How manifolds affect pump performance

[ShakeNBake444]

This question is more for understanding then looking for a specific answer. But I'll start by describing my system a little, which i have a little schematic attached. I have a big pump supplying flow to a series of manifolds that coil copper coils that have a sizable amount of current going through them and therefore need to be cooled. This schematic just shows a small section of my system but there are more manifolds spanning out to 80 meters. So my main question is as I attach more manifolds to my system, how does the pump performance change? Does the flow change? I usually calculate the amount of flow im getting through the manifold outlets with they darcy weibach equation and just solving for Q since i know everything else. The pressure drop i use is the 20 psi I am showing in the diagram. Once i get the flow i check whether or not it meets the cooling requirements. But I'm not sure if this is correct to just solve for flow because the flow will also depend on how many manifolds are connected to it, right? as the 2" pipe becomes longer we install booster pumps to supply more pressure to the farther away manifolds. But this puts stress on the main pump and sometimes overheats it. Any idea why this would occur? Thanks in advance!

 

[BigInch]

More manifolds means more pressure drop (becoming greater than the 20 psi you have assumed) and probably less flow as that pressure drop increases. As flow drops you eventually get to a point where the flow is insufficient to adequately cool the pump.

 

[Compositepro]

Adding manifolds will almost certainly increase the amount of fluid flow. Although the manifolds are technically in series, the flow loops supplied by the manifolds are all in parallel. This would also explain the overheating of the pump motor due to the increased flow. Use of booster pumps with the additional manifolds will certainly increase total flow through the main pump.

 

[lilliput1]

Get the pump curves at its operating rpm. Know that the pump will operate where the pump curve intersects the system flow vs pressure drop curve. Check the system requirement. The cooling required will be a % of the power input to the electrical equipment cooled. You can reduce pressure drop and increase flow by connecting another 2" pipe close to the pump discharge and connecting it just upsteam of half the number of manifolds in series. Now how is the flow to each coil controlled? Is it a manual valve or an automatic temperature controlled valve, meaning you have variable flow. Also is the flow recirculated or not (flow discharged out of the system after going through the coil)? Is it a closed loop and the water from the coils is returned, cooled then go back to the pump suction? If flow is variable and may drop, pump will overheat if the flow is stopped or considerable reduced. If closed loop and insufficient cooling is done, pump will also overheat. Provide adequate cooling. Make sure pump flow do not go below minimum by either putting a differential pressure control valve to discharge the minimum water flow when it set pressure is reached, or by omitting the automatic control valve to the most remote branches that would equal minimum allowable flow. Verify it is ok for coil temperature to drop down to the minimum supply water temperature if the automatic temperature control valves are omitted. You have to modify the system curve to give you the flow you need and be on the correct operating point on the pump curve.

 

[LittleInch]

shaknbake,

I think you've answered a few of your own questions but see my answers to your questions below:

How does the pump performance change? - See the pump curve. Most centrifugal pumps operate within their working flow range with a pressure which doesn't change much beyond 20%. However they have a maximum flow at which point pressure falls and power increases as efficiency falls also.

Does the flow change? - Frankly this one answers itself. You're adding more units which take flow so what do you think???

The question is how are you controlling flow? Normally with a long header system, you need to restrict flow through the units closer to the pressure source to prevent excess flow and not enough flow going to those further away. I would suggest you control on flow via a control valve and monitor flow directly rather than trying to calculate.

As said above and by Lilliput, see your pump curve. Once you go beyond the end of the curve, you've reached maximum flow for this pump and need to add another one in parallel, not series.

Without knowing any more details it's difficult to comment much further, but I suspect your pump is running off the end of curve and the 2" pipe is too small to feed lots of manifolds 80m away. If you're adding a booster pump in series you're doing the wrong thing - this is normally used when you need to increase pressure at a fixed flow, but you need to increase flow. Clearly to get more fluid down the same pipe you also need to increase pressure or make your pipe bigger, or stage them so one pump feed the first 4 and a second pump feed the next 4 or dual up your headers so one pump feeds manifolds 1,3,5,7 and the other 2,4,6,8. multiple ways to solve this, but it needs more info.

Normally for a header type system, you want a large header, which at the extremity only uses up 10-20% of overall friction losses leaving the majority for the units which also helps as the pressure for all of them should be within a fairly small band ( 10-15%).

Also like Lilliput says - where is this heat from the coils going? If the system is a loop system then excess heat ( pump or motor overheats?) is possible.


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

 

[lilliput1]

I forgot to mention that to generate the system curve calculate the system pressure drop for the required flow then estimate the pressure drop at other flows using P2 = P1 x (GPM2/GPM1)^2

For open circuits the typical design maximum gpm for each pipe size (with allowance for open system pipe aging):
Pipe Size Inches GPM
1/2 1.4
3/4 3.0
1 6.0
1.25 12.0
1.50 17.0
2 35.0
2.50 55.0
3 95.0

For closed circuits the typical maximum gpm for each pipe size (with allowance for closed system pipe aging):
1/2 1.9
3/4 4.0
1 8.0
1.25 15.0
1.50 20.0
2 45.0
2.50 75.0
3 125.0

Remember that in balancing the pressure drop from the pump discharge through each branch must be equal because for open systems the end is all at athmospheric pressure while for closed systems all the branches eventually converge to a point in the piping. So balancing valve in each branch have be adjusted to create the pressure drop needed to make the pressure drop equal to that of the worst branch.

