Case Histories of Liquefaction Failures

MISCELLANEOUS PAPER S-76-4 CASE HISTORIES OF LIQUEFACTION FAILURES

Paul A. Gilbert

Soils and Pavements Laboratory U. S. Army Engineer Waterways Experiment Station P. O. Box 631, Vicksburg, Miss. 39180

April 1976 Final Report

Approved For Public Release; Distribution Unlimited

. ± ¿ L Z «

Prepared for Office, Chief of Engineers, U. S. Army TA Washington, D. C. 2 0 3 14 7 .W34m Under Project No. 4AI6II02B52E, S-76-4 Task 04 1976 LIBRARY

um 2 5 1981

Waters Por-- P o l i r e - ■■ Iteiiv^r, UGìOfiiuò

Destroy this report when no it to the BUREAU OF RECLAMATION DENVER LIBRARY 92044012

SECURITY CLASSIFICATION OF THIS PAGE (When Data Entered) READ INSTRUCTIONS REPORT DOCUMENTATION PAGE BEFORE COMPLETING FORM 1. REPORT NUMBER 2. GOVT ACCESSION NO. 3. RECIPIENT'S CATALOG NUMBER Miscellaneous Paper S-76-4

4. T IT L E (and Subtitle) 5. TYPE OF REPORT & PERIOD COVERED Final report CASE HISTORIES OF LIQUEFACTION FAILURES 6. PERFORMING ORG. REPORT NUMBER

7. AUTHOR!» 8. CONTRACT OR GRANT NUMBER*»

Paul A. Gilbert

9. PERFORMING ORGANIZATION NAME AND ADDRESS 10. PROGRAM ELEMENT, PROJECT, TASK AREA & WORK UNIT NUMBERS U. S. Army Engineer Waterways Experiment Station Soils and Pavements Laboratory Project 4A161102B52E, P. O. Box 631, Vicksburg, Miss. 39180 Task 04

11. CONTROLLING OFFICE NAME AND ADDRESS 12. REPORT DATE April 1976 Office, Chief of Engineers, U. S. Army Washington, D. C. 20314 13. NUMBER OF PAGES 24 14. MONITORING AGENCY NAME & ADDRESS!’// different from Controlling Office) 15. S E C U R IT Y CLASS, (of thia report) Unclassified

15«. declassification / downgrading SCHEDULE

16. DISTRIBUTION STATEMENT (of thia Report)

Approved for public release; distribution unlimited.

17. DISTRIBUTION STATEMENT (of the abatract entered in Block 20, if different from Report)

18. SUPPLEMENTARY NOTES

19. KEY WORDS (Continue on reverae aide if neceaaary and identify by block number) Liquefaction (Soils) Noncohesive soils

20k. ABSTRACT (Cbnttaum an reverae aid* ft neceeeary and identify by block number) Liquefaction of loose, saturated, cohesionless soils is a phenomenon in which the soil mass suddenly loses shear strength, behaves as a fluid, and acquires a degree of mobility sufficient to permit large movements. This report reviews various case histories to determine common characteristics associated with liquefaction failures. A review of case histories reveals that liquefaction failures are dependent upon (a) a collapsible soil structure, (b) a saturated and undrained condition, and (c) a triggering mechanism. Typically collapsible soils which liquefied were fine, uniform, loose sand deposits with D 10 sizes ranging from 0.05 to 1.0 mm and a coefficient of uniformity ranging from 2 to 10. Saturated-undrained conditions provided a situation conducive to high pore pressure development upon collapse of the soil structure. Generally, water was the pore fluid; however, several (Continued)

DO I JAN 73 1473 ed itio n OF I NOV ss is o b so l e t e Unclassified SECURITY CLASSIFICATION OF THIS PAGE (Witen Data Entered) ______Unclassified______SECURITY CLASSIFICATION OF THIS PAGEQWin Data Entered)

20. ABSTRACT (Continued).

unusual cases were reported with air as the pore fluid (termed fluidization). A variety of triggering mechanisms, including monotonically changing stresses, earthquakes, explosive blasts, and cyclic vibrations, were found to cause liquefaction failures. However, monotonically increasing shear stresses and earthquakes are the most common triggering mechanisms. Density is the most important property controlling the susceptibility of saturated-undrained sands to liquefaction. It was found that sands which liquefy when subjected to earthquake shakings do not become significantly more stable against reliquefaction. Conversely, vibrations appear to alter the sand structure, making a deposit less susceptible to liquefaction than indicated by density increases.

Unclassified SECURITY CLASSIFICATION OF THIS PAGE(TWien Data Entered) PREFACE

This study was funded by the Office, Chief of Engineers, U. S. Army, under Project No. 4A161102B52E, Task 04, “Research in Military Engineering and Construction.” The work was conducted during the period September 1974-May 1975 at the U. S. Army Engineer Waterways Experiment Station (WES) by Mr. Paul A. Gilbert under the direct supervision of Dr. Frank C. Townsend, Chief, Soils Research Facility. The study was under the general supervision of Mr. Clifford L. McAnear, Chief, Soil Mechanics Division, and Mr. James P. Sale, Chief, Soils and Pavements Laboratory. The technical monitor of this study at OCE was Mr. A. F. Muller. The Director of WES during this study was COL G. H. Hilt, CE, and the Technical Director was Mr. F. R. Brown.

1 CONTENTS

Page

PREFACE ...... 1 CONVERSION FACTORS, U. S. CUSTOMARY TO METRIC (SI) UNITS OF MEASUREMENT ...... 3 PAR TI: INTRODUCTION ...... 4 Background ...... 4 Objective and Scope ...... 4 PART II: NECESSARY CONDITIONS FOR LIQUEFACTION FAILURE ...... 5 Loose Deposits, Fine Sand ...... 5 Soil Saturation ...... 7 Triggering Mechanisms ...... 8 PART III: CASE HISTORIES ...... 9 Liquefaction Induced by Monotonically Changing Stresses ...... 9 Liquefaction Induced by Cyclic Vibrations ...... 12 Flow Failures Induced by Earthquakes, Blasts, and Wave Action ...... 14 PART IV: DISCUSSION OF CASE HISTORIES ...... 19 Soil Type and Triggering Mechanisms Causing Liquefaction ...... 19 Effects of Density on Liquefaction Potential ...... 19 Liquefaction Potential ...... 21 PART V: CONCLUSIONS ...... 22 REFERENCES ...... 23

2 CONVERSION FACTORS, U. S. CUSTOMARY TO METRIC (SI) UNITS OF MEASUREMENT

U. S. customary units of measurement used in this report can be converted to metric (SI) units as follows:

Multiply By To Obtain feet 0.3048 metres miles (U. S. statute) 1.609344 kilometres acres 4046.856 square metres acre-feet 1233.482 cubic metres cubic yards 0.7645549 cubic metres tons (short, 2000 lb mass) 907.1847 kilograms pounds (force) 4.448222 newtons tons (force) 8.89644 kilonewtons miles per hour (U. S. statute) 1.609344 kilometres per hour tons (nuclear equivalent of TNT) 4200 megajoules degrees (angle) 0.01745329 radians

3 CASE HISTORIES OF LIQUEFACTION FAILURES

PART I: INTRODUCTION

BACKGROUND

1. By definition, liquefaction of cohesionless soils is a phenomenon in which a soil mass suddenly loses shear strength and behaves as a viscous fluid in that it undergoes large shear deformations without recovery of shear resistance. This loss of shear strength is due to a collapse of the soil structure which transfers load carried by the grain structure onto the pore water, thereby increasing the pore water pressure and decreasing the effective stress within the soil mass. Hence the process requires (a) a collapsible soil structure, (b) a nearly saturated and undrained condition, and (c) a triggering mechanism. A collapsible soil structure implies a low density and little or no cohesion to restrict movement of the soil grains. In this context, most liquefaction failures have been observed in sand deposits. A nearly saturated-undrained condition is implied since if air were present (i.e., partial saturation) or drainage were to occur, the induced pore pressures would quickly dissipate and the effective stress would increase, allowing the mass to recover strength. A variety of triggering mechanisms exist which cause shear strains leading to collapse of the soil structure; these may be monotonically changing shear stresses, cyclic vibratory loads, or random shock waves.

