The Mathematical Institute, University of Oxford, Eprints Archive

The Influence of Unstructured Mesh Type on the Prediction of Convoluted Shear Layers

Salman, Hayder and McGuirk, James J. and Page, Gary J. and Moinier, Pierre (2000) The Influence of Unstructured Mesh Type on the Prediction of Convoluted Shear Layers. Technical Report. Unspecified. (Submitted)



Lobed mixers are used in gas turbine engines to enhance mixing between hot and cold streams and to reduce noise. Computational modelling of such systems has previously been carried out on structured meshes, although mesh generation difficulties have encouraged the use of unstructured tetrahedral meshes. However, the ability of numerical schemes to predict the mixing behaviour correctly on tetrahedral meshes has not been studied and is the subject of this work. Three different mesh types for the mixing region resolution have been studied: purely hexahedral, purely tetrahedral, and a mixed mesh combining hexahedra, tetrahedra and pyramids. Results are presented for the evolution of both a planar and a convoluted turbulent shear layer. In regions of high shear, misalignment of control volume faces has a major influence on spurious numerical spreading of the shear layer. For the tetrahedral mesh, there is an initial rapid mixing, followed by a reduction in mixing rate. The smoothing terms present are triggered by the combination of a high gradient across a control volume face and a velocity normal to that face; this occurs on the diagonal edges of tetrahedral meshes. The magnitude of the spurious smoothing is diminished by increasing the aspect ratio of the cells. For lobed mixer predictions, a mixed mesh with aligned high aspect ratio hexahedral elements in the shear layer region and pyramids and tetrahedra linking to the outer domain provides a good compromise between ease of mesh generation and quality of solution.

The authors would like to acknowledge funding from the Engineering and Physical Sciences Research Council (UK), Grant No. GR/L17863. Financial assistance and technical review monitoring was also provided by Rolls-Royce and DERA; the authors would like to thank in particular Dr Leigh Lapworth (RR), Dr Jens-Dominik Müller (Oxford University) and Prof Mike Giles (Oxford University).

Item Type:Technical Report (Technical Report)
Subjects:H - N > Numerical analysis
Research Groups:Numerical Analysis Group
ID Code:1273
Deposited By: Lotti Ekert
Deposited On:01 Jun 2011 08:21
Last Modified:29 May 2015 18:58

Repository Staff Only: item control page