The ongoing US shale oil boom has operators rewriting the field development playbook.
One question that has not yet been settled is how close you can space hydraulically fractured horizontal wellbores before they start to interfere with one another’s production.
600 foot wellbore spacing has become commonplace in field development, but many operators are also testing 300 foot spacing. This is important for several reasons, however one in particular stands out for those of us studying the evolution of shale oil plays.
If you can anticipate what wellbore spacing best practices the industry will ultimately adopt based on the interplay between well interference and total drilling locations, it’s much easier to determine the long-term productivity of plays like the Bakken, Eagle Ford and Marcellus.
Well Interference in the Bakken
To shed light on what spacing regime might win out, let’s leverage Drillinginfo’s comprehensive US production data.
We can compare production volumes in adjacent wells to see if we can find evidence of well interference.
First, we’ll filter the data down to pairs of wells based on three criteria:
- Nearest neighbors to each other and no more than 10 miles apart.
- Drilled by the same operator.
- Brought into production more than a specified minimum number of months apart.
The first two filters will help control variations in geology and completion practices between the well pairs. The last will ensure the first (primary) drilled well’s production isn’t impacted by the second (secondary) drilled well.
We’ll narrow our scope to the Bakken because it has significant production history, with extensive unconventional development beginning in 2004.
Running the Shale Oil Numbers
For each well pair, we’ll calculate the ratio of six month total production between the secondary and primary wells and plot the median ratio as a function of wellbore distance.
For each pair, the secondary well begins production at least six months after the primary. To keep an eye on the data set size I colored the distance intervals by the number of data points in each interval.
If well interference is occurring, we expect the secondary well to produce less than the primary well over the same period, resulting in a production ratio less than 100%.
As you can see, the chart suggests well interference occurs when you space your wellbores less than roughly 2000 feet apart. For tightly spaced wells, the secondary wells are only producing about 85% as much as the primary wells.
At greater than 2000 foot spacing, the secondary wells consistently out-produce the primary wells. You would intuitively expect this to happen in the absence of well interference. All things being equal, an operator with six more months of unconventional drilling experience is generally going to drill more productive wells.
Now that we see the behavior at six months, let’s extend the time frame to look at the longer term implications. We will compare 12 month total production for well pairs drilled at least 12 months apart. Here you see the same trends as with 6 months, but they are more pronounced.
Again, 2000 foot wellbore spacing is a breakpoint. For wells spaced greater than 2000 feet apart, secondary wells significantly out-produce primary wells, which implies no interference. For well spacing less than 2000 feet, secondary wells lag further behind primary wells than they did at six months. The production ratio here is 70% to 80% for wells with the tightest spacing.
Finally, let’s take an even longer view and look at 24 month total production for wells drilled at least 24 months apart. The size of the dataset becomes more of an issue because of the filtering criteria, but we see the same trends as before. And this time they are even more apparent. At the tightest spacing, secondary wells only produce about 60% of the amount primary wells produce.
The Net
Admittedly, to make more quantitative rather than qualitative conclusions, we would need a larger body of data. But waiting around for beautiful datasets to appear is a luxury we do not have in this industry. So we are left with three conclusions:
- Well interference in the Bakken appears to occur for hydraulically fractured horizontal wellbores spaced closer than roughly 2000 feet.
- The magnitude of well interference on production appears to increase over time.
- The full impact of well interference in the analysis above is likely somewhat masked since operators become more proficient in drilling and completion techniques over time. As we saw, secondary wells over-perform when spaced wider than 2000 feet.
The 300 foot to 600 foot spacing regime operators are implementing in the Bakken is also commonly observed in the other major shale plays, such as the Eagle Ford and Marcellus. If the Bakken is any indication, most shale oil wells are or will be subject to significant well to well interference. But if the rates of return are high enough, it can make perfect sense for operators to sacrifice individual well productivity in the name of maximizing the total amount and rate at which an area produces.
All of this leaves us with one key takeaway people making decisions in the boardroom and out in the patch must remember. If you want to accurately estimate your well’s production type curve, you can’t forget to account for interference based on your spacing strategy.
Your Turn
What do you think? Is it worth the per-well productivity hit to minimize well spacing and maximize total drilling locations? How do you feel tighter well spacing will affect the overall productivity and decline of the shale oil plays around the country? Please leave your thoughts in the comments below.
Enlightening. Thank you.
You are welcome Chuck. Thanks for reading.
Enlightening. Thank you.
You are welcome Chuck. Thanks for reading.
Is your data sufficiently detailed to allow determination of distance between the horizontal portions of wells? If not, any difference in the length of horizontal legs will affect the accuracy of your results.
Mike, the distance between wellbores was estimated using the mid-points of each well’s horizontal length. The horizontal length was defined by drawing a straight line from the well’s surface hole location to the bottom hole location. While not perfectly accurate, this is a reasonable approximation given that I’m binning by 500 ft distance intervals.
The effect of different horizontal lengths on wellbore distance is pretty minor because lateral lengths are so uniform in the Bakken, with by far the most common lengths being around 10,000 ft. That said, what you’ve suggested is one of a couple of additional filters I could apply if I wanted to draw more quantitative conclusions. The challenge is whether there would be any data left after all the possible filters had been been applied. Thanks for the comment.
Thanks for the explanation Kevin. This is a very interesting first cut.
Is your data sufficiently detailed to allow determination of distance between the horizontal portions of wells? If not, any difference in the length of horizontal legs will affect the accuracy of your results.