 

[ShakeNBake444]

These are all great responses and apologize for not posting sooner but I was away for the weekend. The facilities engineer is away on vacation so I can't pick his brain on the particulars at the moment. I'm hoping to do all my homework by the time he gets back so we can find the best solution. Having said that I will try to paint a better picture. But first, I should answer some of your questions:

How is the flow controlled?

at each unit that needs to be cooled, it's basically not controlled actively but rather knowing the pressure drop across the equipment we have an idea of the flow through it. The water coming off of the equipment flows to a reservoir and then gets pumped (by a totally separate unit) to a chiller. So the water leaving the equipment is at atmospheric. So the cooling supply circuit is an open loop. This is the case near the equipment that needs to be cooled. I don't know what helps control flow near the pressure source but will find out.

I work in a particle accelerator facility so cooling problems are abundant. I'm new so bear with me. I design magnets and usually never deal with facilities problems like these but with these new magnets we plan on installing, we want to design coils that can carry more current but not have to upgrade the entire cooling system. so i have to ask these questions up front so i can make something that meets those requirements. These coils carry several hundred amps of current totaling about 10 kW per coil. They are hollow conductors whos coolant flows directly through the center of the conductor along the entire length of the coil. the cooling channel is a circle. There are several separate coils per magnet due to the length of the coil (Energized in series, cooled in parallel). at each output on each manifold you have a coil, hence the diagram. I've been told at the location we intend to put these magnets there will be 20 psig available to us. This is essentially all the info i have at this point. So my first natural idea is to design a coil length and cross section that has a pressure drop below 20 psig. I'm also told that "our system cannot handle any coil with a higher pressure drop than 20 psig". This statement makes no sense to me for the following reason... The facilities engineer does not care how cool or hot my magnet is. It's my job to worry about that. He just cares about how the cooling system is operating. Knowing the supplied pressure in the supply line and pressure drop in the coil, i can calculate the amount of flow i can expect through the coil. If it is higher than a certain value, then we are good. So i offer the following thought experiment in reference to the first schematic i showed: In the schematic, all the flows are acceptable and the pump is behaving as designed and everyone is happy. But then i want to install another coil that needs cooled. Except this coil has an infinitely small cooling channel and therefore an infinite pressure drop (like adding a dead leg). If my understanding of this problem is correct, then nothing will flow through the coil and it is as if I have done nothing to the cooling system. So this makes me think... well the larger the pressure drop in what i add, the better! - at least for the facilities engineer.. so why would he say that his system cannot tolerate anything larger than a 20 psig pressure drop?

 

[lilliput1]

Isn't there danger of electrocution and short circuiting?
I suppose 20 psig is the current pressure at the pump discharge. But look at the fan curve and see what flow is available at this pressure. You have to modify the system piping so the new system curve will intersect the pump curve at the correct total flow and system maximum pressure drop. There has to be balancing valves at each branch otherwise more flow will occur at branches nearest the pump. You have to determine the cooling effect required and supply the correct flow. If the cooling load at each branch is the same the temperature differential across each coil and the gpm should be the same. If the load is different and variable, the flow should be automatically controlled to maintain the design coil leaving water temperature.

If the chilled water temperature differential is 10F (verify) and if all 20kw converts to heat (verify) the required gpm to each coil is"
GPM per coil = 20 x 3412.142 /(500 x 10) = 13.65

This seems high and 2" pipe size for main will not be enough.

 

[LittleInch]

shaknBake,

It's sometimes easier to think of this set of circuits as an electrical circuit.

What your facilties engineer is telling you is that the voltage (=pressure) of your system is set at a certain voltage and to change it would need a different cable ([pipe) and a new transformer (pump).

Your cooling coil is like a resistance back to ground ( I know resistances heat but bear with me in this analogy. Your current set up works OK, but the resistance nearest the power source actually takes more current than it should, but if this is ok then no problem.

However as you add more and more parallel resistances the voltage reduces, the cable and transforme heat up and no one is happy.

Hope this helps in understanding a bit more, but also
Like I said above, your supply poressure is essentiall fixed within a fairly small band. If you calculated that to get the cooling flow you needed through your coils you needed say 35psi, then that pressure isn't there. You would get some flow, but not enough to do your cooling.

Whilst you may not care about excess cooling, the reality is that this is essentially cooling flow which can't be used for any other manifold.

What you should be doing is working out how much flow you actually need and then control the water flow to use only that amount and no more. then you might have the capacity in the pump / pipe to add another manifold or two.

If you want to add flow (cooling), but not change anything, then you need to make the best use of what you have.

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