Data-Driven Prediction of Complex Flow Field Over an Axisymmetric Body of Revolution Using Machine Learning

Abstract Computationally efficient and accurate simulations of the flow over axisymmetric bodies revolution (ABR) have been an important desideratum for engineering design. In this article, field ABR is predicted using machine learning (ML) algorithms (e.g., random forest (RF), artificial neural network (ANN), convolutional (CNN)) trained ML models as surrogates classical computational fluid dynamics (CFD) approaches. The data required development were obtained from high fidelity Reynolds stress transport model (RSTM)-based simulations. approximated functions x y coordinates locations in velocity at inlet domain. optimal hyperparameters are determined validation. can predict rapidly exhibit orders magnitude speedup conventional CFD results pressure, velocity, turbulence kinetic energy compared with baseline data. It found that ML-based surrogate predictions results. This investigation offers a framework fast scenario critically