eduKateSG · How Science Works
Choose your reading route
Follow grain contacts, pore pressure and effective stress. Work an explicit stress budget and examine field evidence without mistaking a teaching model for a site assessment.
Understand the internal mechanism · Chapters 1–4 · Start this route
Calculate and examine consequences · Chapters 5–7 · Start this route
Test evidence and limits · Chapters 8–10 · Start this route
Full contents · Check your understanding · How Science Works master guide
A road can split, a building can tilt and water carrying sand can emerge from the ground during an earthquake. In some places the underlying sediment has lost much of its ability to resist deformation. Yet the mineral grains have not melted, and the earthquake has not turned rock into a new liquid.
The key lies between the grains. In susceptible water-saturated sediment, shaking can raise the pressure of water in the pore spaces. That pressure reduces the effective stress carried through the grain framework, weakening its resistance to shear. This process is earthquake-induced soil liquefaction.
Here we examine the mechanism, its measurements and its limits. Seismology supplies the earthquake and ground-motion background. The existing Seismic Ground Map article addresses planning decisions. The calculations below are learning models, not assessments of any building or parcel of land.
Contents
Understand the internal mechanism · Chapters 1–4
1. Ground can be a framework of solid grains
2. Separate total stress from effective stress
3. Shaking creates a conflict between rearrangement and drainage
Calculate and examine consequences · Chapters 5–7
5. A worked pressure and friction model
Test evidence and limits · Chapters 8–10
8. Pressure records reveal what photographs cannot
CHAPTER 1 OF 10 · Understand the internal mechanism
1. Ground can be a framework of solid grains
A sand deposit contains mineral grains and spaces between them. Those spaces, called pores, may contain air, water or both. In saturated soil, water occupies the pore space. The material’s behaviour depends on the grains, their arrangement, the pore fluid and the forces applied to them together.
Imagine magnifying a small region of sand. Neighbouring grains press against one another through a network of contacts. That framework can resist forces because grains must slide, roll or rearrange for substantial deformation to occur. Friction and interlocking contribute to its resistance.
A pile of individually hard grains is therefore not mechanically equivalent to one continuous block of rock. A change in contacts can make the assembly behave very differently without changing the mineral phase. This is why the phrase “the ground became liquid” needs interpretation.
USGS defines liquefaction in terms of loosely packed, water-logged sediment losing strength during strong shaking. It is a change in the behaviour of the sediment-water system. A grain recovered afterwards can still be solid quartz or another mineral.
The distinction between material and assembly also explains why visual firmness before an earthquake is insufficient. A paved surface hides the sediment below. A soil layer can support ordinary conditions yet respond differently under repeated rapid loading. Its previous appearance does not reveal all the variables governing its earthquake response.
CHAPTER 2 OF 10 · Understand the internal mechanism
2. Separate total stress from effective stress
Stress is force per unit area. In saturated soil, a simplified normal-stress relation is effective stress = total stress − pore-water pressure. It is often written σ′ = σ − u, where σ is total normal stress and u is pore-water pressure.
Total stress represents the overall loading on the region being considered. Effective stress describes the part relevant to the load-bearing grain framework in this soil-mechanics description. Water pressure acts within the pore space and changes how strongly the grains are effectively pressed together.
This is not a claim that water is weightless or irrelevant to total loading. Water contributes weight, and changes in water conditions can affect more than one term. The equation becomes useful only when all stresses and pressures are defined consistently at the same location.
Consider an original hypothetical soil element with total vertical stress of 100 kilopascals and pore-water pressure of 40 kilopascals. Its vertical effective stress is 100 − 40 = 60 kilopascals. If total stress is held fixed while pore pressure rises to 85 kilopascals, effective stress falls to 15 kilopascals.
The applied total loading has not disappeared. Its distribution between the pore-fluid pressure and the grain framework has changed. The final effective stress is one quarter of the starting value, a 75% reduction.
USGS’s engineering-geology treatment uses this same subtraction when describing saturated subsurface layers. Our round-number calculation is only an illustration of the relation. Real earthquake loading involves varying stresses, directions and material response; one vertical stress value is not a complete strength model.
