Olivier Roussette

Experimental Study of a Cavitating Centrifugal Pump During Fast Startups

The start-up of rocket engine turbopumps is generally performed only in a few seconds. It implies that these pumps reach their nominal operating conditions after only a few rotations. During these first rotations of the blades, the flow evolution in the pump is governed by transient phenomena, based mainly on the flow rate and rotation speed evolution. These phenomena progressively become negligible when the steady behavior is reached. The pump transient behavior induces significant pressure fluctuations which may result in partial flow vaporization, i.e. cavitation. An existing experimental test rig has been updated in the LML laboratory (Lille, France) for the start-ups of a centrifugal pump. The study focuses on cavitation induced during the pump start-up. Instantaneous measurement of torque, flow rate, inlet and outlet unsteady pressures, and pump rotation velocity enable to characterize the pump behavior during rapid starting periods. Three different types of fast start-up behaviors have been identified. According to the final operating point, the start-up is characterized either by a single drop of the delivery static pressure, by several low-frequency drops, or by a water hammer phenomenon that can be observed both a the inlet and outlet of the pump. A physical analysis is proposed to explain these three different types of transient flow behavior.

An Experimental Study of the Flow Field Inside the Diffuser Passage of a Laboratory Centrifugal Pump

Measurements are processed on a centrifugal pump model, which works with air and performs with the vane-island type diffuser of a real hydraulic pump, under five flow rates to investigate the internal flow characteristics and their influence on overall pump performance. The mean flow characteristics inside the diffuser are determined by using a miniature three-hole probe connected to an online data acquisition system. The flow structure at the inlet section of the diffuser is analyzed in detail, with a focus on the local pressure loss inside the vaneless gap and incidence angle distributions along the hub-to-shroud direction of the diffuser. Some existing calculations, including leakage effects, are used to evaluate the pressure recovery downstream of the impeller. Furthermore, particle image velocimetry (PIV) measurement results are obtained to help analyze the flow characteristics inside the vane-island diffuser. Each PIV measuring plane is related to one particular diffuser blade-to-blade channel and is analyzed by using the time-averaged method according to seven different relative positions of the impeller. Measurement results show that main loss is produced inside the vaneless part of the diffuser at low flow rates, which might have been caused by the strong rotor–stator interaction. When the impeller flow rate is greater than the diffuser design flow rate, a large fluctuating separated region occurs after the throat of the diffuser on the pressure side. Mean loss originates from the unsteady pressure downstream of the diffuser throat. For better characterization of the separations observed in previous experimental studies, complementary unsteady static pressure measurement campaigns have been conducted on the diffuser blade wall. The unsteadiness revealed by these measurements, as well as theirs effects on the diffuser performance, was then studied.

Leakage Flow Influence on SHF pump model performances

This paper deals with the influence of leakage flow existing in SHF pump model on the analysis of internal flow behaviour inside the vane diffuser of the pump model performance using both experiments and calculations. PIV measurements have been performed at different hub to shroud planes inside one diffuser channel passage for a given speed of rotation and various flow rates. For each operating condition, the PIV measurements have been trigged with different angular impeller positions. The performances and the static pressure rise of the diffuser were also measured using a three-hole probe. The numerical simulations were carried out with Star CCM+ 9.06 code (RANS frozen and unsteady calculations). Some results were already presented at the XXth IAHR Symposium for three flowrates for RANS frozen and URANS calculations. In the present paper, comparisons between URANS calculations with and without leakages and experimental results are presented and discussed for these flow rates. The performances of the diffuser obtained by numerical calculations are compared to those obtained by the three-holes probe measurements. The comparisons show the influence of fluid leakages on global performances and a real improvement concerning the efficiency of the diffuser, the pump and the velocity distributions. These results show that leakage is an important parameter that has to be taken into account in order to make improved comparisons between numerical approaches and experiments in such a specific model set up.

Leakage flow simulation in a specific pump model

This paper deals with the influence of leakage flow existing in SHF pump model on the analysis of internal flow behaviour inside the vane diffuser of the pump model performance using both experiments and calculations. PIV measurements have been performed at different hub to shroud planes inside one diffuser channel passage for a given speed of rotation and various flow rates. For each operating condition, the PIV measurements have been trigged with different angular impeller positions. The performances and the static pressure rise of the diffuser were also measured using a three-hole probe. The numerical simulations were carried out with Star CCM+ 8.06 code (RANS frozen and unsteady calculations). Comparisons between numerical and experimental results are presented and discussed for three flow rates. The performances of the diffuser obtained by numerical simulation results are compared to the performances obtained by three-hole probe indications. The comparisons show few influence of fluid leakage on global performances but a real improvement concerning the efficiency of the impeller, the pump and the velocity distributions. These results show that leakage is an important parameter that has to be taken into account in order to make improved comparisons between numerical approaches and experiments in such a specific model set up.