Simulations and experiments of turbulent diffuser flow with hydropower applications

Abstract: Most hydropower plants in Sweden were built more than 50 years ago. The design tools were model tests and one-dimensional analytical methods. Since then the field of Computational Fluid Dynamics has evolved into an important engineering tool. To use CFD in a design or redesign stage of hydropower waterways would be very attractive due to its flexibility and cost-effectiveness. The hydropower plants operate under different circumstances today, since the electricity market has been deregulated and the nuclear power will be phased out. Turbines are thus run in a different manner than before with high power output during peak demand hours when the electricity price is high. Hence, efficiency improvements over a wide range of operating conditions are of interest. Predictions of runner flow have been established since long. Turbulence is of secondary importance which makes flow modeling relatively easy. The draft tube, however, is responsible for a large portion of the total hydraulic losses for low and medium head turbines and it is also the most challenging part of the waterways to predict numerically. The draft tube flow is characterized by the interaction of many complex flow features such as diffuser effects, unsteadiness, swirl, streamline curvature, separation and impingement. The Turbine-99 workshops assessed the ability of CFD to predict the flow features and engineering quantities of a model draft tube flow. It was found that many different problems contribute to the overall uncertainty in the results. The implementation of unknown boundary conditions and the post processing of the results were identified as error sources. The performance of the different turbulence models could not be evaluated, due to the numerical uncertainties. In the present work, factorial design was applied to investigate the sensitivity of the numerical solution to the unknown boundary conditions. It was shown to objectively give valuable information about the importance of different input parameters and to simplify the assessment of different turbulence models. To validate turbulence models for draft tubes, carefully selected validation cases need to be developed. This work presents two experimental validation cases that incorporates some of the complex phenomena in the draft tube, but in a well-controlled environment where detailed near-wall turbulence data can be measured by laser Doppler velocimetry. The effect of a 3-D straight asymmetric diffuser on a developing duct flow is investigated both for steady and pulsating flows. It was found that the average flow field is unaffected by the pulsation, while the oscillating turbulent quantities were influenced depending in their frequency.