Optimisation and efficiency improvement of pelton hydro turbine using computational fluid dynamics and experimental testing

• Main Author: Zidonis, Audrius
• Other Authors: Aggidis, George
• Published: Lancaster University 2015
• Subjects: 333.79
• Online Access: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.677256

The aim of this PhD research was to develop a generic optimisation method for Pelton turbine runners and assess the key design parameters using Computational Fluid Dynamics (CFD). This optimisation was applied on a modern commercial Pelton turbine runner taken as a base design. The design together with the field knowledge and experience was provided by a turbine manufacturing company Gilbert Gilkes and Gordon Ltd. to establish the state of the art starting point. The work described in this thesis can be divided into three main parts: 1) developing of numerical modelling technique by combining current commercial CFD models with engineering assumptions to produce results of acceptable accuracy within reasonable timescales and verifying this technique, 2) optimising the Pelton runner provided by Gilkes to produce better efficiency and simplify its design, 3) manufacturing of original and optimised design model runners and experimentally testing them. The numerical techniques created during part 1) included many numerical and physical assumptions to simplify the problem. This was necessary because accurate modelling of impulse turbines (Pelton in this case) that include complex phenomena like free surface flow, multi fluid interaction, rotating frame of reference and unsteady time dependent flow is a challenge from a computational cost point of view. These simplifications included the usage of symmetry plane and modelling of only two consecutive buckets to reduce the size of the computational domain. Casing and any backsplash effects were not modelled at all expecting that a runner with higher hydraulic efficiency would reduce these effects since the remaining energy in the water that leaves the bucket would be reduced. For domain discretisation it was decided to use two types of mesh sizing. Fine mesh simulation was mesh independent but the required time to solve was still unfeasible for parametric optimisation. Therefore, this fine mesh sizing was used only at the key points to verify the design changes. Coarse mesh simulation was not mesh independent but reduced the timescale by the factor of 5; therefore, making it possible to acquire the results within a reasonable timescale. It was observed that the coarse meshes slightly underpredict the efficiency as compared to the fine mesh simulations. However, it was assumed that this underprediction is going to be constant when comparing small changes in geometry. Based on this assumption the coarse mesh simulations were chosen for design optimisation. In part 2) some of the design parameters were expected to be interrelated and therefore were grouped together and analysed using Design of Experiments technique, some of the parameters were assumed to have low relation to other parameters and were analysed individually. In the end, CFD was predicting a 2.5 % increase of the original efficiency. Moreover, a reduction in the amount of buckets to 15 (originally the runner contained 18 buckets) was investigated and provided some promising results. This reduction can be very beneficial from the manufacturing complexity and cost point of view. In part 3) which was the final stage, three model runners were manufactured and experimentally tested in the Laboratory of Hydraulic Turbomachines at the National Technical University of Athens. It was decided to manufacture the original runner, the runner that contains 18 optimised buckets and the runner that contains 15 optimised buckets. The experimental results confirmed the increase in the efficiency and proved this optimisation technique to be valid.