Common Discharge Pumps 4

[EVEN01]

Hi,

I am analyzing a system that consists of three identical centrifugal pumps. I am debating weather the pumps are in parallel because the literature indicates that for a series of pumps to be considered in parallel they would have to share a common suction and a common discharge. This is not my case, the three pumps share the discharge. From the design point of view two pumps should be enough in order to move water from an open tank, however all of them are constantly running in order to keep the level in the tank constant even when the impeller size was increased. Now, I suspect the limiting factor in this case is the discharge pipe. Has anybody ever seen a problem like this, and if so, what’s the best way to try and approach a solution. I should add that we are about to expand our operations and thus we would like to try and add an extra pump, but find if it’s the line what’s causing the restriction.
I've attached a drawing for better undestanding.

 

[mtngreen]

The pumps are installed in parallel. They share a "commom" suction (ie the tank) and a discharge.

More info on flow and pressure would be helpful. Also, what's controlling the level in the tank? The control valve?

 

[BigInch]

Yes as mtgreen notes, they are definitely in parallel.

Can you plot a system curve? Plot pressure at the header converted to head in feet vs the total flowrate of all running pumps? Find that pressure for as many flowrates as you can. If you want to add another similar pump, you will have to extend that curve to however far it goes plus the added flowrate of another pump. Compare that "system" curve with one pump curve. If the pump curve shows sufficient head is still available (head where the pump curve crosses the system curve) at a flowrate equal to 4 pumps, you might be able to get away easy (without increasing the size of the pipe).

**********************
"Pumping accounts for 20% of the world’s energy used by electric motors and 25-50% of the total electrical energy usage in certain industrial facilities."-DOE statistic (Note: Make that 99% for pipeline companies)

http://virtualpipeline.spaces.live.com/

 

[EVEN01]

mtngreen,

The level in the tank has to be kept at 60% in order to maintain the level in the priming chamber (the tank on the suction of the pumps). The control valve maintains the level and is currently 45% open. The flow going into the pit is approximately 4500 GPM. TDH is 120 ft, NPSHA is 18 ft and NPSHR is 16.93. The pump capacity is 2640 GPM.
The pressure on the common discharge header is 60 psig.

 

[JJPellin]

I have a few comments. First, the NPSH margin is not adequate to protect the pumps. If the NPSH (required) is based on a 3% head loss (which is typical), then you are cavitating the pumps constantly. You only have about a 1 foot margin. Our pump selection standards would require a minimum 5 foot margin for water. Based on the diagram, I would be most suspicious that the control valve is the pinch point. But, you indicate that the control valve is only 45% open. You say you have to run all three pumps. I assume this means that if you shut down one pump, the control valve will go to 100% open.

In addition to the excellent advice given by BigInch, I would suggest a pressure survey of your system. If you have the taps available, check the pressure at:

* The pump suction
* The pump discharge
* The common 18" header
* The inlet to the control valve
* The outlet from the control valve

This should make it clear if the problem is an excess pressure drop in the piping or not. I would also verify that you have adequate submergence on the suction line. Even if you have enough NPSH available, it is still possible to have vortexing that could allow gas entrainment in the suction of the pumps.


Johnny Pellin

 

[BigInch]

As JPellin says, NPSH is close, so keep that tank full, esp when you open that valve more.

**********************
"Pumping accounts for 20% of the world’s energy used by electric motors and 25-50% of the total electrical energy usage in certain industrial facilities."-DOE statistic (Note: Make that 99% for pipeline companies)

http://virtualpipeline.spaces.live.com/

 

[EVEN01]

JJPellin,

During previous analysis it was determined that the conditions previously mentioned (pit level and valve position) are optimum to maintain a constant level in the priming chamber and thus avoid cavitation, the data I included previously is from the data sheet. You are right about the condition of the valve if one of the pump goes down, it indeed opens up 100% to try to maintain the level down.
Thanks for the suggestions, there are several places where I can get the pressure including the discharge of the pumps and upstream of the discharge point from the second pump. No luck with the suction side and so I will have to rely on calculations.

BigInch,

One more question, I am not very clear as to how to vary the flowrate in order to plot pressure vs. flowrate. These are very good suggestions and I will try to implement them. Thank you for the assistance.

 

[Artisi]

Can you post the pump curve - this gives us some additional info to play with.

 

[EVEN01]

Here is the pump curve

 

[Artisi]

seems to be a confusion of data -

"The flow going into the pit is approximately 4500 GPM. TDH is 120 ft, NPSHA is 18 ft and NPSHR is 16.93. The pump capacity is 2640 GPM.
The pressure on the common discharge header is 60 psig."

Total 120 ft --- discharge header 60PSI ?????????.

Think you neeed to sit back and rationalise the data as pointed out by Johnny Pellin - plus establish the flow rate from each pump.

 

[HEC]

Looking at the curve and a quick look at Re, suggests that with increased flow the system pressure drop (excluding the control valve) will increase quite substantially. Current line velocity is 1.8m/sec. It is worth noting that the pumps are currently operating left of the BEP. Based on information supplied flow of 4500USGPM/3 is 1500USGPM from each pump. This is a flat section of the pump curve, hence any small increase in total head will result in a decrease in volumetric flow! Short answer is adding an additional pump will not achieve increased flow.
The duty point on the curve suggests that two pumps should operate to achieve required flow. (4500USGPM/2 is 2250USGPM each). this operating point is at a lower discharge (header) pressure therefore the control valve must open to allow this.
If infact the system pressure has increased to the 170 feet predicted by the information supplied, a better solution would be to install a 14" impellor in each pump and increase the motor size accordingly.
To verify the problem and therefore validate the solution the pump suction and discharge pressures are required.

