Jilles van den Beukel and Enno Peters
Every month the EIA produces an update of US tight oil and shale gas production. For each major play, the key figures are the total production and the added new well oil production per rig. In this paper we try to analyse and better understand these figures, focusing on the new well oil production per rig for the Bakken.
Whilst the increase in Bakken production to over a million barrels/day is impressive (at the least), we find the increase in the added new well oil production per rig (from about 100 to over 700 barrels/day) the most impressive – and intriguing. If the oil production resulting from a month of drilling can increase by such an amount over a period of 10 years (and if oil in place figures for a single play like the Bakken indeed run in the tens or hundreds of billion barrels) then this seems to carry a great promise for the future. What chance does OPEC have to reign in shale oil production in the long term if technology can give us these kind of productivity gains?
But is technology the key driver? The question we have asked ourselves is: what lies behind these productivity gains? To what extent is it a better understanding of geology and the location of production sweet spots? To what extent is it technology, the ability to drill longer horizontal sections that are fracked in an increasing number of stages with larger proppant volumes? Are increasing efficiencies an important component (in other words: are we just drilling faster)? To what extent does high grading play a role (reducing drilling activities to the very best areas onlyin a low oil price environment)?
We have based this work on data available in the public domain (EIA, NDIC oil and gas division), the relatively limited amount of recent overview papers on the Bakken petroleum system that we were able to find (e.g., Grau and Sterling, 2011) and have used shaleprofile.com (the website on US shale oil production built by one of us (Enno Peters)).
The main target interval for Bakken production is the silty and dolomitic Middle Bakken. This layer is situated in between two shale layers, the Upper and Lower Bakken. These shales, with high organic content, are the source rocks for the Bakken petroleum system. Additional production comes from the Three Forks member, immediately underlying the Bakken Formation.
Hydrocarbon generation in the Upper and Lower Bakken shales has resulted in overpressure generation and fracturing (more intense in the centre of the basin and gradually decreasing towards its margins; the centre of the basin having just entered the gas generating window). This has enabled updip migration of oil through the Middle Bakken towards the basin margins. Migration has for instance been taking place towards the SW (Elm Coulee area, pinchout stratigraphic trap towards the SW), towards the E (Sanish Parshall sweetspot area, diagenetic trap towards the E) and towards the N where more intense faulting and fracturing on the Nesson anticline has enabled oil to (partially) migrate out of the Middle Bakken. As a result oil in the centre of the basin is locally generated, whereas oil in the more peripheral sweetspots tends to be a mix of locally generated and migrated oil. A close correlation exists between oil production, oil saturation (inferred from the amount of water in early production) and overpressure (Theloy and Sonnenberg, 2012).
The overall distribution of oil productivity and sweetspots is given in more detail in the figure below (from Theloy and Sonnenberg, 2013 (except for annotation)). Except for the Northern part of the Nesson Anticline, the pattern is not overly complicated and most production comes from a limited number of sizeable areas. It is clear that the location of a well has a large bearing on its expected EUR. The completion of a well will obviously have an influence as well but seems unlikely to be able to fully compensate for a lack of “good geology”.
The large number of wells that have been drilled in the Middle Bakken post 2005, and the geographic spread, ensure that the overall pattern of producing areas and sweetspots must have been well known by 2009 (and probably quite a bit earlier; by the end of 2009 about 1000 wells had been drilled). A key event was the discovery (EOG’s Parshall 1-36H well) of the Parshall area in 2006 after approximately 50 wells targeting the Middle Bakken in North Dakota had been drilled. This area is the only area that is charactised by overpressures that fall above the regional trend (thus being a seperate pressure cell) and is the most prolific sweet spot.
In short: the location of a well is of key importance to its EUR. The hydrocarbon productivity pattern was already well established by 2009. Establishing this pattern (and in particular the Parshall discovery) was key in getting the play of the ground. Increased knowledge of the hydrocarbon productivity pattern/sweet spot location cannot have been responsible, however, for the major post 2009 advances.
The figure below shows the number of North Dakota wells spud, for a 30 days period, per active rig. With the recent large drop in the number of active rigs we do not want to read too much in the large swings for the last few months.
Based on these data it is clear that there has been a large increase in drilling efficiency in the 2011-2015 period, of roughly a factor 2. This period coincided with a very high level of drilling activity with about 200 active rigs. This is not surprising; increasing levels of activity result in gaining experience and increasing efficiencies in virtually every industry.
