Computational Fluid Flow Simulation on Body Fitted Mesh Geometry with FPGA Based Emulated Digital Cellular Neural Networks

摘要

The analog CNN-UM can be used to solve the Navier-Stokes equations quite fast. But using in engineering applications it can not be sufficiently accurate and reliable because noises front the environment, such as power supply noise or temperature fluctuation. With the proper Field Programmable Gate Array (FPGA) we can gain sufficient (adequate) computation speed with high precision. The dedicated hardware elements of the FPGA can highly accelerate the computations on curved surface. Consequently it can be used in industrial applications where fluid flow simulation around complex shapes is required. In the paper the implementation and optimization of a new Computational Fluid Dynamics (CFD) solver architecture, which can work on Body Fitted Mesh geometry, on FPGA is described. The proposed new architecture is compared to existing solutions in terms of area, speed, accuracy and power dissipation.