Unsteady Rotor-Stator Simulation of the U9 Kaplan Turbine Model Olivier Petit, Håkan Nilsson Chalmers University 6 th OpenFOAM workshop Penn State University, USA 13-16 June 2011
Develop OpenFOAM as the future first choice for industrial CFD in water turbine applications ( validate new implementations useful for Turbomachinery applications) Interact with the experimental project, in order to improve the flow in the U9 model
Prototype located in Porjus, consist in a Kaplan turbine. Designed to be able to perform detailed measurements on a real scale turbine/ generator unit. 1:3.1 scale model of the U9 turbine is set up at Vattenfall Research and Development in Älvkarleby. U9 unit, Porjus, Sweden U9 scale model, Älvkarleby, Sweden
Create a detailed measurements database that can be used to validate future numerical simulations. Investigate the impact on the flow of the curved pipe at the inlet of the spiral casing.
U9 spiral casing: mesh of 5 millions cell, block structured hexaedral, realized with ICEM-HEXA. Y+ value between 50-100.
U9 Runner: mesh of 100 000 cell, block structured hexaedral, realized with ICEM-HEXA. Y+ value between 50-100.
U9 draft tube: mesh of 1 million cell, block structured hexaedral, realized with ICEM-HEXA. Y+ value between 50-100.
A-A A A
At the inlet of the spiral casing, velocity profiles measured with LDA technique at 5 horizontal profiles(py 1-5), and 1 vertical (Pz). Results presented in dimesionless form using R inlet =0.316m, and vbulk=q/ Ainlet, Q=0.71 m.s -1 Comparison between experimental and numerical results are shown for the best efficiency point, Q=0.71 m/s, and alpha=26 (angle of the guide vanes)
Chalmers University of Technology
Similar behaviour between experimental and numerical results. Difference between numerical and experimental prediction of the flow can be due to the turbulence model: the k-ε model is not accurate enough to give a good prediction of the flow in a curved pipe.
SI SII Reference point is at the wall for both windows.
The reference point is located on the centreline. Axi-symmetric profile is taken at the inlet from experimental data provided by Berhanu Mulu Inlet boundary conditions
P1_a P1_c P3_a P3_c
Inlet boundary condition needs to include the unsteadiness of the wakes to predict accurately the flow in the draft tube Radial velocity must be included to get a good approximation of the flow Better turbulence model should predict more accurately the flow features.
mesh of 12 million cell, block structured hexaedral, realized with ICEM-HEXA. 7 GGI interfaces, linking different part of the U9 model TransientSimpleDyMFoam, OpenFOAM-1.5-dev, k-ε model Inlet boundary condition: plug flow with Q=0.71 m/s Turbulence parameters calculated so that I=0.1, and νt/ ν=10. CFD tool is OpenFOAM. Pressure and velocity probes are created to monitor the flow in the spiral casing and in the draft tube.
Chalmers University of Technology
The prediction of the flow before the runner is rather accurate A separation of the flow occurs after the runner blades, pushing the flow against the shroud. The y-plus value needs to be lowered in the runner area.
Focus on the U9 runner to get proper y+ value, and proper flow behaviour Pressure, velocity analysis of the whole U9 model and comparison with the measurement data. Thank you for your attention!