DUC Datathon: A Look Back

Hands-on Tutorials

Back in August 2020, I was fortunate enough to participate in a datathon hosted by the Society of Petroleum Engineers Calgary and Untapped Energy on the topic of drilled but uncompleted (DUC) wells. The premise was simple enough: what is a DUC and how can one identify it in a company’s inventory of wells? Our presentation was time limited, so I thought it would be valuable to go through, in more detail, our working methodology and analysis of this uniquely Canadian problem.

Introduction

In the US, the US Energy Information Administration (EIA) publishes a monthly list of predicted DUCs. In Canada however, there is no such information, not even by province.

We went with the strict definition of drilled but uncompleted well, which is a well that has not registered or recorded a completion activity yet. There was a minimum time period we required between rig release and the completion activity because sometimes there might be delays in construction or completion, and we wouldn’t qualify that as a DUC since it was never intended to be one to begin with.

Methodology

My group and I had to set some criteria for how we classified and treated DUCs. Here’s the breakdown of what we used:

• A well required at least 3 consecutive reporting periods (roughly 3 months) to be considered a well. We didn’t want wells that produced for a bit and never produced again to really be counted as an asset or something of value to a company. This is different than wells that have initial testing and further production later on.
• Anything after 2019 was also not included because the data given to us was only current up to the very beginning of 2020. There was no way to validate if those wells were operational for data received in 2020.
• We used a little more than 2 months between the rig release date and the completion date as the qualifying criteria for a DUC. The EIA’s own definition of a DUC is as follows:

"The metric uses a fixed ratio of estimated total production from new wells divided by the region’s monthly rig count, lagged by two months."

• Our methodology ignored whether the dataset given to us showed whether or not a well was active or inactive. That is the status of the well at the time of exporting the data. Given the historical nature of the dataset, one could find out whether or not a well was a DUC at one point by noting its rig release date versus the completion date as noted above.

License Grouping

There can be multiple wells that are drilled at the same time or over long stretches of time at a common surface location. When there are multiple wells (or more often known as multi-well pads) drilled at the same time, a typical rig will stay on the site until everything is complete. In another scenario, a producer might want to drill another well at the same surface location years down the road How do we capture this? For our analysis, we grouped wells together by a common surface location. I went a bit further when we were looking into this by grouping wells with common surface location and by common license date (by month, year). Doing this allows us to roughly estimate which group of wells were drilled and completed at the same time. However, there can be errors like when a producer decides to license 3 wells but drills 2, or if they add an additional well to their license the next day after.

DUC+

While the strict criteria of a DUC was a lag time of two or more months between rig release and completion, we felt this did not cover all types of well activities. We designated wells that "looked and acted" like a DUC by two additional criteria:

• Well is completed, but there is more than 2 months of no production between rig release date and the first date of production.
• Well is completed and starts to flow with production, but there is a gap in production within the first 3 reporting periods (or roughly first 3 months) of said production. This indicates there was some testing involved, but the well was shut-in for a period of time for reasons unknown to us.

So what’s so special about DUC+ wells and why did our group care? The main idea of exploring these particular wells was because true DUCs have not had the capital money spent on the completion of a well nor on the facility that’s built around them to allow for production. However, with DUC+, either capital has already been spent on completions or capital has been spent on both completion and the above-ground facilities and piping. If we can use some sort of rule of thumb for capital costs based on where the well is at in its lifespan, then we could in theory (and in a very crude way), estimate the value of those assets or liabilities.

The second reason is that because they act in the same way as a true DUC, in which they are not actively producing, it gives us an opportunity to evaluate how much surplus production a company can have in theory if they are short on capital to drill new wells.

Results

We found around 1400–1500 cumulative DUCs over the span of 5 years in about a dataset that had roughly 10,000 unique wells. This means at some point in their timeline, 15% of wells were a DUC, which from my point of view, isn’t an outrageous number. You can see that there is a rough correlation between oil price and the number of DUCs being held by producers. What is very important to note that the graph starts at 0, but also ends at 0 cumulative DUCs. We have no information if there were an inventory of DUCs before. We also know the dataset doesn’t constitute all of the producers, and we are limited to data published after 2015.

