In July the Insurance Institute for Highway Safety (IIHS) published crash test results for the “small overlap front crash test,” in which only one of the 12 cars tested earned a “good” rating. These results, combined with the results of earlier tests, led Consumer Reports to publish a headline reading, “Most Small Cars Bomb New Small-Overlap Crash Test.” At the same time, Nissan presented an excellent paper on the computer simulation challenges of the small overlap test. This is an example of the increasingly stringent requirements for automotive safety and why the use of high performance computing (HPC) for crash safety simulation is critical in the automotive design process.
There are dozens of crash load cases evaluated in the vehicle development process (including front, side, offset and more). This number is growing and the new load cases are increasingly challenging. Introduced in 2012, the small overlap test replicates what happens when the front corner of a vehicle collides with another vehicle or an object such as a tree or utility pole. (See “Small Overlap Crashes: New Consumer-Test Program Aims for Even Safer Vehicles“). What makes this crash test so challenging is that it bypasses the primary crumple zone in the front of the car, concentrating the impact load in the front suspension and along the side of the vehicle. That results in much more cabin deformation and potentially reduced driver safety. (For an example, see Crash Test IIHS: Small Overlap VS Moderate Overlap Crash Test).
A recent technical presentation by Nissan at the ESI Global Forum outlined the technical challenges in simulating the small overlap test. Since the load path for this crash test bypasses much of the front structure, the energy of the collision is carried through the suspension and wheel of the vehicle to the side structure. Hence, accurately modeling tire behavior, rim and suspension is critical to predicting the overall crashworthiness. The Nissan presentation states the goal of “accurate wheel behavior without tuning” (i.e., the ability to predict results without first adjusting parts of the model to match a prescribed result). This requires a high level of fidelity in the simulation to capture local response such as the failure mode of the suspension, tire deflation and rim rupture. The presentation outlines significant improvement in simulation results by building a more detailed model of the vehicle’s wheel and suspension system (e.g., 2mm solid mesh in the steering gear). However, this comes at a cost of increased compute requirements. “Calculation cost reduction”—reducing turnaround time for higher-fidelity simulations—is highlighted as one of the remaining challenges. As discussed in previous blog posts, the only practical way to significantly improve simulation turnaround time is via improved parallel performance. Significant progress has recently been made in the scalability of explicit structural applications, and Cray continues to work with ISVs in this area.
The crashworthiness of automobiles has improved dramatically in recent years. The industry should be applauded for these enhancements, but small overlap test results are an indication of how design challenges keep growing. Tougher safety standards combined with increased fuel efficiency regulations (54.5 MPG by 2025) and extreme competitive pressure present an incredibly difficult product development challenge. Cray’s goal is to supply the hardware and software tools that enable the auto industry to meet this challenge and continue to improve vehicle safety.