OBJECTIVE AND SCOPE

2. The objective of the study was to determine the common conditions associated with liquefaction failures caused by the various triggering mechanisms. Published case histories are reviewed and evaluated to establish the common conditions. Each of these conditions is studied to determine its relative importance in the evaluation of liquefaction susceptibility.

4 PART II: NECESSARY CONDITIONS FOR LIQUEFACTION FAILURE

3. This part of the report discusses the field conditions necessary for a liquefaction failure to occur. Categorically, these are: a. Deposits of loose, cohesionless soil. b. A high groundwater table causing a saturated or near-saturated soil.

c. A triggering mechanism.

LOOSE DEPOSITS, FINE SAND

4. When a granular material experiences a shear strain, it tends to change volume. Generally, if the material is “loose,” it tends to decrease in volume; and if “dense,” it tends to increase in volume. This fact is illustrated in Figure 1. This figure shows that in dense sands, as shown in Figure la and b, the grains are so tightly interlocked that deformations will not be possible unless accompanied by an expansion of the sand. If the sand is confined so that it cannot expand, then shear strains can occur only if the crushing strength of the grains is exceeded. Conversely, loose materials, as illustrated in Figure lc and d, tend to decrease in volume when subjected to shear strains. Typically, an increase in volume, i.e., dilation as observed in dense sands, causes a decrease in pore pressure, while a decrease in volume as observed in loose sands causes an increase in pore pressure. Therefore, dense cohesionless materials are not susceptible to liquefaction since a disturbance decreases the pore pressure, increasing the apparent strength and rendering the material more stable. In loose materials the pore pressure increases and the material loses strength as undrained shear occurs.

c. LOOSE SAND BEFORE SHEARING <•. LOOSE SAND AFTER SHEARING

Figure 1. Effect of shearing on volume of cohesionless soils (after Casagrande’)

5 5. Since there is greater potential for volume change with loss of strength in very loose materials, the danger of liquefaction is

NV also greater. Terzaghi2 describes the formation I of a loose, unstable soil structure which he calls a metastable structure. According to Terzaghi, - H d h ~ a soil particle with weight Q (Figure 2) settling through water arrives at the surface of a deposit and tends to roll or slide into a stable position However, this tendency is resisted by a couple, Mc , produced by adhesion at the point of first contact. The moment Mc is independent of particle size, but the overturning moment Mq varies directly* as the fourth power of the particle diameter. Therefore the probability that particles will assume stable positions decreases with decreasing grain size. 6. A theoretical relationship between settling velocity of soil grains in a viscous fluid and grain size was established by G. G. Stokes in 1850. If the assumptions used in the derivation are accepted, then Stokes’ equation indicates that the terminal settling velocity of soil particles in a fluid varies as the square of the diameter of the particle. This means that in settling, larger diameter particles have considerably more momentum and kinetic energy to hammer themselves into stable positions than do smaller particles. (Momentum and kinetic energy are proportional to the fifth and seventh powers, respectively, of the diameter.)** Here again, it is probable that smaller particles will assume less stable positions.

* Note: In Reference 2 Terzaghi states, “ Mq decreases in inverse proportion to the fourth power of the diameter of the grain.” This writer believes that the word “ inverse” should read “ direct,” since: C i,C 2,C 3 = constants of proportionality D = average diameter of the grain Q = weight of the grain -

d = moment arm of the grain (a linear function of the diameter) = C2D Mq = overturning moment = Qd

Mq = Qd = (C i D3)(C2D) = C3D4

which indicates that Mq varies directly as the fourth power of the diameter D . ** By Stokes’ law, the terminal velocity v = C jD ^ . Then

KE = j mv2 = | 1 (C2d 3)J (Ci D2)2 = C3D?

P = mv = (C2 d 3)(Cj D2) = C4D5 where K E = kinetic energy = 1/2 mv^

m = particle mass (a function of the cube of the particle diameter) = C2D^ P = momentum = mv

6 7. This is not to indicate, however, that structural stability continues to decrease indefinitely as particle size decreases. There is a very definite range of particle sizes which are most susceptible to liquefaction. For example, as the particle size of cohesionless soils increases, the permeability of the soil becomes so great that an undrained condition could not be maintained long enough for liquefaction and large movements to occur. At the opposite extreme, as particle sizes become very small, most soils begin to develop cohesion and become less susceptible to liquefaction flow. 8. D’Appolonia3 suggests that soils exhibiting all of the following physical properties are most susceptible to liquefaction: a. The percentage of silt and clay-size particles is less than 10 percent. b. The particle diameter at 60 percent passing on the grain-size curve, D60 , is between 0.2 and 1.0 mm. c. The uniformity coefficient (Cu = D60/ D io), which is a measure of the variation in particle size, is between 2 and 5. d. The blow count from standard penetration is less than 15, indicating that the material is loosely deposited.

Other investigators such as Whitman,4 Seed and Idriss,5 and Terzaghi and Peck6 report similar findings on the relationships between physical properties of soils and liquefaction susceptibility. 9. Based upon studies of flow slides due to liquefaction along the lower Mississippi River conducted under potamology studies beginning in 1947, the Corps of Engineers established criteria for slope instability.7 All major flow failures occurred in point bar deposits with a somewhat cohesive overburden underlain by fine uniform sands (upper sand series) and deeper coarse sands (lower sand series). The upper sand series is subdivided into fine loose sand (zone A sand) and a relatively coarser and denser sand (zone B sand). The criteria suggested for slope instability due to liquefaction flow slides of these Mississippi River embankments are as follows: a. The zone A sand is at least 20 ft* in thickness. b. The ratio of the overburden thickness to the zone A thickness is 0.85 or less. c. More than 20 percent of the overburden soil passes the No. 200 sieve. d. At least 50 percent of the upper sand passes the No. 40 sieve with 25 percent or more of zone A sand passing the No. 60 sieve and less than 25 percent of the zone B sand passing the No. 60 sieve. e. Less than 50 percent of the lower sand passes the No. 40 sieve.

SOIL SATURATION

10. In this review, all investigators cited were in agreement that two major requirements for liquefaction to occur are saturation of the soil in question and an undrained state in the zone being liquefied. Theoretically, liquefaction does not require complete water saturation, but if the volume of gasses is sufficient, their high compressibility in the pores of partially saturated soil would prevent the buildup of significant pore pressure and consequently lessen the probability of liquefaction. Very fine grained cohesionless soils with air as the pore fluid have been reported to flow by Casagrande.8

* A table of factors for converting U. S. customary units of measurement to metric (SI) units is presented on page 3.

