COMPUTATIONAL FLUID DYNAMICS (CFD) plays a key role in the design and optimization of many industrial applications. In the case of aeronautics, aircraft design has been traditionally based on costly and time-consuming wind tunnel tests. Computer-based flow simulations would enable much faster and less expensive tests, significantly reducing design costs and allowing for the exploration of new airfoil geometries. More importantly, CFD would also enable shape optimization, thus facilitating the development of safer, less polluting and less fuel-consuming aircrafts. Unfortunately, the huge computational costs of CFD prevent it from being a valid tool for the entire design process. CFD is currently used only at some design steps, and wind tunnel tests are still essential. These huge costs come from the Navier-Stokes equations that govern the air flow motion. These Navier-Stokes equations derive from the physical laws of mass, momentum, and energy conservation, and they cannot be solved analytically except in concrete cases, so their solutions must be approximated numerically. The drawback with CFD simulation in aeronautical design is that, in many situations, flows develop two physical phenomena: shock waves and turbulence. Such cases require using a fine discretization of the space to obtain accurate results, so the time required to compute the solutions becomes prohibitive, even in the best high-performance computing (HPC) clusters. Because of this, researchers have expended considerable effort to try to accelerate the execution of these algorithms. Developed algorithms include computing parallelization, GPU computing, and FPGA solutions.
展开▼
