Numerical prediction of maldistribution in a series of T-junctions

Abstract

This work assesses the possibility/feasibility of utilizing the CFD to predict water-air separation in a mini-manifold at ambient conditions. The manifold consists of a header with three reduced T-junctions. A combination of the manifold size and the operating parameters yields a computationally demanding surface tension dominant flow. A computational model of separation is based on the interface tracking technique and the FVM framework. The results are validated against the experiment. Findings show that the separation can be successfully predicted, even under such conditions. Analysis of computational cost revealed that the long computational time is the only limiting factor for regular CFD use in such applications. The paper's novelty is in the successful prediction of phase separation. Such an approach requires flow pattern prediction capabilities in the straight parts of the manifold and physically sound interface split prediction at the T-junctions that have not been explored previously in such a way.