CHAPTER 3 OF 10 · Understand the internal mechanism
3. Shaking creates a conflict between rearrangement and drainage
Loose granular arrangements can have a tendency to become more compact when sheared. If the pore water can escape sufficiently quickly, grains can rearrange into a smaller overall volume and the layer can settle.
During rapid earthquake loading, however, water may not be able to drain away fast enough. The tendency towards a smaller pore volume then raises pore-water pressure instead. The soil is often approximated as undrained over that short loading interval: substantial water does not leave the soil element during the process being analysed.
“Undrained” does not mean the deposit is sealed forever. It is a statement about competing timescales. A layer can retain pressure during seconds of loading and dissipate it later. Drainage depends on permeability, distance to a drainage boundary and the surrounding layers, among other factors.
Draw two panels showing the same initial loose saturated grains. In Panel A, mark an available drainage route and enough time for water to leave as the grains compact. In Panel B, mark rapid shear and insufficient drainage, with a pressure gauge rising. Label the second panel “pressure increases while volume change is constrained”. Do not draw it as water suddenly appearing from nowhere.
The chain is now explicit: imposed deformation, a contractive tendency, restricted drainage, excess pore pressure, reduced effective stress. Missing any link can produce a misleading explanation. Shaking alone does not guarantee liquefaction, and the presence of water alone does not guarantee it either.
Laboratory research on pore-pressure rise in saturated sands during cyclic loading examined how loading intensity, prior loading and repeated cycles influence the response. This is why “it shook” is a beginning, not a complete description of the experiment.
CHAPTER 4 OF 10 · Understand the internal mechanism
4. Why not every wet soil responds alike
Soils differ in density, grain-size distribution, grain shape, plasticity, fabric and stress history. Their drainage conditions differ too. The simple loose-saturated-sand picture identifies an important mechanism without turning all wet ground into one material.
A denser granular assembly can show a tendency to dilate, meaning it tends to increase volume during shear. Under undrained conditions that tendency can oppose pressure increase and temporarily restore effective stress. Soil response can therefore vary over a loading cycle rather than following one smooth decline to permanent zero strength.
This helps distinguish different behaviours grouped within discussions of liquefaction. Some loose materials can undergo large flow deformation when their reduced resistance is insufficient for the stresses already acting. Other soils experience repeated episodes of low effective stress and accumulating deformation during cyclic loading, with resistance changing as they deform.
Neither pattern means the ground has become an ordinary frictionless liquid. “Liquefied” does not specify a universal viscosity, zero resistance under all conditions or a guaranteed depth of sinking. Those consequences require a more detailed material and loading description.
A USGS study of liquefaction-related flow and deformation distinguishes contractive and dilative responses and relates continued flow to the stresses acting on the weakened sediment. The valuable principle is conditional: losing strength and undergoing a large destructive movement are linked, but they are not identical questions.
For the beginner, retain three separate ideas. A soil may be susceptible to pressure buildup. A particular loading history may trigger severe weakening. The geometry and loads then influence what deformation follows. Collapsing those steps into one label hides much of the science.
CHAPTER 5 OF 10 · Calculate and examine consequences
5. A worked pressure and friction model
Return to the hypothetical total vertical stress of 100 kilopascals. Start with pore pressure 40 kilopascals and effective stress 60 kilopascals. Suppose three later snapshots have pore pressures of 60, 85 and 95 kilopascals, while total stress remains fixed for this exercise.
The corresponding effective stresses are 40, 15 and 5 kilopascals. Relative to the initial effective stress, the last is 5/60, approximately 8.3%. A large change in contact-related loading can occur without a change in the chosen total load.
To connect this to shear resistance, introduce an intentionally simplified friction model: available frictional resistance equals an effective normal stress multiplied by a chosen coefficient, here 0.6. We neglect cohesion and all changes in geometry, fabric and coefficient. At effective normal stresses of 60 and 15 kilopascals, the calculated resistances are 36 and 9 kilopascals respectively.
