Virtual Weapons of Mass Destruction

A challenging example is the simulation of an underwater explosion, the data from which can be used to protect ships. Many complex physical processes must be modeled: a detonation wave through explosive material, a shock wave into water, the pulsating behavior of gas bubbles (which wreak most of the damage), and the interaction of these shocks with nearby structures. These processes occur over vastly different time scales — from seconds to millionths of a second, making typical numerical models expensive if not impossible, even for today's massively parallel supercomputers. Additionally, the dramatic variation between the properties of the explosive material and water, in both liquid and gas phases, can easily lead to the meaninglessness and collapse of the simulation.

SAIC's Hong Luo and Joseph Baum together with Rainald Löhner have solved this demanding computational fluid dynamics problem in a novel way, applying a method never before used on a problem of this complexity. Rather than explicitly calculating the movement and properties of the entire multi-material flow, which would have been computationally expensive and unstable, the team focused on the interfaces between the different phases of materials. "We do not allow the materials to mix within individual cells of the grid. Pockets of liquid and gaseous water and explosive material are kept distinct by explicitly tracking the interfaces between them," explains Luo. "The bulk of the flow 'goes along for the ride,' which simplifies the computation while making it more robust."

The team specifically tested three variations of the Arbitrary Lagrangian Eulerian (ALE) method, which is typically used to analyze static bodies that interact with fluid. The team adapted the method for their highly volatile multi-material flow problem with excellent results from two of the three ALE schemes. The models achieved sharp resolution along the material interfaces and handled strong property and velocity discontinuities between materials. In addition to successfully modeling a series of underwater explosions, the methods also effectively simulated a supersonic missile over water. Subsequent research will predict damage on structures with more complex geometries, as virtual WMDs will eventually mirror their real-life counterparts.

"On the computation of multi-material flows using ALE formulation" appeared in Volume 194 of Journal of Computational Physics. This research was sponsored by the Defense Threat Reduction Agency.

Related Information

• Homeland Security