CHAPTER CONCLUSIONS AND RECOMMENDATIONS 6.1 Conclusions In the present study, six different types of centrifugal pump impellers are selected to simulate their internal flows by using a three-dimensional Navier-Stokes numerical model. The first two model impellers called M1 and M2 are radial ones with five straight blades that were also used by Kosyna and Kecke (2002) in their research work. Both single-phase (water) and two-phase (water/air) flows through these two impellers are considered and the numerical results are compared with experimental and numerical data given by Kosyna and Kecke (2002) for validation. The third model impeller called M3 has four straight vanes and the last three model impellers called M4, M5 and M6 have six, six and five twisted vanes respectively. The numerical results from these pumps were compared with the experiment data except for model impeller M6 and the comparison focus on the pump performance curves. In the first research stage, the single-phase flow model with the standard k − ε turbulence model, RNG k − ε model, the Wilcox k − ω model and the shear stress transport (SST) model were used to simulate the internal flow of six different types of model impellers. The important findings can be summarised in the following paragraphs. (1) The computed head-flow H-Q curves for model impellers M1-M5 are compared with the experimental data at various rotational speeds. A good agreement 166 Chapter Conclusions and Recommendations between numerical and experimental result is achieved over the entire flow range. However, it is also found that the present numerical results show a little higher prediction than the experimental data. This is because the flow losses in the pipes and volute are neglected in the current calculations. (2) The comparison of computational and experimental η -Q curves is also made for model impeller M3, M4 and M5 and shows good consistency. The predicted maximum efficiency also agrees well with the impeller design value. The numerical deviations between predicted and design value for model impellers M3, M4 and M5 are below 10%. Therefore, the present efficiency prediction is relatively accurate. The comparison also shows that the efficiencies of model impellers M4 and M5 are much higher than that of model impeller M3. It may suggest that efficiency of twisted blade pump will be higher than that of straight blade pump. Besides, it is found that the maximum efficiency occurs near the pump design point, and thus the impeller design is acceptable. (3) The present numerical study shows the flow feature at design points for impeller M1-M6 with various vane numbers. It is found that the flow in the mid-plane of the passage near the design point follows the curvature of the blade well for model impellers M1, M2, M4 and M5. However, re-circulation occurs at design point for impellers M3 and M6. The main reason is for the effect of the vane number. The vane number for model impeller M3 and M6 cannot constrain the flow well and makes through-flow strong, thus the relative vortex can be observed. (4) The present study also shows the flow feature at conditions other than the design point for model impellers M1-M6. It is found that when the flow rate decreases below a certain value of the design flow rate for model impellers M1, M2, 167 Chapter Conclusions and Recommendations M4 and M5, recirculation will occur near the pressure side of blade surface. The change of flow patterns at low flow rate is not obvious for model impellers M3 and M6, but it can still be seen that the recirculation core enlarge as flow rate drops. The reversal flow at off-design point is caused by the re-circulation near shroud side of inlet section and secondary flow from hub to shroud. The existence of the vortices in the passages at small flow rates will increase the loss of flow and thus reduce the efficiency of the centrifugal pump. Although some of these findings were once reported by some researcher such as Liu et al. (1994) and Sun and Tsukamoto (2001), their study was not as thorough as ours because only one impeller was studied. (5) The present study also makes contribution to study the pressure distribution along the impeller blade among the various flow rates for pump impellers M3-M6 at the rotational speeds of 2900 rpm and 1450 rpm. It shows that pressure along the pressure surface varies slightly when the flow rate changes, but it varies largely at the suction side. The pressure difference between pressure and suction surfaces will get large as inflow rate increase. Since these findings are observed among the centrifugal pumps we studied so far, we can thus conclude that they are applicable to the centrifugal pump impellers with small and medium specific speeds. (6) The influence of turbulence models on the flow and pressure field through the centrifugal pump impellers is also investigated. It is found that the results from the Wilcox k − ω model and the shear stress transport (SST) model are close to that from