/thermo_fea

Using a GPU, parallelize computation for large and complex heat transfer transient problems.

Primary LanguageC++

ThermoFEA

ThermoFEA is another small project I completed which utilizes a computer's GPU to parallelize computation for a large and complex thermodynamic finite element analysis transient problem, consisting of thousands of nodes.

The objective of the project was to dramatically decrease and improve runtime for the implementation found from MathWorks' webinar. This was successfully accomplished, decreasing the runtime from the original script by over 900 times, using OpenCL and parallelizing repetitive calculations using my computer's GPU. I wrote a kernel to concurrently execute the program in parallel over thousands of GPU work-items.

To visualize the output from the OpenCL project, I wrote a MATLAB script to display every time steps' heat map to a figure. The following image is an example of the steady-state thermodynamic solution of a long aluminum bar's square cross-section. The boundary conditions are constant temperatures applied to each side of the long metal bar. The axes plot the node numbers in the x and y directions.

Dependencies