Since early 2015, many people have asked about the refrac market, wondering if it is beneficial to reenter a well to complete new zones or to refrac existing perforations. People have many ways of identifying, defining and filtering their data to focus on these refracs. This potential ambiguity makes it vital to understand the set of data used to research this topic.
To set the record straight for this article, the filtering process is as follows: Populate all wells with more than one reported completion. Keep any that filed a second completion from January 1, 2014, through June 30, 2015. Focusing on refracs before June 2015 increases the likelihood that selected wells will have multiple months of reported production post-refrac.
The combination of these constraints left more than 1,700 wells across North America. Showing that group of refracs on a map highlights the large population of refracs in and around Texas, as well as the Northeast and North Dakota, with a smattering scattered throughout Canada.
The Leading Basin for Refracs
Among the most popular basins for refracs is the Williston Basin in North Dakota. The top four companies performing refracs in the Williston combine to represent 67 percent of the refrac market in that basin from January 2014 through June 2015. Their completion activity in the Williston, Marathon Oil Corporation (Marathon), Whiting Petroleum Corporation (Whiting), Continental Resources Inc. (Continental) and ConocoPhillips Company (Conoco) represents 34 percent of the overall initial completion market from January 2014 through June 2015. Displayed on a map, the recompletion activity of these four main operators shows that Marathon has focused its refracs in the southeast portion of the Williston Basin, while other operators such as Conoco and Whiting are not as concentrated (see Figure 1).
What Is the Current Completion Trend?
Before diving into refrac methods and their results, let’s take a step back and look at current completion techniques being used by operators in the Williston as well as the completion methods of these four highlighted operators.
Analyzing the completions from January 2014 through June 2015 shows that Marathon, Whiting, Continental and Conoco represent between 25 and 44 percent of the completion market each month in the Williston Basin. Overall, the four selected operators appear to use slightly different completion methods than other operators in the area, but no significant variation was present.
Operators in this region rely heavily on raw sand, sand and ceramic, and ceramic-only proppant types when performing initial completions. One of the slight differences is the market share for ceramic mass, which was higher among the four selected operators than among the remaining Williston operators. Another important difference is the slightly higher proppant intensity per well for the selected operators compared with others active in the Williston Basin. The main frac job types returned no major variations, because both groups of operators were likely to use slickwater crosslink-hybrid fracs or crosslink fracs, representing a combined market share that regularly exceeded 75 percent of the market. This is probably to be expected in this the Williston Basin because of the drop in oil prices and the cost savings associated with these types of frac jobs.
How the Refracs (Re)Measure-Up
Keeping the initial completion methods in mind, we will focus on the refracs performed by the four selected operators—Marathon, Whiting, Continental and Conoco—and we will remain concentrated on the Williston Basin in North Dakota. The refrac market among these operators peaked with the completion market in Q3 2014 before experiencing fluctuation in 2015. All four seemed to rebound in Q1 2015 over Q4 2014, but refracs almost completely disappeared in Q2 2015.
Looking into different refrac methods reveals that crosslink fracs continued to hold the majority of the market, followed by slickwater and crosslink-hybrid fracs and then a small percentage of slickwater fracs. Analyzing each operator’s choice in frac job type reveals that Marathon, with 41 percent of the Williston refrac market during this timeframe, used crosslink fracs for all of its refracs. Whiting, with a 10 percent share of refracs, used hybrid slickwater-crosslink fracs for 55 percent of its refracs, crosslink fracs for 27 percent of its refracs, and the remaining percent consisted of solely slickwater fracs. Conoco used exclusively hybrid slickwater-crosslink fracs, which were also Continental’s primary choice; however, Continental also used crosslink and slickwater separately in 22 percent and 11 percent of refracs, respectively.
The proppant used in refracs also remained similar to the overall trends of these four primary refrac operators, showing an overall focus on raw sand proppant, which was pumped in 77 percent of the refracs performed by these four operators. The second most common proppant was sand and ceramic, following the general trends seen across the Williston Basin. Compared with current completion methods, the four selected operators’ refracs use less water, which is not uncommon for refracs. Compared with current completion methods that use between 2.5 million and 5 million gallons of water, these refracs consistently used 2 million gallons of water. Although on the surface these trends may seem unsurprising because they are similar to the initial completion trends, they are meaningful.
We know that wells are typically refracked because the initial completion had a poor design or other inefficiencies. By going back and refracking wells with more proppant or premium proppants, or altering any number of completion techniques, operators hope to increase production. Because of the age of many wells in this analysis, the ability to obtain proppant and other downhole consumable details is limited.
So the remainder of our research will rely solely on wells that have reported information to the state for these initial fracs, allowing for production comparisons to be made. This reduces the population from 74 refracs by the four selected operators to 53 refracs.
It’s a Marathon, Not a Sprint
Companies that have held assets and been around for a while, such as Marathon and Conoco, tell the dramatic story of how much technology and fracking have advanced the industry. After pulling in wells with initial completion proppant details, we were left with a market still controlled by Marathon, representing 85 percent (an increase from 62 percent) of the four selected operators’ market in the Williston. The remaining 15 percent consists of Conoco with a 9 percent share among the four selected operators and Whiting with a 6 percent share, while Continental fell completely from the list. Of the remaining wells, 79 percent of them were initially fracked in 2008 or 2009, 6 percent were fracked in 2007, and the remaining 15 percent were originally fracked in 2010 and 2011.
