Testing a Turbine Pump

[safiamoiz]

My company has purchased a 10 stage Turbine pump, 10 HP. The pump manufacturer has provided the pump curve. Their persons visited our site for commissioning of this pump. The problem is that the data point of commission test comes a bit lower than the pump curve. Can any body suggest why this is happening. Else the pump is running smooth.

Moiz khan

 

[CH5OH]

the pump curve being not the pump curve of your pump but a generic pump curve for that pump, meaning slight differences in casting may cause your pump to be slightly different.

 

[dabluffrat]

Most industry standards allow for some +/- margin on the rated conditions. Test tolerances too may come into play in field testing, the location of the gauges used relative to ideal test lab conditions probably don't exist (i.e. straight runs of pipe up and down stream of flow meter).

10 stage & 10 HP means this is a fairly small pump... small castings are difficult to predict performance due to variations in vane thickness and geometry. Your company should have purchased a test at the manufacturer if this was a critical application. That would have ensured performance was guaranteed.

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

 

[Artisi]

"comes a bit lower than the pump curve" it would be nice to know how much variation - as stated it could well be within test code tolerances.

 

[safiamoiz]

I am attaching the pump curve with the test points shown in circle.

 

[JJPellin]

You don't provide enough information about the service or the system. If this is a hydrocarbon pump, then I the most common discrepancy I see is the incorrect use of hot specific gravity when converting pressure into head. If the pump is in water service, then I might suspect a dropped stage, a restricted suction strainer, or some pretty rough castings.

Johnny Pellin

 

[Artisi]

The pump curve represents the turbine only ie, from inlet to outlet of the turbine, it does not include losses for any discharge pipe/ column etc from the turbine to the point where you are measuring.

As pointed out by Johnny P, you need to sketch or describe more fully the installation and show the measuring points etc - this is the only way we can get an idea of what is being measured.

 

[dabluffrat]

Artisi,

Not necessarily. The Head axis on the curve may either say Bowl Assembly Head, or Total Pump Head.

In the first case, only the head generated at the last discharge stage impeller/bowl flange is shown. This is the most efficient curve, independent total pump design losses (type of discharge elbow, column size, etc) and often the mfg Sales Curve.

The second case is the actual Pump Head at the discharge flange. This is usually a mfg Test Curve and will include the head loss through the column and across the discharge head.

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

 

[dabluffrat]

I just looked at the curve, and yes this looks to be the typical sales curve.

I would have considered an 11 stage pump before a 10 stage at full dia.

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

 

[Artisi]

dabluffrat,
I would say this is a standard sales curve for 1 stage of turbine ( that is why it has the derating % for a low number of stages) and does not include any discharge head or column losses.
The test plot looks to be typical of a bowl assembly together with the losses for the discharge elbow, inlet strainer and column - deduct the losses and the turbine could well be on curve.

More info from the OP who seems to have vanished would help with a better understanding of what is actually being measured.

 

[safiamoiz]

Thanks to all. Now the pump is running smooth.

 

[Artisi]

It would be nice to know the reason why

 

[QualityTime]

First of all it is not a vertical turbine pump. It is a mixed flow pump. Mixed flow pumps are very unstable even if you operate momentarily on the left hand side of the published curve. That is why they do not publish that part of the curve. Centrifugal pumps, for example, can be started against a closed discharge valve and that is why the pump curve shows a TDH point at zero flow.

As far as for your problem your field test points are not parallel to a pump curve and they droop more and more as the flow increases. I would bet you have some trash caught up in the impeller bowls. Pull the pump out of the pit and take a look. I had a similiar problem back in 1983

 

[QualityTime]

Forgot to add that the pump will run smooth even though the bowls could be partially filled with trash. So just because a pump is running "smooth" it does not mean it is running right. Again I still recommend that you pull the pump out and check if the bowls are partially filled with debris

 

[Artisi]

The curve looks to be pretty stable to me for operation in any normal operation.

