The near-wall non-overlapping domain decomposition (NDD) method proved to be very efficient for turbulence modelling. In this thesis, several variants of the NDD method are implemented into the open source software OpenFOAM for the first time to accelerate the simulation of near-wall turbulence. The main idea behind the NDD algorithm is to split the computational domain into the inner and outer region. The two regions are coupled with the interface boundary. The NDD method can be divided into the approximate NDD (ANDD) method and the exact NDD (ENDD) method. In the ANDD, the inner region is simplified into a thin-layer form which leads to a faster convergence by avoiding the computation of the time-consuming near-wall regions. Although it has been successfully applied to the Reynolds-averaged Navier-Stokes equations, the algorithm has some limitations regarding to accuracy. For the ENDD, on the contrary, the inner region and outer region are solved without any simplification. In the thesis, both ANDD and ENDD methods are implemented in an in-house code and the OpenFOAM framework for the low-Reynolds-number k-epsilon models to simulate channel flow cases. The results demonstrate that the solution of ENDD method is as accurate as the benchmark solution and almost insensitive to the interface boundary position. The results from the in-house code demonstrate the ENDD method can reduce the original computational time by one order of magnitude. In OpenFOAM, the computational time of ENDD is longer than that of one-block case. Several techniques are therefore used to accelerate the convergence of ENDD. As a variant of ANDD, the slip boundary condition is formulated at the wall of the computational domain. It can be solved with a coarse mesh. The algorithm of slip boundary condition is implemented into OpenFOAM and then tested by low-Reynolds-number and high-Reynolds-number k-epsilon models. The result shows the accuracy and computational time of the method is almost the same as the ANDD method.
Some acceleration techniques for near-wall turbulence modelling implemented in unstructured codes
Wang, C. (Author). 31 Dec 2023
Student thesis: Phd