Modelagem Matemática e Computacional da Erosão em Escoamentos Multifásicos com Malhas Dinâmicas

Abstract

In several industrial applications, the phenomenon of erosion is very important and modify the eroded surface, which can, for example, reduce the useful life of pipes and equipment. As the experimental study can be complex or even unfeasible due to its high cost, numerical simulations are an important tool to erosion prediction and avoid losses and/or improve equipment performance. However, in many cases the topology change due to erosion is relevant and requires a computational mesh that reproduces the movement of the eroded surface, requiring dynamic meshes. This type of mesh presents a higher computational cost and the available codes do not have a robust methodology that includes the transient resolution of the flow followed by the movement of the mesh at each time step, while obtaining good results and easy convergence. Therefore, a numerical methodology was developed to meet this need and be applied to any conditions. A difficulty in simulations is usually the generation of negative volumes due to a decrease in mesh quality as mesh deformation increases. In addition, it was observed that erosion results may present non-physical noise due to the statistics of impacts on the cells of the computational mesh. Thus, after studying the effect of several variables and the methodology validation, a smoothing method was implemented in the UNSCYFL3D code in order to smooth the gradient of the eroded surface. It was observed that the mesh quality improved significantly and for the case of an impinging erosive jet the simulation time decreased by up to 58% due to the possibility of using 8 times fewer parcels while maintaining the quality of the results obtained. In addition, the code used in this work allows the use of both static and dynamic meshes, choosing to solve the flow at each time step or use a converged flow field to calculate only the solution of the particles and, consequently, of erosion, considerably reducing the computational cost. When comparing the methodology for static or dynamic meshes used in this study with others available in the literature, it was observed that the results are much more satisfactory, showing differences of 15% in relation to the maximum depth of experimental erosion, while this value reaches 400% using other methodologies. In addition, it is the only methodology with robustness for application to any geometry with the possibility of studying its modification and its influence on erosion combined with obtaining good results.