Abstract

A new mesh adaptation procedure is developed that is well suited to challenging turbomachinery CFD simulations. The adaptation procedure creates a mesh that robustly conforms to the underlying geometry, that respects user-defined locally anisotropic boundary layer mesh refinement for efficiency, and that is driven by an adaptation sensor that accurately resolves large as well as subtle secondary flow features. The method seeks to control numerical error in the discrete solution by minimizing the truncation error based on the CFD solver discretization. An adaptation sensor is developed for a node-centred finite volume CFD solver, including control over the rate at which the mesh size increases with each adaptation step. The adaptation sensor is used to define a target size field for an updated discrete mesh conforming to the underlying CAD geometry. The mesh adaptation procedure is demonstrated across the speedline of a pump for flow ranging from 80% to 120% of the design flow rate. Both the pump impeller and volute are modeled, using a steady-state frame change interface between the rotating and stationary components. Operating point independent adaptation is performed, creating adapted meshes unique to the local flow conditions and features for each operating condition. Adaptation is performed at each operating point until a target average discretisation truncation error is achieved, ensuring that the numerical error is consistently reduced for all operating points across the speedline. In addition to the truncation error, the impeller head, power and volute losses are evaluated across the speedline, as a function of the mesh adaptation cycle. It is seen that off-design operating points require more mesh adaptation relative to the design point flow. Computational effort of the adaptation procedure is summarised and compared to the effort that would be required to achieve a similar level of truncation error by a manual mesh refinement approach.