4. Conclusions
We have presented a technique for generating highly stretched high-order boundary-layer meshes as required by current CFD solvers. The proposed technique is very effective, and modular in the sense that it can readily be combined with a number of existing coarse grid generation techniques in order to produce a mesh which is well-suited to boundary layer flows. Starting from a valid coarse prismatic boundary-layer mesh, our isoparametric approach permits the generation of a sequence of meshes with increased resolution with very little additional cost. This should prove valuable for mesh convergence studies at high Reynolds numbers. We have established requirements of validity for modal elemental shape functions, but the same arguments are also applicable to guarantee the validity of the mesh when using nodal shape functions. The main limitation of the technique, as presented here, is the requirement that the subdivision of the prismatic mesh should be accomplished without affecting the rest of the mesh. Although extending the method to other cases is not a difficult technical issue, it will require the use of transition elements such as pyramids. Whilst this case is beyond the scope of this article, in [28] we give a more detailed description as to how one may split different element types. This goes some way towards solving this problem, and may offer the chance to extend this technique with new functionality, such as introducing variable numbers of boundary layer elements around the profile of an aerofoil.