7 However, the mechanism resulting in flow was the same as in water-saturated soils with air playing the role of water. Casagrande appropriately called this phenomenon “fluidization.” Whereas fluidization remains a dangerous possibility, most cohesionless deposits lie below the groundwater table so that the potential danger from liquefaction is greater simply because of the greater quantity of water-saturated material. For this reason, liquefaction failures will be stressed in this report.

TRIGGERING MECHANISMS

11. A soil structure may be disturbed to the extent that it suffers structural collapse and liquefaction flow by either a monotonically changing shear stress causing shear strains, a cyclic vibratory load, or a shock wave. If a soil is very loose and unstable (metastable), the disturbing mechanism causing liquefaction may be so small that it is unidentifiable. Terzaghi2 called flow failures triggered by such small causes “spontaneous liquefaction.” 12. Monotonic changes in shear stress may be caused by seepage forces, tidal currents, channel currents, sediment deposition, man-made structures, or changes in stress caused by erosion, excavation, etc. Any phenomenon which causes a gradual change in stress could induce structural collapse in cohesionless materials with unstable structures; if the material is water-saturated without adequate drainage, a liquefaction failure could result. 13. It is common knowledge that vibratory loads are used very effectively to compact granular fill material.9 In this context, if loose, saturated granular materials are subjected to vibratory loads, the tendency for the material to densify can result in increased pore pressure and possibly liquefaction. However, the intensity and duration of the vibratory load must be sufficient to densify the soil structure rapidly enough to prevent dissipation of excess pore pressure. Vibratory loads may be caused by reciprocating machinery such as used in railway10 and pile-driving11 operations. Casagrande1 mentions the use of pile driving for the purpose of compacting loose granular deposits to improve their ability to carry loads without undue settlement. Special machines to compact loose granular soils by means of forced vibrations were developed in Germany in the 1930’s. This process in the United States is called vibroflotation. The vibroflotation process described by D’Appolonia in Reference 9 employs a vibrating probe which is water-jetted into a granular deposit. As the probe is slowly pulled out, the vibrations densify material around the probe within a certain sphere of influence. The number of penetrations required to densify a given deposit is a function of the areal extent of the deposit and the dimension of the sphere of influence of the probe. This process has proven very efficient in stabilizing granular deposits against settlement and liquefaction, and demonstrates the effectiveness of vibrations in inducing volume change in granular soils. 14. Shock waves and the accompanying random ground motions and vibrations may induce shear stresses in a saturated granular deposit of such intensity that liquefaction results. This is perhaps the most important category of disturbance since earthquakes are included in this group. As with forced cyclic vibrations, the likelihood of transient vibratory motions causing liquefaction depends on the intensity and duration of the motion. Explosions or blast loads can result in ground disturbances similar to those caused by earthquakes, and for this reason are included in this category.

8 PART III: CASE HISTORIES

15. Although experience has shown that the greatest damage to civil engineering structures due to liquefaction occurs as a result of earthquakes, cases of failures triggered by other mechanisms will also be considered. This part of the report presents case histories of liquefaction failures grouped according to triggering mechanism.

LIQUEFACTION INDUCED BY MONOTONICALLY CHANGING STRESSES

Zeeland Slides 16. The Dutch province of Zeeland consists of a large group of islands separated by wide estuaries. On the shores of the islands in that province, numerous flow slides have occurred. From 1881 to 1946, 229 such slides were reported with a volume displacement of 25 million cu m and an area loss of 660 acres. The soils involved in these slides were sands deposited in shallow water by the rivers Rhine, Meuse, and Scheldt. Grain-size curves reported by Koppejan10 showed these sands to be mostly fine and uniform, with a typical Dio of 0.08 mm and uniformity coefficients, typically, of less than 2. 17. Erosion and seepage pressures resulting from tidal fluctuations of up to 4.6 m were believed to have triggered the slides. According to the description by Koppejan, these slides progressed inland as a succession of small slides starting when the toe of a slope was scoured away by tidal current, releasing lateral support in the slope. Under reduced lateral stress, the soil tended to expand, generating negative pore water pressure which caused inward seepage. Water seeping into the soil relieved the negative pore pressure, which became positive as shear strains occurred and finally became large enough to cause liquefaction flow in that portion of the slope. After a certain amount of material had flowed away in one small slide, the process would begin again and continue in this manner until denser, more stable material was reached at which time the flow process would stop. The process is clearly illustrated by Hvorslev12 in Figure 3. According to Koppejan, the slides progressed inland at a rate of about 50 m per hour.

Mississippi River Flow Slides 18. Numerous flow slides have occurred along the banks of the Mississippi River.12 The materials involved in these slides were uniform fine sands with rounded to subrounded particles and uniformity coefficients from 1.5 to 1.8. Those slides occurring in revetted point bar deposits have been carefully studied, but slides have occurred in unrevetted reaches of the river as well. The slides have damaged revetments and levees built for flood protection. On 24 March 1949 a slide involving the loss of a portion of levee occurred at Free Nigger Point near Baton Rouge, Louisiana. This single slide involved the volume displacement of 4.4 million cu yd of material and occurred in less than 12 hr. The slides occurring on the river are described by M. J. Hvorslev12 to be similar to those occurring in Zeeland, Holland, in regard to form, dimensions, slopes, duration, material, and groundwater conditions.

Partial Failure at Fort Peck Dam 19. A granular shell of a dam constructed by the hydraulic fill method is likely to be deposited in a loose condition, and therefore susceptible to liquefaction failure. The Fort Peck Dam in Montana is an embankment with shells of this type. On 22 September 1938 a partial failure occurred involvinga 1700- ft-long section of the upstream shell, portions of which moved some 1500 ft in a period of about 4 min.

9 OVERBURDEN A = FAILED SECTION. PARTIAL LIQUEFACTIONJ FLOWING OUT THROUGH NECK

SECTION B TENDS TO EXPAND CAUSING NEGATIVE PORE WATER PRESSURE AND INWARD SEEPAGE AS TEMPORARY STABILIZING FORCES

AFTER IN ITIA L INWARD SEEPAGE^ LATERAL EXPANSION AND SHEARING STRAINS OCCUR; CAUSING GRADUAL CHANGE OF NEGATIVE TO POSITIVE LATERAL DIS PORE WATER PRESSURES IN SECTION B PLACEMENT AND SHEAR STRAINS

OVERBURDEN

THE POSITIVE OR EXCESS PORE WATER PRESSURES INCREASE WITH INCREASING STRAINS AND FIN ALLY BECOME LARGE ENOUGH TO CAUSE FAILURE OF SECTION B

EXCESS PORE WATER PRESSURES 1 1 1654 S

Figure 3. Dutch hypothesis for progressive flow failures (from Hvorslev12)

10 20. The dam was constructed of river sands and finer grained alluvial soils on a foundation of alluvial sands, gravels, and clays with a total thickness of up to 130 ft. Beneath the river alluvium is Bearpaw shale which contains layers of bentonite.13 In a discussion of a paper by Middlebrooks,14 Gilboy expresses the opinion, which is shared by Casagrande, that liquefaction was the predominant failure mechanism which was triggered by shear strains in the foundation shale. This failure involved the movement of about 10 million cu yd of material. 21. The failure of the hydraulic-filled Calaveras Dam in California was similar to that of the Fort Peck Dam and was attributed to liquefaction of the sands in the shell by Allen Hazen15 in 1920.

