parallel piping system with uneven flow rates

[jhabetz]

I have an existing system with a known flowrate and pressure drop consisting of a pump supplying CW to 5 parallel coolers. It is running at 60 gpm and 200psi. I want to divert some of the flow from the system (the pump is running near dead head)to increase the flowrate across some other coolers. The other coolers would require an additional 30gpm. The increased capacity would not significantly lower the discharge pressure. The mawp of the added coolers is 150 psi. The two paths meet at a common return header at 35 psi. I was thinking of using a pressure regulator to accomplish this, but I don't know how to figure the set point. It's probably not that complicated, although I'm still drawing a blank. Any suggestions would be greatly appreciated.

 

[rbcoulter]

I am not sure what your question is here. Adding the other two coolers would increase flow and decrease pressure as you mentioned. Getting 30 gpm into the other coolers is a different matter. You are at the mercy of the pump curve and the resistance of the coolers.

 

[BobPE]

rbcoulter

I am glad I was not the only one confused...I just was afraid to admit it...LOL....

BobPE

 

[yogibear1]

One thing you would want the pressure regulator to do is limit the working pressure to the new set of coolers. As you say one system is running at 200 psi, but the new coolers are limited to 150 psi; so on the tap to the new coolers the PRV should be set for a downstream pressure of not greater than 150 psi. As rbcoulter said you will be at the mercy of the pump curve. Depending upon the pump curve, if you don't have much or any excess pump capacity left, the PRV may have not to pinch down too much.

 

[plainwater]

As the pump is presently running at almost shutoff head, it seems a bigger pump was installed considering the future requirement of additional coolers. As the pump now is to work at 150 % flow while keeping the head more or less same, first check whether the motor capacity is OK for meeting the addtional load. For balancing the pressure in the hot return water proper detailed balancing calculations are to beperformed. Selection of pressure regulator or any other device will depend on the operation logic and the process parameters which are not clear from the question.

Hope it helps you

 

[jhabetz]

Thanks for your comments. I guess what I am trying to figure out is how do I split the flow so that I get 30 gpm in one direction and say about 55 gpm in the other. From the pump curve my discharge pressure will only drop about 5 psi. How do I calculate how much flow is going in each direction?

 

[rbcoulter]

You need to figure out the "resistance" of the coolers.
This may be provided by the manufacturer. It might be expressed as a C value. The resistance of valves and pipes in each branch most also be considered and then K values calculated for each branch. Now each branch will have the same pressure drop (DP) & the pump must operate on its curve. Use a spreadsheet & Darcy's equation. Try different DP values (goal seek) until the overall flow rate & DP agrees with a point on the pump curve.

 

[Bbird]

I would have thought that you would need a valve for isolating the new coolers from the existing units and could use this valve to “tune” the system. Basically I agree with Plainwater that you need a pump with a bigger capacity because your existing coolers have so much friction as equivalent to the shutoff head of the pump. That friction will not diminish unless you accept a lower flow to the cooler.

If you fit the additional coolers, with a valve as suggested, you can turn the valve until the combined friction loss of the valve and the new coolers matches the friction loss of the existing units. Then your pump will operate same as before and each cooler should receive nearly identical flow. Mind you the total flow has not been increased. If you open the valve fully, as though the new coolers have been fitted without the valve, more water will pass into the new coolers leaving the existing coolers starved of water.

You can play around with the valve to vary the flows in the cooler but you are not going to get more flow out of your pump unless the total resistance is reduced.

Calculation is possible as described by rbculter but you need to have a fairly good idea of the friction characteristic. Both the pump curve and the system characteristic (combined friction losses + any static lift) obeys the H=KQ*Q (H=head, Q=discharge, K=constant) and you can plot the two out. The crossing point of the two curves is your operating point. My suggested valve, equivalent to your pressure regulator idea, simply gives a means to vary the K value of your system characteristic. A quick way to appreciate your problem is treat K1=200/(60*60)=1/18 as your existing system. The new cooler will have K2=150/(30*30)=1/6. If you combined the two together then the combined K=1/18+1/6 = 2/9. Plot these graphs in an Excel spreadsheet and superimpose them with you pump curve (x=Q, y=H). You will see your pump is not man enough to do the trick you have got in mind.

 

[Bbird]

Correction

The previously suggested summation of the two friction constants is incorrect as it works only for pipes connected in series instead of in parallel. Thus flow increase is possible beyond the 60gpm. I have checked this by a computer program.

There must be a lot of computer software for flow through parallel pipes in steady state. Unfortunately I seldom use them as I work on transients (waterhammer) mainly. Others may be able to offer a more direct method of calculation to you.

 

[ivanlocke]

Download epanet (free).

It can evaluate a system in parallel.
It will allow you to input pump curves.
You may have to determine equivalent lengths of pipe from friction coefficients, though I've only used it a couple times.

Do a google search for epanet.

 

[fvincent]

Depending on the size of your system (or how much you can simplify it) I would propose you to give a look on the freeware version of FNESS, a software my company has released. It is based on the Finite Element Method and you can model pumps, lines, and so on. You can find it in

http://www.figener.com

fvincent
Figener S/A