The model illustrates why a reduction in effective normal loading can weaken frictional resistance. It does not calculate a foundation’s capacity or reproduce a complete cyclic soil test. Vertical effective stress is not automatically the normal stress on every possible shear plane; the model treats one specified idealised plane only.
A second useful quantity is the excess pore-pressure ratio in this simplified constant-total-stress setting. The pore-pressure increase from 40 to 85 is 45 kilopascals. Divide by the initial effective stress of 60 to obtain 0.75. Here that means 75% of the initial effective stress has been offset by added pressure.
If pore pressure reached 100 kilopascals under the same assumptions, effective stress would approach zero and the excess ratio would reach one. This is an idealised indicator, not a rule that no important deformation can occur before the ratio reaches one. Real definitions, stress paths and triggering criteria must be stated when interpreting measurements.
CHAPTER 6 OF 10 · Calculate and examine consequences
6. What weakened ground can do
One consequence is loss of support. A loaded foundation can settle or rotate if the underlying material no longer provides sufficient resistance. How much it moves depends on the structure, the layers and the evolving soil response, not simply the word “liquefaction”.
Another is lateral spreading. Ground can move horizontally towards a free face, such as a riverbank, or along a gentle slope. The surface may stretch and crack as the underlying material deforms. A nearly level-looking site is therefore not necessarily limited to vertical settlement.
Water and sediment can also be expelled through openings, leaving sand boils or other ejecta at the surface. “Boil” describes the appearance; it does not imply that the water has reached its thermal boiling point. Pressure-driven flow can transport grains without melting them.
As excess pressure later dissipates, sediment can consolidate into a denser arrangement. Additional settlement may follow. The return of stronger grain contacts does not reverse a building’s tilt, repair a ruptured pipe or restore the earlier ground surface.
USGS’s investigation at Oceano after the 2003 San Simeon earthquake combined subsurface exploration and aftershock monitoring to examine liquefaction and lateral spreading. Such investigations distinguish ground failure from the separate question of amplified shaking. Both can occur in one area, but they are not the same mechanism.
A photograph of damage is therefore a clue rather than a complete diagnosis. Settlement can have other causes; a ground crack can arise from several processes. Interpretation improves when the surface pattern is connected to subsurface layers, groundwater and a recorded loading history.
CHAPTER 7 OF 10 · Calculate and examine consequences
7. A real earthquake tests the sediment explanation
The 1989 Loma Prieta earthquake produced conspicuous ground damage around parts of San Francisco Bay. Its effects provide a case study because investigators could compare damaged sites, sediment properties and earthquake shaking rather than relying on one dramatic image.
A primary USGS investigation of East Bay sites documented liquefaction in uncompacted artificial-fill deposits. Field testing identified low penetration resistance in cohesionless sandy and silty fill at affected locations. The study examined several sites and compared observations with methods for assessing liquefaction susceptibility.
The evidence supports a connection among material state, loading and failure. It does not establish that every artificial fill deposit will liquefy, that all natural deposits are safe, or that one historical earthquake defines future performance. Fill is a history of emplacement, not a complete mechanical specification.
The broader USGS Loma Prieta reports also distinguish shaking and ground-failure effects. Soft sediments can alter the motion arriving at the surface, while saturated susceptible layers can undergo pressure-related weakening. Those processes can interact, but each requires evidence of its own.
An original comparison clarifies the logic. Suppose two hypothetical sites experienced similar shaking, but only one showed sand ejecta and settlement. The difference suggests local conditions matter. It does not tell us whether density, saturation, layer thickness or drainage was decisive until those variables are measured.
Likewise, comparing two sites with different shaking and different sediment does not isolate a single cause. Good case-history work documents the differences rather than claiming that an earthquake provides a perfectly controlled experiment.
CHAPTER 8 OF 10 · Test evidence and limits
8. Pressure records reveal what photographs cannot
The mechanism predicts a changing internal state. To test it directly, researchers need measurements of ground motion and pore-water pressure, ideally at relevant depths and with instruments whose response is understood. Surface damage alone cannot supply that time history.
