91–108.

Rajendran, K., C. P. Rajendran, M. Thakkar, and M. P. Tuttle (2001), The2001 Kutch (Bhuj) earthquake: Coseismic surface features and theirsignificance, Curr. Sci., 80(11), 1397–1405.

Rauch, A. F. (1997), An empirical method for predicting surface dis-placements due to liquefaction-induced lateral spreading in earth-quakes, 331 pp., Va. Polytech. Inst., Blacksburg, Va.

Rossetti, D. D. (1999), Soft-sediment deformation structures in late Albianto Cenomanian deposits, Sao Luis Basin, northern Brazil: Evidence forpalaeoseismicity, Sedimentology, 46(6), 1065–1081, doi:10.1046/j.1365-3091.1999.00265.x.

Sawicki, A., and W. Swidzinski (2007), Simple mathematical model forassessment of seismic-induced liquefaction of soils, J. Waterw. PortCoastal Ocean Eng., 133(1), 50 – 54, doi:10.1061/(ASCE)0733-950X(2007)133:1(50).

Schnabel, P. B., J. Lysmer, and H. B. Seed (1972), SHAKE: A computerprogram for earthquake response analysis of horizontally layered sites,102 pp., Earthquake Eng. Res. Cent., Univ. of Calif., Berkeley, Calif.

Seed, H. B. (1979), Soil liquefaction and cyclic mobility evaluation forlevel ground during earthquakes, J. Geotech. Eng. Div. Am. Soc. Civ.Eng., 105(2), 201–255.

Seed, H. B. (1980), Soil liquefaction and cyclic mobility evaluation forlevel ground during earthquakes-Closure, J. Geotech. Eng. Div. Am.Soc. Civ. Eng., 106(6), 724–724.

Seed, H. B., and I. M. Idriss (1970), Soil moduli and damping factors fordynamic response analyses, Univ. of Calif., Berkeley, Calif.

Seed, H. B., and I. M. Idriss (1971), Simplified procedures for evaluatingsoil liquefaction potential, J. Soil Mech. Found. Div. Am. Soc. Civ. Eng.,97(SM9), 1249–1273.

Seed, H. B., I. Arango, C. K. Chan, A. Gomezmasso, and R. G. Ascoli(1981), Earthquake-induced liquefaction near Lake Amatitlan, Guatemala,J. Geotech. Eng. Div. Am. Soc. Civ. Eng., 107(4), 501–518.

Seed, H. B., K. Tokimatsu, L. F. Harder, and R. M. Chung (1985), Influenceof SPT procedures in soil liquefaction resistance evaluations, J. Geotech.Eng. Div. Am. Soc. Civ. Eng., 111(12), 1425–1445.

Shtivelman, V. (1996), Kinematical inversion of first arrivals of refractedwave-A combined approach, Geophysics, 61, 509–519, doi:10.1190/1.1443977.

Skempton, A. W. (1954), The pore pressure coefficients A and B, Geotech-nique, 4, 143–147.

Studer, J., and L. Kok (1980), Blast-induced excess porewater pressure andliquefaction experience and application, paper presented at InternationalSymposium on Soils Under Cyclic and Transient Loading, Swansea,Wales, 7–11 Jan.

Takahashi, T., T. Takeuchi, and K. Sassa (2006), ISRM suggested methodsfor borehole geophysics in rock engineering, Int. J. Rock. Mech. Min.Sci., 43(3), 337–368, doi:10.1016/j.ijrmms.2005.09.003.

Terzaghi, K. (1943), Theoretical Soil Mechanics, 510 pp., John Wiley,Hoboken, N. J.

Tohno, I., and S. Yasuda (1981), Liquefaction of the ground during the1978 Miyagiken-Oki earthquake, Soil Found., 21(3), 13–34.

VanEekelen, H. A. M., and D. M. Potts (1978), The behaviour of DrammenClay under cyclic loading, Geotechnique, 28(2), 173–196.

Van Genuchten, M. T. (1980), A closed-form equation for predicting thehydraulic conductivity of unsaturated soils, Soil Sci. Soc. Am. J., 44,892–898.

Vucetic, M., and R. Dobry (1991), Effect of soil plasticity on cyclicresponse, J. Geotech. Eng., 117(1), 89–107, doi:10.1061/(ASCE)0733-9410(1991)117:1(89).

Wierenga, P. J., R. G. Hills, and D. B. Hudson (1991), The Las Crucestrench site: Characterization, experimental results, and one-dimensionalflow predictions, Water Resour. Res., 27(10), 2695–2705, doi:10.1029/91WR01537.

Wildenschild, D., and K. H. Jensen (1999), Numerical modeling ofobserved effective flow behavior in unsaturated heterogeneous sands,Water Resour. Res., 35(1), 29–42, doi:10.1029/98WR01959.

Yaalon, H., and J. Dan (1974), Accumulation and distribution of loess-derived deposits in the semi-desert and desert fringe areas of Israel,Z. Geomorphol. Suppl., 20, 91–105.

Youd, T. L., et al. (2001), Liquefaction resistance of soils: Summary reportfrom the 1996 NCEER and 1998 NCEER/NSF Workshops on Evaluationof Liquefaction Resistance of Soils, J. Geotech. Geoenviron., 127(10),817–833, doi:10.1061/(ASCE)1090-0241(2001)127:10(817).

Yuan,H.M., S. H.Yang, R. D. Andrus, and C.H. Juang (2004), Liquefaction-induced ground failure: A study of the Chi-Chi earthquake cases, Eng.Geol., 71(1–2), 141–155, doi:10.1016/S0013-7952(03)00130-3.

Zilberman, E. (1992), Late Pleistocene sequence of the northwestern Negevflood plains-A key to reconstructing the paleoclimate of southern Israel inthe last glacial, Isr. J. Earth Sci., 41, 155–167.

Zilberman, E., N. Porat, and J. Roskin (2007), The middle to late Pleistocenesand sheet sequence of Kerem Shalom, western Negev-An archive ofcoastal sand incursion, 23 pp., Geol. Surv. of Israel, Jerusalem.

��H. Gvirtzman, Institute of Earth Sciences, Hebrew University, Jerusalem

91904, Israel.Y. H. Hatzor, I. Orian, and I. Wainshtein, Department of Geological and

Environmental Sciences, Ben-Gurion University of the Negev, Beer-Sheva84105, Israel. ([email protected])


22 of 22

B02311

Documents

Induced liquefaction experiment in relatively dense, clay ... et al JGR 2009.pdf · Induced liquefaction experiment in relatively dense, ... In this paper we report results from a