Modeling polydispersed droplets in non-equilibrium condensing CO2 flows through turbine cascades using moment-based methods for efficient energy utilization

• Numerical simulations of non-equilibrium condensation in carbon dioxide and steam. • Droplet size distributions are accurately modeled using moments-based methods. • Larger expansion rate increases the dispersion of the droplet sizes. • Higher isentropic efficiency in turbines with carbon dioxide versus steam. • Wetness losses: 1.42–1.91% for carbon dioxide and 3.58–4.01% for steam. The efficiency of a power cycle is significantly influenced by the condensation process and the size distribution of the droplets, as these factors directly affect heat transfer rates and overall energy conversion efficiency. Accurate numerical modeling of thermodynamic non-equilibrium condensation is crucial for predicting droplet nucleation and growth in carbon dioxide flows. This study applies moment-based polydispersed droplet models, such as the quadrature method of moments, to account for the droplet size spectrum. Additionally, discrete methods based on a predefined shape of the polydispersed droplet size spectrum are utilized and evaluated to reduce numerical complexity. Our results demonstrate that polydispersed models provide higher accuracy compared to monodispersed models when simulating steam and supercritical carbon dioxide flows in converging–diverging nozzles. In turbine cascades, the choice of model significantly influences the predicted mean droplet size and efficiency. Isentropic efficiency evaluations reveal higher values for carbon dioxide compared to steam, with efficiencies of 94.6–95.8% versus 91.8–92.9% for wet cases and 96.5–97.2% versus 95.8–96.5% for dry cases. Notably, spontaneous droplet nucleation slightly affects the efficiency of the blade cascade in carbon dioxide flows, whereas steam flows experience efficiency reductions between 3.58% and 4.01%. This work advances the understanding of droplet condensation processes in turbine cascades, highlighting the superior performance of polydispersed models and providing quantitative insights into their impact on efficiency.