Pump Specific Speed And Four Quadrant Data In Waterhammer Simulation – Taking Another Look
AFT Impulse™ Technical Paper
Authors: Trey Walters, P.E., Applied Flow Technology, Trygve Dahl, Ph.D., P.E., Intelliquip, David C. Rogers, P.E., Rogers Engineering Hydraulics
Presented at the ASME 2020 Pressure Vessels and Piping Conference (Virtual)
Abstract
In some situations, it is possible for flow to go backwards through a pump during a transient waterhammer event. Sustained reverse flow will lead to reverse rotation. Understanding and predicting the pump behavior during waterhammer under these conditions is typically accomplished using previously published four-quadrant pump data. Historically, the selection of which data to use is based on the similarity of pump specific speed.
The weaknesses of using specific speed are described and an improved method of selecting appropriate four-quadrant data is given based on fundamental curve shapes for head and power in the normal operating zone.
CONCLUSION
Four quadrant pump data available in Suter form has been the state-of-the-art for over five decades. Correlation of data based on pump specific speed has been the typical approach for selecting four quadrant data. An improved selection method is described based on comparisons to the shape of head and power curves in the zone of normal operation and comparisons to standard curve shapes from Stepanoff [11].
Highly referenced data from Knapp [3] and Donsky [7] was originally reported as specific speed of 1,800 US / 35 Metric.
Because the pump was a double-suction pump, their data was later claimed to be valid for a specific speed of 1270 US / 25 Metric and has been used for that specific speed for roughly the last 40 years by virtually all waterhammer analysts. The analysis here indicates that the Knapp data should be recategorized back to 1,800 US / 35 Metric. A similar situation applies to lesser known data by Kittredge [16] which he originally reported as specific speed 1500 US / 29 Metric.
Below is an excerpt. Use the link above to view the full paper.
Introduction
The possibility of reverse pump flow or rotation during a waterhammer event requires the use of methods for estimating pump behavior in all four quadrants. For the last five decades the state-of-the-art has been the Suter Method (Marchal, Flesch and Suter [1], and Suter [2]). This method correlates pump data sets into dimensionless form. Except in the rare cases when engineers have four quadrant data for their pump, the typical modeling choice is to use data for a previously tested pump with a similar specific speed. Some publications over the years have questioned the correctness of using pump specific speed as a correlating factor. This paper explores this concern.
First, in order to correlate four quadrant pump data based on specific speed, true dimensional and dynamic similarity between pumps must exist. Modern insights of pump performance and pump specific speed will be given to assess the effectiveness of the assumption of correlation.
Second, some of the excellent four quadrant data used today was first published as far back as the 1930’s – see Knapp [3] and Swanson [4]. The researchers could not have envisioned how their data would be used in modern waterhammer simulation. Indeed, many of the data sets used today were taken before Suter published his method in the 1960’s and before digital computers were created.
As a result, it is not clear that the researchers placed a strong emphasis on appropriately calculating and reporting the specific speed of the pumps used in their testing. This paper will attempt to confirm or disconfirm whether the correct specific speeds were reported.
Finally, the Suter Method is a convenient and creative dimensionless formulation of pump data referenced to the Best Efficiency Point (BEP) of the test pump. However, researchers over the years have taken to only publishing the dimensionless data and not the dimensional data behind the test. This makes it impossible to compare the waterhammer analyst’s pump to the original pump’s dimensional data, especially when the analyst’s pump is not operating near its own BEP – which is often the case. See Walters, Lang and Miller [5] for a comprehensive summary of the history and evolution of four quadrant pump research in the literature, a good overview of the Suter Method (in Part 1 of their paper), and a thorough explanation of the implications of off-BEP operation.
These issues are explored and recommendations are given to improve how pumps are modeled in waterhammer simulation.