In a closed loop system with a pump, how can we control the pressure?

[Chano]

TL;DR Summary

Control of water pressure in closed loop system with a pump.

How can we control the pressure of the water inside a closed loop system (chiller system for example)?
Let´s say, we have a pump curve and an system resistance curve that can be modified (through opening or closing some valves)

In everywhere, what I see is that the intersection of the system curve with the pump curve marks the operating point of the system.
My question is, at the operating point, what´s the water pressure in the closed system? I would say that it is 0 barg everywhere as the energy given by the pump is cancelled out by the resistance. If that is right, what can we do if we want the system pressure to be 2 barg (we want the water to be pressurized)?

Thanks!

 

[hutchphd]

Would the cooling system of an automobile be considered such a system?

 

[Chano]

Would the cooling system of an automobile be considered such a system?

I would definitely say so, since there is no phase transition and there is a pump and some pipes.

 

[hutchphd]

So a "pump curve" is outflow velocity vs pressure?

 

[Chano]

So a "pump curve" is outflow velocity vs pressure?

kind of, it´s head versus flow rate. From my understanding, the head or equivalent pressure is the resistance that the pump has to overcome. The lower the head (resistance), the higher the flowrate (mass flow).

here´s an example

 

[Lnewqban]

The static pressure within the loop is maximum at the pump outlet, decreases along the loop, and reaches its minimum value at the pump inlet.
As the different users close or modulate individual flow, either a bypass, a self-balancing valve or a VFD system acts to keeps the pump happy and the chiller from freezing.

 

[DaveE]

Are you asking about the average pressure of the entire system? The (gauge) pressure at a specific location in the system? That curve tells you what the pressure differential is across the pump on the vertical axis. I'm confused about which pressure you are asking about.

You also haven't told us much about your system. The chillers I worked with had a reservoir that was at atmospheric pressure so it could be big and easy to open to refill. This would be at the inlet to the pump. So the pump outlet is the highest pressure as shown in the P vs. F graph. The pressure elsewhere in the system depends on the details of construction.

 

[Chano]

The static pressure within the loop is maximum at the pump outlet, decreases along the loop, and reaches its minimum value at the pump inlet.
As the different users close or modulate individual flow, either a bypass, a self-balancing valve or a VFD system acts to keeps the pump happy and the chiller from freezing.

I thought that maybe due to the closed nature of the system, the static pressure would be uniform in the system, guess I was wrong.

I am curious about the pressure at the inlet of the pump in the case where a too powerful pump is selected and the static pressure it adds is higher than the loss, what would happen? Will the static pressure at the inlet gets higher and higher? My question is about whether the pump changes its behaviour if the static pressure of the fluid it is pumping is more than 0 barg, what is the steady state of the system?. Assuming we don´t have pressure relief mechanism.

 

[Chano]

Are you asking about the average pressure of the entire system? The (gauge) pressure at a specific location in the system? That curve tells you what the pressure differential is across the pump on the vertical axis. I'm confused about which pressure you are asking about.

You also haven't told us much about your system. The chillers I worked with had a reservoir that was at atmospheric pressure so it could be big and easy to open to refill. This would be at the inlet to the pump. So the pump outlet is the highest pressure as shown in the P vs. F graph. The pressure elsewhere in the system depends on the details of construction.

Sorry about not being clear. I would like to know if the gauge pressure is the same in every part of the system (if we can neglect dynamic pressure because the fluid velocity is about 3 ft/s). Or it depends on the part of the system where you place the pressure gauge.
Here´s a simple scheme of the system

Both dry cooler and heat exchanger are merely coils where the water pass through. So this is a completely closed system full of water.
The system is fully closed because it actually contains glycol solution (50% water, 50% glycol), which is toxic.

My another question is that if the pressure added by the pump is not compensated by the head loss, does the pump keep adding pressure even if the pressure of the fluid is already high?

 

[Lnewqban]

I thought that maybe due to the closed nature of the system, the static pressure would be uniform in the system, guess I was wrong.

I am curious about the pressure at the inlet of the pump in the case where a too powerful pump is selected and the static pressure it adds is higher than the loss, what would happen? Will the static pressure at the inlet gets higher and higher? My question is about whether the pump changes its behaviour if the static pressure of the fluid it is pumping is more than 0 barg, what is the steady state of the system?. Assuming we don´t have pressure relief mechanism.

If there is substantial flow, there must be a pressure differential between the two ends of the pipe system (where the pump is located).

Coils and valves are the big pressure drops that the glycol suffers in your system.
Centrifugal pumps naturally compensate for what the system demands from it.
If more pressure differential is needed (a valve is closed, for example), the flow gets reduced, and vice-verse.

