Volute throat area and wall modelling influence on the numerical performances of a very low specific speed pump

Abstract

Low specific speed pumps find applications in a broad range of domains, but suffer from a low efficiency and a risk of head instability close to shut-off. The numerical computations on these pumps performed in the last years have shown discrepancies with experimental results. Recent studies suggest that the use of wall-functions underpredicts the losses of these pumps, especially at overload. The reason has been attributed to a detachment zone downstream the volute tongue, not well captured with the wall-function approach. This paper focuses on the influence of the volute casing on a pump with a specific speed of 8.9 on two issues. First, the influence of the wall modelling approach relatively to the low-Reynolds number method on the performance prediction is discussed. The results are, as expected, an underprediction of the losses when the wall-function approach is used. With a larger volute, the difference between the two wall modeling approaches is smaller. Secondly, the influence of the volute throat area enlargement on the pump performances is discussed. Both the head and efficiency are improved at the design point and at overload with an increased volute throat area. However the part-load head decreases and the head flattens. The study of the flow at part-load, in the region at the outlet of the impeller reveals that with a larger volute, larger flow exchanges are present, contributing to additional mixing losses and head loss.Low specific speed pumps find applications in a broad range of domains, but suffer from a low efficiency and a risk of head instability close to shut-off. The numerical computations on these pumps performed in the last years have shown discrepancies with experimental results. Recent studies suggest that the use of wall-functions underpredicts the losses of these pumps, especially at overload. The reason has been attributed to a detachment zone downstream the volute tongue, not well captured with the wall-function approach. This paper focuses on the influence of the volute casing on a pump with a specific speed of 8.9 on two issues. First, the influence of the wall modelling approach relatively to the low-Reynolds number method on the performance prediction is discussed. The results are, as expected, an underprediction of the losses when the wall-function approach is used. With a larger volute, the difference between the two wall modeling approaches is smaller. Secondly, the influence of the volute throat area enlargement on the pump performances is discussed. Both the head and efficiency are improved at the design point and at overload with an increased volute throat area. However the part-load head decreases and the head flattens. The study of the flow at part-load, in the region at the outlet of the impeller reveals that with a larger volute, larger flow exchanges are present, contributing to additional mixing losses and head loss.