(CFD) simulations intertwining inferences from deep neural networks (DNN). The strategy
leverages the local spatial data of the velocity field to leverage three-dimensional
convolutional kernels within DNN. The hybrid workflow is composed of two-step cycles
where CFD solvers calculations are utilized to feed predictive models, whose inferences, in
turn, accelerate the simulation of the fluid evolution compared with traditional CFD. This …