The industry has learned a lot about refracturing vertical wells. Some wells in the DJ basin have been refraced five, six, or seven times, or even more. The initial hydraulic fracturing of a lot of the earlier horizontal wells was ineffective.

Applying what the industry has learned could result in increasing return on investment and net present value without having to drill new wells.

“We’ve done a few more than 300 refracture stimulations in the Barnett where we’re not reperforating the laterals. We’re restimulating the perforations and fracture systems that were there,” said Steve Ingram, North American technology and marketing manager, Halliburton. “What we will continue to see over and over is some form of increase in production up to 50% to 60% over initial production.

“That is a production increase without having to drill a new well or not having to do new perforations. What we are really talking about is having a relatively small capital investment to procure a frac spread, use some interesting diverting materials, do a low-cost intervention, and get a tremendous return on investment,” he continued.

“I do believe that 2014 and beyond will be the time frame for these restimulations. There are hundreds if not thousands of wells in the Eagle Ford producing around 50 b/d with very, very ineffective initial hydraulic fracturing. It doesn’t pay to drill a new well in the Haynesville. But if a company can take one-tenth, one-eighth, or one-sixth of the cost of drilling a well and get 60% of the initial production, operators can truly increase their margins,” Ingram emphasized.

Randy LaFollette, director, applied reservoir technology, Baker Hughes, and Mohan Kelkar, professor and chairman of the Petroleum Engineering Department, University of Tulsa, joined Ingram as part of a technology panel at the Summer NAPE conference in Houston on Aug. 14.

LaFollette asked the audience, “How do we prove what is the better technology? What are the right materials or injections rates? How do we manage to make the most of what Mother Nature gave to us in any given well location?”

As he pointed out, the first horizontal well was drilled in the Barnett in 1991. However, the well was given the thumbs down and “that more or less killed the technology for several years. The point is never to try anything once and give up on it. You need statistically valid experimentation to understand what is really happening.”

Horizontal wells finally took off in the early 2000s with an exponential increase in well productivity. Vertical well locations were replaced by horizontal wells. Between 2004 and 2009, the perforated length of laterals went up dramatically. Volumes per perforation foot of treatment went down dramatically and proppant quantity increased by about 50% during that time period, LaFollette explained.

He described a boosted tree performance analysis. The company “sliced and diced” the directional surveys in over 3,300 wells in a study of the Barnett. “We looked at well location since that makes a pretty decent proxy to reservoir quality issues in the unconventionals. We looked at the completion type, the number of stages, how many fracs were counted in that data on a per well basis, aspects of treatment where it was appropriate, and production management as well.

“In this case when you are boosting the decline of the well, you could see the most important factor was the well’s location. Another critical factor was the gross perforated interval. If you had less gross perforated interval in the initial completion, you were going to have a poorer well. If average injection rate was not up to snuff, you were going to have a poorer well,” he continued.

In applying the boosting tree in the Bakken shale, the company found the most influential factors on production in order were location, proppant quantity, and fluid volume. “A lot of people pay attention to the stage count in the Bakken. However, stage count is not subject to good statistical analysis because there hasn’t been enough experimentation with variable proppant quantities in the stage count,” LaFollette said.

If you look at barrels of oil per foot of lateral, the most influential variables are 1) completed lateral length (CLAT), 2) location, 3) proppant quantity, 4) fluid volume, and 5) proppant concentration. The best wells in terms of barrels of oil per foot are on 640-acre spacing. This can also be seen in other shale plays as well when data mining is conducted, he added.

“There are issues with being able to effectively complete and clean up the longer laterals. These are also more expensive to work on, if anyone hasn’t figured that out yet,” he said. “That does not speak to the economics of 640 acres versus 1,280 acres. I suggest that you put a pencil to that to see what it can do for you.”

As LaFollette emphasized in his summary, “I think we demonstrated that analyzing data is a good measure to leverage technology. Everything matters. It is just that some things matter more than others. For example, longer is not necessarily better in terms of lateral length. In the Barnett and other plays, we found that it is a nonlinear relationship. Another reason to go to boosting-tree type analysis is that you can take into account sweet spots.”

Kelkar focused on a new method for estimating unconventional reserves that addresses underlying geological continuity, interference between wells, and the relationship between volumetric analysis and production performance. In looking at data from wells the Haynesville, digital well logs from old vertical wells drilled through the same formation were combined with normalized IP data from a large number of horizontal wells.

The 64 vertical wells were not drilled in the same locations as the horizontal wells. After filtering the raw IP data and contours along with the log data, the correlation between the filtered log data and the performance data were strikingly similar.

As Kelkar noted, “The production data from unconventional resources is not random if appropriate scale is incorporated in the analysis. The fracturing network created at new wells is not completely independent. The production performance of in-fill wells can be influenced by surrounding wells and can depend on the spacing. There is connectivity in the fracture networks.”

Contact the author, Scott Weeden, at