ID 336

Benefits of AMR for atomization calculations

 

Abstract:

Adaptive mesh refinement (AMR) has been introduced as an attractive means of significantly improving computational efficiency for a variety of two-phase problems. In the current study, the benefits of AMR are investigated for the case of liquid jet atomization. The evaluation consists of a systematic analysis of results from the interDymFoam (AMR octree) and interFoam (static octree) codes, both of which form part of the family of solvers distributed within the opensource OpenFoam C++ Toolbox. The two-phase flow treatment is based on an algebraic VoF methodology. As a preliminary set of exercises, cases for pure advection, stationary wave dynamics, and Rayleigh-Plateau breakup of a cylindrical liquid element are considered. The results from these exercises confirm the expected trend of higher numerical efficiency in AMR, while still retaining essentially the same level of accuracy as the static mesh solutions. However, for the liquid jet atomization, the behavior is a bit more complicated. First, at lower levels of injection velocity, we observe a similar trend as the preliminary exercises. At higher injection velocities, due to a noticeable increase in interfacial area density, the computational burden of AMR for locating interfacial regions and performing refinement/coarsening increases drastically, potentially offsetting the benefits of this approach. In fact, at much higher velocities, for instance, those pertaining to Diesel injection, the results suggest that a static mesh would provide better computational efficiency. However, this conclusion depends on the target lowest level of numerical resolution, Δxmin. The current work shows how the efficiency of AMR suffers from increasing interfacial area density, and how this can be alleviated via a decrease in Δxmin. Various test cases are presented to illustrate this effect.