We welcome comments on more specific reasons why drilling efficiency has increased so much. Potential components we could think of are more experience on how to drill (“learning while doing”), more sharing of best practices throughout the industry, better equipment (rigs, drilling bits, motors, etc.), less time spent on keeping producing sections exactly horizontal, less time spent on hole cleaning, increased use of batch drilling from a single location, etc.
If we now look at the average cumulative production per Bakken well (all formations, all counties) then we interpret this figure in the following way:
– the large increase in well productivity in the early years we attibute to learning the basics on the geology and the location of sweet spots. Pre 2008, during this geology learning phase, a relatively large proportion of wells was still being drilled in what we now know to be areas of low production.
– the limited increase in well productivity post 2008 (figure below) is in striking contrast with the large increase in rig productivity post 2008 (EIA figure at the beginning of the paper). The significant increase in drilling efficiency (which does influence rig productivity but has no bearing on well quality) is the main reason for this. A large part of the increase in rig productivity post 2008 is not drilling better wells but simply drilling them faster.
Nevertheless, the figure above does show some real increase in well productivity (post 2008) as well. In the following we want to look at potential reasons for this.
High grading vs well quality
High grading we here define as focusing on the best areas and intervals in a low oil price world. An additional component may come from keeping only the best performing rigs and crews in such an environment. Well quality we here define in a narrow sense: the well productivity for a particular area and stratigraphic interval. An increase in well quality in this narrow sense is due to technological advances (such as longer producing sections, more and larger fracs) rather than geological advances (increased knowledge on where the best producing areas are located) or high grading (focusing on sweetspots out of financial considerations). In addition, depletion may result in a decrease of well productivity if some oil of the oil in the targeted area has been produced by an existing well. It may be masked by an increase in well quality; a true decrease in well productivity (for a given interval and area) must come from depletion, however.
The figure below gives average cumulative Bakken well production for the Middle Bakken in the two counties with the best producing wells (Mountrail and McKenzie). In contrast to the previous figure (which showed production for all counties and intervals), this figure shows no systematic increase of well productivity with time. The production for a given area and interval seems to be relatively constant (if anything the data seem to suggest a slight decrease with time; more so for the long term production and less so for the initial production). This suggests that high grading is the main reason for the post 2008 increase in well productivity rather than technological advances.
Well productivity for other individual areas shows varying results. In some cases there is an increase in well productivity over time – be it that this tends to be strongest in the earlier years and tends to be less pronounced than the increase in well productivity for all counties combined.
The final figure gives the number of wells spudded (as a fraction of the total) in the different counties. It illustrates the increasing focus on the best counties (McKenzie, Mountrail, Williams). The effect is not that pronounced, however, and we suspect that only keeping the best performing rigs also plays an important role in high grading.
The way we now look at the EIA figure of Bakken new well oil production per rig is given in the figure below.
For the Bakken new oil production per active rig we see the following timeline:
- The basics of the technology, horizontal wells and fracking, were developed in the Barnet shale in the 1990’s.
- Subsequently, the potential of the Middle Bakken was recognised. By 2009 the overall picture of well productivity and production sweet spots was well established. A key milestone was the drilling of the 2006 EOG wells that established the Sanish Parshell sweetspot area. Up to 2009 increases in new well oil production per rig came primarily from increased knowledge on geology and sweet spot location.
- Upon a brief interlude (2008-2010 oil price low and reduced drilling activity) the Bakken play took off in earnest. During 2011-2015 the active rig count was in excess of 200. During this period increases in new well oil production per rig came primarily from increased drilling performance.
- During the subsequent low oil price world, starting in mid 2014, further increases in new well oil production per rig were primarily due to high grading (drilling in the most productive areas with the best performing rigs).
The road ahead
In our opinion there are no major further advances in new well oil production per rig to be expected. That is perhaps a bold call to make after 10 years during which it increased by a factor 7. Had the increases in new well oil production per rig over the last 10 years been due to technological advances (an increase in well productivity for a given area and stratigraphic interval) we would have been much more reluctant to make this call. But they are not. They are due to better geological knowledge, faster drilling and high grading. Here, we see much less scope for further improvements.
Geologically, the play is now well established. We expect drilling efficiencies to have reached its limit, upon years of high activity followed by two years of low oil prices and intense competition in the service industry. We see no scope for further high grading now that the active rig count is down to about 30. In short: we think that the Bakken shale oil industry is now as competitive as it can possibly be.