Data-Driven RANS Closures for Trailing Edge Noise Predictions

Abstract

The ability to conduct high fidelity numerical simulations of aerospace flows of practical interest is becoming a reality in academic research. However, industry still demands computationally inexpensive tools for iterative design, and this is particularly true in the field of aeroacoustics, where noise prediction incurs additional costs. This paper uses the current research trend in using data from high fidelity simulations to train more accurate low fidelity models for trailing edge noise predictions. A symbolic regression tool based on Gene Expression Programming is leveraged to find new expressions for stress strain relationships that are easily implemented in open-source computational f..