Fetkovich Aquifer results not same as Eclipse

[discover-volve]

I have a model run in Eclipse and in OPM flow. The OPM flow run gets completed and no major issues are faced during the run. However, the Fetkovich aquifer used in the model shows a higher water flux when the Wells begin to have high water cut. Can you please do a check on the Fetkovich aquifer algorithm that may be causing this error. I have attached an image depicting the error.

[GitPaean]

Hi, @discover-volve

Thanks for the information while I am afraid that it is too little information to have some clues regarding the potential problems.

[discover-volve]

Sure, I understand. I just want you to try this out internally on any model and check if the problem exists in the simulator or is it model specific. Fetkovich is a common aquifer model used in simulation studies.

[EdmundStephens]

[v2021.10] See attached a comparison between ECL and Flow. The model show a very exact comparison, so at least in this case the AQUFETP / AQUANCON check out (*).

Test model TEST_AQUFET_compare.zip

Comparison 1

Comparison 2

(*) There is a tiny syntax discrepancy, the AQUFETP entry in Eclispe takes a fixed number of aquifer input lines with no terminator '/'.

[bska]

Just a follow-up question here, in light of the underlying issue being addressed in #3911. Does your Fetkovich aquifer mode default the initial aquifer pressure at the datum depth? In other words, is Item 3 of AQUFETP defaulted for this aquifer?

Since the rates match initially between the two simulators I would not assume that the initial conditions are too different, but I'd like to know nonetheless. I'd also really like to see timeseries for the aquifer pressures in both simulators—i.e., the AAQP summary vector if at all possible and preferably with some sense of how much the two curves differ. E.g., maximum difference is <5% or >200% or something to that effect.

[discover-volve]

I just saw this query. Yes, the fetkovich aquifer mode is defaulted at the datum depth. For the cumulative injection chart - The peak aquifer influx from ECL is around 0.9 million m3 and for OPM the peak is 1.2 million m3. For the daily influx plot, the peak influx calculated in ECL is 500 m3 and the OPM calculated value is 1172 m3. I do not have access to the AAQP summary vector of the model.

I believe the issue lies in the way cell connections occur with the analytical aquifer.

[GitPaean]

Thanks for the feedback. From my side, I will need a case to reproduce the problem then I can look into it and fix it.