Implementation of Direct Simulation Monte Carlo Algorithms using FPGAs
Direct Simulation Monte Carlo is a ubiquitous simulation method for low-density flows. It involves statistical simulation of representative populations of molecules in rarefied gas mixtures. The technique can simulate many physical processes very accurately at much less computational expense than direct molecular dynamic modelling. However, the computational cost is still very high for a range of practical fluid problems. As the technique mostly consists of simple random number generation, the method has the potential to be used very efficiently using field-programmable gate arrays. These programmable hardware devices can operate with highly efficient inter-process connections and can generate random numbers at very fast speeds.
This project involves an evaluation of the suitability of the FPGA architecture for implementation of the DSMC method, outlining the feasibility and limitations of the architecture for solution of difficult problems such as diffusion of gases in capillary tubes.
The basic equipment for the project is available. There is no external funding source for the project.
Associate Professor Sean O’Byrne,
GF07, Building 17,
School of Engineering and IT,
Campbell, ACT, 2600