An experimental investigation of draft tube flow

Abstract

This thesis presents a study of the flow in draft tubes on hydro turbines, with different runner cone configurations, operating in the part load operating range. Dynamic wall pressure measurements on a model Francis pump-turbine and two prototype Francis turbines have been performed. The velocity distribution in the draft tube cone has been captured using two-dimensional Laser Doppler Velocimetry (LDV) on the model pump-turbine. Video recordings of the visualized flow in the draft tube cone were performed on the model, to investigate the typical behaviour of the characteristic rotating vortex rope.

Time-dependent wall pressure measurements for eleven different runner cone configurations and a grid of twelve operating points in the part load operating range were executed for the model. A semi-tapered cone were attached to the hub of the runner for ten of the runner cone configurations, whilst the eleventh beeing the original runner cone configuration. The attached cones were tested with three diameters and each at three different lengths. 2D velocity measurements with LDV were performed for two operating points in the part load operating range for four of the runner cone configurations. This involved the three different diameter semi-tapered cones with length 2.95D and the original runner cone. Two components of the velocity vector, the longitudinal and the tangential, were measured in two measurement planes in the draft tube cone.

A Rotating Vortex Rope (RVR) in the draft tube cone at part load is a periodic phenomenon. Each velocity measurement obtained by LDV must be referenced to the phase of the periodic phenomenon. This is achieved by phase-averaging the velocity data based on wall pressure data, i.e. the static wall pressure values acquired simultaneously with the LDV-samples. Video recording was performed at the two part load operating points for all eleven runner cone configurations. Visualization of the RVR was achieved by lowering the pressure downstream the runner to vaporize the vortex rope.

For the original runner cone configuration, the results show a strong periodic behaviour for the pressure- and velocity field in the draft tube cone. It is suggested that the RVR is a phenomenon occurring after vortex breakdown, and is analog to a precessing vortex core. The semi-tapered cones influence the flow field in the draft tube, thus in the runner as well. The adverse pressure gradient in the draft tube is affected through the change in the cross-sectional area. The pressure gradient decreases as the diameter of the semi-tapered cone increases, hence this leads to a suppression of the RVR and a reduction of draft tube backflow zones. For the largest semi-tapered cone, the flow field shows almost no sign of the RVR, i.e. the flow field is stationary rotating.

Wall pressure fluctuation measurements was carried out on the draft tube wall of two prototype Francis turbines. The measurements were performed for different runner cone configurations, throughout the operating range for the turbines. For the Litjfossen power plant the results indicate that both of the attached devices have a damping effect on the pressure fluctuations in the draft tube cone for high part load operation, compared to the original runner cone. For lower part load operation the devices are too small to affect the RVR significantly. For the Oksla power plant the attached cylinder is too small to give any significant damping of the pressure fluctuations on part load.