What is different than the Canadian DUCs that we observed was that the overall accumulation of DUCs down in the US was positive rather than net zero. What we are observing, albeit with limited data, is that Canadian producers operate very differently than their US counterparts when it comes to when they tap into their DUC inventory.

If we broke down true DUCs, DUCs+, and other well pattern types, this is the type of rough distribution we could estimate:

Here is the rough definitions of what I meant by the above legend:

• Completed DUC with a long completion gap: This is a well that is completed after a month but no longer than 2 months between rig release and completions. I would consider this a grey area for a DUC+ type well, as there are other factors at play such as delayed construction, maybe a completion done between years, holidays, etc.
• Completed DUC with long production gap: This is a well that is completed but there is a gap of more than two or more months between completion and initial production.
• Delayed Well: This well has a rig release date but no completion date, so we don’t know what is going to happen to it.
• Shut-In: This is a misleading legend title. This a well where the time between rig release and completions is less than 2 months, so we did not include them as they might not have intended to be a true DUC.
• True DUC: As described before, this well had two or more month gap between rig release and completions, so at some point in its lifetime it was a DUC.
• True DUC with long production gap: This is a true DUC, but it was identified that it had a two or more month gap between rig release and completion, AND a two or more month gap between when it was completed and its first recorded production date.

Production

One topic that wasn’t discussed in great detail was whether or not there was so financial/economic reason why certain wells would be designated as DUCs and others not. What we can see in the figure below, is that it depends on which formation you look at. For Montney, Viking, and Cardium, non-DUCs produce more than DUCs, whereas it is the opposite for the Duvernay wells. When you look at the overall average of DUCs versus non-DUCs production, the aggregate is similar to that of the Duvernay formation. In short, the inconsistencies do not give a clear answer why one well would be chosen as a DUC over another if we were to solely focus on potential production alone.

Proven Assets

One of the key metrics that investors, banks, and interested parties can obtain from the knowledge of DUC counts is the proven assets of a given company. However, we didn’t look at the DUC production decline curve (for each formation) more closely. A typical well production decline curves looks as follows:

There are two great possibilities that we could have extended our analysis as a value added item in the datathon:

1. For true DUCs, we could have used the surrounding proven wells to predict the production and decline of the DUCs and see how closely we could match it. Now, you not only have a proven asset in terms of a well that is ready to produce, you also can predict with confidence what its production will look like once it is completed.
2. For wells that have been completed but either have not started production or have had some initial test production but then shut in shortly afterwards, we can also use machine learning to predict if a) what the production decline curve will look like or b) if a well will match the test production once a producer restarts the well.

Top Operators by DUC Count

1. Baytex Energy
2. ARC Resources
3. Seven Generations Energy
4. Petronas Energy Canada
5. Tourmaline Oil
6. Teine Energy
7. Ovintiv Canada
8. Paramount Resources
9. Advantage Oil and Gas
10. Shell Canada

We found that Baytex Energy had over 200 DUCs during the 5 year period, while the next producer, ARC Resources, had half as many around 100. Shell, coming in at 10th, had only around ~30. Without further details, we were unable to answer the question of why and what explains how DUCs are determined and what factors influence their numbers.

Financials

One of the major hurdles we encountered during the very end of the datathon was to try and incorporate financials of the companies that had the most DUCs.As we collated the data from TSX, we began to realize that this was probably not feasible (at least with the time limit that we had). Consider that Baytex Energy was predicted to have the most DUCs in Canada at 200+. For simplicity’s sake, its annual income statement from 2015–2019 hovered around $1.1–1.8 billion CAD.

Now consider Royal Dutch Shell which had a revenue that is 300 times that of Baytex. Also note that there is no way to quickly figure out the portion of revenue that is specific to Shell Canada rather than Royal Dutch Shell as a whole. We predicted Shell to have around 30 DUCs, so obviously there are significant more factors at play than what we had time for.

I want to note that we tried to comb through the finances of all the public companies that were represented in the dataset, and go through their assets and liabilities. As described above, this did not work in our favour and we had to abandon a good 5–10 hours of work. This did not represent time lost or a failure in my opinion. We were able to conclude that it would not possible to derive any meaningful conclusions from the information gathered, and we were able to also exclude this in order to get our primary messages across.

