I've picked up from several postings that some are saying that the consultants used by OSH assumed a water drive mechanism. Is that engineering information as to how they arrived at that available anywhere?
If one goes from pure depletion drive to full water drive in a gas reservoir, recovery efficiency can drop by 50+%. A good reservoir engineer would give you a better number, and that is quite different for fractured carbonates (like here) vs. sandstone depositions.
Production tests are conducted in part to determine the drive mechanism and thus recovery efficiency. However, that is more difficult to ascertain when the extent of the reservoir boundardies are not well known and the reservoir volume is huge. A material balance is conducted measuring the reservoir pressure before and after the production test. This information is used to determine the drive. However, if the volume of the reservoir is highly uncertain (our case), then the outcome is also highly uncertain. Since the aquifer presence is well established, then I have to believe that there is some degree of water drive. The question is how much.
Since relatively speaking, the total volume produced to date is miniscule in reference to the total reservoir volume, I presume we are dealing with tiny, tiny fractions. After all of these production tests, the reservoir pressure should have sequentially declined in the case of perfect depletion and bubble tight reservoir seals. The data I've seen (if memory is right) is the pressure decline was a very small fraction of a psi. In a water drive or partial water drive, reservoir pressure will have recovered vs. time and will not plot on a log/log straight line plot of pressure vs. time.
So the last production tests provided more information along those lines. When you have a water drive in a heterogenous rock like this, then recovery gets very sketchy. The type, size, and distribution of the void space (pores and fractures), well type, well placement and production rate will have a major impact on recovery efficiency. You can design and operate the wells in a way to significantly improve recovery. Well placement is critical. It is possible to "outrun" the water since gas has much higher mobility and thereby improve revovery.
Reservoir models have to be built to see how this might turn out. But all it takes is a very conductive fracture through the reservoir down to the aquifer to blow the model away. So I suspect they are relying heavily on analogue fields to bracket the range of outcomes. So in summary, there may be a good bit of subjectivity rather than objective analysis in picking the recovery factor. A probablistic approach for all of the critical variables makes the most sense. But I am not a reservoir engineer, but just happened to work with some of the best.