This conference paper was submitted for presentation at the NAFEMS World Congress 2025, held in Salzburg, Austria from May 19–22, 2025.

Abstract

Crash simulations are inherently complex and involve significant uncertainty due to the dynamic and nonlinear behavior of materials during impact events. These uncertainties arise from factors such as contact mechanics, friction, variability in impact conditions, and sensitivity to initial conditions. Even small variations introduced during manufacturing can trigger unexpected structural responses, often leading to costly and time-consuming modifications shortly before the start of production (SOP). Addressing these challenges efficiently is critical to ensuring the safety, reliability, and robustness of vehicle designs. Traditionally, quantifying such uncertainties requires a large number of standard simulations to generate sufficient data for accurate analysis. However, this approach is both computationally expensive and time-intensive, posing significant challenges for meeting tight automotive development schedules and cost constraints. To address these limitations, SEAT has developed AQUA (Artificial Quantification for Uncertainty Anomalies), an innovative tool that uses advanced machine learning techniques to revolutionize uncertainty quantification in crash simulations. AQUA leverages data from VPS/Pamcrash simulations to train machine learning models that can accurately predict a broad range of structural behaviors. Unlike conventional methods, AQUA achieves this using a limited number of virtual tests, significantly reducing the computational resources and time required for analysis. This advanced methodology allows engineers to evaluate uncertainties under a variety of conditions, assess the robustness of designs, and proactively identify potential structural weaknesses before production begins. By enabling comprehensive and reliable analysis without the need for physical prototypes, AQUA supports a zero-prototypes goal, eliminating costly trial-and-error iterations. This not only ensures efficiency but also shortens development timelines, reduces expenses, and supports a more sustainable approach to vehicle design. AQUA represents a transformative breakthrough in crash simulation techniques. By combining the predictive power of machine learning with traditional simulation processes, it enhances accuracy, efficiency, and reliability. This groundbreaking tool empowers engineers to develop safer, more robust, and cost-effective automotive products, setting a new standard for innovation in the industry.