Lab Home | Phone | Search
Center for Nonlinear Studies  Center for Nonlinear Studies
 Home 
 People 
 Current 
 Executive Committee 
 Postdocs 
 Visitors 
 Students 
 Research 
 Publications 
 Conferences 
 Workshops 
 Sponsorship 
 Talks 
 Seminars 
 Postdoc Seminars Archive 
 Quantum Lunch 
 Quantum Lunch Archive 
 P/T Colloquia 
 Archive 
 Ulam Scholar 
 
 Postdoc Nominations 
 Student Requests 
 Student Program 
 Visitor Requests 
 Description 
 Past Visitors 
 Services 
 General 
 
 History of CNLS 
 
 Maps, Directions 
 CNLS Office 
 T-Division 
 LANL 
 
Monday, May 09, 2011
10:00 AM - 11:00 AM
CNLS Conference Room (TA-3, Bldg 1690)

Seminar

Two new developments for accelerating the simulation of incompressible multiphase flow problems. Joint work with: Yaohong Wang, Svetlana Simakhina, Austen Duffy, and Alan Kuhnle.

Prof. Mark Sussman
Department of Mathematics, Florida State University

The numerical simulation of incompressible multiphase flow can be beneficial in predicting flows around a swimming whale, atomization of fuel in an internal combustion engine, and flows in microfluidic devices. We have made two new developments which significantly accelerate the simulation of these flows: (1) We have developed a hybrid level set and volume constraint method (LSVC) for representing dynamic, complex, interfaces with zero volume fluctuation, and (2) we have extended the multigrid preconditioned conjugate gradient method (MGPCG) so that one can solve elliptic equations with singular source terms and discontinuous coefficients on a hierarchy of grids (block structured adaptive mesh refinement) with a guarantee for convergence. The LSVC method enables one to easily simulate multiphase problems on unstructured meshes, or on meshes in which a gas/liquid interface is not wholly contained on the finest adaptive level. Also, for multiphase problems with an embedded solid undergoing solid-body motion or perhaps an embedded deforming solid, it is trivial to conserve liquid/gas volume with the LSVC method without complicated refluxing procedures in cut cells containing liquid,solid, and gas. The new MGPCG method for adaptive grids has the same advantages over MG (multigrid) or ICPCG (incomplete Cholesky preconditioned conjugate gradient method) that the original MGPCG method (developed by Tatebe) had on a single grid. We prove that the new adaptive MGPCG method is guaranteed to converge regardless of the coefficients or source terms. We show by example that the adaptive MGPCG method is 3 times faster than PCG or MG on adaptive grids. The adaptive MGPCG method becomes especially more efficient as one increases the number of levels of adaptivity, or if the air/water interface is not wholly contained on the finest adaptive level. The new adaptive MGPCG solver is so fast, that the pressure projection step is no longer the step that consumes the most CPU time insolving the Navier-Stokes equations for inncompressible multi-phase flow.

Host: Mikhail Shashkov. shashkov@lanl.gov, 667-4400