Well design in the Williston Basin has changed significantly since the first horizontal wells in North Dakota and Montana. These wells typically targeted the middle Bakken using very short lateral lengths, a handful of stages and very little water and proppant. Operators have found they can use laterals in excess of 10000 feet with up to 40 stages. These wells now used 100000 barrels of water and up to 9 million pounds of proppant. Improvements have decreased depletion rates, and increased EURs. Decreased depletion will increase the cumulative production derived from fracking. If this is the case, current models may be too conservative.
When looking at Williston Basin production, it is important to remember this is a stacked play. It is focused on the middle Bakken, but the Three Forks may have more upside. The Three Forks is comprised of a series of benches. In some areas a total of four can be found. This is important as four wells could be drilled, producing a total of 16 potential targets not including the middle Bakken. The Three Forks has seen less activity, so it should have more upside. The middle Bakken has been the focus and has seen the most activity. Comparing the two is difficult, but not impossible. Middle Bakken thickness is over a hundred feet thick in some areas, but the Three Forks reaches three hundred feet. The Three Forks is also deeper. It is very likely all of these benches can be drilled from the same pad as Kodiak (KOG) is doing in its Smokey and Polar Prospects.
Models are currently used to determine EURs. These are just estimates, and results could vary significantly. The operator, well design and geology are all independent which is why results have been inconsistent. Historical production is a better method, but there is too little data in most areas. The Sanish and Parshall fields are the only areas with enough data to model using historical production. In order to calculate EURs, depletion must be figured. Early well depletion is the most important and the curve flattens out in two to three years. After this period we can use a consistent depletion rate until production slows to a point where the well is not economic. By analyzing these wells we can see how other wells will deplete, or at least produce an estimate. This is not a perfect comparison, but should provide answers to current depletion rates.
In order to determine Table 1 provides current well designs (subscription required) by different operators. Companies have changed to longer laterals, larger chokes, increased stages, water and proppant.
Differing Well Designs By Operator In The Bakken (Table 1)
|21378||EOG Resources (EOG)||6475||24/64||32||79855||6867099|
|22868||Whiting Petroleum (WLL)||8437||48/64||22||24731||1716416|
|21182||Kodiak Oil & Gas||9304||34/64||19||50574||1546067|
|21884||Continental Resources (CLR)||9597||30/64||39||68090||3828310|
|22350||Oasis Petroleum (OAS)||10020||68/64||28||66254||3507779|
|19738||Triangle Petroleum (TPLM)||9657||30/64||31||82504||3918293|
The above table shows differences in well design. A longer lateral length provides increased surface area of the source rock to be stimulated. The choke size is very important, and a good description is found here. A tight choke allows less resource out of the well, but keeps well pressures high. Because it is more restrictive, IP rates are decreased. In general, tighter chokes are used in wells with higher gas production. It is believed this improves longer-term production, but is not as important in wells with high percentages of liquids.
Well design affects depletion. The number of feet per stage are directly correlated. The shorter the stage, the more effective the hydraulic horsepower is in stimulating the source rock. This creates better fracturing. Over time fractures start to close up under the weight of the formation unless water and proppant are used. The water pushes the proppant deep into the fractures. The proppant "props" the fractures open allowing the resource to flow. Ceramic proppant is more resilient, and does not crush as easily as sand does. Its consistent shape provides more space for crude to flow.
Table 2 is of the Bakken's best wells as of January first of 2013. I have calculated depletion starting at the average of the first 90 days. On the far right of the table, I have the total barrels of oil produced. Remember, these wells have only produced for three to four years. It will continue to produce for an estimated 30 years. It is not inconceivable some of these wells will produce a total of over 2 million barrels of oil in its life time.
The Bakken's Best Oil Producers (Table 2)
|360 Day||720 Day||1080 Day||1440 Day||Bbl Oil|
When modeling production, there are some averages I use. Wells that produce for 5.5 years are estimated to have produced 50% of its total production. In ten years or 120 months, 85% of well production occurs. These time frames can be used to predict future well production. From the 90 to 360-day IP rate we see a high rate of depletion. Years two and three have a slower, consistent depletion. In the fourth year, depletion decreases significantly.
