On the Dynamic Mode Decomposition of the Flow Over an Idealized Ground Vehicle at High Reynolds Number

Abstract

In fluid flow analysis, Dynamic Mode Decomposition (DMD) has shown the ability of linearizing a large dataset of high dimensional sequential measurements by decomposing it into low-dimensional coherent structures and their associated time-dynamics. The low-rank dataset thus obtained can then be applied to a linear regression model to predict the future state of the fluid flow. Existing literature suggests that the success of this algorithm is limited to low Reynolds number applications. Given that most flows of engineering interest are high Reynolds number flows, this paper explores and presents DMD analyses of the flow around an idealized ground vehicle, the Ahmed body, at a Reynolds number of 2.7 million. The high dimensional dataset explored in this paper was collected from a Computational Fluid Dynamics (CFD) simulation using the SST k-omega based IDDES approach. DMD algorithms currently found in literature were not able to produce the desired outcome for this flow case. Hence, enhancements to the existing algorithm were explored and a modified DMD algorithm, which is applicable to high Reynolds number, separation dominated flows was developed. The effectiveness of the proposed algorithm was tested by comparing DMD-based reduced order model predictions of the force and moment coefficients, and surface pressure field to CFD simulation data.