The curve that the manufacturer shows for such pumps is the result of experiments where pressure differential is artificially change, only to measure how much flow results for each case.
The curve of the system (excluding the pump) shows only how the flow varies as more pressure is pushing the fluid to go through it.

An economical selection of a centrifugal pump considers both curves and try to locate the normal operation point of the system as close as possible to the point of maximum efficiency of the pump (more delivered energy into the fluid with same consumption of electricity).

Please, see:
https://en.wikipedia.org/wiki/Centrifugal_pump_selection_and_characteristics

https://www.michael-smith-engineers.co.uk/resources/useful-info/centrifugal-pump-selection

https://www.pumpfundamentals.com/download/book/chapter4.pdf

 

[Chano]

If there is substantial flow, there must be a pressure differential between the two ends of the pipe system (where the pump is located).

Coils and valves are the big pressure drops that the glycol suffers in your system.
Centrifugal pumps naturally compensate for what the system demands from it.
If more pressure differential is needed (a valve is closed, for example), the flow gets reduced, and vice-verse.

The curve that the manufacturer shows for such pumps is the result of experiments where pressure differential is artificially change, only to measure how much flow results for each case.
The curve of the system (excluding the pump) shows only how the flow varies as more pressure is pushing the fluid to go through it.

An economical selection of a centrifugal pump considers both curves and try to locate the normal operation point of the system as close as possible to the point of maximum efficiency of the pump (more delivered energy into the fluid with same consumption of electricity).

Please, see:
https://en.wikipedia.org/wiki/Centrifugal_pump_selection_and_characteristics

https://www.michael-smith-engineers.co.uk/resources/useful-info/centrifugal-pump-selection

https://www.pumpfundamentals.com/download/book/chapter4.pdf

I see, if the pressure is uniform, there will be no flow at all. On the other hand, I do realize that the head loss through the pipes is actually negligible compared to the pressure drop in coils and valves.
And I think I was interpreting the centrifugal pump curve wrong, the pump performance actually depends on the flow conditions and it self adapts to them, meaning that if the resistance of the closed loop system is low, the pump would just pump more flow (adding kinetic energy), it will not stick to a fixed flow and add more pressure. Until the point where a balance between the pressure drop and the differential pressure of the curve match for a certain flow rate (because the pressure drop depends on the flow rate too), then that flow rate is the system flow rate in steady state.

We are choosing a pump taking into account the friction losses only, not the total dynamic head because it´s a closed system. We have a design flow rate, we were just not sure whether the pump we have chosen would overpressure the system if the real head loss (we have estimated one) is lower than the pump´s head for that given flow rate, now we know that the pump would just pump more flow rate and we can actually modify valve openness (forgive me for using this word) to add resistance and make sure that the flow rate reduces and matches the required one.

Thanks for the links, we will check it out!

By the way, Lnewqban, is it necessary to implement a pressure relief valve that allow the flow at the discharge part of the pump to go back to the lifting side? I think it is not necessary as long as the pump shut-off pressure(at 0 flow) is lower than the maximum system operating pressure, are there any other reason a pressure relief valve might be needed in a centrifugal pump system?

 

[Lnewqban]

Normally, conditions of zero flow are to be avoided, as water inside pump is quickly heated up by the rotating impeller and mechanical damage to pump’s internal seals could follow.
Another bad thing to avoid is excessive restriction or high temperature or height on the suction side, because that could induce cavitation, which would damage the impeller.

Expansion tanks or relief valves in the loop are used more with dilatation of the fluid due to temperature and the water hammer phenomenon in mind than eventual pump overpressure.
Pump motors controlled by variable frequency drives is the modern way to modulate pump flow as needed while saving electricity.

Please, see:
https://en.m.wikipedia.org/wiki/Cavitation

https://en.m.wikipedia.org/wiki/Variable-frequency_drive

You are very welcome.

 

[Chano]

Normally, conditions of zero flow are to be avoided, as water inside pump is quickly heated up by the rotating impeller and mechanical damage to pump’s internal seals could follow.
Another bad thing to avoid is excessive restriction or high temperature or height on the suction side, because that could induce cavitation, which would damage the impeller.

Expansion tanks or relief valves in the loop are used more with dilatation of the fluid due to temperature and the water hammer phenomenon in mind than eventual pump overpressure.
Pump motors controlled by variable frequency drives is the modern way to modulate pump flow as needed while saving electricity.

Please, see:
https://en.m.wikipedia.org/wiki/Cavitation

https://en.m.wikipedia.org/wiki/Variable-frequency_drive

You are very welcome.

Thank you very much, Lnewqban, we were worried about the project and after your help, things became clearer!

