Chilled Water System - Centrifugal Pump - Effect of Closing Isolation Valves 1

[RobertBarbe]

Apologies for the simplicity of the following question; however, I'm trying to confirm my understanding of how centrifugal pumps work.

I have a closed loop, chilled water system that has a centrifugal pump with pressurized supply and return lines. There are multiple consumers (i.e. heat exchangers, cooling coils, etc.) within the system each having its own isolation valve.

The question is: what is the effect on the flow rate through the remainder of the system if one or more of the isolation valves close? My gut feel is that the flow rate in the system should increase due to the reduction in total system resistance (i.e. the fluid has one or more less high resistance consumers to go through). I also assume that as the flow rate increases, the system pressure would increase back to its original pre-isolation valve closure level. If this is all true, does this mean that systems with centrifugal pumps maintain constant pressure (ignoring transients) since the flow self-adjusts?

Any help would be appreciated.

Thanks,

Rob

 

[DubMac]

You're throwing raw meat to the dogs here.....this post will end up with 40 or 50 answers. It takes intelligence to ask such a question, you have struck to the heart of pump theory; no apologies necessary.

I will just nibble a bit at your question and let some of the others tear it to shreds. I might also say that there are a few recent posts on this very subject...look for posts on system curves, performance curves and the like.

I first encourage you to look closely at the performance curve of the pump in question and remember that the pump is always going to operate somewhere on that curve (as long as speed is constant). ALWAYS. No matter what you do downstream with your valves, resulting flow and head will fall somewhere on that curve.

As much as I would like to get into system curves, friction vs static heads, and curve intersections, I will remain gracious and release the chained dogs to finish this one off.

I suspect Little Inch will pounce on this one quickly and efficiently.

 

[LittleInch]

If you look through some of the similar posts on this forum you'll probably find most of your answers, but try this one for atarters..

http://www.eng-tips.com/viewthread.cfm?qid=343261

It is very difficult to say what the impact of any particular valve being closed without a sketch of the system and understanding what the size, length and set-up is, but for the system you describe, the overall system flow will probably be very roughly proportional to the number of consumers providing you've sized your centrifugal pump and the pipework right.

Centrifugals are normally chosen such that at the "duty point", normally the max flow and differential head, the differential head (discharge minus suction) (pressure) should be around 10 to 15% below the no flow head. Thus the presusre drop accross any one piece of equipment should be more or less the same and hence flow goes in "lumps".

Of course the reality of any system of more than two bits gets much more complex than that and unless all the flow pipes between each element are identical then the pressure drop accross each item and its impact on the rest of the system may vary a lot.

Sketch your system and upload and you might get a bit more repsonse, but this is a very general question so gets a very general answer.

My motto: Learn something new every day

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

 

[LittleInch]

Just to clarify here - Dub Mac got in whilst I was writing my response....

My motto: Learn something new every day

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

 

[LittleInch]

I usually like to throw in an analogy or two. These sort of systems are analogous to electrical systems. Think of the pump as your supply voltage and the loads as well loads. If you size your cables (pipes) right then you won't notice much of a difference if you turn loads on and off. If you don't size it right then you notice the lights dim or get brighter when you turn something on or off, but essentially the electrical load (flow) is dependant on the number of loads you have connected. If you connect too many then the system trips the circuit breaker (pumps exceeds max flow). It's not quite right, but I think gets the general principle accross....

My motto: Learn something new every day

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

 

[BigInch]

If your chillers are all parallel and you shut down one of them while keeping the same flowrate in the system, resistance in the remaining flow path will always be GREATER. Therefore flowrate will try to go down since resistance increased, but you have a closed system, so it can't go down, there is no flow control. Consequently pressure will increase as the same flowrate is now forced through a system with a higher resistance.

Continuing with the electrical analogy. Two parallel resistors, 10 Ohm and 50 Ohm. 2 Amp Current. Let's say your chillers are 50 Ohms each, but you only have one 50 Ohm chiller and a 10 Ohm bypass in this circuit.
Net resistance is 1/[(1/10) + 1/(1/50)] = 1/0.12 = 8.33 Ohm
Required Voltage (ie. pressure) = 2 * 8.33 = 16.67 Volts

Required Voltage with the chiller removed, Ohms = 10, Current = 2 A, Voltage = 20 V

Adding any additional parallel looping bypass, even if it is 1,000,000 Ohm additional parallel resistance, or a 1/2" diameter pipe, will decrease the net resistance of the equivalent circuit. Removing one loop increases resistance and requires a higher voltage (pressure) if the flowrate remains the same.



Independent events are seldomly independent.

 

[RobertBarbe]

Thanks for all of the responses. I did look at the threads before I posted the question (plus various internet sites) but it was like drinking from a fire hose. I felt like I was looking at a TV screen 1 pixel at a time and couldn't put the whole picture together. Hence the general nature of the question.

