Water Hammer in Flowlines from Lufkin Pumps 2

[denniskb]

I am working on a field with Lufkin pumps producing from a reservoir about 2000 m below the surface and into flowlines around 2000 m long to a separator station. Production is typically three phase but I suspect that it is sometimes 100% liquids.

Flow calculations for the flowlines indicates wellhead end pressures for the flowline should vary from 500 to 1100 kPa as the pump flow goes from 0 (bottom of stroke) to maximum (half way up stroke) but we are seeing pressure peaks up to 2300 kPa. I have run transisent calculations (water hammer) which predict pressures up to 2000 kPa when the lines are liquid filled and believe the system may be experiencing this.

Problem is that we have pressure switches installed to trip the wells at 1850 kPa to protect the 150 lb piping system.

Is anyone aware of water hammer happening in these systems? Dennis Kirk Engineering

http://www.ozemail.com.au/~denniskb

 

[KernOily]

Yes. This happens all the time in the fields around here (central California). The operators here sort of accept the fact that the pressure peak will momentarily exceed the ANSI 150# flange rating. I personally have watched PIs at the test manifolds go over 300 psig when the pump is on the up-stroke and the wellhead check valve slams.

Alse remember when doing your waterhammer calcs that you had to assume one set of fluid properties in order to do the calcs. In my experience you never know a priori what the nature of the fluid is in the flowline: it may be 100% crude, 100% water, all even fractions, etc. And that doesn't even start to address the emulsion issue. For a design basis, then, all you can do is assume the fluid is all one or the other and then bracket the solution. Thanks!
Pete

 

[denniskb]

Thanks Pete,

The clients engineers and the field operators are not aware of the transients and I have yet to convince them so I have forwarded your response, any other field contacts you can refer me to would be useful if you are able to do so. I can arrange for the client to call and confirm the problem for himself.

The issue is very important as we have 600 km of CS flowlines to replace with GRE for 500 wells and if we don't get the line sizes right we will leave the field with a huge problem. The PSHH is required to shut the wells down as they are SCAD controlled and the link does not have the rquired reliability.

You might be interested to know that we tried to model the two phase flow with PipeSim in order to size the lines to avoid slug flow and to reduce wellhead backpressures but we could not establish a reliable model. We had to fall back on a minimum velocity approach based on various calculations and texts. I believe we settled on 4m/sec to suit the GOR and fluid properties. Dennis Kirk Engineering

http://www.ozemail.com.au/~denniskb

 

[KernOily]

Dennis - A couple things to check are:

- piston or ball check valves will aggravate the situation. You might recommend installing swing checks for the wellhead check valves and the check valves at the test manifold. There are also special check valves available for which the closing rate can be controlled/adjusted to increase the duration time of the pressure transient. This will decrease the magnitude of the reflected wave and might get you below the PSHH setpoint.

- See if they can lower the test vessel pressure or whatever is controlling the final sink pressure, e.g. the FWKO dome pressure. If your lines are sized properly you shouldn't be seeing large dP's across the gathering system.

- we use PipePhase in this office for 2Ø and 3Ø flow calcs. I am not familiar with PipeSim but it looks like a PipePhase competitor from what I saw on the web. Once again, since the nature of the flow regime is not known a priori, all you can do to get a defensible solution is to bracket the solution assuming all water on one end, all crude on the other, and your best guess at GOR to get your gas rates. In my experience it is very difficult to do accurate 2Ø flow calculations in the oil patch due to the dynamic nature of the process. There are just too many variables.

- you could also install an accumulator vessel or slug catcher. Or, install de-surgers to reduce the magnitude of the transients. There are several types out there. This addresses the symptom, not the problem though. Thanks!
Pete

 

[denniskb]

Thanks again Pete,

I believe the water hammer is a direct result of the pump and flowline being 100% liquid filled (no gas break out) so the Lufkin pump movement has to accelerate the entire pump column and flowline contents from stationary to its maximum rod velocity at the middle of the upstroke and then bring it all to a stop again at the top of stroke.

The downhole pressures are low so solution gas breakout is occuring in the casing and gas production is from the casing valve, this results in 100% liquid production from the pump. Occasional low casing pressure will lead to the flowline also being 100% liquid filled.

The speed of this change is faster than the return time of the pressure wave so a maximum pressure wave peak is generated. The simple physics of the forces required to accelerate the mass from stationary to the displacement velocity confirms that the water hammer pressures are present.

The check valves will have no affect as the pressure waves originate downhole at the pump. Reflected waves in the flowline are unlikely as a small amount of gas breakout will occur in the lines and quickly attenuate the wave.

Surge accumulators may well do the trick if we locate the pressure switch on the gas side of the accumulator. Buying 500 of these will not make me popular but we may be able to provide a number to install on problem wells only.

Our line sizing plan will address all extremes of product flow and as you put it "bracket the solution".

Do you have any oilfield operations contacts that know about this problem who we could talk to so I can sell this theory to the client? Dennis Kirk Engineering

http://www.ozemail.com.au/~denniskb