Adaptive forcing distance in an immersed boundary method for internal flow simulation at high Reynolds numbers

Future advanced aero-engine features highly complex internal flow mechanisms, such as small gap blade row interactions and coupled fluid-structure interactions. Simulating these is challenging for convectional body-fitted computation fluid dynamics methods. As a first step towards this ultimate goal, this paper presents applications of an immersed boundary (IB) method for internal-flow problems at high Reynolds numbers. A hybrid mesh strategy, with a single-block and structured mesh to fit the physical domain and an IB method to model the internal walls, is adopted to the internal-flow problems. To achieve this, an adaptive forcing distance (AFD) technique is proposed for determining the forcing point locations when large-aspect-ratio cells are involved near the walls. This extends the IB method to curvilinear grids so that the domain boundaries can be resolved by a body-fitted mesh, which significantly reduces the total cell number needed to resolve the boundaries compared to the usage of a Cartesian grid, and thus improve the computational efficiency. The present methodologies are firstly validated on a Cartesian grid through a 2D NACA0012 test case. Subsequently, the methods are tested for two internal flow problems with increasing complexity: a 3D subsonic compressor cascade and a well-known transonic rotor, NASA Rotor 37. These case studies prove that the present method can achieve the same overall accuracy for the pressure distribution with the traditional body-fitted simulations at the attached-flow regions, and further studies are still necessary to improve the accuracy of predicting the friction coefficient and separated flows.