The Energy Geosciences Division at the Lawrence Berkeley National Laboratory, or Berkeley Lab, is working in partnership with Lawrence Livermore National Laboratory on a project to create a next-generation supercomputer that can predict the effects of earthquakes.
The project is called the Exascale Computing Project and is sponsored by the U.S. Department of Energy to develop more advanced computing to run complicated earthquake simulations. According to David McCallen, a guest senior scientist at Berkeley Lab’s Energy Geosciences Division, current supercomputers aren’t able to predict earthquake ground motions and their effects on buildings.
McCallen, the project’s lead researcher, added that the year-old Exascale Computing Project is a numerical experiment in which the supercomputer in development — nicknamed Cori — is used to run repeated simulations of an earthquake in the Bay Area and learn more about how buildings respond to earthquakes in the process.
McCallen’s research aims to use a supercomputer model to predict earthquake effects rather than rely on historical records.
“We are developing a software application to do a better job of simulating earthquake ground motions and structural response, enabled by next-generation supercomputers,” McCallen said.
Supercomputer simulations help researchers understand the distribution of earthquake ground motions and which areas will be more affected, McCallen said. He added that this greater understanding will in turn allow us to design future structures better while improving existing structures in at-risk areas.
According to McCallen, the next-generation supercomputer should be online by the early 2020s.
“We would like to have a regional-scale Bay Area model that everyone can use.” McCallen said.