Flow Slide at Aberfan Village in Wales 22. Aberfan is a mining village, situated at the foot of a low mountain, the side of which had been used to deposit waste materials from coal mining operations. The waste materials were deposited in hillocks which were locally called “tips.” From 1915 to October 1966, seven tips had been formed, and on 21 October 1966, the most recently constructed tip failed by liquefaction sending about 200,000 cu yd of material flowing down the mountainside and into the village. In this catastrophe, 146 children lost their lives in an elementary school in the flow path. The tip that failed had been deposited over natural springs issuing from outcrops of the underlying sandstone. At the time of failure, material was still being deposited on the tip. Heavy rainfall, common to this area in October, probably eroded the base of the tip, reducing lateral stress and causing internal shear strains. This condition, coupled with the natural groundwater conditions of the mountainside and the continued loading of the tip, was believed to have triggered the liquefaction failure.16 23. Hutchinson16 describes the material involved in the flow slide as consisting chiefly of shale fragments and mudstones ranging from clay size particles to large boulders as much as 3 ft in diameter. For the grain sizes small enough to be sieved, the average D io size was between 0.1 and 0.8 mm and the coefficient of uniformity was approximately 18. Hutchinson points out that the Aberfan material was neither fine grained nor poorly graded. He describes tailing materials from other sites which failed in a similar manner as having high uniformity coefficients and questions the uniformity of soils as a criterion for liquefaction susceptibility.

Submarine Slide at the Folia Fjord in Norway 24. The occurrence of submarine failures very often goes unobserved unless indicated by a change in the shoreline or a temporary, unusual agitation of the sea. If a disturbance is indicated, the extent can be determined only by sounding and mapping changes in the ocean floor. 25. In a personal communication to Terzaghi in 1956, L. Bjerrum described a submarine slide which occurred in the Folia Fjord in Norway, which Terzaghi reported in Reference 2. The Folia Fjord is about 1000 m wide, is roughly rectangular in shape, and is located about 120 miles north-northeast of Trondheim in Norway. On 9 January 1952 a dredge was docked in the fjord and stayed with two cables and three anchors. In calm sea, one cable and one anchor chain snapped as a result of a local sand slide. A wave 4-5 ft high coming from the head of the fjord passed the dredge a few minutes later, after which the sea again became calm. About 14 min after the first cable and chain snapped, the remaining cable and one of the two remaining anchor chains snapped. The one remaining chain and anchor dragged the dredge about 1000 ft toward the center of the fjord. It was discovered later that a pier at the head of the

11 fjord and a ferryboat landing about halfway between the dredge and pier had been damaged by slides. 26. These observations led Terzaghi to conclude that initially, shallow liquefaction occurred near the dredge and started sand movement there. As the liquefaction spread toward the head of the fjord, the depth of material affected became much greater and precipitated a major slide there. From the head of the fjord, liquefaction proceeded again toward the dredge, this time affecting a much greater quantity of sand, and terminated with a major slide beneath the dredge. Terzaghi suggests that about 300,000 cu m of material were displaced and that the propagation speed of the liquefaction was about 5 mph.

LIQUEFACTION INDUCED BY CYCLIC VIBRATIONS

Liquefaction of Railway Embankment Near Weesp, Holland

27. Koppejan10 describes a flow slide in the approach to a railway bridge over the Merwede Canal near Weesp, Holland, in 1918. The embankment which failed had been formed by dumping moist sand. The sand had consequently “bulked,” forming a very loose and unstable structure. After the water level in the canal rose and saturated the sand, the embankment was subjected to vibrations from a passing train and collapsed with the train causing heavy casualties. It is probable that capillary forces present in the moist sand served to give the embankment initial stability. As saturation occurred, the capillary forces disappeared and the stability of the embankment decreased to the extent that vibrations from the passing train were sufficient to cause liquefaction. The coefficient of uniformity of the sand involved in the slide was about 2 and the D io size was about 0.14 mm. Geuze17 determined that the in situ density of the material was lower than the critical density defined by Casagrande.1 According to Koppejan,10 liquefaction triggered by vibrations occurs much more rapidly than liquefaction by monotonic stress changes because the entire mass subjected to vibrations liquefies almost simultaneously. This railway accident was so serious that as a result, the formal practice of soil mechanics was begun immediately in the Netherlands.

Vibratory Compaction Study by Prakash and Gupta

28. Since field investigations to study the vibratory compaction characteristics of sands are expensive and difficult, such studies are frequently made in the laboratory. Prakash and Gupta18 conducted such a study to investigate the behavior of sand under vertical and horizontal vibrations. The study was conducted on a fine uniform sand from India called Solani River sand with a uniformity coefficient of 2.06 and a Dio of 0.077 mm. For dry sands, the general conclusions reached in the study were as follows: a. Horizontal vibrations are more effective than vertical vibrations as a densifying mechanism. b. The acceleration applied to the sand determines the speed and degree of densification. c. There is an optimum frequency for each amplitude of vibration at which maximum densification is achieved. In vertical vibration, the density increases as the frequency increases up to the optimum frequency; then rarefaction occurs when the optimum frequency is exceeded. In horizontal vibration, the density increases up to the optimum frequency, then remains essentially constant as the frequency is increased past optimum. d. A particular static load or surcharge is required to produce the most efficient densification

12 at a given frequency. Static loads too large or small will result in a lower density at the given frequency. e. Moist sands will not be densified as much as dry or saturated sands by vibratory compaction. Surface tension between grains due to the presence of water will hamper densification. / Coarse and well-graded sands can be compacted with much less effort than uniform fine sands. 29. For saturated sands in this study, it was reported that the sands initially liquefied before densifying when vibrations were applied. If vibratory effects in dry sands are similar to those in saturated-undrained sands, then conclusions c and d suggest interesting possibilities. Conclusion c would suggest that a certain frequency would cause liquefaction more readily in a given material than any other, and conclusion d would suggest that liquefaction would be more likely to start at a certain depth under a given vibratory excitation. 30. Conclusion a suggests that horizontal ground motion resulting from earthquake loading would be more likely than vertical motion to induce liquefaction. Conclusion b indicates that the intensity and duration of ground motion determine whether or not liquefaction will occur during an earthquake. Conclusions e and/ relate to the stability of compacted sands: e suggests that moist sands tend to form deposits of low density;/suggests that coarse, well-graded sands will be compacted to a higher density and consequently form a more stable structure than uniform fine sands. The conclusions which can be drawn from this investigation are all consistent with actual field observations except for c and d for which there is no available information for comparison.

Vibrations In duced by Pile Driving 31. In an effort to develop a field test to evaluate earthquake liquefaction potential, Ishihara and Mitsui11 drove piles into a sand foundation and measured the induced pore pressure in the vicinity of the pile. According to Ishihara, this approach is superior to blasting tests since vibrations caused by blasts are very short in duration compared to the duration of an earthquake, and blasting cannot be performed in congested urban areas. 32. The site selected for the investigation was near Tokyo. The foundation material was a hydraulically placed, fine uniform sand about 8 m in thickness. The uniformity coefficient of the material was between 2 and 5 and the Dio size between 0.05 and 0.20mm. The standard penetration blow count was approximately N = 5 . 33. To measure the induced pore pressure and acceleration, piezometers and accelerometers were installed 4 or 8 m deep in the foundation in a circular configuration about the point at which the pile would be driven. The pile was driven at the preselected spot by a vibratory generator capable of producing a variable oscillating force up to 35 tons at a frequency of 17 Hz. The oscillating driving force was adjusted to produce a penetration rate of 1-2 m per minute. The pile was a closed-end steel pipe with an outside diameter of 40 cm. Measurements during the driving of the pile indicated that during some tests, the pore pressure became temporarily equal to the overburden pressure, but quickly dissipated. Acceleration was measured only in the vertical direction. Measurements showed that there was a unique relationship between sand density and acceleration level which produced a given pore pressure near the pile. The test results show that even though 100 percent pore pressure response occurs only for short periods, the excess pore pressure at substantial distances from the pile was slow to dissipate and resulted

13 in a significant reduction in strength of the foundation material. Such a strength reduction, even if not sufficient to cause liquefaction, could result in large settlements or even bearing capacity failures in adjacent structures. Casagrande1 mentions that pile driving frequently does result in settlement of neighboring structures resting on loose sands and suggests that pile driving may be used to densify and stabilize sand foundations.

