In an earthquake, the ground doesn’t always just shudder and roll. In some places, solid earth turns liquid and can swallow cars and roads, or even topple buildings that may be resting on top of the weakening soil. This “liquefaction” can be the most destructive part of the disaster. Scientists have long tried to study the phenomenon, but now researchers say that many of these experiments are flawed and fail to accurately represent real-world conditions.
In sediments like sand, the grains stack on top of one another and create a skeletal network. The energy of an earthquake jostles and squeezes the sediment grains together and the skeleton collapses. This collapse forces all the grains closer together, closing in the walls around ground water in between grains. The water molecules push back against the sand, raising the water pressure so it’s strong enough to force the grains out of contact with one another, while the shaking allows grains to readily move. These two things make it so every bone in the skeleton has been rattled and pushed apart, and the ground becomes like a slurry and liquefies.
But not all sediments are equal. Scientists and engineers have devised methods and models to test different sediments and predict where liquefaction is most likely to occur. That would help city planners avoid liquefaction-susceptible areas when placing important infrastructure like dams or gas pipelines that can burst and start fires in an earthquake. Such knowledge could also help scientists devise methods to mitigate liquefaction risks.
But geotechnical engineer Amy Rechenmacher, an associate professor at the University of Southern California (USC) in Los Angeles, says that in these tests “the understanding of the physics was not complete.” People weren’t thinking about liquefaction correctly, adds Rechenmacher’s colleague, Daniel Lakeland, a Ph.D. graduate in civil engineering at USC. “Historically, people considered this to be an undrained process,” meaning the water inside sediment pore spaces doesn’t flow away. Scientists knew that water would flow through sediments during earthquakes, but Lakeland says “the assumption was that flow was not important.” The idea was that pressure from the earthquake builds up far faster than water can flow away to relieve it, almost 10 times faster, so it didn’t matter if water was moving.
To understand which sediments liquefy, scientists could use a test where they shake a waterlogged sediment wrapped in a watertight skin to prevent flow and measure the pressure when the sediment liquefies. But Rechenmacher says that this approach makes the incorrect assumption that liquefaction would occur locally in some part of a sediment deposit. If water and pressure never migrated, then soils would liquefy all at once right where they were.
To get a better sense of how liquefaction works, the team built a model that more accurately replicated what happens in the real world. The model showed that once water pressure ratchets up in sands, water can begin draining almost as quickly as the pressure builds, so pressure actually diffuses through the sediment pores, the researchers report in the May issue of the Proceedings of the Royal Society A. This movement creates a wave of pressure that travels through the sediment until it eventually piles up on itself, and liquefaction initiates somewhere else, possibly somewhere unexpected.
“This flow is placing where liquefaction will occur,” Lakeland says. If you ignore the flow, then all that matters is how hard the earthquake is squeezing the sediment “and you won’t see a difference in the location from one place to another as to what the pressure looks like.” Rechenmacher says not capturing this part of the physics led to “a lot of inability of those lab tests to predict liquefaction.” Using these tests, scientists weren’t able to accurately pinpoint what areas would be the safest or the most dangerous during an earthquake, creating difficulties when trying to mitigate earthquake damages.
There are other possibilities, too. If the water is moving through the ground, it could load up against some more impermeable material, like silt or clay, and cause liquefaction against that boundary, too. If the clay or silt is on a slope, that would result in catastrophic landslides.
Geotechnical engineer Ross Boulanger, a professor at the University of California, Davis, says the authors make a good point. “I think their model is doing a good job of capturing this important part of the physics.” Because the lab tests can’t account for fluid flow at all, Boulanger says the tests can’t reflect liquefaction behavior in the real world, and a lot of people ignore that. But he also says there were some people who did know fluid flow is important and had tried to account for it in the past.
Lakeland admits that people had seen the effects of fluid flow in some specialized experiments, but he says that “nobody had come up with an explanation for it.” He says now we can “be confident that we’ve made a model that captures all the physical phenomena.”
Rechenmacher hopes their new model will help engineers and city planners improve analysis of liquefaction risks and identify places “that previously weren’t thought to be a problem.” Lakeland thinks the work could one day help “prevent enormous potential damages” by aiding planners in designing safer cities and softening the blow from a powerful earthquake.
*Correction, 24 March, 8:29 a.m.: This article previously stated that Daniel Lakeland is a civil engineer at USC, and that Amy Rechenmacher felt the research would help identify where liquefaction is most likely to occur. This has been corrected to a Ph.D. graduate in civil engineering at USC, and that Rechenmacher hopes the model will help improve analysis of liquefaction risks.