CFD Based Stochastic Optimization of Pelton Turbine Bucket in Stationery Condition

Computational fluid dynamics (CFD) and stochastic optimization are both highly computationally expensive processes. These processes may not produce the same unique result every time and demand large computing resources. The outcomes are determined as the final if the results repeat themselves for some predefined number of iterations causing convergence. Due to this expensive and non-deterministic nature, research on CFD optimization using stochastic optimization method such as Genetic Algorithm has been limited. This paper presents a noble method in which the CFD codes can be used together with genetic algorithm to optimize the shape of a responsive surface such as a Pelton turbine bucket. An existing Pelton bucket's model has been acquired and a set of random surfaces have been created as the initial population to optimize the shape of the bucket in stationery condition. The results show that an increase in efficiency by 13.21% to the normalized efficiency of existing design can be obtained by incorporating the changes suggested.