Publication

Matrix-based techniques for (flow-)transition studies

Song, W., 2019, [Groningen]: University of Groningen. 147 p.

Research output: ThesisThesis fully internal (DIV)Academic

Copy link to clipboard

Documents

  • Weiyan Song
In this thesis, numerical techniques for the computation of flow transitions was introduced and studied. The numerical experiments on a variety of two- and three- dimensional multi-physics problems show that continuation approach is a practical and efficient way to solve series of steady states as a function of parameters and to do bifurcation analysis. Starting with a proper initial guess, Newton’s method converges in a few steps. Since solving the linear systems arising from the discretization takes most of the computational work, efficiency is determined by how fast the linear systems can be solved. Our home-made preconditioner Hybrid Multilevel Linear Solver(HYMLS) can compute three-dimensional solutions at higher Reynolds numbers and shows its robustness both in the computation of solutions as well as eigenpairs, due to the iteration in the divergence-free space. To test the efficiency of linear solvers for non-flow problems, we studied a well-known reaction-diffusion system, i.e., the BVAM model of the Turing problem. The application to the Turing system not only proved our program’s ability in doing nonlinear bifurcation analysis efficiently but also provided insightful information on two- and three- dimensional pattern formation.
Original languageEnglish
QualificationDoctor of Philosophy
Awarding Institution
Supervisors/Advisors
  • Veldman, A.E.P., Supervisor, External person
  • Waalkens, Holger, Supervisor
  • Wubs, Fred, Co-supervisor
  • Verstappen, Roel, Assessment committee
  • Klawonn, A., Assessment committee, External person
  • Dijkstra, H.A, Assessment committee, External person
Award date25-Jan-2019
Place of Publication[Groningen]
Publisher
Print ISBNs978-94-034-1353-2
Electronic ISBNs978-94-034-1352-5
Publication statusPublished - 2019

Download statistics

No data available

ID: 74020523