Optimization of end-wall fence in turbine based on response surface methodology and genetic algorithm

Abstract

Secondary flow loss accounts for a large proportion of the internal flow loss in turbine stages. The use of end-wall fences can effectively reduce secondary flow loss. In this study, the White cascade was taken as the research object, and the position of the end-wall fence was parameterized. Based on the response surface method, the mapping relationship between the fence position and the isentropic expansion efficiency was obtained, and a surrogate model was constructed. Finally, single-objective and multiobjective optimizations were carried out using genetic algorithms to obtain the optimal fence position parameters. The results showed that the existence of the fence can not only reduce secondary flow but also effectively reduce shock losses when the fence is located at the rear of the passage. Therefore, the optimization effect of the blade position is best when it is located at the end of the passage. When the flow deviates significantly from the design condition, the end-wall fence can significantly reduce the low-speed region on the pressure side of the fence. Therefore, the fence can play a greater role in low-load conditions with significant deviations from the design condition. After optimizing the design conditions, when H is 128.6 mm, V is 66.93 mm, and A is 56.48°, the isentropic expansion efficiency is the highest, reaching 96.160%. After optimizing the multi-inlet angle at low load conditions, when θ is 56.07°, R is 129.7 mm, and A is 64.36°, the comprehensive optimization result is the best, and the isentropic expansion efficiencies at inlet angles of 0°, 10°, and 45° are 96.098%, 96.050%, and 93.930%, respectively. The research results can provide a reference for the design of turbine flow.