Pump distribution line issue

[Mike4chemic]

A process line requires connection to 3 pumps, only two work at the same time (see attached picture). The problem is that the fluid doesn’t seem to distribute evenly between the working pumps. This is not specific to a single pump, when 2 pumps operate there seems to be a preference to the right pump of the two, no matter which is working. For example: When operating the right pump and the middle pump the right pump takes most of the flow and when operating the middle pump and the left pump the middle pump takes most of the flow.
My current assumption is that the distribution line to the pumps is not properly designed and should be enlarged. I would appreciate any thoughts on the matter or any tips for correctly sizing the distribution line.

 

[LittleInch]

In theory this should work as the piping is balanced on the discharge side.

However balancing parallel pumps is notoriously difficult.

What is your flow velocity and what do the pump curves look like?

Are all three identical?

If the curves are very flat then any small difference makes a big difference in flow.

How are you measuring flow rates and what is the difference?

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

 

[Mike4chemic]

The three pumps are identical. We noticed the difference in flows when looking at the electric current each pump takes which is ~300 amps for the "dominant" pump vs. 100 amps for the 2nd pump.
The required flow is 2300 gpm and the "dominant" pump always stays at its max flow point on the curve:~1900gpm (see attached).
When the 2nd pump is turned on cavitation/vibration is noticed. Flow velocity on the line entering the pump is ~5 ft/sec.
I should note that the head at the distribution line height is ~3 feet.
There is a reducer from 16" to 12" just before the distribution to the pumps, do you think it affects the flow?

 

[bimr]

Yes, the reducer is incorrectly located. It should be located after the takeoff for the second pump. Pump suction pipes are generally designed for 4-6 ft/sec (including the header pipe).

When the second pump is activated, your pump system is trying to pump more than 2,300 gpm. This scenario is operating past the flow capacity of the piping and causing the vibration issue.

However, I suspect that the problem is the pump was incorrectly sized for a higher discharge head than what the pump is seeing in the field.

Your pump is supposed to operate at 1,150 gpm and instead at least one of the pumps is operating at 1,900 gpm.

Note that the pump may be overrunning the motor and this may cause the pump motor to overheat and fail.

You need to determine the operating head and develop a pump system curve. Then, either install an orifice plate on each pump or throttle the pumps to increase the discharge head. You should also consider installing a smaller impeller.

You can simulate this by throttling the shutoff valve.

 

[georgeverghese]

Setting up a FC loop on each pump discharge with a master flow ratio controller will help if you want an automated solution to this problem.

 

[Mike4chemic]

It does look like the main problem here are the oversized pumps, but for the sake of limiting the repair costs, do you think changing the distribution line (larger, with reducer correctly placed) will allow even flow to the pumps?

 

[bimr]

Installing a piping system that is hydraulically balanced will provide a more even pump discharge.

Note that the same hydraulic issue may exist on the pump discharge side.

The easiest fix is probably installing orifice plates.

 

[LittleInch]

The inlet piping issue for pumps of this size and head is not your issue. If there is no flow control on the individual pumps you Errol not solve your problem. I haven't had a close look at the data and curves but running one unit first at a point off the end of the curve shows the pumps are too big.

I can't work out how one pump can continue to operate and the other so much less but it sounds like you're having npsh issues so the second pump might just be cavitating.

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

 

[georgeverghese]

If your operators dont mind going out to this location once in a while, a butterfly valve on each pump discharge line for light duty flow balancing referencing motor current and a FT - FI will do just as well.

 

[georgeverghese]

An even more basic solution would be to use the existing discharge block valve as the flow balancing valve - throttle this valve to even out the motor current between the pumps that are running. So no need for butterfly valve or FT - FI.

 

[bimr]

It is feasible to simulate throttling of the pump with the shutoff valve. However it is not a good solution for the long term. Most likely, the valve was not designed for this application and you will have two different flow streams to balance at the same time.

 

[LittleInch]

I've had another look at your data and there's something seriously wrong with your data.

If 300 amps equals around 300hp, then 100 amps is arid 100hp, but the lowest power on the curve is 200hp/amps.

It does say this is a 13 stage unit, are you sure each of the pumps has the same number if stages fitted as with that data I cant see how you have such a huge variation from "identical" pumps. Whatever is causing this is not the inlet pipe.

At this sort of head loses, you need proper control devices, not throttling on off valves, you would need to shut them virtually closed to get any appreciable pressure Los and would destroy your valves in double quick time.

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

 

[georgeverghese]

Sorry, its after reading other postings that it appeared that there was something else other than a flow balancing problem here.

From the pump curve, it is evident that flow balancing is not the cause of this problem, since the pump Q-h curve is reasonably sloped.

Ideally, this should have been a 16inch feeder line till the last pump on the left in the photo ( v = 1.2m/sec). We now have v = 2.2m /sec at 12inch on the feeder at 2500gpm. Have we done an NPSHa calc on this feeder line at this pentane flow with a vapor press of 20.3psia at 120degF? - what does it tell you? If this is rundown from a column, are you still at 120degF at 2500gpm? Or maybe there are some lighter components in the feed that could have raised the vapor pressure?

 

[Mike4chemic]

Thank you all for your help.

Mike

 

[LittleInch]

So what are you going to do/ have discovered??

There seemed to be quite a bit here which is now left unanswered.

Were glad we helped, but there is no conclusion here. A bit more feedback would be nice......

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

 

[Mike4chemic]

I have no updates and I'm afraid its going to take some time before starting the actual work, but the planned course of action (in gradual order from no cost to high cost) is:

1. increase the head at the distribution line (to negate flash in the pipe or cavitation due to low NPSHA).
2. experiment with the valves at the pumps exit in order to reach equal flows from each pump.
3. replace the distribution line with a larger pipe.
4. de-stage the pumps to match the current system requirements.

 

[georgeverghese]

The pump datasheet fluid vapor pressure for this fluid is indeed for pure n-pentane, so if these pumps are running on a de C5 column or similar, and there is a small amount of iC5 creeping into this stream at higher flows, you would have an elevated vapor pressure, and this might throw you off your feet on NPSHa margin to NPSHr.