Fully Automatic Mesh Generation Strategy for CFD Simulations Integrated in an Optimisation Workflow

Carlo Zelioli, Nicola Rossi, Maurizio Barbato

University of Applied Sciences of Southern Switzerland

 

When dealing with Computational Fluid Dynamic (CFD) simulations one is always challenged with the time-consuming task of mesh generation for the computational domain. In cases where few simulations are required it is admissible to invest a certain amount of time in order to generate the suitable computational grid by hand using available mesh generation software products. On the other hand, if for example the goal is to perform an optimization based on CFD simulations, where the number of configurations to analyse significantly increases, it is preferable, if not unavoidable, to adopt a fully automatic strategy to generate the mesh.
The problem, from where the strategy presented in this paper was derived, is related to Compact Heat Exchangers (CHE). The computational domain for simulations is the smallest periodical part of a channel of a plate-fin CHE, which comprises the solid zone of fins and the fluid zone. Several configurations of fins have to be analysed and this leads to a large amount of time dedicated to mesh generation. Therefore a strategy was developed, which is based on ANSYS Meshing and the use of a macro that is created by means of Matlab scripts. Basically, the macro handles Named Selections (NS) built with worksheets which are used to set up mesh controls for the mesh generation and define zones names necessary to assign materials and set boundary conditions in the CFD solver. In addition, the macro manages certain parameters of the mesh controls in order to obtain the desired mesh refinement. All the inputs for the macro are provided by the Matlab scripts, which are prepared in accordance with the chosen computational domain.
The exploitation of this strategy allowed to generate hexahedral meshes of good quality and suitable for the physical modelling of the CFD simulations. Furthermore, this strategy can be adapted to other mesh generation approaches and to computational domains whose topology changes to a certain extent.

Back to Program