 

[Lnewqban]

Thank you very much, Lnewqban, we were worried about the project and after your help, things became clearer!

My pleasure.
Happy to help.

 

[Chano]

My pleasure.
Happy to help.

Hi Lnewqban, can I ask you about NPSHa in closed systems?
From our previous analysis we think that the pressure on the suction side should always be the static pressure (atmospheric pressure) and NPSHa would be Patm - Pvapor, which is independent of the friction.

 

[Chestermiller]

If it is a truly closed loop, the exterior pressure is immaterial.

 

[Chano]

If it is a truly closed loop, the exterior pressure is immaterial.

Yes it should be independent of the exterior pressure. We said atmospheric pressure because we though that the fluid pressure at the moment the circuit was first time filled would be the atmospheric pressure at the location where it is filled (let´s say, 1 bar). The later changes in the height (let´s say, you transport it to the mountains) would not change the internal initial pressure, it is still at 1 bar.
By that I mean that it is more correct to say that NPSH would be Pcharge - Pvapor, where Pcharge is the pressure at which the circuit is filled? Can it be atmospheric?

 

[Lnewqban]

Hi Lnewqban, can I ask you about NPSHa in closed systems?
From our previous analysis we think that the pressure on the suction side should always be the static pressure (atmospheric pressure) and NPSHa would be Patm - Pvapor, which is independent of the friction.

Please, see this old thread:
https://www.physicsforums.com/threads/calculate-pump-height-by-using-the-bernoulli-equation.985469/

I would recommend to get help from a local mechanical engineer who could evaluate the whole system and provide good solutions, not only for design, but for installation and operation of the pump, valves and pipes of your system.

 

[andrew s 1905]

The classic solution is a kickback loop (value outlet to inlet) with a control valve in the loop. By changing the flow rate (through the kickback loop) you control the outlet pressure via the pump curve.
Regards Andrew

 

[gmax137]

is it necessary to implement a pressure relief valve

If the system is built to some code (local or national, eg ASME) then maybe.

 

[DaveE]

If the system is built to some code (local or national, eg ASME) then maybe.

Or if there isn't an applicable code, but you don't want bad things to happen. Regulations aren't a substitute for doing your own worst case FMEA analysis. So yes, maybe, IDK.

 

[Flyboy]

By the way, Lnewqban, is it necessary to implement a pressure relief valve that allow the flow at the discharge part of the pump to go back to the lifting side? I think it is not necessary as long as the pump shut-off pressure(at 0 flow) is lower than the maximum system operating pressure, are there any other reason a pressure relief valve might be needed in a centrifugal pump system?

Expansion tanks or relief valves in the loop are used more with dilatation of the fluid due to temperature and the water hammer phenomenon in mind than eventual pump overpressure.

I'll chime in here with a bit of input from aviation, especially with fuel and lubrication systems.

With a centrifugal pump, there is no need for a bypass, because, as others have mentioned, it'll just backlog and not drive as much fluid through. On the other hand, there's a reason we use centrifugal pumps as fuel boost pumps: If the pump motor fails, fuel can still be delivered past the pump by gravity or suction.

On the other hand, displacement pumps do need some sort of bypass or relief system to prevent damage. With an axial piston pump with a variable swashplate, like the kind used in the Constant Speed Drives of airliner ending driven generators, you can link the swashplate to a pressure driven actuator, set up so a spring pulls it to the maximum angle allowable, and you use the actuator to reduce the swashplate angle. It then becomes self regulating depending on the force of the spring used to counter the actuator.

Fixed displacement pumps, like a gear pump or a simple piston pump, will require a bypass or relief valve, usually recirculating to either the reservoir or to the pump inlet, to provide pressure control. Aircraft engine oil systems use this system, especially the piston engines on light aircraft. You can adjust the oil system pressure by adjusting the backing screw for the relief valve spring, though this is almost never needed in the field.

Likewise, any spots where you could accidentally develop a blockage under expected operations should have a bypass valve to ensure continued flow of the fluid. Oil filters, for example, sometimes have a spring-loaded port that can open to bypass oil past a clogged filter. After all, contaminated oil is better than no oil. And in the case of oil coolers, the bigger warbird engines will sometimes have a thermostatic bypass valve at the oil cooler to allow the oil to bypass part of the oil cooler (usually by flowing through the outer jacket) when cold, as aviation oil can become very viscous when cold. As the engine warms up and the oil flows around the bypass, it will gradually warm up the oil in the cooler and the bypass valve, allowing the oil to start to flow through the cooler.

