Abstract

Predictive sampling of compressible flows is an important aspect of aerodynamic design, analysis, and optimization. The process is usually done by generating flow fields from computational fluid dynamics (CFD) simulations and solving governing evolution equations, from which quantities of interest (QoI) are obtained by post-processing. With this study, we propose a data-driven approach for the predictive sampling of compressible flows around airfoils. This paper demonstrates the potential of generative adverserial networks (GANs), particularly the Pix2Pix model, to predict supersonic flow fields around airfoils. Building upon the similarities between our generator architecture and FlowGAN, which also employs the Pix2Pix model, our work extends the capabilities of FlowGAN to encompass supersonic flows and higher-resolution predictions. Our methodology introduces a multi-variable prediction framework, enabling the simultaneous prediction of various flow variables, including pressure, velocity, and density. We achieve the generation of flow fields with a relative error of less than 1.5%. We utilize the same framework to train a model capable of generating synthetic schlieren images, exhibiting a similarity structure index greater than 0.97, underscoring the high-fidelity generation capacity of our model. Addressing the challenges of generalization to unseen Mach numbers and angles of attack, we adopt transfer learning. By fine-tuning a pre-trained model on a limited number of examples from the unseen parameters, we obtain an improved prediction of shock wave positions. Furthermore, we present a practical application of our methodology by reconstructing flow fields from schlieren images. This application could prove valuable in real-world scenarios, providing a useful tool for studying and comprehending complex flow phenomena.

Full Text
Paper version not known

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call

Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.