Parallel distributed, reciprocal monte carlo radiation in coupled, large eddy combustion simulations

Update item information
Publication Type dissertation
School or College College of Engineering
Department Chemical Engineering
Author Hunsaker, Isaac L.
Title Parallel distributed, reciprocal monte carlo radiation in coupled, large eddy combustion simulations
Date 2015-12
Description Radiation is the dominant mode of heat transfer in high temperature combustion environments. Radiative heat transfer affects the gas and particle phases, including all the associated combustion chemistry. The radiative properties are in turn affected by the turbulent flow field. This bi-directional coupling of radiation turbulence interactions poses a major challenge in creating parallel-capable, high-fidelity combustion simulations. In this work, a new model was developed in which reciprocal monte carlo radiation was coupled with a turbulent, large-eddy simulation combustion model. A technique wherein domain patches are stitched together was implemented to allow for scalable parallelism. The combustion model runs in parallel on a decomposed domain. The radiation model runs in parallel on a recomposed domain. The recomposed domain is stored on each processor after information sharing of the decomposed domain is handled via the message passing interface. Verification and validation testing of the new radiation model were favorable. Strong scaling analyses were performed on the Ember cluster and the Titan cluster for the CPU-radiation model and GPU-radiation model, respectively. The model demonstrated strong scaling to over 1,700 and 16,000 processing cores on Ember and Titan, respectively.
Type Text
Publisher University of Utah
Subject combustion; parallel computing; radiation; ray tracing
Dissertation Institution University of Utah
Dissertation Name Doctor of Philosophy
Language eng
Rights Management © Isaac L. Hunsaker
Format Medium application/pdf
Format Extent 27,149 bytes
Identifier etd3/id/4020
ARK ark:/87278/s6pc69qx
Setname ir_etd
Date Created 2016-08-02
Date Modified 2017-09-11
ID 197570
Reference URL