connection of pumps in parallel 9

[Abhi1712]

Hi, I am new to Pump systems. I am working with small scale company. Recently i got a project related to pump parallel connection. Our company is having 6 centrifugal pump needs to be attached in parallel. I have to design pipe having 1 inlet from tank and 6 outlets going toward 6 pump suction lines. And another pipe having 6 inlets from pump discharge and 1 outlet going toward tank. This system I have to set up on trailer for movable purpose. I am new to such kind of project. Can you please guide me how to proceed? What precaution should I take while designing pipe? I am totally clueless about how to start this project.

 

[Artisi]

I suggest you start with some reading of pump applications, piping configuration, operating of pumps in parallel, pipeline losses, control systems etc etc - it should only take you 6 -12 months of intensive study to get the feel of what you are doing and to be able to recognise if you are on the right track.

Or else hire someone experienced in multi-pump installations.

 

[sheiko]

I presume the pumps are centrifugal type? Will they be identical (same curve)?

Normally, each pump will operate at the same Head, but share the flowrate with the other pumps.

Pressure drop in parallell paths must be equal.

Piping arrangement should also include:
- a straight run of 5 to 10 times the pipe diameter to the suction nozzle
- both sets of piping to be, at least, one size larger than the pump nozzle
- use of an eccentric reducer on the suction side, with the flat side up
- ...



"We don't believe things because they are true, things are true because we believe them."

 

[ccfowler]

Abhi1712,

Unless the number of pumps reflects a need to operate at several distinctly different flow rates and against a significant head (such as some sort of charge pump), it seems strange to choose to use so many pumps rather than one, two, or at most, three pumps.

Take careful note of Artisi's comments. His advice is probably the most valuable that you will receive in response to your question.

If you can state your problem more completely, better advice is more likely to be possible. Having worked with many centrifugal pumps operating in parallel, I can assure you that the selection of this arrangement by choice should only be made after very careful and well reasoned consideration of other alternatives.

Are you pumping water, a clear liquid, a slurry, ...?

Valuable advice from a professor many years ago: First, design for graceful failure. Everything we build will eventually fail, so we must strive to avoid injuries or secondary damage when that failure occurs. Only then can practicality and economics be properly considered.

 

[Abhi1712]

To Artisi: I am very grateful for your valuable guidance! There are lots of websites and books available in market for pump study. Will you please recommend me a website / e-book / book for this type of project as mentioned above?


To sheiko: first off all thank you for your valuable guidance!
Those pumps are centrifugal pumps. They are not identical (different curve). You said "Pressure drop in parallell paths must be equal." How do I verify it? Also you suggested for "a straight run of 5 to 10 times the pipe diameter to the suction nozzle" I do not understand it. Will you please elaborate on it? Please find attached document.

To ccfowler: I am very grateful for your valuable guidance! Let me explain my project in detail....
We are going to fix those pumps on trailer. This system will be mobile. Those pumps will be used for "Crude Oil" transfer in small scale refineries. (Transferring crude oil from one tank to another tank via pumps...) Those are special types of pumps having low capacity. So as you described, in some refinery we might require 3 pumps in parallel. Other might require 5 pumps in parallel due to large volume crude oil transfer. 6th pump will be 'standby'. What factors should I consider for designing such pipe for suction / discharge side for pumps as per your opinion. Please find attached document.

 

[Artisi]

here are 3 sites to start with.

http://www.cheresources.com

http://www.mcnallyinstitute.com/index.html

http://www.engineeringtoolbox.com/pumps-t_34.html

 

[sheiko]

Another thread where you will find answers:

http://www.eng-tips.com/viewthread.cfm?qid=284476&page=1

"We don't believe things because they are true, things are true because we believe them."

 

[Drexl]

Hi,

I don't understand why you have 6 pumps in parallel as I assume they are not that big. Is there a limit in motor size or are you controlling flow by starting up/shutting down pumps?

A few recommendations to consider however
a) all pumps should be the same and have "similar" inlet/outlet lines
b) pump outlet pipes should have check valves to avoid backflow through stand-by pumps
c) find out the max flow, make a reasonable line size assumption based on velocity and check the needed head based on a pressure drop calculation.
d) add the worst case pressure difference between the inlet/outlet of the system to the pressure drop to get the needed pump head (which will be the same for all pumps). The flow is divided on the number of active pumps in the system.

 

[ccfowler]

Abhi1712,

Looking at the schematic that you provided, I recommend the following for consideration:

1) Assuming that you plan a similar arrangement for the pumps, the inlet to the suction manifold should be nearest to the pump at one end, and the exit from the discharge manifold should be nearest to the pump at the other end of the line of pumps. This arrangement will minimize the variation in differential pressure that the several pumps will experience.

