Affinity Law for Multi-Stage Pumps 1

[Chance17]

I am evaluating a 10-stage Pacific pump originally purchased in 1957. The rated head at shutoff was 960-ft.

Twenty years later the client removed three stage. There is no pump curve for 7-stages. The engineer in 1977 estimated the 7-stage pump to be 440-ft at shutoff. Today I field verified the shutoff being with 10% of 440-ft.

I am not aware of any pump affinity laws regarding multiple stages. My current view is the original 10-stage pump would be 96-ft per stage. Removing 3-stages would reduce the head by [3*96]=288-ft resulting in 662-ft at shutoff.

My question: How can one predict going from 960-ft to 440-ft? I hope the question is not too dumb.

 

[Artisi]

If ten stages of the same diameter gave 960ft then as you said each stage was producing 96ft at the CV.

To achieve 440ft at the CV you will probably need 5 stages with a diameter trim on 1 or 2 stages.
As the head has reduced what about the flowrate? if there was any change to flow - then you need to draw a system curve and reselect the number of impellers and impleller diameters at the flow / head required.

 

[rmw]

Are/were all stages equal stages? Some multiple stage pumps have different configurations of stages within the mix.

rmw

 

[ione]

A multistage pump can be considered as a series of centrifugal pump (flow rate is constant, while head is the sum of the head of each stage).

 

[Chance17]

To Artisi and RMW - all stages are suppose to be equal.
The records are vauge and I guess some undocumented impellor trimming could have occurred.

The data is taken at shutoff - meaning the flow is not changed.

Artisi - You seem to confirm my 96-ft per stage analysis.

Meaning the affinity law for pressure rise per stage is proportional??

 

[JJPellin]

Running a 10 stage (or 7 stage) pump at zero flow, even for a few seconds, is risky and I would not recommend it. Do you have data at higher flow rates to compare between the original curve and the current condition? Many of these old pumps used a first stage impeller that was either double suction, different diameter or had a much larger impeller eye to reduce NPSH required. If the first stage was different than the others, it might explain some of the discrepancy, but not all of if. If all stages were identical, I would expect the head at a given flow rate to be reduce to 7/10ths of the original head. You should verify that you are using appropriate values for actual (hot) specific gravity. You should re-check your calculation for head. Have your suction conditions changed since original installation? If they removed 3 stages, I would wonder if this was not done because of some major change in suction conditions or product composition.

Johnny Pellin

 

[rmw]

JC Pellin has described the pump configuration I had in mind when I posted earlier-the double suction first stage.

Multistage pumps are also designed with stage configurations so that individual stages and especially the higher pressure stages oppose each other in order to balance the thrust forces. Therefore, it is difficult to imagine how they removed an odd number of stages and didn't over-thrust the shaft.

Something doesn't add up here unless the third stage was the double suction first stage that JC described.

That old pump must have been a beast to have survived this abuse (and most of those built back then were just that, beasts.)

rmw

 

[dabluffrat]

If all impellers are equal (no double suction 1st stage, all same dia) then yes the head per stage calc is proportional.

There are losses associated with removing impellers depending on the type of pump and how they are removed. The engr in 1977 likely took these losses into account, though 200+ ft of losses is very high. Vertical pumps will typically leave the bowls empty, horiz. pumps will be fitted with de-stage pieces and sleeves to blank off volute passageways and balance thrust.

The affinity law is used to refigure pump performance from one speed to another... has the RPM of the pump changed as well?

Did you know that 76.4% of all statistics are made up...