Development of Machine Learning Models for Predicting Wind Fields Around a Military Ground Vehicle

During multi-day missions, military vehicles face different environmental conditions. Calculating high-fidelity flow fields for these varying conditions in real-time is an impossible task due to the significant computational time required. This paper discusses a machine learning (ML) based approach to predict the flow fields faster than real-time. The testcase for this ML model is taken as the FED-Alpha vehicle geometry, and the training data for the ML model is taken to be the high-fidelity simulation data from computational fluid dynamics studies involving various wind directions using Ansys/Fluent. The surface temperature of the vehicle is calculated based on the operating conditions of the vehicle using the software TAITherm from ThermoAnalytics, Inc. Three different ML models were tested to estimate the accuracy of the predictions and time requirements.