Aerodynamic characteristics and ventilation losses of turbine in a compressed air energy storage system

Abstract

Compressed Air Energy Storage (CAES) systems frequently operate turbines under part-load or low-load conditions, resulting in substantial energy losses. This study investigates the evolution of flow fields and loss distributions in air turbines operating across 70 operating conditions, ranging from optimal to low-load regimes, using validated numerical methods. Results reveal a critical transition from turbine to compressor mode when the flow coefficient drops below 0.12, marked by the development of uniform spanwise blockage in stator passages and rotor root separation bubbles that redirect flow toward the tip. Aerodynamic and entropy analyses indicate that rotor-airflow interactions are the primary contributors to losses, accounting for about 90 %–95 % of total entropy production. A one-dimensional ventilation model, developed through the integration of Random Forest regression and dimensional analysis, effectively predicts turbine performance across varying conditions. The model highlights airflow velocity (26.0 % contribution) and density (31.6 %) as key predictors, achieving errors of less than 15.0 % for most operating scenarios. By explicitly incorporating flow coefficient effects, the model surpasses classical prediction approaches, offering valuable insights for optimizing turbine performance in low-load CAES operations.