This article also appears in
Subscribe now »

The CFD predicted pump performance curves using the steady state MRF and the transient sliding mesh approaches are shown. The results show a considerable difference in the characteristic curve obtained using the two approaches.

Tata Technologies researchers use CFD to predict cavitation in liquid ring pumps

Liquid ring pumps are used in aircraft fuel systems in conjunction with main impeller pumps. These pumps are used for priming the pump system as well as to remove fuel vapor and air from the fuel. Prediction of cavitation in liquid ring pumps is important as cavitation degrades the performance of these pumps and leads to their failure. As test-based assessment of cavitation risk in liquid ring pumps is expensive and time consuming, recent approaches have been to assess and predict the risk of cavitation using CFD methods with the goal to quicken the design process and optimize the performance of these pumps.

CFD models have demonstrated the ability to be used as a cost-effective tool to analyze cavitation phenomenon in pumps for aerospace industry. Pump reliability is of utmost importance in both commercial and military fixed wing and rotary wing aircrafts due to their need of vapor or air free fuel that is required to be supplied to their engines at all flight missions. As liquid ring pumps serve the function of removing fuel vapor and air from the fuel, their reliable functioning plays a critical role in determining the safe operation of aircrafts during flight. As cavitation has the potential to severely limit the operability of these liquid ring pumps and in severe cases may lead to their structural failure, accurate prediction of cavitation in liquid ring pumps is extremely important to design these pumps for safe operation.

The cavitation phenomena occur in regions where large pressure drops cause the local pressure to fall below the vapor pressure resulting in formation of vapor bubbles. Typically for pumps, cavitation occurs in the suction side of the pump blades that in turn results in a reduction of effective area of blade thereby diminishing the efficiency of the pumps. The formation of vapor bubbles and their subsequent bursting creates pressure impulse on the blade surfaces, which leads to vibration and fatigue induced structural damage leading to pump failures.

Researchers from Tata Technologies Ltd. used steady state Multiple Reference Frame (MRF) methodology and the transient sliding mesh methodology to assess cavitation, pump performance, and Net Positive Suction Head (NPSH) in liquid ring pumps using ANSYS-Fluent CFD software.

Results of the research show a considerable difference in the characteristic curve obtained using the two approaches. The results obtained using the MRF approach show a sudden slope change in the performance curve when the flow through the pump exceeded 30 m3/s. This unphysical behavior that is not observed in the transient sliding mesh attests to the possible inaccuracy that could result when simulating cavitation using the steady state MRF approach.

Typically cavitation occurs near the hub surface, and investigation of the pressure distribution in the hub area is important to understand and analyze the cavitation phenomenon. Cavitation occurred in the first, second, and fourth quadrant of the hub. The cavitation region extends from around 0° to 50° and from around 270° to 360°. The results show the appearance of pressure spikes that coincide with the fluid compression and subsequent ejection through the outlet port. The pressure spikes result in an implosion or collapse of the vapor bubble, which is formed during the cavitation process. The large magnitude of the pressure spikes, value of which is as much as 2.25 MPa for the pump configuration considered in this study, creates pressure impulse load at the impeller surfaces, which may lead to structural failure. Furthermore, the cyclic nature of the pressure impulses leads to fatigue of the impellers that further expedites their structural failure.

To compare the results obtained using the steady state MRF approach and the transient sliding mesh approach, the volume fraction distribution at the pump mid-section plane are shown in Figure 1. These distributions of volume fraction and pressure are shown for varying values of outlet pressure ranging from 0.4 MPa to 0.6 MPa. The results demonstrate that cavitation occurs in the region between the inlet and the outlet port along the direction of rotation of the impeller. Furthermore, the results show that the cavitation area shrinks with in an increase in outlet pressure. The distribution of the vapor fraction shows the accumulation of liquid fuel around the periphery due to centrifugal forces and the accumulation of the vapor bubbles in and around the hub region as these regions experience the low pressure regions that lead to cavitation.

The vapor fraction distribution predicted by the MRF approach indicates an unrealistic unphysical location of cavitation. The MRF model predicts that the cavitation occurs near the inlet port which is physically unrealistic.

Figure 2 shows the distribution of absolute pressure at the pump mid-section plane as predicted by the transient sliding mesh and the steady state MRF methodology, respectively. As expected, regions of low near the inlet port and regions of higher pressure near the outlet port are observed.

The results indicate that though the computation efforts are cheaper for the steady state MRF model, the results obtained are unphysical. The computationally expensive transient sliding mesh approach results in realistic predictions. Due to unavailability of experimental data, a quantitative validation of the sliding mesh approach for cavitation prediction could be not be performed, but the trends observed in the results show promise in this approach as compared to the MRF approach. Further investigation along with experimental validation would be required to refine the prediction fidelity of the transient sliding mesh based cavitation model for liquid ring pump applications.

Continue reading »