A HYBRID REDUCED ORDER MODELING APPROACH FOR RAPID TRANSIENT THERMAL SIMULATION AND SIGNATURE ANALYSIS

This paper presents a hybrid CFD and reduced order modeling (ROM) approach for fast and accurate flow and thermal analysis of vehicles to enable rapid thermal signature prediction. The modular hybrid ROM solver includes several key components, such as the turbulence modeling, CFD full order model (FOM) customized for vehicle thermal analysis, FOM/ROM alternation, proper orthogonal decomposition (POD) for basis vector construction, and online model switch decision maker for coupled simulation, which are all developed in an integrated framework. Several case studies of Army relevance at increasing complexity levels are undertaken. The proposed hybrid ROM solver is able to accurately analyze flow, turbulence, and thermal phenomena under time-varying operating conditions with unprecedented computational performance. Quantitatively, the relative error of our hybrid CFD FOM/ROM simulation stays below 0.35% and the absolute error is less than 4 K. The ROM has a much smaller model order (typically 10-15) relative to the CFD FOM. Speedup ratio up to 27× is achieved in the present study without the use of any hardware accelerate technique.

Citation: Y. Wang, W.C. Krolick, A.L. Kaminsky, N. Tison, Y. Ruan, V. Korivi, K. Pant, “A Hybrid Reduced Order Modeling Approach for Rapid Transient Thermal Simulation And Signature Analysis”, In Proceedings of the Ground Vehicle Systems Engineering and Technology Symposium (GVSETS), NDIA, Novi, MI, Aug. 10-12, 2021.