Chesapeake Teams With Analytics Firm To Improve Asset Performance

The Oklahoma City independent has a new-look portfolio and new operational and financial priorities. And now it has enlisted an energy research firm to leverage advanced analytics and machine learning to help get the most out of its assets.

A Chesapeake-operated pump jack in Wyoming's Powder River Basin, the company's burgeoning "oil growth engine."
Source: Chesapeake Energy.

Chesapeake Energy is partnering with RS Energy Group to improve operational efficiency and capital discipline by employing advanced analytics and machine learning.

RS Energy is a Calgary-based energy research firm founded in 1998 covering more than 150 operators in the major North American and international oil and gas regions, including the US shale plays. It provides technical analysis of basins, including completions and production, as well as asset evaluations for operators considering acreage additions. All of this is done within the context of shifting capital markets.  

Chesapeake announced the pact fresh off its $4-billion merger with WildHorse Resource Development, which bolstered its position in the Eagle Ford Shale of South Texas. The Oklahoma City-based independent has overhauled its portfolio in recent years and now is focused on just a few major US basins, increasing its companywide share of oil production, and reducing debt.

Doug Lawler, Chesapeake president and chief executive officer, said in a news release that he believes RS Energy’s “deep understanding” of the industry, Chesapeake’s asset portfolio, and physical and financial markets will help the operator “achieve our strategic goals and create additional value."

During a recent SPE Gulf Coast Section event, Brook Papau, RS Energy managing director, explained that his firm has found that, for operators, investing $1 million in data science yields results many times greater than investing that sum directly into the well.  

Papau said RS Energy polled its operator clients asking how much well-by-well improvement they believed they could capture by employing data science initiatives such as advanced analytics and machine learning. “The majority said 10%, and the vast majority said 5% or greater,” he said.

Data-Driven Case Studies

To illustrate his point—and make a case for his firm—Papau cited examples in which RS Energy leveraged advanced analytics and machine learning to glean development insights from big US operators and plays.

In one instance, the company looked at publicly available data on Parsley Energy-operated Wolfcamp B wells and tried to figure out the optimal amount of wells drilled per section on its land. The question was whether the operator should drill very few wells and “jack up” its net present value (NPV) per well, he said, or “go extreme” with 30 wells/section where interwell communication becomes a problem and NPV/well drops.

RS Energy first found NPV for a 5 wells/section was $10 million. With 8 wells/section, RS Energy saw declines in NPV/well, estimated ultimate recovery (EUR)/well, and type curve while the EUR/section and gas-oil ratio (GOR) increased. Most importantly, NPV/section improved, which is what ultimately should be maximized, Papau said.

At 17 wells/section, the type curve decreased but NPV/section went “way up,” he said. With 26 wells/section, NPV/well declined, “so we figured out kind of where the optimization would be—16 to 18 wells. And our EUR/well actually got better.”

On the higher end of the analysis, with 32 wells/section, “the GOR went way up and these wells started to actually gas out,” while the EUR/well and EUR/section dropped off.

RS Energy also has observed how operators are attempting mitigate lost productivity from child and parent wells. “And we know time and time again—every time we write a research report—the biggest indicator of whether or not there’s going to be a parent-child interaction is the time it takes between the initial well and the subsequent wells.”

In another example, Apache had to drill on a piece of acreage to hold it and then waited 10 months before drilling again. When the operator returned, it drilled two more wells that were both fractured at 1,650 lbs/ft of proppant, and “the child wells were underperforming by quite a bit,” he said.

“So they had to figure out how they were going to complete the remaining wells on this planned pad,” Papau said. “And what they came up with was actually really novel, and I think we'll be seeing a lot more of it.”

This involved fracturing “sacrificial wells” at a lower proppant intensity before fracturing the remaining wells at a higher intensity. The sacrificial wells underperformed, but that tactic made the most of production from the pad as a whole.

“This is an example of advanced analytics,” he said. “We've got all the data at our fingertips. We can run it through models and visually compare” activity in the field.

RS Energy’s machine learning approach uses a random forest decision tree model with completion, spacing, and geological inputs. To glean economic insights from wells in the Wattenberg field of the DJ Basin, the firm set up a cost framework. On the low end, it estimated that fracturing a mile-long lateral with lower proppant and fluid intensity averaged $3.1 million, while, on the high end, fracturing a 10,000-ft lateral with higher proppant and fluid intensity averaged $7.5 million.

The team learned that an operator could get the same NPV/acre by either pumping a high-intensity frac at wider spacing or a low-intensity frac at tighter spacing. However, the internal rate of return metric favored the wells with lower proppant intensity. 

Operators Adopting Data Science

Papau said that RS Energy has always maintained a staff that mirrors that of its operator clients so that it “thinks about the world in the same way.” This means having engineers, geoscientists, and financial professionals on the payroll. In the last couple of years, however, the script has flipped as operators add the data science component. For example, EOG added a chief technology officer and Anadarko now has an entire data science team of almost 20, he noted.

“Now it doesn't mean that [operators] need to go and hire a data scientist,” he said. “But if you want to be future-proofing your team, if you want to be able to compete for assets, your engineer might need to wear the data science hat,” meaning he or she “also needs to code in MySQL.”

This will require overcoming the industry’s pervasive and longstanding data quality challenge. Papau said RS Energy is working “more and more closely with clients who want us to host their data on a separate cloud” and organize the data. When it comes to public data, “every single jurisdiction that we look at has a data quality issue,” he added. “North Dakota is probably the best. Oklahoma's probably the worst.”

But once operators have their data organized and a team in place capable of analyzing the data, Papau said, their digital proficiency tends to improve exponentially.