Assessment and improvement of conjugate heat transfer CFD code models of wall condensation from a gas–vapour mixture in a vertical channel

Abstract

Predicting wall condensation in the presence of a non-condensable gas is of vital importance in the design of processes for refrigeration systems and in modelling passive containment cooling systems of nuclear reactors. The non-condensable gas causes a significant reduction in condensation rate and heat transfer. No consensus exists, however, on which CFD modelling approach should be used for this application.

To address this, fully conjugate heat transfer models are developed and assessed for condensation from a downward turbulent steam–air mixture flow in a primary vertical channel separated by a metal plate from a turbulent water counterflow in a secondary channel, and compared in a benchmark study. Two wall condensation modelling approaches are used in each of the commercial codes STAR-CCM+ and Fluent: (1) a sub-grid film model, which solves simplified versions of the liquid film governing equations; and (2) a user-defined wall-adjacent volumetric source term model, which neglects the fluid film. Grid-independent results for the CONAN separate effects test experiment were obtained.

A new improvement to the energy source term in the second modelling approach is presented that prevents non-physical predictions for the conjugate analysis. The Fluent custom wall-adjacent source term model reaches grid-based convergence fastest. Similar computational effort is required by all models. Model accuracies varied case-by-case, with total condensation rate maximum errors ranging from 4 to 28%. Analysis of the non-dimensional boundary layers indicate $y^{+}\lesssim 1$ is required to prevent excessive error. New results for the local heat transfer coefficient in the cooling channel are also presented.