Simulation-based design and optimization of Francis turbine runners by using multiple types of metamodels

In recent years, optimization started to become popular in several engineering disciplines such as aerospace, automotive and turbomachinery. Optimization is also a powerful tool in hydraulic turbine industry to find the best performance of turbines and their sub-elements. However, direct application of the optimization techniques in design of hydraulic turbines is impractical due to the requirement of performing computationally expensive analysis of turbines many times during optimization. Metamodels (or surrogate models) that can provide fast response predictions and mimic the behavior of nonlinear simulation models provide a remedy. In this study, simulation-based design of Francis type turbine runner is performed by following a metamodel-based optimization approach that uses multiple types of metamodels. A previously developed computational fluid dynamics-based methodology is integrated to the optimization process, and the results are compared to the results obtained from on-going computational fluid dynamics studies. The results show that, compared to the conventional methods such as computational fluid dynamics-based methods, metamodel-based optimization can shorten the design process time by a factor of 9.2. In addition, with the help of optimization, turbine performance is increased while cavitation on the turbine blades, which can be harmful for the turbine and reduce its lifespan, is reduced.