Acceleration of combustion simulations using GPUs

Damm, K. A., Gollan, R. J. and Veeraragavan, A. (2015). Acceleration of combustion simulations using GPUs. In: Yi Yang and Nigel Smith, The Australian Combustion Symposium 2015 Proceedings. Australian Combustion Symposium, Parkville, VIC, Australia, (148-151). 7-9 December 2015.

Attached Files (Some files may be inaccessible until you login with your UQ eSpace credentials)
Name Description MIMEType Size Downloads
Author Damm, K. A.
Gollan, R. J.
Veeraragavan, A.
Title of paper Acceleration of combustion simulations using GPUs
Conference name Australian Combustion Symposium
Conference location Parkville, VIC, Australia
Conference dates 7-9 December 2015
Proceedings title The Australian Combustion Symposium 2015 Proceedings
Place of Publication Melbourne, VIC, Australia
Publisher Combustion Institute Australian and New Zealand Section
Publication Year 2015
Sub-type Fully published paper
Open Access Status Not Open Access
ISSN 1839-8170
Editor Yi Yang
Nigel Smith
Start page 148
End page 151
Total pages 4
Collection year 2016
Language eng
Formatted Abstract/Summary
The solution of the chemical kinetics problem in reacting flow simulations requires the integration of a large set of Ordinary Differential Equations. This paper explores the use of general purpose graphic processing unit (GPGPU) programming to accelerate this costly computation via the implementation of a chemical kinetics ODE solver on a GPU using OpenCL. The GPU solver is coupled into the University of Queenslands in-house Navier-Stokes code Eilmer. Three reaction mechanisms are simulated for two case studies: a well-stirred reactor and a shock-induced combustion flow. Results from the well-stirred reactor simulation show that the GPU implementation outperforms an identical CPU implementation for problem sizes exceeding 64 to 256 cells, depending on reaction mechanism complexity. The performance is shown to scale with both number of cells and reaction mechanism complexity. The results of simulating shock-induced combustion showed that branch divergence limited the performance of the GPU implementation.
Keyword Chemical kinetics
Reacting flow simulation
Q-Index Code E1
Q-Index Status Provisional Code
Institutional Status UQ
Additional Notes

Version Filter Type
Citation counts: Google Scholar Search Google Scholar
Created: Fri, 05 Feb 2016, 08:53:04 EST by Brianne Mackinnon on behalf of School of Mechanical and Mining Engineering