Marathon and Conoco both significantly increased their proppant use from the initial frac to the refrac. Marathon increased proppant per completion by an average of more than three times (327 percent) in their 45 refracs, while Conoco increased its proppant per completion by 2.5 times (248 percent) in its 5 refracs. Marathon, with the largest set of refracs meeting our criteria, experienced huge success with its refracked wells.
From January 2014 through June 2015, Marathon completed 45 horizontal refracs in the Williston that had proppant mass reported for their original fracs. Analysis of their production shows that the shortest interval from the first month of initial production to refrac was almost four years (45 months). At year six, 87 percent of wells were still producing, which means that 13 percent of the 45 refracs occurred within the first six years.
Many of Marathon’s refracs took place between six and seven years from the first month of initial production. After being refracked, peak month production of these 45 wells averaged 78 percent higher than the peak month of the initial completion.
Furthermore, the refrac production did not decline as quickly as the initial production. Eight months after peak production, the refrac production rate declined 43 percent, while the initial production rate decreased by 47 percent. After 12 months, the refrac production had declined by 52 percent and the initial production had declined by 45 percent, showing that the refrac production was falling off at a higher rate. This comparison is based on the full set of 45 refracs as each refrac reported at least 12 months of production and 49 percent of these refracs reported at least 20 months of production, allowing for a full comparison.
Though Conoco does not have as many refracs meeting the above criteria, with only five horizontal refracs during this timeframe, the company successfully increased production after pumping almost 250 percent more proppant than its initial completions. The original completion in each of these five wells produced for almost 6 years (70 months) before the refrac occurred. All five wells were refracked between 70 and 77 months of production, and each of these wells reported at least 17 months of production after being refracked.
Within the small population of refracs, Conoco was able to increase peak month production for its refracs by 208 percent over its initial production rates. Another benefit of Conoco’s refracs was the slower production decline experienced by refracked wells compared with the initial fracs. The refrac production declined 52 percent from peak after eight months and 63 percent from peak after 12 months, while the initial production declined 68 percent after eight months and 77 percent after 12 months.
Lastly, Whiting has refracked three wells, meeting the above criteria. Contrary to the previous operators, Whiting did not increase the amount of proppant from the initial frac to the refrac as significantly. Though its initial completions used around 1.6 million pounds of proppant per completion (more than Marathon or Conoco), Whiting’s refracs used less than 1.8 million pounds of proppant, an 11 percent increase over the initial fracs (using an average of 35 percent less proppant per refrac compared with Marathon and Conoco).
Whiting began refracking its wells after three years (35 months) of production from the initial frac, and within four years all three of its wells had been refracked. Each refrac well has reported at least 21 months of production. Overall, among its three refracs, Whiting did not experience as much success in increasing production as Marathon and Conoco, but it did experience an increase in production immediately before and after the refrac occurred.
Comparing month 35 of initial production to month 4 of refrac production, Whiting was able to increase its production by 84 percent; however, it did not improve peak month production. In fact, the refrac’s peak month production was 64 percent lower than the peak month of the initial frac.
Adding it All Up
Figures 5 through 7 take into account the total production for all refracked wells both before and after refracking started: 45 wells for Marathon, five wells for Conoco and three wells for Whiting. Each chart analyzes the cumulative production prior to any refrac.
Each operator completed its first refrac at different times with respect to initial production:
- Marathon completed its first refrac 45 months after the well began producing.
- Conoco completed its first refrac after 70 months.
- Whiting’s first refrac was after 35 months.
Regardless of the actual date of the refrac, all production related to the refrac was time-set back to month 1 once the refrac occurred. For example, after 45 months, Marathon refracked its first well, meaning that month 46 for this well would be the first month of refrac production, but its second refrac was not until month 51.
In these figures, month 52 for this particular well would still be represented as “Post-Refrac Activity” month 1. Using this tool, we can analyze the effect on cumulative production caused by refracking for each operator. Similar to the decline rate charts (Figures 2 through 4), these charts evidence the overall difference between production pre- and post-refrac.
Marathon, with 45 producing wells through month 45, experienced a significant jump in cumulative production once it began its refrac efforts. Conoco, with the largest increase in peak month production, also experienced a significant increase in production once its refrac activity began, after 70 months of initial production. Finally, Whiting, which did not see as large of a return as the other operators, still saw a slight increase in cumulative production once the refrac program began.
As operators find wells that have reached a consistent rate of production from month to month, many have started to wonder if refracking a well, or a group of wells, would be beneficial. With improved completion methods and frac designs available, this idea of elevating production without spending the full amount to drill and complete a new well can prove to be successful, but it leaves one crucial question: What is the main reason for completing a refrac? After reviewing the data and seeing the success experienced by Marathon and Conoco, it is easy to believe that many operators perform refracs to receive higher production outputs, but other benefits are possible.
Sure, the profits of receiving production from wells producing minimal oil and gas each month are undoubtedly beneficial, but what if operators could constantly change one variable of their refracs in order to identify small pieces that lead to more success, providing them with crucial knowledge of how to improve the initial completions they will begin? What if operators could plan ahead and keep this possibility in mind when performing initial completions or—better yet—leverage this information to further improve their peak production from day one?
With a collage of older wells making up many of the portfolios of the more seasoned operators, a refrac might be all you can do. But as we continue to see more efficient techniques leveraged on initial completions, especially from many of the younger operators hopping into the market while prices are low, the production data will look drastically different than it does now.