Ricardo Taborda

This research entails the development of methods, capability, and software for high-fidelity and physics-based simulations of entire urban regions to assess the engineering impacts of large magnitude earthquakes on buildings, transportation systems, and underground civil infrastructure. This application is a unique science driver for petascale systems because of its multiple length scales with different physics, large data volumes, and need for highly scalable parallel visualization and data querying. The science problem is to improve understanding of impacts of earthquakes in large urban regions by developing new highly scalable computational capability for end-to-end simulations of earthquakes, from the fault rupture, through the earth and surficial soil layers, to the coupled nonlinear dynamic response of inventories of buildings, bridges, and underground pipelines. Simulations of complete systems will produce on the order of tens to hundreds of terabytes of data for one earthquake simulation, and petabyte datasets for ensembles of earthquakes. This massive scale drives the need for scalable visualization and data analysis in order to provide users and decision-makers an unprecedented multi-resolution view of the earthquake impacts.

The project is making new advances in a hierarchical, multi-scale methodology, for petascale simulation. Going from a large regional, earthquake simulation to multiple subregions is allowing us detailed modeling in a highly scalable manner to capture site response, and structure response accurately for an entire inventory. The Domain Reduction Method (DRM), which allows us to subdivide the overall problem of determining the earthquake response of an urban region into two subproblems. The first involves the evaluation of ground motion in the absence of the structural region, under assumptions of linear soil behavior, and the second introduces the urban setting and the inelastic behavior of the soil. The DRM is being applied for the first time to using ground motion from a regional simulation as input for the populated subregions. These subregions include models of highly nonlinear soil and detailed models of buildings and bridges along with embedded foundations. A new scalable implicit-explicit time integration method has been developed to provide optimal computational performance for the complex earthquake simulations for a subregion. Data analysis and visualization capability is under development to run on the same parallel processors as the simulation, drastically reducing the need to move data. Using this approach for data-intensive supercomputing, in-situ visualization and a new computational database system will allow unprecedented ability to understand earthquake impacts.