Large Eddy Simulation of Turbulence in Helically Coiled Pipes: Effects of Mesh Strategy

Abstract

In this thesis we investigate the difficulties associated with the prediction of the behavior of turbulent flow in straight and helically coiled pipes, due to their prevalence in real world scenarios. To simulate the turbulent flow, we used Computational Fluid Dynamics with Large Eddy Simulation turbulence models. Due to the requirements on the mesh from the numerical code and turbulence models, we investigated which strategy for meshing of pipes provided the most accurate result, as well as the parameters of importance in mesh creation. To reduce the computational time needed for these simulations we compared the accuracy between coarse meshes with wall models accounting for small-scale phenomena, to meshes with a higher resolution where all scales of motion are resolved. Furthermore, we studied the effect the choice of turbulence model had on the development of turbulence in pipes. The results were compared to data from literature to examine the difference in accuracy between the techniques. The results from the aforementioned investigations were extrapolated to helically coiled pipes, where three different configurations were used to identify characteristics of turbulent flow within such geometries. We used fully turbulent flow from the results of the pipe simulations to insert accurate data to the inlet of the coils. The results showed a shift of the location of highest velocity towards the outer-wall, with a low velocity region appear- ing at the inner-wall. A spanwise pressure gradient was observed, inducing a centrifugal force on the flow, which caused secondary flow phenomena to occur. We found that the central meshing parameter for straight- and helically coiled pipes was the skewness of the cells, and that a mesh with uniform size and shape is important towards the walls. Using wall models on a coarse mesh, gave satisfying results for a straight pipe, but may cause inaccuracies in more complex flow scenarios in the case of a helical coil. Moreover, the requirement of grid resolution from a straight- to helically coiled pipe seems to be non-transferable due to the increased complexity of the distribution and diffusion of turbulent kinetic energy.