Another general question: when looking at pump curves, I assume that I can equate pump head (i.e. work against graviatational potential) to system resistance (i.e. work against kinetic friction)?

Also, to push my question to it's logical limit: what if I closed all of the isolation valves in the system? I assume the total system resistance would drop dramatically and the pump would move towards its pump run-out point (i.e. max flow before damage).

Last question, why do people refer to centrifugal pumps as constant pressure if typical pump curves show that pressure & flow are related?

Rob

 

[DubMac]

Centrifugal pumps are not constant pressure machines. They are used in "constant pressure systems" or booster systems used to keep a constant pressure on a system no matter what the volume requirement is; such as in a hotel or other multi-use building. Maybe thats what you are thinking of??

Read some of the stuff written on generating system curves.

To start simply, the system curve will be made up of two types of pressures; static or elevation change pressures, and friction losses. The friction loss portion of the curve is what changes the shape of the curve as more flow is delivered through a system of pipe. Very approximately, the friction losses will increase as the square of the flow increase.

 

[LittleInch]

Ummmm, not quite. A pump curve is the curve that shows what the pump can produce for head vs flow. What is more difficult to produce is the system curve of the connected system. For a closed system like a chilled water loop, there is no static head, only dynamic friction losses. As flow increases the friction losses increase as you open up more chillers. This creates several curves which correspond to adding more chillers or exchangers. You get steady flow when the pump curve meets the system curve. I'll sketch something up and post it tomorrow to show what i mean.

If you closed all valves the resistance is essentially unlimited, not zero and therefore the pump will be at no flow and have the highest head.

As I said earlier, centrifugal pumps have a pump curve which is essentially flat, but drops off as flow increases, but by no more than 10 to 15% as long as you don't go past the duty point.

My electric analogy works here. If you turn all the switches (valves) off in your house you have no current flow. Equally the voltage in your house is essentially constant, but not quite, so long as you don't overload the system. If you connected 5 electric showers all at once then the lights would dim and the incoming cable would melt. Pump systems are similar.

Are we helping you get it?

My motto: Learn something new every day

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

 

[LittleInch]

Apologies for my scribble attached hopefully.

I've assume your system is relatively simple and has chillers in line with AHUs or HXs then return lines back to the pump.

The presusre drops through the equipment are assumed to be realtively constat for a contant Differential pressure as noted on the line diagram.

The pump curve is the top line and then the system curve for 1 circuit, then 2 then 3... There are multiple other option such as 2 chillers, 3 AHUs etc, but that's too many and would just be lines inbetween other system lines.

Where the system line crosses the pump line is where you find your flow.

As the pressure at low flow is a litle higher, then the single unit has the highest flow per unit. Each unit then adds slightly less flow each time as the pressure difference drops slightly. No flow (all units isolated) means no flow. Not good for the pump by the way so don't try it for more than a minute or so max

Hope this helps?

My motto: Learn something new every day

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

 

[BigInch]

Frictional resistance can be measured in pressure lost, or converted into head. It is the energy required, or that delivered by a pump, it is not work. It only becomes work when that amount of energy is delivered or expended during a given time. Since both are energy, and can be expressed in equivalent units, they can be equated. Pump head is that energy needed to overcome the equivalent frictional resistance of the piping system.

Closing all valves to the chillers? I guess you didn't understand my answer. What will happen is the opposite of what you said. Closing any fluid path of any number of parallel paths of any individual resistance always increases net resistance of the system. If flowrate remains the same when the resistance increases, pressure will have to increase, so the pump moves towards shutoff head, NOT TO RUNNOUT.

"People" say a lot of things that are only what they believe to be true. There often is quite a difference from reality. Perhaps they are referring to a pump with a discharge pressure control, or a variable speed drive set to hold the same discharge pressure. As you have found out, centrifugal pumps alone are not constant pressure, at least if the flowrate changes they're not.

Independent events are seldomly independent.

 

[LittleInch]

Big Inch as ever is completely correct, but in terms of "constant pressure" then a centrifugal pump, to most people who only come into contact with pumps occasionally, the pressure would appear to be "constant" providing that flow is less than its rated flow, within normally accepted margins of "constant", ie. 10 to 15%.

To take my electrical analogy further, most people would say that the voltage coming into their house / office is "constant". If you ever plotted it over a decent period, you would find it varied quite a bit, especially during periods of low / high load, but you don't really notice it. Same sort of thing here

Of course when you look at it in more depth, and different types of pumps can have much bigger changes in pressure with flow, then this generalisation becomes a bit ragged and incorrect unless , as Big Inch says, you do something to actually CONTROL the pressure such as a control valve or speed control to maintain an absolutely fixed pressure.

My motto: Learn something new every day

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