 

[QualityTime]

Pumps WILL perform according to the impellor diameter cut of the pump curve. If the point measured point is off the pump curve the reasons could be:

1. Incorrect method of determining the points
2. Not enough minimum submergence and you are drawing air
2. Clogged pump suction
3. When you have a clogged pump suction you are in effect reducing the NPSHa which in turn could cause cavitation. The more the flow is, the more the reduction in NPSHa and,if you are cavitating, the worse the cavitation will be and the pump performance will become poorer as evidenced by the drooping curve as you go further to the right of the curve. I have marked up the points on the pump curve that was previously available to illustrate my point.

The problem with turbine pumps is that you can't "hear" any problems because they are down in a hole

 

[Artisi]

Thanks for the lessons in pump hydraulics and performance with is academic anyway as the OP has advised he is happy with the pumps operation.

 

[QualityTime]

I think you have two problems here. The clogged suction issue that I mentioned in my previous post and it looks like their is too much gap between the impellor and the wearing rings. I have only worked with vertical pumps once before but I think the gap can be adjusted by pulling up on the shaft. If it is not a gap problem i think they cut your impellors to the wrong diameter.

 

[dabluffrat]

QualityTime,

I don't know where to begin to correct you...

Mixed flow pumps run against a liner, though I'm not convinced this is even a mixed flow pump.

Blockages at the suction, cavitation, or increased clearance would be assoc. with high vibration and noise, which wasn't mentioned.

Forget looking into any bowls for trash... 10 stages and only 10 hp motor indicates these are very small bowls and impellers (even smaller hydraulic passages).... actually the curve shows the impellers are only 5" in diameter.

What is more likely, and I have seen in my many years of pump testing, is variations in castings (vane thickness and geometry) on the smaller impellers contribute to greater performance variations (+/- 1/4" deviation on a 5" dia impeller is a greater portion of the total area then on a 35" impeller).... or even more likely, the setup and test equipment in the field isn't ideal.



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

 

[QualityTime]

Hi dablufratt,

I will respond to your comments:

1. The pump is certainly not a centrifugal pump. If it was, the curve would have started at zero flow on the x axis. If you look closely at the pump curve the pump manufacturer does not show the left hand side of the curve all the way back to the y axis.

2. You won't hear cavitation or anything because the bowl is submerged in a well. I also did not say there was any cavitation. It depends how severe the blockage is

3. I have had practical experience with blockages on vertical turbine pumps and the curve looked the same. In one of the municipal water treatment plants that I have built in my career the high lift vertical turbine pump was not performing to its curve. I had them pull out the pump and send to back to the pump shop. There was construction debris in the pump bowl. Once cleaned out, field tests showed the pump performing exactly to the pump curve. I think if you looked in some pump handbooks they will show you a classic curve of a blockage on the suction side and this curve looks pretty close to my experience. BTW, the same installation had mixed flow pumps there also...

4. While it is possible there could have been some manufacturing issues I would suggest that pump manufacturers invariably get their pumps to perform EXACTLY to a published pump curve. Otherwise what is the point of publishing a pump curve if you cannot depend on it performing to its curve. How can anyone design anything in this world if the pump does not perform as advertised.

5. By the way I have done many witness tests on all sizes of pumps all the way up to 4000 HP. In the 4000 HP instance, Fairbanks Morse could not get their pump to perform during witness testing according to the commercial curve that they tendered. It followed the left hand side of the curve exactly and then started to droop more and more the further it went to the right. They spent so much money trying to get the pump to perform and at the end of the day were ready to give up and started to babble about fitting the witnessed points to some 3rd order curve fitting equation???!!!. The potential liquated damages were stupendous. I did my own calculations at the Kansas City witness test site and told them the pump was not performing the further they went on the right hand side of the curve because there was not enough suction lift capability. I went over the calculations with them on a white board and immediately they looked at their own spread sheets and saw where they made the computational error. I told them the pump does perform and I would release the pumps for delivery. We retested the pumps on site where we had flooded conditions and the pump performed according to its curve. END OF STORY.