Thermo-fluids applications are studied, both experimentally and numerically, due to its applications in various engineering and scientific problems. With the growing computer technology, numerical simulations are often employed to analyze thermofluid phenomenon. In this paper, we study the effect of heat source and sink in mixed convection in a lid-driven square cavity. Many applications require efficient computation of flow and temperature field data for cooling and other control purposes. We develop an explicit numerical methodology, both on CPU and GPU platforms, to solve the governing equations. The numerical scheme is implemented sequentially on a CPU and in parallel on a GPU using CUDA programming language. We vary the size and locations of the heated wall, along with the inclination of cavity, to study its effect on convection. In terms of parallel performance, GPU provides speed-up ranging from 3 to 13 depending on the problem configuration.

This content is only available via PDF.
You do not currently have access to this content.