Do you want to avoid filling the test separator with sand? If so, one solution might be to test the well on a reduced choke, so that the rate is below the sanding rate and then use a programme like Prosper or Wellflo to calculate what the production would be on a larger choke (you will have to calibrate the Propser model with an accurate PI and with tubing pressure loss information - a spinner survery, that kind of thing). Or just clean out the sand in the separator periodically- it's a test separator, so you could do it without interupting production.
Great DrillerNic, we are going to clean it and then come back on production. The thing is the BHP of all wells are monitor by Gas lift and we do believe that a correct Gas lift optimisation must be performed to check the efficiency of wells management. thsi may reduced the incoming sand.
1. Clean the test separator
2. Get a static and a flowing pressure survey done on all your wells as a minimum (to make sure you know where you're lifting from). You should be doing these regularly anyway!
3. Ideally, get information about the PI and reservoir pressure from each well- a build up test, for example. Don't assume that the measured PI when the well was originally drilled is OK: scale, tubing fill etc can all alter a well's PI.
This should then let you build your Prosper/ Wellflo model accurately. Then you test on a reduced choke, match the test production numbers to the Propser or Wellflo model with the test Flowing Tubing Head Pressure and then re-run the model forward on the producing FTHP for allocation purposes.
The main sources of error in this method are:
1. friction losses in the tubing may not be modelled correctly; one way around this is to get a plt log done and then tune the tubing loss correlations with the plt data.
2. the match to the test results: with a gas lift completion, assuming you know things like water cut, flow rate, gas lift rate and gas lift gas composition accurately, you can alter many unknown parameters to get a match (reservoir PI, reservoir pressure, gas lift entry point etc), giving you lots of different solutions. Knowing some of these parameters accurately (reservoir PI, gas lift entry point for example), means you reduce the number of variables you can alter to get a match.
This is the main problem with this method: I once spent ages matching every monthly production test done of a well over the previous 2 years, getting a nice consistent story of declining reservoir pressure. I proposed a GLV changeout to change the lift point to get more production, only to find after the GLV changeout that all my matches had been wrong due to a shallow hole in the tubing that had been gradually growing so that the lift point had gradually become shallower, and changing the GLVs didn't alter this....
