Centrifugal pump parallel operation 1

[TheElv]

I'm looking to run 2 large (600kW) centrifugal pumps in parallel (duty / assist). One of the pumps is new the other is 50 years old but has similar head / flow characteristic although lower efficiency now. I'm concerned that the pump curves are not continuous rising to closed valve. Are there any recommendations on controlling these pumps, particularly when the assist pump starts to prevent hunting? I should add that the pumps will be speed controlled (via VSDs) and my intention at the moment is to always start the pumps at minimum speed and use current set-point control to run the pumps at the same speed but would be grateful for any advice. Thanks

 

[BobPE]

I don't think running the pumps at the same speed will work, they are both different and will behave diffenently in the system, making speed a variable that shouldnt be used for control. I am assuming that you are adding a second pump to maintain system pressures that have fallen for some reason. Having the pumps controlled to maintain a set pressure would make more sense. How are you designing the second pump to come on? Based upon pressure?

I also dont understand what you mean when you say the pump curves are not continuous rising to closed valve? Please give us some more information if you can...

Bob

 

[TheElv]

Bob

Fair comment re the speed - I agree speed control wouldn't work. However using current set-point control the controller would operate the pumps at the same current therefore the pumps could indeed run at slightly different speeds to ensure the pumps are operating at the correct point on the head / flow curve.

The second pump would start when flow demand is above a pre-set point in the water delivery pipe and after a delay would start the assist pump. I should add that this is an existing system which has worked OK - what I'm looking to do is replace one of the old pumps with a new pump which will have slightly different characteristics.

My concern about the curves is that if the pump head flow curves are continuously rising ie the head developed is at its peak at closed valve condition, then the system with parallel pumps should be stable. My problem is that the curves on the existing pump is not continuously rising ie there are points on the curve where the pump could pump the same head but different flow hence my concerns about potential for hunting.

Regards

Jon

 

[BobPE]

Jon:

That's an odd pump curve characteristic for a centrifugal pump in it's normal operation range, I guess that's why I am confused. If your pump does indeed pump two flows at one head in its normal operating range, then it should have been doing it already and someone should have tried to make it work and in essence, follow their lead.

As for the curent draw to control the pump, you are right, but this vaiable also takes into account motor performance too, which which could induce current loadings that may have nothing to do with your flow hydraulics, making you pump operate where it shouldn't.

Sounds like an odd problem. Are you looking at the pump curves where they will normally operate? Maybe looking at them outside thier realistic range is giving you this curve characteristic?

Just trying to help....

bob

 

[vanstoja]

Drooping head-flow curves toward shutoff flow are not all that uncommon in mid and low specific speed centifugal pumps. In my opinion, parallel operation of pumps with low flow head droop should not be a problem unless you're at or close to two-phase suction flow conditions where fluid compressibility effects can get you into system flow instability and flow surging problems. Boiler feed and condensate pumps are in that category while pumping hot water with relatively low suction pressures. In pressurized systems with conservative Minimum Suction Pressure requirements that include dissolved air, other gases and water vapor suppression provisions or even unpressurized systems with adequate NPSH available, low flow operation of pumps with drooping characteristics can usually succeed. If you're operating on the negative slope portion of the head flow curve of all paralled pumps(ie, near rated and higher flowrates) then near shutoff head droop should not affect stable system operations.

 

[abeltio]

You are correct about your concern...the problem is a classic... there are many articles written on the subject...basically what they say is:
operate the system at a pressure that ensures you are to the right of the point where the pressure is equal to the shutoff pressure (i.e. beyond the "hump" in the drooping pump head-flow characteristic).
Otherwise you could have one pump at a very low flow and the other at a high flow and both delivering the same pressure...
How do you ensure that you operate beyond that point?
1. Select the operating pressure to be below the shutoff pressure (zero flow)
2. Have a bypass line discharging back to the suction vessel (not the pump suction), to discharge the excess flow and achieve a back pressure - shutoff pressure = 10 psid
3. have equalizing lines from the pump suction back to the suction vessel (this is a must for vacuum service).
4. the check valves at the discharge of the pumps operate correctly and with very low differential (1~2 psid)

Regarding control you can get as fancy as you'd like... yes, pump speed is a result... you should be controlling discharge pressure and then flow.
e.g. control first that the pressure delivered by each pump is the same, the total flow should be higher than the required flow, then control flow to the required setpoint.

Depending on what is the most critical feature (pressure or flow) the control system will follow.
Also if you have VFD (variable frequence drives) the pumps will be operating on different curves... and the shutoff pressure will depend on the pump speed. That's where the check valves come into the game...unless the speed (developed pressure) is high enough to overcome the check valve the 2nd pump will not deliver any flow.
HTH
Saludos.
a.