FLOW FAILURES INDUCED BY EARTHQUAKES, BLASTS, AND WAVE ACTION

34. liquefaction induced by shock waves and the resulting random vibrations is potentially a more serious threat to civil engineering structures than that induced by other disturbing causes. Shock waves from earthquakes and blast loads or explosive charges are included in this general category. Earthquakes present probably the most serious liquefaction threat because the tremendous amount of energy released during an earthquake can disturb very large areas almost simultaneously. Consequently, structures built on or with cohesionless soils in seismically active areas must be designed considering the probability of earthquake shaking.

Examples of Fluidization Flow Induced by Earthquakes

35. Fluidization, i.e., flow of an essentially dry, cohesionless material with air as the pore fluid, can also be induced by shock waves. Terzaghi19 states that vibrations resulting from earthquakes are often sufficient to break the connection between soil particles in soils with slightly cemented grain aggregates such as loess. For example, on 16 December 1920, a catastrophic earthquake occurred in the Chinese province of Kansu in the heart of large loess deposits. A description of a landslide due to this earthquake was given by Close and McCormick in the National Geographic Magazine.20 According to Close and McCormick, “In each case the earth which came down bore the appearance of having shaken loose clod from clod and grain from grain, and then cascaded like water, forming vortices, swirls, and all the convolutions into which a torrent might shape itself.” Fluidization of this material is evident from the description. In this catastrophe the area of greatest destruction was about 100 miles by 300 miles in extent and contained ten large cities as well as numerous small villages. Nearly 200,000 lives were lost during this event and hundreds of towns and cities totally destroyed. 36. Casagrande in Reference 8 cites three examples of fluidization probably induced by seismic motions. The first example given was a prehistoric 400-million-cu-yd slide of marble breccia which occurred at Blackhawk Canyon in the San Bernardino Mountains of Southern California. The slide mass was 5 miles long, 2 miles wide, and between 30 and 50 ft thick. The second fluidization example cited was a 40-million-cu-yd slide across the Sherman Glacier in Alaska triggered by the 1964 earthquake. The slide formed a blanket of broken rock 2 miles wide and 10-20 ft thick. The third example was a slide in volcanic ash in a high mountain desert in Chile. The slide was roughly circular in shape with an effective diameter of 1500 ft. Pore air pressure during this slide dropped quickly enough that the slide was arrested before it had moved very far.

Liquefaction During the Niigata Earthquake of 1964

37. On 16 June 1964 a violent earthquake shook Niigata, Japan, inflicting severe damage to the

14 city. The recorded magnitude of this earthquake was 7.3 on the Richter scale, and the epicenter was about 35 miles north of the city.21 Niigata is located on the west coast of Japan where the Shinano River enters the Sea of Japan. The city is underlain by about 100 ft of fine subangular alluvial sand with a uniformity coefficient between 2 and 10 and a Dio size between 0.04 and 0.2 mm. Damage due to the earthquake resulted primarily from liquefaction of the loose sand deposits in low-lying areas. Buildings not imbedded deeply in firm material sank in the resulting quicksand or tilted toward the direction of the center of gravity. One apartment building is reported to have tilted 80 deg. Underground structures such as septic tanks, storage tanks, sewage conduits, and manholes floated upward to extend as much as 10 ft above the ground surface. In low-lying areas, sand flows and mud volcanos began ejecting water and sand 2-3 min after the shaking stopped and continued for as much as 20 min. Sand deposits 20-30 cm thick covered the entire city as if it had been hit by a sand flood. Five simply supported girders of the Showa Bridge across the Shinano River fell when pier foundation piles deflected due to the loss of lateral support. 38. The greatest devastation by the quake occurred in the low-lying areas near the river. An extensive soil survey after the quake revealed that the standard penetration blow count was between N = 6 and N = 12 in the upper 30 ft in the areas suffering the greatest damage. The penetration resistance was about the same over the entire area in the upper 15 ft; but in areas where the resistance was measurably higher in the lower 15 ft, notably lighter damage occurred. The recorded peak acceleration during the quake was 0.16 g. The water table was about 5 ft below the ground surface in areas of heaviest damage and 9 ft or more in areas of lighter damage. 39. It was reported by Seed and Idriss21 that Niigata had been shaken by an earthquake of the same intensity 130 yr before the event of 1964. This indicates that areas which have been shaken by an earthquake are not immune to further damage unless remedial measures are taken.

Valdez Slide, 1964 Alaska Earthquake 40. During the 1964 Alaska earthquake an extensive landslide believed to be due to liquefaction occurred at Valdez.22 The city is located on a deltaic deposit of silt, fine sand, and gravel, with the silt and fine sand occurring in layers within the coarser sand and gravel. The sand particles were generally subangular in shape and the uniformity coefficient greater than 10. The standard penetration resistance of the Valdez material varied between N = 7 and N = 25 . The magnitude of the earthquake was 8.3 on the Richter scale with the epicenter about 40 miles away. The landslide resulting from the quake involved approximately 98 million cu yd of material and extended about 500 ft inland from the coastline, destroying the harbor facilities and near-shore installations. This slide is reported to have occurred so rapidly and with such violence that eye witnesses were unsure of what actually happened directly beneath their feet. The dock structure at Valdez supporting several large buildings was reported to have broken off and disappeared into the sea. The slide was responsible for the loss of 30 lives. About 40 percent of the buildings of Valdez were seriously damaged. Lateral movement toward the sea occurred as much as 3600 ft inland, causing settlement, Assuring, and ground breakage. In addition to buildings, the water and sewage systems were severely damaged by ground movement. In many instances, fissures occurring in the foundation of buildings resulted in basements being pumped full of sand and water. Open cracks spurting sand and water along the streets substantial distances inland indicated that liquefaction was not limited to the coast. Since the year 1899 submarine slide damage was reported to have occurred at this location on at least five other occasions, probably caused by seismic activity.

