Pump change over and non return valves 1

[WilliamTz]

Hello

I have 3 x large centrifugal pumps arranged in parallel feeding into a common header, running off in one direction. 1 duty / 2 standby. 90kw motors on each pump and a variable speed drive on each. Each pump has a non return valve on it (just a normal flap working under gravity). System is at 6 bar and 140 degC water. The pumps maintain a constant DP across the flow and return.

The trouble is when we are doing a weekly change over, the standby by pumps ramps up and then the duty pumps ramps down and switches off, and we get water hammer.
When the duty pump is about to go off the non rerun valve slaps down and ends up in generating awful water hammer. I tired increasing the ramp down time to be more gentle but it actually makes it worse as the non return valve starts to flap over a longer period of time as the pump goes thro its ramp down (1 min). I have since decreased that to 30 seconds.

Another interesting thing is if the pump going through ramp down has a pump to the left of it running there is no issue, it is as if when there is flow in the header in front of it there is no water hammer. e.g. If Pump 1 is running and Pump 2 comes on for a change over, when pumps 1 goes off we get water hammer. If however pump 2 is running and pump 1 comes on for a change over, as pump 2 goes off we dont get water hammer. Why is this I wonder?

I understand there are spring assisted non return valves that are good for controlling water hammer on the market and I may change them for that. Do i still need the VSD to ramp down or do I let the VSD coast with these type of non return valves?

Your thoughts?

 

[quark]

Have you tried ramping up of Pump2 at a slower rate?

 

[WilliamTz]

quark,

The current swap over method ramps up Pump 2 first so Pump 1 and 2 are running to maintaining DP....then Pump 2 ramps to off.

 

[LittleInch]

I think trying this with just simple NRVs won't work.

The action of the VFD on ramp down will reduce the head instantly so the changeover ramp speed is critical. It may be a even faster ramp rate up and down works better than slower.

The flowing thing is interesting.

I think the issue is that for the NRV to seal and slam then it needs reverse flow.

For pump 1 that reverse flow hits the non return and bounces back.
For pump 2 the pressure is always there and the flow into pump 2 to close the NRV is just a very small part of the flow so there is no real "impact" and any pressure fluctuation is absorbed better.

In the long run though sitting pumps against NRVs usually runs into back flow / recircualtion issues. Usually far better to install isolation valves and then on changeover you end up just opening one and closing the other whilst the pump is still running. Thus no back flow causing the NRVs to slam close.

I'm fascinated about the fluid here. Water at 140C has a vapour pressure of 3.6 bar. so allowing a bit of spare to avoid flashing the water or cavitating inside the pump your inlet pressure is what? 4 bar? 4.5 bar And the pumps raise it only 2 bar or less?
Or is the inlet 6 bar?

If so what is the differential pressure here?



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

 

[quark]

If you could ramp up pump2 at the same rate of ramp down of pump1, the problem might not be severe, but this is not an option for you. Non-slam check valves help the situation as they start getting closed when the flow start reducing rather than against reverse flow.

The question still remains as to why this doesn't happen when pump1 is running and pump2 ramps down. I faced these kind of problems with pumps at header end. When pump 2 starts pumping, flow happens on either side of the discharge tap off and since pump1 is on the other end and there is header blind, there may be lot of irregularities in flow. I am not sure but that is what I thought. My practice is to use direct elbow connections to end pumps rather than tee. I can only think of a kind of venturi effect when pump1 runs and pump2 ramps down.

 

[WilliamTz]

Little Inch
|Suction pressure 4.5 bar and developing 2 bar approx so 6.5 bar on discharge
B

 

[WilliamTz]

I think the safest thing to do for now is ramp down the duty pump to off, get conformation it is off and then ramp up the next pump to come on line. It can be done in 10-15 seconds.

 

[quark]

If the system is not pressurized, hot water can flash. That may be the reason why you are first running one pump at full speed before stopping the other one.

 

[georgeverghese]

There are a few problems here, I think:
1) Bad piping design - swing check valve oriented vertically with flow going up - closes prematurely as soon as upstream pressure starts to decay. Should be horizontally oriented, preferably dual plate check valve.
2) When one of the pumps begins coasting down, it produces zero flow (it is running dead head) as soon as speed drops below that which is required to enable some forward flow. There should be a min flow line upstream of the check valve (for each pump exit line) to maintain flow at above the minimum corresponding to full speed operation when a pump starts going off line. Use an FIC - FCV to be sure, not a backpressure PIC - PCV.

It could be either one or both of these.

BTW, why is your pump suction connected directly to pump return? Why doesnt the return go through a vapor disengagement drum?

 

[TugboatEng]

Is there any reason the second pump has to be brought online before the first pump is stopped? If the system has any mass it can surely tolerate a moment of stopped flow.

I would also question the necessity of a weekly changeover.

 

[WilliamTz]

TugboatEng.... I think im going to go down that route... If I can ramp down in 3-5 seconds and ramp back up in 5 seconds problem solved....the system is serving a huge factory...