TY - GEN
T1 - Development of an amr octree DSMC approach for shock dominated flows
AU - Sawant, Saurabh S.
AU - Jambhunathan, Revathi
AU - Tumuklu, Ozgur
AU - Korkut, Burak
AU - Levin, Deborah A.
N1 - Publisher Copyright:
© 2016, American Institute of Aeronautics and Astronautics Inc, AIAA. All rights reserved.
PY - 2016
Y1 - 2016
N2 - The present work is a continued effort of the development of the SUGAR code for simulating a hypersonic flow over a double wedge at a Knudsen number of 0.000284. The current work describes verification of the code for Borgnakke-Larsen continuous rotational relaxation model and computation of surface coefficients for hypersonic flows over a hemisphere and a double wedge at Knudsen numbers of 0.277 and 0.020, respectively. Further, the paper mainly describes attempts to solve two main performance bottlenecks in making the code simulate at continuum-like conditions, first, the scalability of the code for more than 128 processors by reducing the communication and evenly balancing the computational load, and second, improve the algorithmic performance of the code by eliminating the expensive recursive tree traversal inherent in Octree based mesh structure. In order to resolve the first issue sophisticated graph-partitioners have been use, however, with no success. However, significant performance improvement has been obtained by linearizing the Octree using Morton-Z space filling curve. Based on these findings, a prediction is made of time required to run the main experimental case.
AB - The present work is a continued effort of the development of the SUGAR code for simulating a hypersonic flow over a double wedge at a Knudsen number of 0.000284. The current work describes verification of the code for Borgnakke-Larsen continuous rotational relaxation model and computation of surface coefficients for hypersonic flows over a hemisphere and a double wedge at Knudsen numbers of 0.277 and 0.020, respectively. Further, the paper mainly describes attempts to solve two main performance bottlenecks in making the code simulate at continuum-like conditions, first, the scalability of the code for more than 128 processors by reducing the communication and evenly balancing the computational load, and second, improve the algorithmic performance of the code by eliminating the expensive recursive tree traversal inherent in Octree based mesh structure. In order to resolve the first issue sophisticated graph-partitioners have been use, however, with no success. However, significant performance improvement has been obtained by linearizing the Octree using Morton-Z space filling curve. Based on these findings, a prediction is made of time required to run the main experimental case.
UR - http://www.scopus.com/inward/record.url?scp=85007569785&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85007569785&partnerID=8YFLogxK
M3 - Conference contribution
AN - SCOPUS:85007569785
SN - 9781624103933
T3 - 54th AIAA Aerospace Sciences Meeting
BT - 54th AIAA Aerospace Sciences Meeting
PB - American Institute of Aeronautics and Astronautics Inc, AIAA
T2 - 54th AIAA Aerospace Sciences Meeting, 2016
Y2 - 4 January 2016 through 8 January 2016
ER -