15 Slides in the San Fernando Dams 41. On 9 February 1971 an earthquake shook southern California resulting in a major landslide in the upstream slope of the lower San Fernando Dam.23 This earthen dam is about 140 ft high and provides 20,000 acre-ft of reservoir capacity. The water level in the reservoir was about 35 ft below the crest of the dam at the time of the event, and the resulting slides on both the upstream and downstream slopes left about 5 ft of freeboard. Because of the possibility that aftershocks might cause further sliding and possible failure of the dam, 80,000 people living downstream were evacuated until the reservoir could be drawn down to a safe level. 42. Upstream from the lower dam is the upper San Fernando Dam which is about 80 ft high and provides a reservoir of 1850 acre-ft. During the earthquake, the crest of this dam moved 5 ft downstream and settled 3 ft, but no water was lost from the reservoir. Had water from the upper dam been released, the lower dam would have been overtopped and consequently would have failed. 43. This near catastrophe prompted an immediate investigation of the safety of earth dams with respect to earthquake loading. The event of 9 February 1971 has been assigned a Richter magnitude of 6.6 with the epicenter located about 8-1/2 miles northeast of the dams. The maximum acceleration in the vicinity of the dams was of the order of 0.55-0.60 g. 44. Both the San Fernando dams were old structures, being constructed in the period between 1915 and 1925 by the hydraulic-fill method. The dams were constructed directly on alluvial soil with no stripping prior to the placement of the embankment fill. Field investigations after the quake revealed that penetration resistance (N values) was very low in the silt and clay core of the hydraulic fill, but was somewhat higher in the outer sands and gravels of the shell. The underlying alluvium was very heterogeneous, exhibiting an erratic blow count in most of the drill holes. Laboratory investigations showed that the hydraulic-fill material was finer, more uniform, and less dense than the alluvium. The hydraulic-fill lower dam sands were coarser than the upper dam sands, D 50 being between 0.05 and 0.8 mm for the upper dam sands and between 0.05 and 1.0 mm for the lower dam sands. The average relative density for all the hydraulic-fill material investigated was about 54 percent. The uniformity coefficients ranged from 4 to 6 for the upper dam materials and from 7 to 10 for the lower dam materials. 45. As a result of the 1971 earthquake, the upstream embankment and about 30 ft of the crest of the lower San Fernando Dam moved about 70 ft into the reservoir, the movement being due to liquefaction of the hydraulic fill near the base of the embankment. Liquefaction was evidenced by observed increases in pore pressure in the embankment, the occurrence of large horizontal displacements, and the formation of sand boils in the slide debris. 46. It was concluded in Reference 23 that a major catastrophe very nearly occurred. If any of a number of conditions had been less favorable, the lower dam could have failed resulting in the sudden release of 10,000 acre-ft of water over a heavily populated urban residential area. Laboratory tests and analytical techniques indicate that the instability of the lower dam was due to insufficient density in the water-saturated shell materials. 47. Other dams in the area were subjected to shaking of lesser intensity (0.2 g maximum acceleration) with no detrimental structural damage. Another hydraulic-fill dam with no water in the reservoir (unsaturated) was subjected to shaking of the same intensity as the San Fernando dams with no detrimental damage. These observations indicate that hydraulic-fill dams are not inherently unstable but become so if the embankments are saturated and experience shaking of sufficient intensity and duration.

16 Ekofisk Tank in the North Sea 48. In June 1973 the 490,000-ton Ekofisk oil storage tank was installed in the North Sea.24 The structure is a reinforced concrete cylinder roughly circular in shape, about 93 m in diameter, and 90 m tall on a mat foundation in 70 m of water. The tank imposes a vertical stress of about 2.72 kg/cm2 on the foundation which is a 26-m-thick stratum of medium-to-dense, fine, uniform sand. The sand particles were mostly an gular in shape with a D 50 of 0.11 mm and a uniformity coefficient of 2. There was concern that the sand under the tank might liquefy due to stresses induced in the foundation by the action of storm waves against the walls of the tank. 49. The loading that the foundation of the tank would be subjected to was thought to be similar to earthquake loading with the following notable differences: a. Storm waves have periods considerably longer than earthquake loadings. b. The duration of ocean storms is much longer than that of earthquakes. c. It is probable that a structure in the ocean would be subjected to a number of minor storms with intermittent periods of calm before the occurrence of the design storm.

The structure was designed for a 100-yr storm producing a wave 23.8 m from crest to trough. The design wave would exert a lateral force of 35 million lb on the tank. 50. Cyclic load triaxial tests of the type described by Seed and Lee25 were performed on the foundation material and the procedure described by Seed and Idriss5 used to evaluate liquefaction potential. Because loading from wave action was of a considerably longer period than for earthquakes, the rate of loading was reduced from 1 Hz, typical for earthquake tests, to 1/12 Hz. The laboratory test procedures were modified to account for the fact that during long ocean storms and during periods between storms, pore pressure dissipation would occur in the permeable sand. This was done by subjecting the test specimen to several cycles of undrained cyclic loading until some desired pore pressure response was reached, then stopping the loading and reconsolidating the specimens. After several cycles of such loading and reconsolidation, the test specimen was cyclically loaded to 100 percent pore pressure response (zero effective stress). It was found that the stability of the specimen increased substantially after several cycles of loading and reconsolidation. The increase in relative density due to reconsolidation was from 1 to 4 percent and would not explain the drastic increase in stability. Finn et al.26 and Bjerrum27 mention that in addition to an increase in density, there is a change in the packing structure of the sand resulting from cyclic loading and reconsolidation which may explain the disproportionate increase in stability. 51. This study revealed that the supporting sands of this tank could withstand a 100-yr storm without liquefaction. During the 10-month period following the installation of the tank, storms approaching the design storm occurred on at least three occasions with both the storage tank and foundation performing satisfactorily.

Failure of the Swir III Dam 52. During the spring of 1935 construction of the Swir III Dam in the Union of Soviet Socialist Republics was completed and the reservoir filled for the first time. The Swir III Dam was a concrete dam with a sand embankment next to the concrete section. The embankment, containing a till core, was formed by dumping moist sand on both sides of the core. 53. With the reservoir filled, blasting operations were being performed about 600 ft upstream of the dam when the sand embankment started to liquefy and flow away. According to Terzaghi2 the failure

17 started at the contact surface between the concrete section of the dam and the embankment. In less than 1 min, liquefaction had spread more than 1000 ft to the outer end of the embankment. The failure reduced the embankment slope from IV on 2H before failure to IV on 10H after failure. The embankment sand had been deposited in a moist state and was probably in a bulked and very loose condition. The embankment became saturated as the reservoir filled, and the blasting was evidently sufficient to trigger liquefaction.

Sand Boils at Operation Snowball 54. On 17 July 1964, a 500-ton hemispherical TNT charge resting on the ground was detonated over a glacially deposited silt with underlying beds of clay, sand, and gravel. The deposit consisted of a friable silty clay to a depth of 12 ft, a brown silty clay from a depth of 12-27 ft, an interbedded stratum of coarse sand and gravel (Dio = 0.15 mm and Cu — 3-4) from 27-32 ft, and a gray silty clay from a depth of 32-67 ft. The water table was located 23 ft below ground zero at the time of detonation. 55. Postshot surveys and reconnaissance reports28 showed that the detonation produced a crater immediately after the shot measuring 14 ft deep, with a radius of 140 ft. High pore pressures due to compressive stresses were indicated by water flowing both inside and outside the crater. Sand boils were also observed to occur in and around the crater. The outflow of the water, sand, and silt caused subsidence of the area and circumferential tension cracks as far as 155 m from ground zero. These descriptions and the occurrence of sand boils suggest that the underlying layers of sand and silt liquefied and flowed to the surface.

Hiroshima and Nagasaki 56. Obviously there are limited data available for the evaluation of liquefaction susceptibility of foundation materials subjected to nuclear explosions. However, damage reports29 for the cities of Hiroshima and Nagasaki were examined for evidence of damage from liquefaction, but no firm conclusions could be drawn from the examination. The damage survey party was primarily interested in fire and blast damage and did not describe foundation conditions simply because it did not seem pertinent at the time. Consequently it could not be determined if liquefaction occurred in these cities as a result of the nuclear blasts.