Mark Hutton

 

[BigInch]

Looking at that fast fall off in TDH, it doesn't look good for adding another pump, because with any more flow you won't get enough head from the pumps, but you might be able to recover some head thats being burned off by the control valve now, if the control valve can be opened. What's the head cut at that control valve now?

Right now we only have one point on the system curve, the blue dot at 138 ft (I added 18 ft suction head to the 120 TDH to get the system head equivalent plot on the pump curve chart) and 2650 gpm.

I took a guess at what the system curve with the control valve in the present position MIGHT look like. Open the attachment. What we really want to know is what the system curve looks like with the control valve full open. Maybe thats something like the purple line. To get the system curve without the control valve effect, we need to change the position of the control valve, measure the discharge pressure at the header, convert to head, and then subtract the head burned off by the control valve to arrive at one point on the true system curve (w/o any CV effect) at your tested flowrate. If you can adjust the position of the CV and read or estimate the head burned by the CV for a couple of points you can get several tested flowrates and a good relative handle for drawing in the system curve (purple line) where its supposed to be by plotting the discharge header Head - CV head value with each flowrate you check.

If you can do that, then you'll know if you can add another pump just by opening the CV, rather than by increasing the discharge pipe diameter. I extended the pump curve green line to some other higher flowrates, so wherever that green line intersects the purple line will be the running point at any other flowrates. The individual flow through any and all identical pumps will always be that flow read under that intersection point of the green and purple lines. To get total flow, just multiply by the number of pumps in parallel.

**********************
"Pumping accounts for 20% of the world’s energy used by electric motors and 25-50% of the total electrical energy usage in certain industrial facilities."-DOE statistic (Note: Make that 99% for pipeline companies)

http://virtualpipeline.spaces.live.com/

 

[JohnGP]

EVEN01, without flow meters you might be able to plot a couple of points by opening up the control valve fully and with 2 pumps operating measure the rate of filling for your pit, and deduct that from your estimated inflow, to get an idea of what the pumps are putting out. Also with 3 pumps, you can maybe measure the rate the pit lowers, and add that to your inflow, again with control valve fully open.

John

 

[Artisi]

without accurate flows and pressures it's anyones guess what is going on.

Get 1 accurate flow and pressure (at the pump discharge or corrected to the discharge point) together with the actual static head and an estimated resistance curve can be calculated.

 

[JJPellin]

As noted in many of the replies, you still have a lot of homework to do on this. But, it is my belief that adding an additional pump would be pointless. Your existing pumps should be easily able to deliver 6000 gpm or more. The only reason that they can't is that you are restricted on the discharge relative to the pumps head capability. Increase the size of the control valve and the piping just before and after the control valve. If this is not possible, install larger impellers.

BigInch,

I have to take exception to a few of your comments. First, the system curve with and without the control valve cannot be parallel as you have them drawn. They have to intersect at zero flow since the static head is unaffected by the presence of the control valve. Second, the flow estimate for a single pump cannot be multiplied by the number of pumps. Since the system curve always has slope, you will always get less flow from the second pump than you got from the first and less flow from the third pump than you got from the second. Each successive pump added will contribute less and less.


Johnny Pellin

 

[BigInch]

Johnny, you are correct (as ususal) in that the two curves should not have been drawn parallel and should intersect at zero flow. My mistake from working too late <1am> at night on that thing.

And your second comment about N x Flow, yes I realized that when I said it, but without knowing very much about what exactly is out there in relation to the piping config and how it affects the system curve, I thought it wasn't going to be too very far away from what we can reasonably estimate right now, but yes it will of course be less than N x.

Thanks for clarifying.

**********************
"Pumping accounts for 20% of the world’s energy used by electric motors and 25-50% of the total electrical energy usage in certain industrial facilities."-DOE statistic (Note: Make that 99% for pipeline companies)

http://virtualpipeline.spaces.live.com/

 

[EVEN01]

Thank you all for the great tips. I will start working on gatherind field data right away. We are scheduled to put pressure gauges sometime today or tomorrow. I will continue posting as we work towards solving this issue. Again thank you and have a great day.

 

[mtngreen]

I haven't followed the thread closely but looking at the pump curve, available NHPSH and NPSHr could you be running the pumps back on the curves to prevent cavitating...at pump design NPSHr is 16.9 - if you back it off to 1500gpm the NPSHr is +/- 11'...

 

[EVEN01]

mtngreen,
You are absolutely right. As a matter of fact according to operations (before my time) those pumps would cavitate quite often. It seems the previous engineer worked on this system but unfortunately the data left with him so it's difficult to say how the system was analyzed in the past.
We are about to expand our operations and so we need to investigate the root cause of this issue to avoid the same predicament. Additionally, even with our current flow conditions it is always desired to keep one pump on stand by.
Next step will be to get a pressure survey on the system, as Johnny suggested, and research the control valve so we can find an optimum point at which two pumps can handle the present flow without cavitating.
Again thank you all for the valuable recommendations as they have clarified much of the confusion I had.

 

[Artisi]

Now we seem to be getting somewhere - nothing like a bit of research to establish what is / has been going on, for years by the sounds of it.