PART 1. LIQUEFACTION AND GROUND FAILUREINTRODUCTIONGround failure caused by liquefaction is a major cause of earthquake damage and casualties. For example, most of the damage to highways and bridges generated by the April 22, 1991 Limon Province Costa Rica earthquake was caused by liquefaction. In that instance, ground failures generated by liquefaction caused floodplains to press into river channels, compressing bridge structures, and caused soils to weaken under highway and railway grades, causing embankment fills to settle and spread laterally. Similarly spectacular damage as a consequence of liquefaction also occurred during 1906 San Francisco, the 1964 Alaska, the 1964 Niigata earthquakes. Because of the potential for damage considerable study of the liquefaction phenomenon has occurred in the past few years to provide criteria for evaluating liquefaction and ground failure hazard. Such evaluations are a major element in earthquake hazard assessment and mitigation studies. |
photo: courtesy EERI |
Liquefaction is a process by which clay-free soil deposits, primarily sands and silts, temporarily lose strength and behave as a viscous liquid rather than as a solid. The actions in the soil which produce liquefaction are as follows: Seismic waves, primarily shear waves, passing through saturated granular layers, distort the granular structure, and cause loosely packed groups of particles to collapse. Disruptions to the particulate structure generated by these collapses cause transfer of load from grain-to-grain contacts in the soil to the interstitial pore water. This transfer of load increases pressure, in the pore water, causing drainage to occur. If drainage is restricted, a transient build up of pore-water pressure will occur. If the pore-water pressure rises to a level approaching the overburden pressure grain-to-grain contact stresses approach zero and the granular layer temporarily behaves as a viscous liquid rather than as a solid and liquefaction has occurred. In the liquefied condition, soil deformation: may occur with little shear resistance. Deformations large enough to cause damage to constructed works (usually more than 0.1 m) are called ground failure.
The ease with which a soil can be liquefied depends on the looseness of the soil, the packing arrangement of soil grains, the amount of cementing between particles, and the amount of drainage restriction. The amount of soil deformation following liquefaction depends on the looseness of the material, the thickness and areal extent of the liquefied layer, the ground slope, and the distribution of loads applied by buildings and other structures on the ground surface.
Liquefaction does not occur at random, but is restricted to certain geologic and hydrologic environments, primarily recently deposited sands and silts in areas with high ground water levels. Generally, the younger and looser the sediment, and the higher the water table, the more susceptible the soil is to liquefaction. Sediments most susceptible to liquefaction include Holocene (less than 10,000 year-old) delta, river channel, flood plain, and aeolian deposits and poorly compacted fills. Liquefaction has been most abundant in areas where ground water lies within 10m of the ground surface; few instances of liquefaction have occurred in areas with ground water deeper than 20 m.
GROUND FAILURE TYPESFour primary types of ground failure are caused by liquefaction: lateral spread, ground oscillation, flow failure, and loss of bearing strength. In addition, liquefaction may enhance ground settlement and lead to eruption of sand boils (fountains of water and sediment emanating from the pressurized, liquefied zone). Lateral SpreadsLateral spreads involve lateral displacement of large, surficial blocks of soil as a result of liquefaction of a subsurface layer. Displacement occurs in response to combination of gravitational forces and inertial forces generated by an earthquake. Lateral spreads generally develop on gentle slopes (most commonly less than 3 degrees) and move toward a free face such as an incised river channel. Horizontal displacements commonly range up to several meters, but where slopes are particularly favorable and ground shaking durations are long, displacements may range up to several tens of meters. The displaced ground usually breaks up internally, causing fissures, scarps, horsts, and grabens to form on the failure surface. Lateral spreads commonly disrupt foundations of buildings built on or across the failure, sever pipelines and other utilities in the failure mass, and compress or buckle engineering structures such as bridges founded on the toe of the failure. Damage caused by lateral spreads, though seldom catastrophic, is severely disruptive and often pervasive. For example, during the 1964 Alaska earthquake, more than 200 bridges were damaged or destroyed by spreading of floodplain deposits toward river channels. The spreading compressed the superstructures, buckled decks, thrust stringers over abutments, and shifted and tilled abutments and piers. Similar damage occurred during the 1991 Costa Rica earthquake and during many previous large earthquakes. |
![]() photo: courtesy of EERI |
Lateral spreads are particularly destructive to pipelines. For example, every major pipeline break in the city of San Francisco during the 1906 earthquake occurred in areas of ground failure. These pipeline breaks severely hampered efforts to fight the fire that ignited during the earthquake; that fire caused about 85 percent of the total damage to San Francisco. Thus, rather inconspicuous ground-failure displacements of less than 2 m were in large part responsible for the devastation that occurred in San Francisco (Youd and Hoose, 1978).
