Deepwater Hard Lessons Learned - "We Tried that and it Didn't Work". Until Someone Else Tried it and it Did.

A number of years back I was contracted by an upstream Operator to perform a reserves audit on their first deepwater development and to report back on my findings before field commissioning. There had been rumors making the rounds that this development might not deliver its sanctioned oil volumes yet expectations remained high for both the host country and the operator. This was the first oil development within this lightly explored country and the host government had hedged its profit oil to secure international loans. This article shares the lessons learned from that audit and describes the pitfalls around building high expectations in an environment of high technical uncertainty.

The depositional framework for the geologic environment prior to my arrival had been constructed from the 3D seismic data set and sparse well control entirely without the aid of semblance analysis to better define sedimentary provenance. No direct hydrocarbon indicators were present. Hydrocarbon filled turbidite reservoir quality sandstone was assumed to cover almost the entirety of the salt dome induced closure. The sedimentary understanding at project sanction was that if one can imagine the above time slice without any of the depositional detail contained between the yellow lines. It existed as a model of sparse well data infilled with complex geostatistics.

I found it unusual that so little geophysical attention had been paid to depositional provenance prior to project sanction so I began inquiries. With each inquiry regarding any previous semblance analysis I was told by several long serving geophysical team members that, "We tried that and it didn't work." Yet there was universal agreement that the 3D seismic data quality was good. None of these long serving team members would perform a second semblance analysis - except for the new geophysicist seconded into the team that knew little of the development's prior history and hype. I'll call him Z.

Once Z got to work within 3 days he had some news. It wasn't that the semblance analysis didn't work, but rather the key geologic horizon - i.e., the unconformity that provided the reservoir seal - had been incorrectly picked in the geophysical data base and was set too high to reveal any of the underlying depositional features using a time slice on this pick. Z was able to fix this and within a week he constructed the correct depositional setting that explained with data rather than geostatistics the discovery and appraisal well results (see above). Z's revised depositional model shows that only the lower one-third of the reservoir lies within the high energy, good quality sandstone feature where the remaining northern two-thirds occurs within a poor quality, low net to gross crevass splay feature. The result was a 55% STOIIP reduction for this USD $720 MM investment. And 1st oil proved Z correct - what was supposed to be a six month, 62,000 bopd production plateau lasted three days and within three months oil production had declined to under 35,000 bopd.

One factor contributing to the publicity and therefore expectations surrounding this development was the prospect had been generated by one guy working out of his house. He became something of a folk hero when the Operator farmed-in to his acreage and he took his equity and went rags to riches when he listed it on the national stock exchange. This gave the development public scrutiny and the new Operator's expectations began to merge with the increasing business media hype. But the signs of trouble could have been identified early in the discovery and appraisal drilling campaign. One sign was the prospect is on the edge of the high energy depositional feature and thus anything on the edge can go from thick good quality reservoir to nothing in a matter of a hundred lateral feet. An early post-discovery semblance analysis with the correct horizon picks would likely have identified this.

Another contributing factor to the high expectation was that three of the four discovery and appraisal wells were drilled into this high energy channel feature. However, one of the three appraisal wells was drilled into the northern crevass splay feature and encountered poor quality reservoir but was explained away during the stage gate approvals process using complex geostatistics. This well was essentially a low net-to-gross bullseye that never faced technical scrutiny. The above type log is representative of the reservoir quality within the high energy channel feature. One can see why it was easy to get excited by the initial results in the absence of additional information that the semblance analysis could have provided.

However, the development drilling reality that matched Z's revised geologic model began to bring these high expectations closer to earth. Z's revised model explains the discovery, appraisal, and development drilling results as most development wells were drilled outside of the high energy channel an encountered low net-to-gross ratio laterally non-connected reservoir sands. Below is an example that is representative of nearly every development well drilled into this field. One interesting aside is that despite the development drilling results and this field's eventual rapid oil production decline, the RE / G&G team refused to accept Z's model and continued to use their 129 MM STOIIP geostatistically derived model to forecast production at 62,000 bopd on a six month plateau claiming that the measured oil rates were in error (they weren't - the first crude offloading proved that).

One final matter complicating development phase reservoir description was introduced by the well testing company early in the development drilling which also could have been resolved by a correct semblance analysis. The young engineer assigned to the well testing operations incorrectly sized the piping for the test vessel and these well tests were constrained to under 5,000 bopd. With a well spread rate exceeding USD $1 MM per day, the decision was made not to extend the test operations to discern reservoir boundaries so only permeability were derived.

The conclusion to my report to the Operator was that a semblance analysis derived from the correct horizons would have reduce net reservoir volume and therefore STOIIP and recoverable reserves to the point that any development concept was uneconomic. So the next time one is confronted with "We tried that and it didn't work", it is plausible that it didn't work because it was done incorrectly. With a great deal of money at stake, a "Sense of Urgency" can sometimes temporarily take a back seat to a "Probable Technical Certainty". That is why I like to sometimes remind people that the oil has been in the ground for 5 million years so it can afford to spend another six weeks there to assure we are making the correct investment decision.