 

[Fisher]

Yes it should be independent of the exterior pressure. We said atmospheric pressure because we though that the fluid pressure at the moment the circuit was first time filled would be the atmospheric pressure at the location where it is filled (let´s say, 1 bar). The later changes in the height (let´s say, you transport it to the mountains) would not change the internal initial pressure, it is still at 1 bar.
By that I mean that it is more correct to say that NPSH would be Pcharge - Pvapor, where Pcharge is the pressure at which the circuit is filled? Can it be atmospheric?

Hello, Chano and everyone.

I totally understand what you are saying, because I have similar question with yours.

Now, I also design a closed loop generation cycle (typical Rankine cycle).
and I am curious how the suction pressure of the pump will be changed from the average pressure
(equilibrium pressure of the cycle when the pump turn off and zero flow) when the pump turned on.
In other words, dose it decrease or increase from the average pressure?

The pump curve just let me know about the differential pressure between the pump. So, I cannot expect the exact value of suction and dischange pressure of the pump.

Chano. Do you have the answer for your question?
please share the final results of this discussion.
and anyone who can help me always welcome.

Thanks.

 

[gmax137]

Now, I also design a closed loop generation cycle (typical Rankine cycle).
and I am curious how the suction pressure of the pump will be changed from the average pressure
(equilibrium pressure of the cycle when the pump turn off and zero flow) when the pump turned on.

In a steam power plant Rankine cycle, the condenser hotwell pressure is a boundary condition, set by the temperature of the water (lake, river, ocean) circulated through the condenser tubes. The pressure at the (condensate) pumps is then determined by the losses (flow and elevation) in the piping from the hotwell.

Not sure if that helps?

 

[russ_watters]

I'd say the inlet pressure has to drop when the pump is turned on, but it is a dynamic situation when starting up the boiler.

 

[Lnewqban]

Welcome, @Fisher !

For defining entering and leaving pressures, you have to consider the changes of fluid temperature and vapor pressure inside the loop.
I believe that there is a big difference between the conditions of the fluid at normal operation of the system and at "equilibrium pressure of the cycle when the pump turn off and zero flow".

The pump is nothing but a tool to transfer electrical (or steam) energy to mechanical (and some thermal) energy of the fluid inside the loop.

A centrifugal pump only "knows" about mass flow and pressure differential, as its performance curve (for certain temperature range) indicates.
The lowest limit to suction pressure is cavitation.

Please, see:
https://es.wikipedia.org/wiki/Cavitación#Bombas_y_hélices

https://en.wikipedia.org/wiki/Net_positive_suction_head

 

[Maher5127]

Yes it should be independent of the exterior pressure. We said atmospheric pressure because we though that the fluid pressure at the moment the circuit was first time filled would be the atmospheric pressure at the location where it is filled (let´s say, 1 bar). The later changes in the height (let´s say, you transport it to the mountains) would not change the internal initial pressure, it is still at 1 bar.

By that I mean that it is more correct to say that NPSH would be Pcharge - Pvapor, where Pcharge is the pressure at which the circuit is filled? Can it be atmospheric?

Hi Chano
i think i understand your question perfectly because this idea had crossed my mind for quite sometime
thought i have no final guaranteed answer yet i have some insights and here they are according to my understanding,

if a closed loop is filled and closed with an X amount of static pressure then physically there is no way for any point in this closed loop to have a static pressure less than this X amount EVEN if we installed a pump and run it

for example lets say we have filled the loop with 3 bar static pressure and installed a pump and run it then then the lowest static pressure in the whole loop will be the pump suction side and this static pressure must always be 3 bar independent of the flow rate

so if equilibrium/steady state flow condition defined by the intersection between system and pump curves was at 100 m3/hr and 50 m head (5 bar approx. in case if water) then the pump discharge will be 8 bar and suction forever 3 bar and the 5 bar differential are the friction losses within the loop
if you throttle discharge valve more the flow will decrease and differential head/pressure will increase say will be 9 bar then at this case still discharge pressure will be 12 bar and suction pressure forever 3 bar and 9 bar friction losses and so on

for this reason i think there is no way according to physics to do NPSH testing for a centrifugal pump in a closed loop because there is no way to reduce static pressure at suction less than static pressure at filling the system unless you install a vacuum pump to reduce suction pressure by reducing ''the initial pressure at filling''

HOWEVER, i have seen NPHS testing before with a closed loop and a throttle valve at suction without vacuum pump and pressure is reduced which makes my doubt my physics understanding illustrated above

this is why i need your insights if someone have a clear answer to correct me if i'm wrong
my thinking is, in case my understanding is right which contradicts the physical fact with actual testing i have seen is that maybe pressure is reduced due to turbulence at suction with mean value staying the same which is the static pressure at filling and the pressure guage is recording a fake/inaccurate reading and the test itself is not accurate in this way