2) Your statement, "Those pumps are centrifugal pumps. They are not identical (different curve)" suggests that you are likely to have a very big mess to handle. It is bad enough to have multiple identical pumps working in parallel. Having pumps with differing characteristic curves operating in parallel usually results in some pumps being shifted to or near shut-off while others hog the load. You would do well to pay very close attention to the characteristic curves of several pumps. You should be able to determine which combinations of the pumps can be operated reasonably under which specific sets of conditions. It would be wise to include some sort of flow indicator for each pump so that you will have a way to determine when a pump is operating at flow rates too small or too large.

3) You may find that the dissimilar pumps may produce interesting stability problems while operating in parallel. You may experience situations such as one pump jumping to shut-off, then other pumps pick up additional flow (with reduced head), then the first pump resumes flow (instantly) and another pump may then jump to shut-off, and then this type of instability may continue to repeat over and over again.

4) If there are solids in the crude, it would be wise to have the branch pipes from the suction manifold to the individual pumps be taken from the top of the suction manifold pipe. This will reduce the formation of "solids plugs" in the branch lines.



Valuable advice from a professor many years ago: First, design for graceful failure. Everything we build will eventually fail, so we must strive to avoid injuries or secondary damage when that failure occurs. Only then can practicality and economics be properly considered.

 

[QualityTime]

The problem with so many pumps in parallel is that if the piping is sized for 6 pumps and you are only running one pump you could have the pump running out on you on the far right of the curve. You have to plot the system head curve and plot the pump curves with one...two..three...all the way to six pumps running. From there you can see whether you will run into pump problems.

Why don't you use a VFD on a single pump. You will get a 2:1 turndown

 

[tr1ntx]

As ccfowler stated, you're going to need flow indicators. The problem with that is that for 6 pumps mounted on a trailer, with a common suction header and a common discharge header, you might not be able get the required piping arrangement for the meters. This is a tough project for someone who is "totally clueless".

On the bright side, it sounds like your company already has the pumps, and is scheming up a quick way to make some money, so you are not all that concerned with maximizing pump operating efficiency and such. In applications where pumps run all day, every day, and are fixed in piping layouts, there is a much greater need to maximize efficiency and minimize operating costs. For you, control valves and pressure gauges on the discharge side of the pumps, along with a general understanding of the pressure range each of the different pumps should operate in, might be enough. Whoever is operating the unit will be required to adjust the valves and keep the system running smoothly, but that shouldn't be too much of a chore for just transferring crude from one tank to another. And like someone mentioned, you'll want check valves in the pump discharge pipes as well so it doesn't just pump in a circle.

 

[Artisi]

Lots of luck, because it is going to be needed.

 

[MortenA]

It seems a bit "happy go lucky" this project.

First you should try to see if all of you pumps will actuallo do any work.

Plot the pump curves in the same graph. You can then actually add flow rate as fixed head values together to get flow as a function of head for the parallel pumps. If you then arrange it so than each pump eg have a different colour (assuming that you are using excel) then you will get an idear if any of the pumps will have zero flow. Now ideally you should calculate your system curve - this would then show if any pumps are actually delivering zero flow (dead heading).

Best regards

Morten

 

[NGLENGR]

Here is a quick and dirty graph I made on a simple 3-pump system we have.

These three pumps are "identical" and have a single common flow measurement. By measuring the DP across the system and the total flow and plotting it against the composite pump curve, we can see what the natural system resistance is.

What will happen, since this fluid is ethane, there is a region of difficulty at approximately 30000 barrels per day, where the head of one pump cannot achieve well injection but at the same rate two pumps will fight each other for flow.

Parallel pumping is always tricky if you lack the instrumentation.

 

[waterpipe]

I would say that you should be careful with the scenarios that you are going to put your 6 pumps in action especially since they have different curves. Take a look to the required heads and flows and define your scenarios. How many numbers of pumps are needed and which pump should work with which other. You have to control the working point of each scenario and control that this point is with in the acceptable working point of each individual pump that is working under that scenario.
I suggest that you use the free software Epanet. You can simply define your pump curves, put them in parallel and check what ever scenario you would like to define. Given the system properties (pipe diameters, lengths and roughness) you will get the working point under each scenario.
As suggested with others, perhaps you would need to take a solid understand by reviewing some preliminary cases before getting to analyse 6 parallel pumps with different curves. There might be other options such as VSD, control valves, etc. to consider as well.

 

[Yobbo]

Tip:
Use PSIM pump software. For a small system you can model the pumps and the piping system.

Karel Postulart, The Netherlands
Nuon Power Generation