Where the ground is flat or the slope is too gentle to allow lateral displacement, liquefaction at depth may decouple overlying soil layers from the underlying ground, allowing the upper soil to oscillate back and forth and up and down in the form of ground waves. These oscillations are usually accompanied by opening and closing fissures and fracture of rigid structures such as pavements and pipelines.
photo: courtesy of EERI |
Flow FailuresFlow failures are the most catastrophic ground failure caused by liquefaction. These failures commonly displace large masses of soil tens of meters and in a few instances, large masses of soil have traveled tens of kilometers down long slopes at velocities ranging up to tens of kilometers per hour. Flows may be comprised of completely liquefied soil or blocks of intact material riding on a layer of liquefied soil. Flows usually develop in loose saturated sands or silts on slopes greater than 3 degrees. Loss of Bearing StrengthWhen the soil supporting a building or other structure liquefies and loses strength, large deformations can occur within the soil which may allow the structure to settle and tip. Conversely, buried tanks and piles may rise buoyantly through the liquefied soil. For example, many buildings settled and tipped during the 1964 Niigata, Japan earthquake. The most spectacular bearing failures during that event were in the Kwangishicho apartment complex where several four-story buildings tipped as much as 60 degrees. Apparently, liquefaction first developed in a sand layer several meters below ground surface and then propagated upward through overlying sand layers. The rising wave of liquefaction weakened the soil supporting the buildings and allowed the structures to slowly settle and tip. |
At 15:57 pm local time, April 22, 1991, a large (M = 7.5) earthquake struck Limon Province, Costa Rica killing 53 people, injuring 198, and causing widespread damage to constructed works (EERI, 1991). In particular, the transportation system in Limon Province was devastated. Both rail and highway traffic was immediately obstructed due to widespread disruption of pavements and roadway grades and the collapse of several bridges. Most of the damage to the highway and railway systems was caused by liquefaction and consequent ground failure, primarily lateral spreading.
The US National Science Foundation awarded a grant to myself and Professor Kyle Rollins at Brigham Young University to conduct a post-earthquake investigation of liquefaction and its consequences following the 22 April 1991 earthquake. The purpose of that investigation was to document ground displacements and related damage caused by liquefaction. We conducted field studies in Limon Province between May 27 and June 6, 1991 (5 to 7 weeks after the earthquake). Our team surveyed three railroad and two highway bridge sites and estimated displacements at two additional sites. We used an electronic total station to measure distances and angles. We also used metric tapes to measure distances and displacements. From this data, we calculated ground and structural displacements and compiled topographic maps for the sites.
Limon Province lies in the eastern part of Costa Rica in a geologic province dominated by a broad plain that gently slopes from the Cordillera de Talamanaca to the Caribbean Sea. That plain is dissected by several large and many small river valleys that generally broaden as they approach the coast. Most liquefaction occurred in alluvial and fluvial deposits that underlie river floodplains or in deltaic, lagoonal or estuarine deposits that underlie lowlands along the coast.
Within the epicentral region, liquefaction was rather pervasive in these lowland areas. For example, the EERI reconnaissance team (EERI, 1991) estimated that about thirty percent of the highway pavement in these areas was disrupted by fissures, scarps and ground settlements caused by liquefaction. Similarly, several segments of railway grade were misaligned by the ground movements. The greatest damage occurred river crossings, however, where bridge decks were thrust over abutments, piers shifted riverward, and fills settled as much as 2m.
