Adaptive Filtering for Hybrid DG-FV methods for Turbulent flows on Unstructured Meshes

  • Yuan, Dean (Cranfield University)
  • Tsoutsanis, Panagiotis (Cranfield University)
  • Nogueira, Xesus (Universidade da Coruña)

Please login to view abstract download link

In this work we develop an adaptive filtering method using the automatic dissipation/dispersion algorithm (ADDA) previously developed in [1,2] for high-order hybrid Discontinuous Galerkin/finite-volume methods on unstructured meshes. By switching to a high-order FV method, we can alleviate the instability issues characteristic of traditional DG methods. The ADDA algorithm is tuned using the linear weight values of the central stencils of the CWENO/CWENOZ type schemes [3], using the energy ratio (as defined in [1,2]) as the determining factor. Depending on the presence of the aliasing errors, or under-resolved frequencies in the DG framework the filtering can be applied either in the DG or FV framework simultaneously to ensure that the right amount of dissipation is present to produce physicallyl meaningful results. The method is implemented in the open-source CFD solver UCNS3D [4] ( and a series of challenging test problems have been employed, such as the subsonic and supersonic Taylor-Green vortex flow problem. The work done may play a key role in the development of high-order, robust DG/FV schemes applied to underresolved DNS, and implicit or explicit LES settings for real-world, high-Reynolds flows involving complicated geometries