Due to the increase of computing efficiency and power, full-resolution two-phase flow simulations have become a practical research tool for model development and analysis of reactor flows. The expansion of state-of-the-art high performance computing (HPC) facilities allows for the use of direct numerical simulation (DNS) coupled with Interface Tracking Methods (ITM) to perform full resolution simulations. Given adequate spatial and temporal resolution, DNS can resolve all relevant turbulent scales, allowing for the extraction of high quality and detailed turbulent and two-phase flow numerical data for use in model development. While larger scale bubbly flow DNS are becoming ever more affordable, it is still computationally expensive due to the requirements of the spatial discretization. This presents the largest obstacle for future applications of DNS.
For this reason, mesh adaptation techniques are sought after to reduce the computational expense of bubbly flow simulations in complex geometries. By fully resolving only the areas of specific interest, the computational costs of DNS can be reduced. Grid refinement can be based on the location of the interface between the two phases, area of greatest turbulent intensity, averaged bulk fluid velocity data, or the prediction of bubble movement. Coupled with an advanced bubble tracking algorithm, the path of individual bubbles moving through the computational domain can be predicted, and the computational mesh refined within the path area. This refinement can create tracks of greater resolution for the bubbles to move through in the domain, while keeping the bulk resolution of the mesh coarser. Through these means, the overall cost of the simulation is reduced, while high quality numerical data is still obtainable. This work outlines the enhancement of existing mesh adaptation algorithms to implement the bubble tracking refinement, and its practical applications to full resolution two-phase flow simulations.