The Wildlife Liquefaction Array in California recorded the 1987 Superstition Hills earthquake sequence. Holzer and Youd’s 2007 analysis attributed pressure generation to both stronger high-frequency motion and later long-period deformation associated with Love waves. Their explanation connected a continued pressure rise with continuing cyclic strain rather than treating the end of the strongest acceleration as the end of all important loading.
That finding illustrates why the largest acceleration alone is not a complete account of loading. Duration, frequency content and the resulting deformation can matter. A modest-looking part of an acceleration record may still contribute to the soil’s evolving state.
Measurement itself remained part of the scientific problem. A published in-place calibration study of Wildlife pressure instruments examined sensor performance, and subsequent analyses discussed interpretation of the records. Instruments require calibration; a recorded curve should not be treated as self-validating merely because a computer plotted it.
An ideal teaching plot would place time on a shared horizontal axis with separate panels for acceleration, pore pressure and deformation. Mark the strongest acceleration and ask whether pressure stops rising at that exact moment. Then ask whether the sensors measure the same layer whose movement is being explained.
This is a comparison of evidence channels. Agreement between a mechanism, independently measured motion and calibrated internal pressure is more informative than a single line or image considered alone.
CHAPTER 9 OF 10 · Test evidence and limits
9. Use a thought experiment instead of a home safety test
Consider two imaginary samples with the same grains, initial packing, saturation and applied shear history. Sample A has drainage prevented during loading. Sample B has an ideal drainage boundary and enough time for water to leave. Predict which is more likely to develop substantial excess pore pressure under contractive loading.
The answer is A under the stated assumptions. B can accommodate some compaction through drainage. But a real experiment must establish that drainage is actually fast enough; merely adding an outlet does not prove fully drained behaviour throughout a sample.
Now change the question. Suppose B is also denser than A. If the outcomes differ, the comparison no longer isolates drainage. Packing and drainage have both changed. To investigate them separately, a study would need additional controlled comparisons and adequate repeated measurements.
A tabletop container of wet sand can illustrate rearrangement, but its scale, drainage boundaries and shaking differ from a deep soil deposit. It cannot certify the safety of a garden, home or neighbourhood. No home demonstration is needed to work through the causal comparisons above.
For a paper-based exercise, create a three-column record: imposed conditions, predicted internal response and observable consequence. Enter “rapid undrained shear of contractive saturated sediment” in the first column, “rising excess pore pressure and falling effective stress” in the second, and “possible deformation, depending on loads and geometry” in the third. The conditional wording is part of the explanation.
If an actual earthquake has damaged ground or structures, follow local emergency directions and professional assessments rather than approaching unstable areas to collect evidence. This tutorial explains a mechanism; it does not determine whether a real location is safe.
Did the soil melt? No. Solid grains can remain solid while their water-saturated assembly loses strength. The important change concerns stress transmission, pore pressure and deformation.
Does water pressure always rise when soil is sheared? No. The response depends on material state and loading, including contractive or dilative tendencies and drainage. The simple explanation deliberately describes susceptible saturated sediment under rapid loading.
Is effective stress the same as shear strength? No. Effective stress is a stress description relevant to the grain framework. Strength depends on it along with friction, fabric, stress path and other material properties. Our friction calculation illustrated a relationship, not an identity.
Does a return to lower pore pressure undo the damage? No. Pressure dissipation and renewed contact forces can occur after permanent settlement or lateral movement. Recovery of resistance is not restoration of geometry.
Can a map or photograph answer the whole question? No. Screening maps, field observations, subsurface tests, groundwater information and loading estimates contribute different pieces. The Seismic Ground Map explains how those pieces enter planning; property-specific conclusions require suitable investigation and qualified interpretation.
The causal chain is compact but powerful: a granular framework tends to rearrange, drainage cannot keep pace, pore pressure builds, effective stress declines, and the weakened material may deform under the forces acting on it. Follow that sequence, check the water and stress conditions, and keep triggering separate from consequences.
Return to Seismology for the waves and measurements that supply the loading side of the story, or explore the How Science Works master guide.