DeepCFD: Efficient Steady-State Laminar Flow Approximation with Deep\n Convolutional Neural Networks

Computational Fluid Dynamics (CFD) simulation by the numerical solution of\nthe Navier-Stokes equations is an essential tool in a wide range of\napplications from engineering design to climate modeling. However, the\ncomputational cost and memory demand required by CFD codes may become very high\nfor flows of practical interest, such as in aerodynamic shape optimization.\nThis expense is associated with the complexity of the fluid flow governing\nequations, which include non-linear partial derivative terms that are of\ndifficult solution, leading to long computational times and limiting the number\nof hypotheses that can be tested during the process of iterative design.\nTherefore, we propose DeepCFD: a convolutional neural network (CNN) based model\nthat efficiently approximates solutions for the problem of non-uniform steady\nlaminar flows. The proposed model is able to learn complete solutions of the\nNavier-Stokes equations, for both velocity and pressure fields, directly from\nground-truth data generated using a state-of-the-art CFD code. Using DeepCFD,\nwe found a speedup of up to 3 orders of magnitude compared to the standard CFD\napproach at a cost of low error rates.\n