the k − ε models except at outlet sections. Since the difference is not obvious, it is hard to conclude whether the Wilcox k − ω model and the SST model are superior to k − ε models in the centrifugal pump impeller simulation. The lack of difference between the various turbulence modelling may be due to the coarse mesh 168 Chapter Conclusions and Recommendations used in the present model. So further research work is required to draw a clearer conclusion. The present study also makes contribution to investigate two-phase (air/water) flow through centrifugal pump impeller by using the implemented Eulerian multiphase flow model. The model impeller M2 is selected to carry out the two-phase flow simulation and the numerical results are compared with those of Kosyna and Kecke (2002) for validation. The main conclusions in this part can be summarized as follows: (1) The present study successfully shows the similar trend in the pump head drop with the experimental and numerical results given by Kosyna and Kecke (2002). The present study also successfully shows three states of pump performance, i.e, when the gas fraction is increased from the state without gas loading, the pump decreases its head continuously in state 1. However, at a centain gas fraction the pump shows an abrupt change (state 2). In state 3, it depends on the operating point whether the pump is still able to operate or not. (2) The pump performance curves at two gas fractions agree well with the results given by Kosyna and Kecke (2002). As the gas fraction increases, it is found that the magnitude of pressure rise coefficient ψ will be reduced and the characteristic curve appears to drop more abruptly. In addition, the present study successfully predicts the trend of pressure distribution along the impeller blade. (3) The accumulation of gas near pressure surface is qualitatively reproduced and it matches well with the photography taken by Kosyna and Kecke (2002) using a high speed camera. It is found that the gas is most likely to accumulate near inlet area of shroud surface. This trend can be seen more clearly when the gas fraction at inlet 169 Chapter Conclusions and Recommendations increases. This is because the inlet recirculation will always occur on the leading edge of the shroud, and it will cause the gas concentration in this area. According to our literature review, only few researchers such as Kosyna and Kecke (2002) have done some research work in this area recently. Some of their experimental results matches well with our simulation. However, their analysis on gas distribution in the impeller is not as deep as ours because they have not presented the gas fraction distribution at various locations and flow rates. Thus, the present study can provide clearer picture about the distribution of gas fraction in the impeller. From the distribution of gas volume fraction, the regions that are easy to wear can always be conveniently found out and this can further guide the design of the impellers. 6.2 Recommendations (1) If possible, some more experiments will have to be carried out in the future study to verify the numerical results. Currently due to the limitation of our experiment, the comparisons between numerical and experimental data focus on the pump performance curves. Other numerical results such as pressure distribution along the blade surface and two-phase flow through the impeller can not be compared and validated. So the test rig may need to be rearranged to perform more sophisticated experiment. (2) In our two-phase model, some assumption has been made to simplify the simulation. For example, lift force and pressure force are considered to be negligible. Therefore, all factors that will affect air/water flow through the pump impeller should be considered in the future study. 170 Chapter Conclusions and Recommendations (3) There is also a need in the future to study the effects of cavitation which is often associated with inlet recirculation but was not considered in the present computation. (4) The present study shows that recirculation flow occurs near the pressure side of blade surface for model impeller M3 and M6 at pump design point. The existence of back flow in the passages will increase the loss of the flow and thus reduce the efficiency of the centrifugal pump. Some modification of the pump design based on the calculation results may be necessary to improve the pump efficiency. 171 . and M6 at pump design point. The existence of back flow in the passages will increase the loss of the flow and thus reduce the efficiency of the centrifugal pump. Some modification of the pump. 166 CHAPTER 6 CONCLUSIONS AND RECOMMENDATIONS 6. 1 Conclusions In the present study, six different types of centrifugal pump impellers are selected to simulate their internal flows. secondary flow from hub to shroud. The existence of the vortices in the passages at small flow rates will increase the loss of flow and thus reduce the efficiency of the centrifugal pump. Although