18 PART IV: DISCUSSION OF CASE HISTORIES

SOIL TYPE AND TRIGGERING MECHANISMS CAUSING LIQUEFACTION

57. An examination of the case histories reveals that with the exception of the Aberfan and Kansu flow slides, the materials liquefying were fine, uniform sands. Typically, Dio sizes ranged from 0.05 to 1.0 mm and the coefficient of uniformity ranged from less than 2 to 10. These observations are in agreement with, but broader than, the criteria of D’Appolonia3 and others.4'5'6 D’Appolonia’s criteria state that liquefaction-prone soils are cohesionless with less than 10 percent fines, fine grained (D60 between 0.2 and 1.0 mm), and uniform (Cu between 2 and 5). In all cases the soils were loosely deposited, and in many cases they were water-saturated. However, water saturation is not absolutely necessary as soils with air as a pore fluid were reported to flow. 58. The case history review also shows that a variety of triggering mechanisms can cause liquefaction. These mechanisms cause collapse of the grain structure, resulting in high pore pressure and subsequent loss in strength and flow. They range from monotonically changing stresses caused by changes in externally applied loads to vibrations and shock waves caused by reciprocating machinery, earthquakes, or blast loadings. However, the most common triggering mechanisms are (a) monotonically changing stresses, such as those causing the Mississippi River and Fort Peck flow slides; and (b) earthquake-induced flow slides, such as those at Niigata and the San Fernando dams. Liquefaction flow slides induced by blasting or cyclic vibrations are less commonly observed and reported. Earthquake-induced liquefaction failures are generally more catastrophic than those induced by other mechanisms due to the tremendous energy released during an earthquake causing almost instant liquefaction over a large area. Conversely, flow slides induced by monotonically changing stresses generally are more limited in area and occur less rapidly.

EFFECTS OF DENSITY ON LIQUEFACTION POTENTIAL

59. The most important property of sands in the evaluation of liquefaction susceptibility is density. Sands may be stabilized against liquefaction by increasing their density. The reason for this is that, in order for liquefaction flow to occur, the static shear strength of a sand must be exceeded and there must be potential to develop high pore pressures. Ambraseys and Sarma30 derived the analytic expression

«h/ 1 ♦«♦V3 -w . ( 1 ) rmax y 0 Z 7 ------1 + ^F"? 3 (A------ï \ ------sin 0'------sin where rmax = shear stress above which the sand becomes unstable p = mass density of the sand = y/g g = acceleration due to gravity h = depth below the ground surface 7 ' = submerged unit weight of the sand 7 = moist unit weight of the sand

19 K = anisotropic consolidation coefficient of the deposit A = Skempton’s A pore pressure parameter in plane strain at failure 0 ' = angle of internal friction of the sand

According to them, a single transient load inducing shear stresses greater than those given by Equation 1 will cause failure and in some cases spontaneous liquefaction. It may Seen by inspecting Equation 1 that r max will decrease as the density and angle of internal friction decrease. Density is the most critical variable in Equation 1 since in addition to being a factor itself it also affects the A parameter and friction angle. 60. Bjerrum et al.31 showed that the value of the A parameter increases sharply as porosity increases (and density decreases), as illustrated in Figure 4. The angle of internal friction also decreases rapidly as density decreases (Figure 5). The large value of the A parameter in loose sand indicates that a tremendous increase in pore pressure

34 38 42 46 50 will occur for a comparatively small change in INITIAL POROSITY, \ void ratio. With the angle of internal friction Figure 4. Pore pressure parameter A at failure versus initial porosity (courtesy of Dunod, Paris31) diminished and the potential to develop large pore pressures, sands of low density will fail

34 38 42 46 S t POROSITY AT FAILURE, % Figure 5. Angle of internaH tvfction of a fine sand versus initial porosity and porosity at failure (courtesy of Dunod, Paris31)

20 and flow at very small strains. Dense sands, on the other hand, are stable against liquefaction as they tend to expand and become stronger when subjected to shear strains in an undrained state due to the development of less positive pore pressures.

LIQUEFACTION POTENTIAL

61. Sands which are initially loose and liquefy when subjected to earthquake shaking do not become significantly more stable against reliquefaction. Ambraseys and Sarma30 suggest that deposits which are susceptible to liquefaction, e.g., 10-20 m thick, have been accumulating in seismically active areas for thousands of years. It is highly improbable that such deposits have never been disturbed by shocks at some stage during their deposition. It is more probable that these deposits have been subjected to a number of earthquakes and have liquefied a number of times and are still susceptible to liquefaction. Historical records for Niigata show three occasions where liquefaction was reported in or near the city.5 This would indicate that earthquakes are incapable of densifying originally loose deposits into a stable mass. This is pointed out by the fact that strong earthquake aftershocks have caused granular deposits to reliquefy after initial liquefaction during the earthquake. It seems then that the existence of deposits in seismically active areas which have been stabilized by earthquake shocks would be the exception rather than the rule. Apparently, after liquefaction occurs, the material subsides again to an unstable structure still quite susceptible to liquefaction. 62. Conversely, soils which are subjected to vibrations and experience an increase in pore pressure but not complete liquefaction become stable against liquefaction far beyond what the increase in density after drainage occurs would suggest. Bjerrum31 suggests that the disproportionate increase in stability stems from the fact that the structure of a sand is altered by vibrations. 63. It appears then that the stability of saturated sands against liquefaction can be increased by increasing the density of that sand. Very generally, the density of a sand must be increased relative to the magnitude of the disturbance which must be resisted by the sand or the duration of that disturbance. Other measures which may be taken to lessen the danger of liquefaction in liquefaction-susceptible soils are: a. To remove the liquefiable soil, replacing it with more stable material. b. To stabilize the material in question with grout. c. To place structures which must be founded on liquefiable material on piles extending through the unstable layers and bearing in stable material or hardpan.

21 PART V: CONCLUSIONS

64. Liquefaction failures are dependent upon (a) a collapsible soil structure, (b) a saturated and undrained condition, and (c) a triggering mechanism. a. Soils which possess collapsible structures are typically fine-grained, loose, uniform sands with less than 10 percent fines, a Dio between 0.05 and 1.0 mm, and a uniformity coefficient between 2 and 10. A loose deposit, i.e., low density, is required for collapse as dense materials increase rather than decrease in volume under shear strains. b. A saturated and undrained condition provides a situation whereby collapse of the soil structure generates high pore pressures. Water saturation is not absolutely necessary as soils with air as the pore fluid have been reported to flow. c. A variety of triggering mechanisms ranging from monotonically changing stresses due to changes in externally applied loads to vibrations and shock waves caused by reciprocating machinery, earthquakes, or blast loads can cause liquefaction. However, monotonically increasing stresses and earthquakes are the most common triggering mechanisms. 65. Density is the most important property in determining the liquefaction susceptibility of saturated-undrained soils. 66. Sands which liquefy when subjected to earthquake shaking do not become significantly more stable against reliquefaction. Conversely, vibrations not intense enough to induce liquefaction, but which increase pore pressure appear to alter the soil structure, making a deposit less susceptible to liquefaction than indicated by density increases.

