Development of a Multi-layer CLSVOF Method for Numerical Simulation of Dense Bubbly Flows

  • Fukuda, Takanari (Japan Atomic Energy Agency)
  • Yamashita, Susumu (Japan Atomic Energy Agency)
  • Uesawa, Shinichiro (Japan Atomic Energy Agency)

Please login to view abstract download link

A coupled level set and volume-of-fluid (CLSVOF) method [1] is widely used for gas–liquid two-phase flow simulations owing to its inherent mass conservation and accurate surface tension modeling. However, conventional CLSVOF formulations employ a single level set and VOF field for all bubbles, which often leads to unphysical numerical coalescence when interfaces come into close proximity. In addition, the conventional level set extension outside interfaces is known to induce spurious attractive forces between neighboring bubbles. In this study, we propose a multi-layer CLSVOF method that integrates the concept of ordered active parameter tracking (OAPT) [2] into the CLSVOF framework. Bubbles are distributed into a limited number of memory layers such that no two nearby bubbles reside in the same layer. Bubble labeling is performed using connected-component labeling, and potential contacts are detected based on bounding-box overlap. A greedy coloring algorithm is employed to assign bubbles to memory layers efficiently, enabling the number of layers to remain small and independent of the total bubble count. All CLSVOF procedures, including VOF advection, level set reconstruction, and surface tension force evaluation, are executed on a layer-by-layer basis, after which the layered VOF fields are aggregated into a single physical volume fraction field. This strategy suppresses unphysical bubble merging and mitigates spurious surface tension forces while preserving the robustness of the original CLSVOF method. The proposed method is validated through simulations of a bubbly flow experiment that we performed. Numerical results demonstrate stable tracking of a large number of closely spaced bubbles without numerical coalescence. Moreover, the predicted bubble size distributions and radial bubble count distributions show reasonable agreement with experimental measurements, which have long been difficult to capture using conventional methods, thereby confirming the validity of the proposed method for qualitatively simulating bubbly flows.