Because of the change in the energy production towards regenerative energy sources with fluctuating production, the demand for pumped hydro power plants has been rising worldwide. Pumped hydro systems with pump turbines are more economical and suitable for future needs like seasonal energy storage or grid stabilization. Current pump turbine rotor designs show unstable behaviour in important operation points. That effects negatively the operation itself and reduces the life time of the whole machine. This study evaluates tools to identify unstable operation behaviour during rotor development to ensure proper behaviour.

For that, two numerical models of a pump turbine model with unstable behaviour that has a small specific speed (nq = 26min-1) and a guide vane angle of 6° has been created. It is the goal of these numerical models to reproduce unstable behaviour accurately and with the shortest possible computing time. After some research it was decided to run URANS simulations with an openFOAM solver and a k-ε turbulence model on one hand and CFX with an EARSM turbulence model and activated curvature correction on the other hand. The k-ε turbulence model should be faster, the EARSM more accurate. The initial conditions for both URANS simulations are evaluated by running a steady state simulation. The initial conditions for the steady state simulation has been measured on a pump turbine testing bench.

For the detection of the unstable behaviour, a tool has been programmed using Python (2.7) as coding language. It can automatically generate values, diagrams, and views out of the simulation results. To implement more functions and outputting views in a simple way, it is coded on a modular base, has an extended documentation and lots of comments in the code. The tool has been developed to fully functionality during this study. As a test, one of the simulated operation points of the numerical model (openFOAM, k-ε) has been analysed.

The result showed, that a boundary condition of a small section in the mesh is responsible that no positive slope in the characteristic is reached and therefore no instability can be detected. In addition, a backflow out of the draft tube inside the runner section is recognised. This backflow might be responsible for one of the vortexes who are typically for instabilities.