Southeast of the community of Matina, the railroad crosses the floodplains and channel of the Rio Matina on a 400-m long bridge. On each side of the river, the bridge over the floodplain is composed of five simply-supported steel plate-girder sections resting on concrete piers. Over the main channel, the bridge consists of three truss sections resting on concrete caissons.
During the April 22 earthquake, liquefaction of sediments beneath the floodplains on both sides of the river caused the ground to spread laterally toward the incised river channel. These lateral displacements carried bridge piers and caissons toward the river causing several plate girder sections to drop off their supports and fall to the floodplain. The truss sections were also pushed off their seatings on the caissons at the river banks, but the trusses did not tip or fall.
By the time of our visit, the railroad had been temporarily repaired by placing the plate girder sections on cribs of timber shoring and by re-leveling and realigning the truss and girder sections. In so doing, the rails had been restored approximately to their pre-earthquake elevations and alignment. This temporary repair allowed trains to cross the bridge at reduced speed and also provided a reference (distance between a seating plate on the plate girder and its former anchorage on the shifted pier) for us to measure displacement of the shifted piers.
We conducted two surveys at the Matina bridge site. First, we used the realigned plate girders as a reference for measurement of horizontal and vertical pier displacements. Visual sighting of the rails across the bridge indicated that no major misalignment (more than a few centimeters) had occurred in realigning that structure, and hence our measurements are probably accurate within a few centimeters. Secondly, we conducted a topographic survey from which we compiled a site map showing positions of bridge foundations, elevation contours, and major fissures and sand boils visible at the time of our investigation. On the northwest side of the river, displacements increased riverward from 11 cm horizontal and 16 cm vertical at the abutment to 44 cm horizontal and 34 cm vertical at Pier M4, the pier nearest the river. On the southeast side of the river, horizontal displacements increased from 75 cm horizontal at the abutment to 120 cm at the Pier M12, the pier nearest the river. (A recent realignment of the bridge now indicates that the latter displacements should be about 30 cm greater for piers M12, M14, and the southern abutment than those determined from our initial investigation.)
At each bank of the river, the girder-truss connections were supported on two 2.8-m diameter steel-lined and concrete-filled caissons. Each of those caissons tilted 4 to 5 degrees with the tops moving 66 cm and 120 cm riverward on the northwest and southeast sides of the river, respectively. Those displacements sheared the bridge connectors from the caisson, but the truss sections remained upright on the caisson. The floodplain soils pushed past the caissons on both sides of the river, leaving gaps as wide as one meter on the river sides of these shafts. The caissons also rocked back and forth during the earthquake as evidenced by circular voids as wide as tens of centimeters the bases of the caissons. The floodplain soils also settled around the caissons by as much as 30 cm.
The railway bridge over the Rio Bananito near Bananito Sur is a 50-m long, single-truss structure supported on elliptically-shaped caissons 1.46 m by 2.16 m across the major axes. The caissons are constructed of a 12 mm cast-steel shell filled with concrete. During the April 22 earthquake, ground displacements caused by liquefaction and lateral spreading pushed the supporting caissons out from under the seating plates on both ends of the bridge. This loss of support allowed the truss to tip downstream or eastward by about 15 degrees.
We surveyed the site and measured caisson displacements beneath the bridge. At the northwest end of the bridge, the tops of the caissons shifted displaced 4.3 m and 5.7 m toward the river and the caissons were tilted 26 and 37 degrees, on northeast and southwest sides, respectively. The 0.9-m high capital on the north caisson pulled off during the earthquake and had fallen to the ground by the time of our visit. A concrete wall in the abutment had shifted 2.8 m toward the river and tilted slightly. These measurements indicate that lateral ground displacements beneath the north end of the bridge were between 2.0 m to 2.5 m.
Beneath the southeast end of the truss, the tops of the two supporting caissons were displaced 2.83 m and 1.90 m, respectively, with reference to seating plates on the truss. The greater displacement occurred on the northeast side, the direction in which the truss tilted. The caissons tilted toward the river 19 and 7 degrees, respectively. A retaining wall in the abutment shifted 1.43 m toward the river and tilted slightly. These measurements indicate that horizontal ground displacements beneath the southeast end of the bridge were approximately 1 m to 1.5 m.
