Mesh generation of curvilinear polygons for the high-order virtual element method (VEM)

We present a proof-of-concept methodology for generating curvilinear polygonal meshes suitable for high-order discretisations by the Virtual Element Method (VEM). A VEM discretisation requires the definition of a set of boundary and internal points used to define basis functions and compute integrals of polynomials. The procedure to locate these points on the boundary borrows ideas from previous work on a posteriori high-order mesh generation in which the geometrical inquiries to a B-rep model of the computational domain are performed via an interface to CAD libraries. Here we describe the steps of the procedure that transforms a straight-sided polygonal mesh, generated using third-party software, into a curvilinear boundary-conforming mesh. We discuss criteria for ensuring and verifying the validity of the mesh. Using an elliptic partial differential equation with Dirichlet boundary conditions as a model problem, we show that VEM discretisations on such meshes achieve the expected rates of convergence as the mesh resolution is increased. This is followed by an illustrative application of the method to the generation of a curvilinear polygonal mesh for an aerofoil geometry. We discuss polygonal curvilinear mesh quality and its enhancement, and use the motion of a cell vertex to appraise three elemental quality metrics, namely convexity, regularity and isotropy, and highlight some of the difficulties associated in their use for mesh quality optimisation. A derivative-free optimisation method is utilised to enhance curvilinear polygonal meshes by maximising a suitable measure of mesh quality. We propose such measure as a combination of the three quality metrics and apply it to optimise a distorted initial mesh for a ring geometry. We show that a suitable version of the convexity metric is effective in untangling invalid meshes. The VEM solution of a model elliptic equation is obtained for a ring geometry where a distorted and an optimised mesh show low errors, indicating that the VEM is robust and relatively insensitive to mesh distortion, and a reduction of the error in the optimised mesh. Finally, we use a more complex geometry, a computational domain for an aerofoil, as a benchmark to further illustrate the ability of the convexity metric to untangle meshes, and also to assess the suitability of two quality measures as optimisation targets to improve the overall quality of curvilinear polygonal meshes.