Car crash simulation is already a challenging supercomputing task, requiring pinpoint estimation of how hundreds of components interact with turbulent forces and human bodies. Spacecraft crash simulation is far more difficult, as those forces are more likely to come from more directions. The stakes are also much higher, as every manned spacecraft landing is, essentially, a crash, and even minor injuries can prevent astronauts from exiting the capsule, leading to a very stringent low-tolerance policy for injury at NASA. Now, a team of researchers from the Wake Forest School of Medicine have leveraged supercomputing to help validate NASA’s spacecraft crash testing practices.
The extensive series of supercomputer simulations modeled a straight-down collision (as in a spacecraft’s water landing) for different kinds of seats and restraints with three different kinds of passengers: the two most popular crash-test dummies (more formally, “anthropomorphic test devices” or ATDs) and a human body model (HBM), which is a computer-simulated human body.
To run these complex simulations, the researchers turned to the Pittsburgh Supercomputing Center (PSC), where they leveraged the Bridges supercomputer. Bridges is an HPE system delivering 1.35 peak petaflops, with a further recent extension called Bridges-DL aimed at deep learning that delivers at least two additional peak petaflops. In total, with Bridges-DL included, the Bridges system boasts nearly 29,036 CPU cores, 216 GPUs and 277 terabytes of aggregate memory.
After running the simulations, the researchers were able to reach a valuable conclusion: they confirmed that the ATDs and HBMs performed similarly to one another in the simulations. For spacecraft designers, this result increases confidence that a limited number of tests with physical crash test dummies will provide sound results for designing seats and restraints that protect against many kinds of impacts.
PSC also released a clip of the collision simulations side-by-side, with the two crash test dummies on the left and the HBM on the right. The clip is embedded below.
About the research
The research discussed in this article was published as “Head injury metric response in finite element ATDs and a human body model in multidirectional loading regimes” in the January 2020 issue of Traffic Injury Prevention. The paper, which was written by James P. Gaewsky, Jeffrey T. Somers, F. Scott Gayzik, Ashley A. Weaver and Joel D. Stitzel, can be accessed at this link.
To read the PSC news release discussing this research, follow this link.