Parallel Pump Selection / Doubts with head losses

[Pucp]

Hi, I have the following problem. We have to select pumps for a big pumping station. We have to deliver 6000 l/s, and we want to use 10 pumps (vertical turbine pumps) in parallel operation (600 l/s each one). I can calculate each pump in its line (with butterfly valves, check valves ) but then the 10 pumps join in one large diameter pipe (parallel operation), and when this pipe is out of the pumping station it reaches a manifold and separates in 4 lines (36" diameter and 1.5 km of length) so each line deliver 1500 l/s. I am not sure about the head losses for each pump, I am calculating the head losses as follows (is the next reasoning correct?):

1)Losses between the pump and the join to the large diameter pipe (600 l/s)
2)When the flow from this pump (600 l/s) join the flow from the next pump (600 l/s), I calculate the head losses with the total flow (1200 l/s) in that section of the big pipe and then divide in two, each loss correspond to each pump
3)I do the same with three, four....ten pumps
4)When I have four lines (1500 l/s each one) I calculate the head losses with 1500 l/s and the total length of the pipe and then I multiply the result by four (four lines) and then divide in ten (ten pumps).
My reasoning is that the energy that the pumps needs to deliver is the sum of all the head losses from all the lines. So the head losses from the four lines (1500 l/s) have to be supply by the 10 pumps (in equal parts). By the way I am using Hanzen-Williams theory. Thanks in advance.

 

[PacificSteve]

Is part of this system already existing with fluid flowing through it? If so, you can use data from that to increase your overall accuracy and improve the selection process.

Ill be glad to help if you can first answer the question above.

With ten pumps, you can rest assured that there will be a problem with at least one of them every year. If your pumping process is critical need, then you should be thinking about some redundancy...at least at a later stage when the initial problem is solved.

PS

 

[MortenA]

I'm not quite sure if you are going about it this way but heres how i would do it:

for the required flow

1) Loss in discharge line (only 1 time for 1/10 of liquid))+loss in line to manifold (for all liquid)+loss in manifold(s) + loss in 1 1.5 km line for 1/4 of the total flow.

Since i read from your text that the system is "symmetric" all lines will have the same flow (or close to)

Find the corresponding head (allowing for any inlet loss differences of ehight etc.) and if you required flow is less than available then your fine (except if your way over but thats more of a control problem.

It sound like its a slightly complicated system and you should maybe consider getting some specialts to advise you at this problem.

Best Regards

Morten

 

[Pucp]

Hi, thank you for your replies. PS, this system does not exist, so I do not have any data to measure. MortenA, what you are doing is different from my reasoning, you said that for each pump the head losses are the sum of the losses in each part of the pipe, it does not depend on the flow rate of each pump (only depends before it reaches the large diameter pipe), so I have to calculate with the actual flow of each part of the total length of the pipe, if it is 1/10, or 2/10 (in the large diameter pipe...or 1/4 (1.5 km length)of the total flow and then sum all the losses and this is the head losses that each pump have to provide. Am I correct?, please for me is a little bit confusing this calculation, so any advice will help. Thanks

 

[abeltio]

You can find an excellent discussion on this topic in:
Pump Handbook
by Igor Karassik, William C. Krutzsch, Warren H.Fraser and Joseph P. Messina (Editors) 2nd Edition
Mc-Graw Hill Book Co.
ISBN 0-07-033302-5

(Guess I'm going to pitch this reference as a FAQ...)

Plot the curves (system and pumps) it will give you the best understanding of what is going on...
E.g.
Plotting the curve of one pump, 2,3,4...10 pumps in parallel
I will give you an idea of what happens with the system with each combination of pumps.
Regarding the system curves:
Plotting the system curves for different sections and adding the curves in series or parallel as appropriate will
give you a fair idea of how the system will behave.

HTH
Saludos.
a.

 

[MortenA]

Pucp

Yes you are right - and i beleive its the way to do it when doing the design - but remember the assuptions:

-The parallel lines are identical (up and downstream your manifolds)
-The discharge in the same elevation and includes (roughly) the same no. of bends valves etc.
-This means that the flow distributes evenly among the lines - and that the pumps deliver the same flow.

Its obvious that as "meeting point&quot e.g. the manifolds. th pressure must be the same for all incomming and exitsing lines. But if the discharge lines to the first merging manifolds are not identical or if the 4 parallel lines are not identical then the assumtions are not true. The flow at the merging point wil offcourse still be the same but flow distribution and pressure at each pump discharge may differ.

This problem could be solved both in a spreadsheet or using a simulation tool if the lines are not identical.

Best regards

Morten

 

[Sw1]

i know that connecting a few pumps in 'series' and 'parallels' yield different results respectvely, in terms of pressure capacity. How and what model (theoretical ref) to explain the result?

 

[Sw1]

Can someone help me calculate the vacuum pressure and/or suction pressure for the following system?
Case 1
1 pump only
Pump capacity = = 4.2m3/min
Hose diameter = 18mm

Case 2
2 Pump connected in series
Pump capacity = = 4.2m3/min
Hose diameter = 18mm


Case 3
2 Pump connected in parallel
Pump capacity = = 4.2m3/min
Hose diameter = 18mm

Case 4
2 Pump connected in parallel
Pump capacity = = 4.2m3/min
Hose diameter = 32mm

And
what would the suction pressure and suction rate if the hose is connected to a suction duct with opening of 20cm x 3cm (rectangular shape)?

System is a vacuum cleaner system for suction of particles from a surface.

 

[MJCronin]

Pucp,

Are you aware that several companies offer sofware to analyze the conditions that you describe?

What you seem to be asking about is various scenarios in a "flow network" ( i.e. variuos pumping conditions and pump configurations.

A very good program is AFT Fathom..... it has been around for a few years, is reasonably priced and can handle evaluation of multiple pumps in many scenarios.

Try

http://www.aft.com/products/fathom/

Let us know what analysis method that you finally decide upon....

Regards


MJC