Data Driven Models for the Design of Rocket Injector Elements

Improving the predictive capabilities of reduced-order models for design injector and chamber elements rocket engines could greatly improve quality early designs. In present work, we propose an innovative methodology that uses high-fidelity numerical simulations turbulent reactive flows artificial intelligence generation surrogate models. The were generated analyzed are deep learning networks trained on a dataset 100 large eddy single-shear coaxial chamber. experiments was created considering three parameters: diameter, recess length, oxidizer–fuel ratio. paper presents developed training optimizing data-driven Fully connected neural (FCNNs) U-Nets utilized as surrogate-modeling technology. Eventually, global quantity, average, root mean square fields used in order to analyze impact length post’s performances obtained behavior flow.