Ever wonder what impact well spacing and other variables such as fluid and proppant loads, water cut or resistivity could have on the EUR of oil when it comes to parent vs. child wells?

Drillinginfo tackled the topic during a recent webinar, using as example about 1,500 wells of varying age in a densely drilled section of North Dakota’s Williston Basin. The area in Mountrail County essentially has only child locations, particularly more-impacted child well locations, left.

Patrick Rutty, a senior product manager for Drillinginfo, cautioned that the model is still a work in progress. But the workflows, he said, make sense and the conclusions are interesting. Essential to the process is distance- and date-dependent well-spacing data.

He used a multivariate linear and non-linear regression model to predict Bakken EUR using footage in zone, water cut, resistivity, proppant per foot, depth, gas-oil ratio (GOR), fluid per foot and porosity for parent wells, less-impacted child wells and more-impacted child wells. Parent wells were defined as those with no pre-existing wells with 5,000 ft, while the less-impacted children have few relatively distant preexisting wells and the most-impacted children are in crowded neighborhoods.

In this area [of the Bakken], interestingly, the earlier child wells have the highest EURs. So not the parent wells and not the most recent child wells or sort of the latest stage of development child wells,” Rutty explained. “That sort of second generation has the highest EURs. Those later child wells, second or third generation, are worse than the parents and worse than the first-generation child well.”

The model, which is actually three models for each well class combined, uses a regression algorithm to generate optimal transform plots to rank the significance of variables on each well class.

Predictive production modeling, however, becomes more difficult with later wells as well-to-well interactions increase complexity, according to Rutty.

“Things like defensive refracs or defensive pressuring up parent wells and more of that is going on later in the game and it gets harder to model,” he said. “In the early days, though, in the parent and the earlier child well, water cut and lateral length are the biggest predictors of EUR.”

When it comes to proppant levels, however, the benefits vary for each well type.

“Proppants in the parent well helps much more than fluid, based on what we see in the data here and the dataset,” Rutty said.

Depth and resistivity dominate models in the later, or most-impacted, child wells. Increasing GOR has a negative impact for the wells.

Looking at input data trends, Rutty said that the most-impacted wells—those that benefited the most from advantages gained overtime—simply did OK regarding production.

The parent wells didn’t have many advantages, he said, noting they were relatively short laterals with not much proppant or fluid but had good porosity.

“They did just fine because they were the first ones to show up,” he said turning to the less impacted well class, which got some advantages but not many. “They were longer, a lot more proppant; they had more fluid. … They really did well and I think largely it’s this lateral length. It’s proppant.”

Considering companies are now drilling more of the so-called “more-impacted child wells, how can designs be optimized. Rutty turned to another multivariate model for insight, using hybrid, slickwater, crosslink gel and linear gel. Based on the model, he said “generally more proppant gives slightly better EURs for both slickwater and hybrid jobs.”

The same goes for fluids.

“In a later child well, fluid is actually more helpful than proppant. But they might be equally helpful in hybrid and slickwater jobs which is mostly getting pumped up there. So probably, depending on economics in terms of physical outcomes, it’s worth pumping more fluid and more proppant.”

 Velda Addison can be reached at vaddison@hartenergy.com.