Accelerating urban scale simulations leveraging local spatial 3D structure

This paper presents a hybrid methodology for accelerating Computational Fluid Dynamics (CFD) simulations intertwining inferences from deep neural networks (DNN). The strategy leverages the local spatial data of velocity field to leverage three-dimensional convolutional kernels within DNN. workflow is composed two-step cycles where CFD solvers calculations are utilized feed predictive models, whose inferences, in turn, accelerate simulation fluid evolution compared with traditional CFD. approach has proved reduce 30% time-to-solution an urban scale study case, which leads generating massive datasets at fraction cost.