Numerical Simulation of Non-Equilibrium Condensation in Supercritical CO2 Compressors

Abstract

There is increasing interest in supercritical CO2 processes, such as Carbon Capture and Storage, and electric power production, which require compressors to pressurize CO2 above the critical point. For supercritical compressor operation close to the critical point there is a concern that the working fluid could cross into the subcritical regime which could lead to issues with compressor performance if condensation was to occur in regions where the fluid dropped below the saturation point. Presently, the question of whether there is sufficient residence time at subcritical conditions for condensation onset in supercritical CO2 compressors is an unresolved issue. A methodology is presented towards providing a validated simulation capability for predicting condensation in supercritical CO2 compressors. The modeling framework involves the solution of a discrete droplet phase coupled to the continuum gas phase to track droplet nucleation and growth. The model is implemented in the CRUNCH CFD® Computational Fluid Dynamics code that has been extensively validated for simulation at near critical conditions with a real fluid framework for accurate predictions of trans-critical CO2 processes. Results of predictions using classical nucleation theory to model homogeneous nucleation of condensation sites in supersaturated vapor regions are presented. A non-equilibrium phase-change model is applied to predict condensation on the nuclei which grow in a dispersed-phase droplet framework. Model validation is provided against experimental data for condensation of supercritical CO2 in a De Laval nozzle including the Wilson line location. The model is then applied for prediction of condensation in the compressor of the Sandia test loop at mildly supercritical inlet conditions. The results suggest that there is sufficient residence time at the conditions analyzed to form localized nucleation sites, however, droplets are expected to be short lived as the model predicts they will rapidly vaporize.