22 REFERENCES

1. Casagrande, A., “Characteristics of Cohesionless Soils Affecting the Stability of Slopes and Earth Fills,” Boston Society o f Civil Engineers, Journal, Vol 23, No. 1, Jan 1936, pp 13-32. 2. Terzaghi, K., “Varieties of Submarine Slope Failures,” Proceedings, Eighth Texas Conference on Soil Mechanics and Foundation Engineering, Special Publication No. 29, Sep 1956, Austin, Tex. 3. D’Appolonia, E., “Dynamic Loadings,” paper presented at the American Society of Civil Engineers Specialty Conference on Placement and Improvement of Soil to Support Structures, Aug 1968, Massachusetts Institute of Technology, Cambridge, Mass. 4. Whitman, R. V., “Resistance of Soil to Liquefaction and Settlement,” 1970, Massachusetts Institute of Technology, Cambridge, Mass. 5. Seed, H. B. and Idriss, I. M., “A Simplified Procedure for Evaluating Soil Liquefaction Potential,” Report No. EERC 70-9, Nov 1970, University of California, Berkeley, Calif. 6. Terzaghi, K. and Peck, R. B., Soil Mechanics in Engineering Practice, Wiley, New York, 1948. 7. Torrey, V. H., “Verification of Empirical Method for Determining Riverbank Stability, 1968 and 1969 Data,” Potamology Investigations Report 12-21, Oct 1972, U. S. Army Engineer Waterways Experiment Station, CE, Vicksburg, Miss. 8. Green, P. A. and Ferguson, P. A. S., “On Liquefaction Phenomena by Professor A. Casagrande: Report on Lecture,” Geotechnique, Vol 21, No. 3, Sep 1971, pp 197-202. 9. D’Appolonia, E., “Loose Sands—Their Compaction by Vibroflotation,” Special Technical Publication No. 156, 1953, American Society for Testing and Materials, Philadelphia, Pa. 10. Koppejan, A. W., Van Wamelen, B. M., and Weinberg, L. J. H., “Coastal Flow Slides in the Dutch Province of Zeeland,” Proceedings, 2d International Conference on Soil Mechanics and Foundation Engineering, Vol V, 1943, pp 89-96. 11. Ishihara, K. and Mitsui, S., “Field Measurements of Dynamic Pore Pressure During Pile Driving,” Proceedings, International Conference on Microzonation for Safer Construction, Research and Application, University of Washington, Seattle, Wash., Oct-Nov 1972. 12. U. S. Army Engineer Waterways Experiment Station, CE, “A Review of Soil Studies,” Potamology Report 12-5, Jun 1956, Vicksburg, Miss. 13. Casagrande, A., “The Role of the ‘Calculated Risk’ in Earthwork and Foundation Engineering,” Journal, Soil Mechanics and Foundations Division, American Society o f Civil Engineers, Vol 91, No. SM4, Jul 1965, pp 1-40. 14. Middlebrooks, T. A., “Fort Peck Slide,” Transactions, American Society o f Civil Engineers, Vol 107, 1942, pp 723-742. 15. Hazen, A., “Hydraulic-Fill Dams,” Transactions, American Society o f Civil Engineers, Vol 83, 1920, p 1713+. 16. Hutchinson, J. N., “Discussion of Flow Slides,” Proceedings, Geotechnical Conference on Shear Strength Properties of Natural Soils and Rocks, Oslo, Vol 2, 1967. 17. Geuze, E. C. W. A., “Critical Density of Some Dutch Sands,” Proceedings, 2d International Conference on Soil Mechanics and Foundation Engineering, Vol III, 1948, pp 125-130. 18. Prakash, S. and Gupta, M. K., “Compaction of Sand Under Vertical and Horizontal Vibrations,” Earthquake Engineering Studies, Oct 1966, University of Roorkee, Roorkee, India. 19. Terzaghi, K., “Mechanism of Landslides,” Engineering Geology, Geological Society o f America (Berkey Memorial Volume), Nov 1950, pp 83-123. 20. Close, U. and McCormick, E., “Where the Mountains Walked,” The National Geographic Magazine, Vol XLI, No. 5, May 1922.

23 21. Seed, H. B. and Idriss, I. M., “Analysis of Soil Liquefaction: Niigata Earthquake,” Journal Soil Mechanics and Foundations Division, American Society o f Civil Engineers, Vol 93, No. SM3, May 1967, pp 83-108. 22. Seed, H. B., “Landslides During Earthquakes Due to Soil Liquefaction,” Journal Soil Mechanics and Foundations Division, American Society of Civil Engineers, Vol 94, SM5, Sep 1968, pp 1055- 1122. 23. Seed, H. B. et al., “Analysis of Slides in the San Fernando Dams During the Earthquake of 9 Feb 1971,” Report No. EERC 73-2, Jun 1973, College of Engineering, University of California, Berkeley, Calif. 24. Lee, K. L. and Focht, J. A., Jr., “Liquefaction Potential at Ekofisk Tank in North Sea,” Journal Geotechnical Engineering Division, American Society o f Civil Engineers, Vol 101, No. GT1, Jan 1975, pp 1-18. 25. Seed, H. B. and Lee, K. L., “Studies of the Liquefaction of Sands Under Cyclic Loading Conditions,” Report No. TE-65-5, Dec 1965, College of Engineering, University of California, Berkeley, Calif. 26. Finn, W. D. L., Bransby, P. L., and Pickering, D. J., “Effect of Strain History on Liquefaction of Sand,” Journal, Soil Mechanics and Foundations Division, American Society o f Civil Engineers, Vol 96, No. SM6, Nov 1970, pp 1917-1933. 27. Bjerrum, L., “Geotechnical Engineering Problems Involved in Foundations of Structures in the North Sea,” Geotechnique, Vol 23, No. 3, 1973, pp 319-358. 28. Rooke, A. D., Jr., et al., “Operational Snowball, Project 3.1, Crater Measurements and Earth Media Determinations, the Apparent and True Craters,” Miscellaneous Paper No. 1-987, Apr 1968, U. S. Army Engineer Waterways Experiment Station, CE, Vicksburg, Miss. 29. URS Research Company, “A Study to Assess the Magnitude of OCD Foundation Problems Pertaining to Nuclear Weapons,” URS-693-7, Final Report, Jul 1970, Office of Civil Defense, Washington, D. C. 30. Ambraseys, N. and Sarma, S., “Liquefaction of Soils Induced by Earthquakes,” Bulletin, Seismological Society of America, Vol 59, No. 2, Apr 1969, pp 651-664. 31. Bjerrum, L., Kringstad, S., and Kummeneje, O., “The Shear Strength of Fine Sand,” Proceedings, 5th International Conference on Soil Mechanics and Foundation Engineering, Vol 1, 1961.

24 In accordance with ER 70-2-3, paragraph 6c(l)(b), dated 15 February 1973> a face 1mlle catalog card In Library of Congress format Is reproduced below,

Gilbert, Paul A Case histories of liquefaction failures, by Paul A. Gilbert. Vicksburg, U. S. Army Engineer Waterways Experiment Station, 1976. 24 p. illus. 27 cm. (U. S. Waterways Experiment Station. Miscellaneous paper S-76-4) Prepared for Office, Chief of Engineers, ü. S. Army, Washington, D. C., under Project No. 4A161102B52E, Task 04. References: p. 23-24.

1. Liquefaction (Soils). 2. Noncohesive soils. I. U. S. Army. Corps of Engineers. (Series: U. S. Waterways Experiment Station, Vicksburg, Miss. Miscel laneous paper S-76-4) TA7.W34m no.S-76-4