RIO ESTRELLA RAILWAY BRIDGEThe bridge across the floodplain near Pandora was constructed of simply-supported plate girder spans resting on the steel piers. The spans over the river are steel trusses supported on caissons. During the earthquake, permanent and transient ground and bridge displacements shifted and tilted the piers causing most of the plate girder sections to fall onto the floodplain. By the time of our visit, the piers had been uprighted by pulling on them with cables attached to the winch on a crane. When the piers were pulled back into a vertical position they were no longer in alignment nor were they equally spaced as they had been. We surveyed the site to determine the positions of the uprighted piers. Assuming that the piers were equally spaced and in line before the earthquake, the survey data indicate that the pier nearest the camera in Figure 15 shifted about 0.8m to the right (in an upstream direction) and the second pier back shifted about 0.8 m inland (away from the river). The other piers shifted as much as 0.15m and some rotated clockwise a few degrees. Fissures, as wide as 30 cm, and sand boils were found in the banana plantation northeast of the damaged bridge. These features indicate that liquefaction and lateral spreading occurred in the vicinity of the bridge. By the time of our visit, however, any fissures or boils under the bridge had been obliterated by construction activities. |
![]() photo: courtesy of EERI |
![]() photo: courtesy of EERI |
The highway bridge over the Rio Estrella incorporated two 75-m long trusses and a 25-m long plate girder section. During the earthquake the ends of the two trusses fell from their common support and dropped into the river. Other damage at the site included spalling of concrete at the tops of the piles supporting the north abutment, and as much as 2m settlement of the fill behind the south abutment. The roadway approach south of the bridge settled, broke up and spread laterally as a consequence of liquefaction of the underlying soils. During our survey, we walked through banana plantations near the south abutment and noted several large fissures that trended parallel to the river. Those fissures are indicative of riverward ground displacements of up to 2 m. We surveyed this site with an electronic total station to determine the post-earthquake positions of piers and abutments. We then calculated distances between these elements and compared them with distances listed on the bridge plans (Table 1). The differences between these distances fall within the range of expected survey and construction error and show that significant permanent displacement did not occur between these components. In particular, the foundation for the southern abutment apparently was sufficiently strong to resist the ground displacements that occurred in the immediate area. |
The building provided space for administrative operations at the Recope oil refinery near Limon. That building was badly damaged during the earthquake due to different resonant frequencies between the roof, which is supported by the tall perimeter columns, and the internal structure which is supported by stiffer internal columns. The damage was primarily to ceiling element; and utilities that formed the connection between the roof and the second story walls.
A secondary cause of damage was a lateral spread that passed beneath the structure and pulled the building apart in extension. The extensional displacements produced several fractures in the lower floor slab, which bears directly on the soil base. We measured the widths of the larger cracks where they were exposed in the floor slab across the front of the building. Those widths were 30mm, 5mm, 1mm, 45mm. and 26mm, respectively, yielding a total extension of 107 mm. That extension was accommodated by fracture and slight tilting of the stiffer exterior columns (those supporting the building and not the roof, as evidenced by circumferential cracks around the bases of several of those columns, and by some distortion of building elements such as window and door frames.
EERI 1991, Costa Rica Earthquake Reconnaissance Report. Earthquake Spectra, v 7, Sup B, 1-127
Youd, T. L., 1984, Geologic effects - liquefaction and associated ground failure, in Proceedings of the Geologic and Hydrologic Hazards Training Program: U. S. Geological Survey Open-File Report 84 760, p. 210-232.
Youd, T. L., and Hoose, S. N., 1978, Historic ground failures in northern California triggered by earthquakes: U. S. Geological Survey Prof. Paper 993, 177 p.
Youd, T.L., Rollins, K.M., Salazar, A.F., and Wallace, R.M., 1992, Bridge damage caused by liquefaction during the 22 April 1991 Costa Rica earthquake: Proceedings, 10th World Conference on Earthquake Engineering, Madrid, Spain, 19-25 July, 1992, Vol. I, p. 153-158.