Centrif Pump with Varying Discharge 2

[zdas04]

I've got some multiple-stage centrifigul pumps installed in a vertical configuration. Normal inlet pressure is 100 psig. Normal discharge pressure is 1,300 psig. Flow is up. Fluid is 60F water (SG 1.03). Pump is electrically driven.

Several times a week downstream upsets cause discharge pressure to drop to about 100 psig (rarely less than suction pressure, but that does sometimes happen) for several hours at a time.

The pump location is inaccessible and the closest location where I could install a backpressure valve is about a mile away (and the source of the upset is between the pump discharge and the location where I could put the backpressure valve).

Is there any significant risk of damaging the pump in this scenario?

David Simpson, PE
MuleShoe Engineering

http://www.muleshoe-eng.com

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips Fora.

The harder I work, the luckier I seem

 

[checman]

Zdas04 It all depends on the pump curve and how sudden the discharge pressure drops off. For hydraulic balance and thrust loads reasons multi stage pumps do not like sudden pressure changes for the most part. If it is not a sudden change and if the pump is still running in an expectable area of the pump curve it should not be a problem. You normally have 1200 psi in differential pressure and if at 1100 psi (2466’ of head) your still on the pump curve all is well. I am a little confused to why you referenced the drop in discharge pressure to the suction pressure if you are sure the problem is down stream of the pump. Most centrifugal pumps build additional pressure above suction pressure. If you think your loss in discharge pressure is do to loosing your suction pressure this could be a big problem if you fall below the pumps NPSHR. Cavitating water at that temperature will cause major pump damage.

Regards checman

 

[JJPellin]

If I understand this correctly, you are stating that the discharge pressure drops to 100 psi. This would indicate that the pump is stone-walling. It is running off the end of the curve to the maximum flow it can produce. This is generally not good for any centrifugal pump. But it can be especially bad for a vertical multi-stage. At the Texas A&M Pump Users symposium last year, there was a very good paper presented on this very issue. The paper found that at high flow rate (low discharge pressure) vertical propane pumps would experience hard up-thrust which would buckle the shafts. The extreme run-out from the buckled shaft would not recover when flow returned to normal. In this case, the shaft run-out was causing the bushings to fail. We have seen this same problem in our pump in tank transfer pumps when they would start-up without having the line liquid packed. In our case, the run-out was causing catestrophic failures of the mechanical seal. You can verify if this is happening by checking the shaft run-out on a freshly overhauled pump and then again after one of these events. If you are finding bent shafts during overhauls I would bet on up-thrust from high flow operation.

 

[checman]

Wo I should have read that a little closer. It does appear that you are opperating at the wall. The amount of damage will have allot to do with the impeller design (mixed or radial ect.)and the inlet design. Multistage pumps flow are determind by the maximum flow of the impeller design the number of the impellers determine the pressure it will build.

Regards checkman

 

[JJPellin]

In general, the damage that I would be concerned with includes bent shafts from up-thrust, impeller damage from cavitation, erosion damage from high velocities and coupling or motor problems from up-thrust and sudden load changes. If a back pressure (pressure control) valve is not practical, there are limited options. The pump could be set up to trip on low pressure with a pressure switch. But then you would have to figure out how to reset the trip and restart the pump when the upset was passed. You could drop an orifice plate in a flange at the pump discharge that would limit it to a reasonable pumping rate when these events occur. But then the orifice plate would limit flow during normal conditions. The best thing would be to find the source of the pressure loss and correct that.

 

[zdas04]

The pump I'm talking about is a downhole Electrical Submersible Pump hanging in a 3,000 ft well bore.

The upset comes from a relatively small amount of natural gas coming with the produced water. If I put 1,000 SCF in the 1.996 ID tubing, then the gas displaces enough liquid to drop the pump discharge pressure from 1,300 psig to about 100 psig.

Downhole gas separators help, but nothing prevents the gas from getting into the pump discharge. Usually the gas is in the form of very small bubbles (that don't tend to collapse and cause cavittion damage due to their size) and/or disolved gases that evolve out of solution as the water moves up the piping (due to temperature changes as the fluid passes through different temperature strata).

When a slug of gas is large enough to fill the pump with gas, you run into heat and friction problems that destroy pumps pretty quickly. I'm trying to determine the damage I can expect if the liquid-full pump sees a rapid decrease in discharge pressure.

Low-pressure trips are not practical because you'd have to put the sensor on the surface and you don't often see surface indications of the problem.

An orifice on the discharge of the pump won't work because the downstream side of the orifice would see such a wide range (i.e., an orifice that gives you 100 psid at normal flow rate would let the pump drop to 200 psig if the column backpressure drops to 100 psig).

The pump is not stonewalling. Every one of them that I've seen has been on its pump curve and able to move across the pump curve as upstream and downstream conditions have changed.

David

 

[rzrbk]

If I understand correctly, the gas is entering the pump suction with the water and drops the specific gravity of the pumped fluid, which drops the delta pressure across the pump to whatever it takes to lift the column of mixed water and gas bubbles 3000’. The gas would expand as it rises, artificially reducing the discharge pressure further, which might cause some flow increase.

If this is the case, then the pump would stay on its curve, flow volume would stay somewhat stable, forces would stay in balance in the pump, and horsepower use would drop during a swing. Damage that would normally be seen when a pump runs too far from BEP would not be present. (vibration, recirculation, shaft deflection, etc) However, I would expect that the less dense mixture being pumped could cause heat/rubbing damage at bushings and wear rings, due to loss of lubricity and a reduced ability to carry away the heat produced in the pump.

If the above sounds right, and the pumps have been seeing these swings with no apparent drop in performance, it could be that nothing needs done except to trend any damage seen as the pumps are routinely repaired. If in service times prove too short, self lubricating materials might be looked at. A no flow switch to trip the pumps may be a good idea to limit damage when they totally gas up.

SWAGs:
Not sure why/how the discharge pressure (measured at the pump) drops so much. Maybe there is some pumping affect by the column of bubbles (like in an aquarium riser). If it is very short duration drops to below suction pressure, maybe the momentum of the water further up the discharge pulls along the lighter fluid for a time, or the pressure transducer overreacts to the quick swing.

 

[zdas04]

rzrbk,
You have it just about right. The big issue is the discharge column simply gets lighter due to the gas in the vertical pipe. We don't see terribly long pump lives, but they aren't rediculously short either. I was thinking that lubricity and heat removal were the big concerns, but I wanted to see if there was any different input from pump guys (as opposed to downhole guys).

A lot of people have tried pump-off-control schemes with these pumps (based on water rate, full-stream reaction to a heated probe, and current) and some of them work well at telling when the pump is trying to pump gas. The big difficulty is deciding what to do with this information--most of the time if you turn the pump off, solids in the produced liquids will settle onto the top of the pump and it won't be able to develop the necessary starting torque.

These pumps are pretty popular in one particular field and I'm working on a class on deliquifying gas wells and I needed to get some additional input on the risks.

Thanks for your input.

David