Well design has been a big factor in producing the best middle Bakken wells. EOG has more wells than any other in the top 20. It was using the number of feet per stage, proppant per stage and water per stage that is currently used today by many of the best operators. Because of this, EOG gets more production per foot on average than any other operator in North Dakota. Table 3 provides data on wells operated by Helis. This leasehold was recently purchased by QEP Resources (QEP) and is thought of as some of the best acreage outside Mountrail County.
Helis Results in NE McKenzie County (Table 3)
Helis has done a great job in northeast McKenzie County. Its two top wells produced more than three hundred thousand barrels of oil in the first year, and were still producing more than seven hundred barrels of oil per day. Helis has shown its acreage in Grail Field can produce 1000 MBoe wells in both the middle Bakken and Three Forks. In Table 3, I have marked wells with a * to show there were production problems.
Two of the Helis wells from Table 3 are top twenty wells using its 360-day IP rate. The Helis wells are depleting faster than the top twenty. There are two possible explanations. The first is choke size. Helis currently uses a 28/64 choke, while most operators in 2008 used a 22/64. This could be the reason for the initial decline from the 90 day to 180 day IP rates. Well number 16059 is a good example. This well didn't begin to deplete until two years of production. PetroHunt used a 5/32 choke which is tight by any standards. This is not the only reason, as well 16059 is currently the sweetest spot in North Dakota. Not only is the best well to date, it is so much better than any other that it is difficult to use as a comparison. The second possibility is the gas to oil ratio or GOR. The Helis wells are in northeast McKenzie County, which has a higher percentage of natural gas than in Mountrail County. In Table 2, the 1440-day IP rate had a depletion of 6.3%. Sometime between the 1080 and 1440 production days we see a marked decrease in depletion. This is matrix production, and the time frame used where a well falls into a lower depletion rate. QEP Resources uses 8%, for its acreage in northeast McKenzie County. In Table 4, I will use this as the standard for the remaining days to 5.5 years. It is possible this depletion data is skewed. If well 16059 is removed from the calculated depletion from the 1080-day to 1440-day IP, depletion is closer to the two years prior. I would imagine matrix production would normally start between year five and seven. This is after a hyperbolic initial decline of 87.6% from a northeast McKenzie well versus 72% in a Mountrail County well. These estimates are for middle Bakken wells, so the Helis upper Three Forks wells we could see different depletion rate. Keep all of this in mind, as these are not perfect comparisons.
Helis Well Model (Table 4)
|EUR Bbls Oil|
I used the depletion rates of Helis wells with 360 days of production from Table 3 to calculate that IP rate. I did this because Helis wells have a different production mix and deplete faster than in Table 2. Of the three wells with production problems, two were Helis's worst producers. These wells were also modeled, as I decided to stay true to its total results. Table 2 data was used for the 720-day, as Helis wells have not been in production long enough to provide proper results. I used 8% as an average terminal decline starting after the third year to the 5.5 year mark. At 5.5 years an estimated 50% of production occurs, and this number is doubled. Keep in mind, all of the results from these tables are in barrels of oil only. These EUR estimates do not include natural gas or NGLs.
In summary, in some areas of the Bakken we are beginning to see enough data to produce EURs from historical production. This is important as we can now see how well design can provide better results for some operators. We can also do the same for poor results. Helis uses more water and ceramic proppant and this is why it outperforms with EURs over 1000 MBoe. It is possible northeast McKenzie County could be a better area than southwest Mountrail County. McKenzie County has very good middle Bakken wells, with an even better upper Three Forks. Additional benches also add upside. There are several ways to play northeast McKenzie. Kodiak, Newfield (NFX), Conoco (COP), Hess, EOG, QEP, Halcon (HK) and Continental are all present in this area. Kodiak and Halcon are more levered to this part of the Bakken.