High Efficiency Computation of High Reynolds Number Flows via Anisotropic Adaptive Mesh Refinement

Award Information
Agency:
Department of Defense
Branch
Navy
Amount:
$79,962.00
Award Year:
2013
Program:
STTR
Phase:
Phase I
Contract:
N00014-13-P-1198
Agency Tracking Number:
N13A-009-0112
Solicitation Year:
2013
Solicitation Topic Code:
N13A-T009
Solicitation Number:
2013.A
Small Business Information
CMSoft, Inc.
566 Glenbrook Drive, Palo Alto, CA, 94306
Hubzone Owned:
N
Socially and Economically Disadvantaged:
N
Woman Owned:
N
Duns:
172390481
Principal Investigator:
Goeric Daeninck
Research Scientist
(650) 614-1101
gdaeninck@cmsoftinc.com
Business Contact:
Francoise Farhat
President
(650) 898-9585
ffarhat@cmsoftinc.com
Research Institution:
University of Washington
Ulrich Hetmaniuk
Dept. of Applied Mathematics
Box 352420
Seattle, WA, 98195-2420
(206) 484-3889
Nonprofit college or university
Abstract
This STTR Phase I project aims to design, implement, and demonstrate a rigorous, practical, fast, and re-usable anisotropic mesh adaptation software module for enabling the efficient computation of high Reynolds number flows in large computational domains. To this effect, it focuses on developing: (a) a set of portable and cache-friendly dynamic data structures that ease the implementation in a hydrodynamic code of fast mesh refinement and coarsening operations, (b) a set of algorithms and corresponding numerical software for performing anisotropic mesh adaptation in general, and isotropic adaptation in particular, (c) a robust, order of accuracy preserving, and computationally efficient treatment of the problem of non-conforming mesh interfaces resulting from mesh adaptation, and (d) optimization strategies for maximizing the efficiency of adaptive implicit flow computations. Because mesh adaptation inherently destroys load balance, this STTR project also focuses on the development of a measurement-based approach for assessing workload unbalance, and a set of smart and portable algorithms for transferring data across cores or processors to evenly redistribute the computational workload in order to achieve maximum scalability on a given massively parallel processor.

* information listed above is at the time of submission.

Agency Micro-sites

US Flag An Official Website of the United States Government