2020 Mergers

It has been quite the rollercoaster ride for oil and gas majors for 2020. If we did a quick analysis and review some of the major mergers and acquisitions within the oil industry (at least in the Canadian context), would these in any way affect their overall predicted inventory of DUCs?

• Cenovus and Husky: Perhaps the biggest merger of 2020, we had no numbers for Cenovus and the DUCs from Husky were minimal.
• Tourmaline buys Jupiter and Modern Resources: Although Tourmaline is #5 on our list, DUCs from Modern Resources was limited, and a lot of the Tourmaline wells are gas wells rather than oil. There was no information on Jupiter DUCs.
• Obsidian Energy and Bonterra: As of the writing of this post, Obsidian Energy has not officially acquired Bonterra but rather suggest Bonterra shareholders agree to an acquisition. It was interesting to note that Bonterra had more DUCs than Obsidian, so there could be something more at play but unlikely.
• Whitecap and NAL: For all intents and purposes, Whitecap and NAL have similar number of DUCs.
• Strath and Cona: Both producers agreed to merge, and their DUCs were limited in number so I didn’t think there would be any difference.

From the 5 case studies, the DUCs would have little affect on any merger and acquisition decisions when looking back.

Regulatory and Environmental

One of the most important items that I always find we miss is the focus on the public as a stakeholder. The environmental and regulatory processes in Alberta and Canada are always under scrutiny. Below is an example article that focuses on the issue of orphaned and abandoned wells in Alberta.

The economic climate has been particularly brutal in 2020. It gives us pause on the impact of possible bankruptcies in the oil and gas sector. Even the pioneer in shale hydraulic fracturing giant, Chesapeake Energy, could not escape this fate back in June 2020 (albeit there was poor management prior and high debt loading). Having a clear picture on DUCs, DUCs+, and number of wells can give us an idea of what the current landscape looks like. It can help regulators like the BC Oil and Gas Commission (OGC) and the Alberta Energy Regulator (AER) better prepare should a bankruptcy occur.

This is hugely important as one would remember the Redwater Energy case. After Redwater Energy, a publicly traded oil and gas entity, went into bankruptcy, the AER took its trustee to the Supreme Court on appeal that it had to pay for its environmental liabilities first before its creditors. In the landmark Supreme Court ruling, it writes

Although GTL remains fully protected from personal liability by federal law, it cannot walk away from the environmental liabilities of the bankrupt estate by invoking s. 14.06(4) . On a proper application of the Abitibi test, the Redwater estate must comply with ongoing environmental obligations that are not claims provable in bankruptcy.

Perhaps of interest in the future is an exploration of Alberta Orphan Well map to see if we can gain any insight through the power of Data Science. This would be an increasingly bigger takeaway than purely financials alone.

Takeaways

1. We found over 1400 DUCs over a 5 year period with the data that was given to us (roughly 10,000 wells)
2. We found more than 2500 DUC+ over a 5 year period, which gives us insight into additional capacity that producers might have. This was not asked of us in the datathon, but it was something we were particularly interested about.
3. We could not conclusive find causation to how DUCs are selected and the reasons why producers would have DUCs at any one time.
4. Can DUCs give us insight into finances and assets? Maybe. It loosely correlated with oil prices, production tied to DUC selection was inconclusive, and impact on balance sheets and M&As were also inconclusive. In the video we presented to the judging panel, we also explored rail and pipeline access versus the formation that the wells originated from, but this is not explored in this particular post as I did not explicity explore this.
5. We are missing some components into how DUCs tie back to the environment and regulatory impacts.

Acknowledgements

In no particular order, I want to thank my colleagues Lahcene K, Dapo A, Tony A, Chris L, Adam E, and Qian G.

Resources

• You can my version of the DUC datathon output in my Github.
• You can find a video to our presentation below:

• Our journey during the datathon was published in the Journal of Petroleum Technology titled "A "DUC Hunt" In Canada Reflects the Engineer’s Will To Evolve" written by Trent Jacobs: https://pubs.spe.org/en/jpt/jpt-article-detail/?art=7770