Mike, the distance between wellbores was estimated using the mid-points of each well’s horizontal length. The horizontal length was defined by drawing a straight line from the well’s surface hole location to the bottom hole location. While not perfectly accurate, this is a reasonable approximation given that I’m binning by 500 ft distance intervals.
The effect of different horizontal lengths on wellbore distance is pretty minor because lateral lengths are so uniform in the Bakken, with by far the most common lengths being around 10,000 ft. That said, what you’ve suggested is one of a couple of additional filters I could apply if I wanted to draw more quantitative conclusions. The challenge is whether there would be any data left after all the possible filters had been been applied. Thanks for the comment.
Thanks for the explanation Kevin. This is a very interesting first cut.
Interesting….
What is your take on the work by Hughes, e.g. Drill, Baby, Drill and his recent update presented here
https://legacy.firstenergy.com/UserFiles/HUGHES%20First%20Energy%20Nov%2019%202013.pdf
Interesting….
What is your take on the work by Hughes, e.g. Drill, Baby, Drill and his recent update presented here
https://legacy.firstenergy.com/UserFiles/HUGHES%20First%20Energy%20Nov%2019%202013.pdf
Interesting….
What is your take on the work by Hughes, e.g. Drill, Baby, Drill and his recent update presented here
https://legacy.firstenergy.com/UserFiles/HUGHES%20First%20Energy%20Nov%2019%202013.pdf
Kevin,
Is there any way you could provide the paired well listing used in the analysis? I am wondering how you discriminated between Bakken and Three Forks completions.
Mike, the Drillinginfo well database I used for the analysis does differentiate between Bakken and Three Forks wells. I included a filter so only Bakken wells were considered. If you’d like further information, you can email me at [email protected]. Thanks.
Kevin,
Is there any way you could provide the paired well listing used in the analysis? I am wondering how you discriminated between Bakken and Three Forks completions.
Mike, the Drillinginfo well database I used for the analysis does differentiate between Bakken and Three Forks wells. I included a filter so only Bakken wells were considered. If you’d like further information, you can email me at [email protected]. Thanks.
Well done! We’ve wrestled with this for several years and are living it right now as we are booking yearend reserves in the Bakken/Three Forks. Of course, the bottom line question is incremental reserve recovery versus acceleration of the existing reserves and it boils down to present value (as you allude to in your conclusions). I can confirm your findings with our own well data set in the Mountrail County, ND area.
Thanks. It’s good to hear you’ve found similar results.
Well done! We’ve wrestled with this for several years and are living it right now as we are booking yearend reserves in the Bakken/Three Forks. Of course, the bottom line question is incremental reserve recovery versus acceleration of the existing reserves and it boils down to present value (as you allude to in your conclusions). I can confirm your findings with our own well data set in the Mountrail County, ND area.
Thanks. It’s good to hear you’ve found similar results.
Very nice empirical analysis and, not the least, experimental design. We have done similar “theoretical” work for Marcellus based on fracture density and natural communication distance. This was inspired by work done by Prof. Engelder at PSU on the risk of aquifer pollution from shale operations. Essentially we find the same break-point at around 2000 ft for top quality shale. My guess is that gas basins will see stronger early losses in well 2.
Poorer shale can tolerate/require tighter spacing for maximum resource recovery. It would be interesting if you could do similar analysis for Barnett and Eagle Ford. I would expect Barnett to have a slightly lower break-point than the Bakken and “north” Marcellus.
Regardless, the results point to a much needed and interesting discussion – also in oil companies – about EUR vs IP and economically optimal well spacing. It seems to me that some companies are now so driven by high rig utilization, and to report ever-lower well cost and high 24hr IPs that they miss NPV.
Kjell Eikland
energyper.com
(PS! Have anyone thought about “rule of capture” in this picture?).
Thanks for adding your insights. It would be interesting to see the results of this analysis in other plays, as you suggest. We are considering doing just that in follow up blog posts. The only issue is that most other plays do not have the same length of production history as the Bakken, so the data is more sparse.
How would rule of capture come into play here? Do you mean looking at well pairs leased by different operators within in the interference interval?
Very nice empirical analysis and, not the least, experimental design. We have done similar “theoretical” work for Marcellus based on fracture density and natural communication distance. This was inspired by work done by Prof. Engelder at PSU on the risk of aquifer pollution from shale operations. Essentially we find the same break-point at around 2000 ft for top quality shale. My guess is that gas basins will see stronger early losses in well 2.
Poorer shale can tolerate/require tighter spacing for maximum resource recovery. It would be interesting if you could do similar analysis for Barnett and Eagle Ford. I would expect Barnett to have a slightly lower break-point than the Bakken and “north” Marcellus.
Regardless, the results point to a much needed and interesting discussion – also in oil companies – about EUR vs IP and economically optimal well spacing. It seems to me that some companies are now so driven by high rig utilization, and to report ever-lower well cost and high 24hr IPs that they miss NPV.
Kjell Eikland
energyper.com
(PS! Have anyone thought about “rule of capture” in this picture?).
Thanks for adding your insights. It would be interesting to see the results of this analysis in other plays, as you suggest. We are considering doing just that in follow up blog posts. The only issue is that most other plays do not have the same length of production history as the Bakken, so the data is more sparse.
How would rule of capture come into play here? Do you mean looking at well pairs leased by different operators within in the interference interval?
now, there is zipper fracking which they claim at least the production is improved significantly after fracking two adjacent wells closely spaced.
now, there is zipper fracking which they claim at least the production is improved significantly after fracking two adjacent wells closely spaced.