Radial basis function based meshless methods for fluid flow problems
Radial basis function based meshless methods for fluid flow problems
This thesis is concerned with the development of meshless methods using radial basis functions for solving fluid flow problems. The advantage of meshless methods over traditional mesh-based methods is that they make use of a scattered set of collocation points in the physical domain and no connectivity information is required. An important objective of the present research is to develop novel meshless methods for unsteady flow problems. Symmetric/unsymmetric radial basis function collocation schemes are proposed for solving an unsteady convection-diffusion equation for various Peclet numbers. Both global and compactly supported radial basis functions are used and the convergence behaviours of various radial basis functions are studied. The performance of the presented schemes is shown by using both uniform as well as scattered distribution of points. Numerical results suggestthat these schemes are capable of obtaining accurate results for low and medium Peclet numbers. Next, two directions have been explored in this thesis for using radial basis functions to solve large scale problems encountered in fluid flow problems. They are namely, domain decomposition schemesand radial basis functions in finite difference mode. These schemes are shown to be computationally efficient and also aid in circumventing the ill-conditioning problem. The performance of both schemes are evaluated by solving the unsteady convection-diffusion problem. The last part of this thesis isconcerned with the solution of the 2D Navier-Stokes equations. Meshless methods based on radial basis collocation and scattered node finite difference schemes are formulated for solving steady and unsteady incompressible Navier-Stokes equations. A novel ghost node strategy is proposed for incor-porating the no-slip boundary conditions. Optimisation strategies based on residual error objective and leave-one-out statistical criterion are proposed to evaluate the optimal shape parameter value in case of the multiquadric RBF for collocation and scattered finite difference approaches respectively.Standard benchmark problems like the driven cavity flows in square and rectangular domains and backward facing step flow problem are solved to study the performance of the developed schemes. Finally, a higher order radial basis function based scattered node finite difference method is proposed for solving the incompressible Navier-Stokes equations.
Chinchapatnam, Phani P.
030a7425-a25f-4e81-905e-85759433a29a
September 2006
Chinchapatnam, Phani P.
030a7425-a25f-4e81-905e-85759433a29a
Djidjeli, K.
94ac4002-4170-495b-a443-74fde3b92998
Nair, P.B.
da7138d7-da7f-45af-887b-acc1d0e77a6f
Chinchapatnam, Phani P.
(2006)
Radial basis function based meshless methods for fluid flow problems.
University of Southampton, School of Engineering Sciences, Doctoral Thesis, 160pp.
Record type:
Thesis
(Doctoral)
Abstract
This thesis is concerned with the development of meshless methods using radial basis functions for solving fluid flow problems. The advantage of meshless methods over traditional mesh-based methods is that they make use of a scattered set of collocation points in the physical domain and no connectivity information is required. An important objective of the present research is to develop novel meshless methods for unsteady flow problems. Symmetric/unsymmetric radial basis function collocation schemes are proposed for solving an unsteady convection-diffusion equation for various Peclet numbers. Both global and compactly supported radial basis functions are used and the convergence behaviours of various radial basis functions are studied. The performance of the presented schemes is shown by using both uniform as well as scattered distribution of points. Numerical results suggestthat these schemes are capable of obtaining accurate results for low and medium Peclet numbers. Next, two directions have been explored in this thesis for using radial basis functions to solve large scale problems encountered in fluid flow problems. They are namely, domain decomposition schemesand radial basis functions in finite difference mode. These schemes are shown to be computationally efficient and also aid in circumventing the ill-conditioning problem. The performance of both schemes are evaluated by solving the unsteady convection-diffusion problem. The last part of this thesis isconcerned with the solution of the 2D Navier-Stokes equations. Meshless methods based on radial basis collocation and scattered node finite difference schemes are formulated for solving steady and unsteady incompressible Navier-Stokes equations. A novel ghost node strategy is proposed for incor-porating the no-slip boundary conditions. Optimisation strategies based on residual error objective and leave-one-out statistical criterion are proposed to evaluate the optimal shape parameter value in case of the multiquadric RBF for collocation and scattered finite difference approaches respectively.Standard benchmark problems like the driven cavity flows in square and rectangular domains and backward facing step flow problem are solved to study the performance of the developed schemes. Finally, a higher order radial basis function based scattered node finite difference method is proposed for solving the incompressible Navier-Stokes equations.
Text
Chinchapatnam_thesis.pdf
- Other
More information
Published date: September 2006
Organisations:
University of Southampton
Identifiers
Local EPrints ID: 72034
URI: http://eprints.soton.ac.uk/id/eprint/72034
PURE UUID: 08a19af4-18cb-432e-8b91-167d69e45cdc
Catalogue record
Date deposited: 15 Jan 2010
Last modified: 13 Mar 2024 20:56
Export record
Contributors
Author:
Phani P. Chinchapatnam
Thesis advisor:
P.B. Nair
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics