calculating Pump Head for 1 suction multiple parallel discharges

[nicwong7]

Hello all,

I am unsure of how to size a pump when i have multiple discharge points. I have always done 1 suction 1 discharge which is easy and straight forward with only 2 points to consider.

Now i have 5 points to consider..

To keep it simple,
One suction point from an atmospheric tank
4 discharge points equal flow equal losses inside the pipes (parallel)
The distances are quite long. But i am not sure the method of calculation.

Tank-> Pump----1.5km--->T off and split into 4 equal pipes.

How do i use the bernoulli equation to size my pump head with multiple points.

P1/pg + Z1 + u1^2/2g + he = P2/pg+ z2 + u2^2/2g + (Straight pipe+ fitting loss)


Thanks in advance!

 

[BigInch]

You can probably neglect u^2/2/g, it's usually very small, and take the highest total of any and all points of Z/p/g + P/p/g

 

[rotw]

I guess you divide the total flow by 4.

That gives you the individual flow for the pump, lets call this: Q.

You calculate the total losses based on 4 x Q for the common suction line, Q for individual discharge pipe.
You should also take into account the piece of pipe between the common suction manifold and the pump suction.

Add to the total losses the geometric head (suction and discharge) based on the highest elevation so this gives you the pump head H.

Then you specify each of the 4 pumps for this Q, H.

"If you want to acquire a knowledge or skill, read a book and practice the skill".

 

[bimr]

What is the application? Is this an industrial application with a loop?

 

[rotw]

Sorry I went a bit too fast ; I thought each of the 4 lines has its own pump.
Seems you have only one pump.

"If you want to acquire a knowledge or skill, read a book and practice the skill".

 

[ccfowler]

Unless very substantial variations from equal flow rates at the four discharges can be tolerated, I would include some suitable means of flow measurement and flow control into each of the four discharge lines to keep the individual flow rates within tolerable ranges for their specific needs. A design that depends on piping geometry alone to provide some specific apportionment of flow to each of several discharges can be expected to produce unpleasant results. Depending on the precision with which the flows must be regulated, throttling losses associated with the use of control valves can be very modest. Allowing for suitable controls in the original design will involve much less bother and expense than trying to correct the problems after the system has been built and placed in service.

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.

 

[nicwong7]

Hello All,

Thanks for the great advices. Yes this is for industrial application. Let me break it down.

It is a chemical dosing pump that is used to pump NaOCL into 4 identical seawater intake pipes(similar depth) which is more than 1 km off shore and 20m deep(It is an open pipe to allow seawater to enter). Dosing end point is at the entrance of the seawater pipes 20m down. Of course the purpose is to prevent shelling from growing and depositing on the surfaces of the pipes.

My main concern is the pressure that i would size this pump which is on the Ground level. pardon me for being mathematical about this as i always look upon the bernoulli's equation.

Now that i understand how the losses is accounted for. (4Q common header pipe loss) + ( 1Q pipe loss because parallel pipe losses are simlar)

Considering Pumping from dosing tank Level to sea water surface Level(Bernouli)
My P1 would still be atmospheric from my chemical dosing tank. My P2 should it be atmospheric? Technically i am discharging it to "atmosphere" seawater. My elevation difference will only be the height of my dosing tank asuming its on the shore.
If i do this way. I will not account for the static head of the pump head to push it 20m deep down the seawater where my discharge point is.

Considering from dosing tank level to 20m beneath the sea,

My P1 is atmospheric but P2 will be 2 bars of pressure and my Z2 will be -20m. My elevation difference will be -20. which means the overall pump head will -20m. Somehow cancelling the 20m head required to pump it down the water.

Which is the correct way?

Thanks!

 

[LittleInch]

Forget bernouilli is my advice and just concentrate on friction and head losses.

Assuming your NaOCL density is similar to seawater, the your negative head (if you like) is the difference between you pump discharge and sea level as the head gain in your pipe is matched, more or less, by the static pressure of the sea 20m down (roughly 2 barg).

Then just work out the friction head losses in ONE of your four "equal" pipes plus the friction losses in your common pipe then subtract the head gain caused by you pump being above sea level and that gives you your discharge head.

Then work out the head gain from the top of the liquid level in your tank minus the friction losses in the pipe from the tank to the pump, add atmospheric pressure as a head and you get the head of liquid arriving at your pump inlet. subtract discharge from inlet and you get your pump differential head.

Note that if your injection pipe is open ended and your pump is more than 10m above sea level, when your pump stops you could easily pull a vacuum on your common line.

whay you need to develop, if you haven't, is a profile sketch. Post that and all will become apparent.

My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way

 

[bimr]

Yes, forget Bernoulli on this one.

You would be better served to procure 4 metering pumps for this application. Metering pumps are relatively inexpensive and it will simplify your pumping requirements.

With a metering pump in a manifold system, you would have to throttle the discharge of all of the other outlets to force the flow into different outlets. The valve balancing equipment would probably cost more than the metering pumps. This would take backpressure valves on each of the 4 manifold outlets. You will also never be able to accurately balance the system. There is also no method to prove that you are pumping the desired flow rate to each outlet.

 

[LittleInch]

bimr - True. What we don't have is the relative length of these 4 "equal" runs compared to the main one. Unless the pressure drop in the four lines is probably over 75% or more of the total then you could very easily get lots more (all?) flowing down one line compared to the others and you would never know without some relatively sophisticated metering and control systems.

By the sound of it the long lie is relatively long compared to the four other lines so running four lines seems like the best plan to me also.

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[nicwong7]

Hello LittleInch and bimr,

Thanks for your inputs on this question, i guess i have been too focused on the bernouli's and putting those away and think logically of the losses is a better option. As i have just been into the industry not too long ago and hope to learn more.

Here's a question, when you size all these pumps, do you put a safety factor of maybe 10% for the losses? Because calculating the losses are theoretic but in actual, sometimes i get a lower pressure then what i sized for even when the pipes are all new.

Also, the feedback on the 4 pumps to 4 lines sounds like a better option.

Thanks.

 

[bimr]

Oversizing pumps depends on the application. For a large centrifugal pump, adding 10% will drop the pump efficiency and cause significant power usage over the long run, so it would not be recommended.

For your metering pump application, it is recommended to oversize the pump because you do not want to be operating at the maximum capacity of the metering pump.