What I saw at ATCE

Perhaps we are in the calm before the storm, perhaps these will become the good old days, but things at ATCE do seem oddly a little different. Even so, it is interesting to see where the industry is focusing its attention.

Themes in the papers tend to show what industry is thinking about, and industry is especially focused on frac hits and on automated and/or intelligent systems. The conference was themed around big data. Separately, authors showed up with a higher proportion of studies using big data or some breed of machine learning. Vendors also seemed to show more prominently their integrated data systems and even to headline a few such buzzwords. The more immediate impact on the bottom line, though, lies in hydraulic fracture interaction.

Big data/machine learning: As pointed out by the conference chairperson, our industry is behind its "peers" in the adoption of data- and machine-based tools. In fact, the oil business has historically ventured into new technologies after they have been demonstrated to be viable by other industries, and it is even slow to adopt new methods developed within the industry. Besides published studies of landmark technologies like 3D seismic and horizontal drilling, my own experience with lesser technologies shows the same pattern. In the early and mid 1990's were were behind other industries when probabilistic methods became fashionable. That fad passed, and engineers now relegate those methods to special circumstances. Neural networks saw a research and publishing heydey around the late 1990s, but I haven't seen a paper or a project on the method for at least a decade. It makes me wonder how well this trend will stick.

I'm betting it will, in part. Last January, Dave Pursell of Tudor, Pickering and Holt (at the time) pontificated that the winners of the future will be those companies with the most data. Sara Ortwein of XTO said almost exactly the same thing in her keynote at ATCE, and it is easy to see why they say so. The next boom may be starting, but there is certainly another bust coming. Those who survive the next bust will be those who invest in the best cost-saving systems--those that gather, reduce, report and even analyze large sets of data without human intervention. What is more, the maturity of the industry in North America makes both the source and need for big data systems.

To wit, I believe that in the immediate future skills with data analysis products like Spotfire, Tableau and Verdazo will be as essential to an engineer's toolkit as spreadsheets and decline curve programs are today.

The machine-learning systems, though, may be slower and harder. Few people in the industry are inclined toward such computationally intense and abstract methods, and even fewer are trained in the methods which are still under development. Catalytic individuals require extensive retraining from petroleum into the methods of data science or else immigration from high tech to the oil industry. To be fair, it takes only a few catalytic individuals to make effective tools which can be propagated across companies. On the other hand, the widest current in the industry is slow to adoption especially of technologies it poorly understands, and the industry has a lamentable track record of developing software. (Exhibit A: ARIES)

Automated analysis may have a future in the oil industry, but barriers are high and strong and not susceptible to the force of keynote addresses. The opportunity may be great, but so far, says Darryl Willis, “One of the things we see at Google Cloud is that it's not the company with the best algorithm that's going to win—it’s the company with the most data.”

Frac hits: More immediate is the impact from fracture interference on today's profitability. Paper topics and the size of their audiences suggest that companies are especially focused on frac hits--the interaction of a new "child" well with a pre-existing, adjacent "parent" well. The effect on parent wells, and its mitigation, have been treated with some excellent papers such as SPE 180200 and SPE 189875. They show that plays have a systematic bias toward improvement or damage to the parent, and a good deal has been written about how to mitigate the impact on parent wells.

Far less has been dedicated to results and mitigation of child wells. Papers like SPE 189875 and SPE 171578 offer direct quantification of the phenomenon, but more skip take notice of the fact and research causes. A 2016 paper leads this way, "Sufficient production and fracture mapping evidence across North America is not available to clearly demonstrate that pairs of delineation and development wells often underperform when there is substantial production time (months or years) between the completions of the two wells." (SPE 181656) (Interestingly, I've not yet seen such an acknowledgement in any investor presentation.) The pattern of multiple papers corroborates my unpublished work; the effect of interference is detrimental to child wells in all plays. Drilling adjacent to older wells commonly causes EURs to drop 20 to 30% and can be 40 to 50%. Longer delays and closer spacing increase the loss.

A one-quarter drop in EUR translates to a one-third increase in break-even price. That is the difference between $54 /bbl and $72 /bbl. Unfortunately, I have found only one technical paper addressing directly methods to prevent the loss of rates and reserves in child wells. Instead, papers show, as theory might suggest, that mitigating effects on parent wells also mitigates effects on child wells. Longer laterals and larger fracs (read "higher CAPEX") have generated equivalent short-term rates, but my read of the literature and the underlying data is that the longer-term decline remains greater than parent wells.

Industry mood: Even though this is the second year of rebounded prices, and those higher than last year, attendance and exhibits seemed to be down. The exhibit hall was smaller and more focused on frac-related services. Apart from markedly more Chinese exhibitors for such services, I did not observe anything new in the hall. The mood was neither relief nor excitement. What is more interesting, the attendees seemed to spend more time in the technical sessions and less in the exhibit hall and, anecdotally, less time talking with exhibitors.

Of the papers I saw, one seemed especially good, others mediocre to poor. (Some of the best-looking papers were to be presented on Wednesday, but business took me away.) Though I got to see a number of friends, there seemed to be fewer than in the past. Overall, though SPE put on a good conference, the response of industry was somehow, oddly different, and it troubles me that I do not understand the significance. (If you have a perspective on the mood or its significance, please share below.)

In any event, it was good to see friends Bill, Cindy, David, Davood, Deanna, Dee, Detlef, Ibe, Jim, Kayli, Lesley, Lori, Mohamed, Mukul, Philip, Randy, Steve, Susan, Tim, Taylor, and others.