Buid-up pressure in a closed loop circuit 1

[Giskard]

We have a closed cooling circuit with a pump and a heat exchanger, (with a by-pass pipe to control the cooling). This circuit is basically closed, because, although the suction of the pump is also connected to an atmospheric tank, the pipe connecting to the suction pump has a check valve, (and this check valve doesn't participate in the fluid recirculation through the pump and the heat exchanger).

I've observed a pressure transient in the circuit that I can not explain:

* When this system is tested, recirculating the fluid through the heat exchanger, the pressure at the pump suction flange is around 2 bar (basically, the pressure with the system stopped and at rest, and equal to the static head from the tank) while the pressure at the discharge flange is around 12 bar, (with a flow of 210 m3/h). These pressures are basically constant, if the temperature of the recirculating fluid is constant. If we keep the fluid to heat up a little, the pressure at both suction and discharge flange increases, (with a constant differential pressure through the pump). Obviously, I assume that this behaviour is due to the fluid dilatation.

* However, after recirculating the fluid for some time, with the water adequately cooled, when we stop the pump, the pressure at the suction flange, in 4 o 5 seconds after stoping the pump, builds up to 5 or 6 bar. That is to say, the pressure at the suction increases from 2 bar to 5 or 6 bar, while the discharge pressure decreases from 12 bar to the same 5 o 6 bar.

* This pressure retained in the circuit slowy decreases with time and, after 15 or 20 minutes, the pressure returns to the original value of 2 bar.

I assume that this decrease is due to a small leakage through the check valve to the atmospheric tank, but ¿what causes the pressure at the pump suction to raise up to 5 bar just when we stop the pump?

¿Has anybody seen a similar behaviour in closed loop circuits?

Thanks in advance
Ferran

 

[KenRad]

Giskard,

In the pump suction line, you describe a connection to an atmospheric tank with a check valve in the line, presumably allowing makeup water to flow into the system if needed. But this doesn't account for expansion when the system fluid heats up. Do you have some kind of expansion path, or are you just counting on check valve leakage or a relief valve to keep from overpressurizing the system?

In the absense of an expansion path, fluid pressure will be highly dependent on fluid temperature. Is there additional fluid cooling during this 15 or 20 minutes following pump shutdown? Any way you could have parts of your system that are really at a higher temperature than where you are measuring?

---KenRad

 

[cawse001]

Could the 5 or 6 bar pressure observed be the discharge static head on the pump, which dissipates as the fluid drains back through the pump (presuming it is a centrifugal) via the small leakage through the check valve?

NB

 

[Giskard]

KenRad,

it is true that the tank is for make-up water and that the system doesn't have an expansion tank, (except for the small leak that the check valve could have to the tank), but we do have relief valves for avoiding the system to overpressurize, [actually, the pressure is not high enough to open these valves].

As we don't have an expansion tank, we see a gradually raise in pressure in the whole circuit if we don't put in service the heat exchanger and the water is recirculating continuously in the circuit. As the water heats up the pressure gradually increases. For example, an increase of 8 degree Celsius (measured at the discharge pump) causes an increase of pressure of 0,5 bar, aproximately.

What puzzle me is the final jump in the suction pressure when we stop the pump.

On the other hand, we don't have additional cooling for the despressurization period of 15 - 20 minutes, during which the pressure returns to its original value.

Regarding the temperatures, we have temperature transmiters af the pump discharge and at the heat exchanger discharge, (downstream the heat exchanger by-pass). I would say that the maximum temperature would be at the discharge pump, however the temperature transmiter is 25 meter downstream the suction flange (more or less), so probably we are registering a lower temperature that the real discharge temperature.

Thanks for your ideas and comments!
Ferran

 

[Giskard]

cawse001,

I'm not really sure what do you mean by the fluid "draining back to the pump", because the system is a closed loop. Do you mean that there could be some kind of reversed flow through the pump?

Thanks for comment!
Ferran

 

[25362]

Please correct me if I'm wrong.

I assume that by stopping flow the pressures equalize at a level concomitant with water's isothermal compressibility [κ] = (1/V)([∂]V/[∂]P)_T = 0.45[×]10^-4/bar for air-free water at 30^oC, and steel's elastic modulus of
2.1 [×] 10⁶ bar for the pipe and exchanger.

Further cooling by half a degree Celsius over half an hour, from, say, 30^oC to 29.5^oC may invove a volumetric contraction, from data on water's cubical thermal expansion [α] = (1/V)([∂]V/[∂]T)_P = 0.28[×]10^-3/^oC, and a drop in pressure explained by the ratio [α] /[κ] = 6.2 bar/^oC which justifies the last drop in pressure from 5 to 2 bar level without invoking any leak.

 

[Giskard]

25362 (Chemical),

I'm afraid I will have to review a little my thermodynamics to clearly understand your point. Thank you very much!

Do you know is this explanation would be very different if there is a little quantity of air dissolved in water? (let's say less than 1% - 2%)

Thank you for yopur comments
Ferran

 

[25362]

Ferran, although in principle air or any other gas dissolved in water can make it more compressible reducing its bulk modulus (the inverse of compressibility) I think the above conclusions wouldn't change much with these small percentages of dissolved air.

BTW, the equilibrium solubility of atmospheric air in water at 25^oC is 0.023 g/L or about 1.7 % vol/vol. Nitrogen and oxygen not in the same proportion they have in air.

 

[IRstuff]

Are you sure it isn't essentially water hammer? You stop the pump and the momentum of the fluid results in increasing the pressure on the suction side and decreasing the outlet side.

TTFN

 

[Giskard]

IRstuff,

humm... I would expect that a pressure increase due to a water hammer would dissipate faster, (and maybe with some kind of pressure oscilation during the transient), but not for 15 or 20 min, with the pressure decreasing steadily and slowy...

Ferran

 

[25362]

Assuming the whole line is, say, 40 m long and the celerity of a pressure wave is around 1300 m/s (considering a Sch 40, 3-in steel line), the time needed for the wave to make a round trip would be about 2[×]40/1300 seconds = 0.06 s ! The pump closure takes longer to stop the flow.
Thus, no water hammer should be expected, right ?

 

[IRstuff]

Assuming that there isn't a check valve in the path.

TTFN

 

[Giskard]

IRstuff,

that's correct, there is no check valve along the recirculation path of the closed circuit.

Ferran

 

[sailoday28]

A possible scenario:
Some hammer effect builds pressure up at the pump suction. The suction check valve(which allows some small leakage) closes. And the system pressure then slowly dissapates.

Regards