[URTeC 2024] Enhancing Production Efficiency: The Impact of Precision Targeting in the Midland Basin

Enhancing Production Efficiency: The Impact of Precision Targeting in the Midland Basin

Technical Paper Details::
Technical Presentation: Monday, June 17th, 2024
Theme 3: What's in Your Rock: Geological Integration that Impacted Go-Forward Business Decisions
Topic: Enhancing Production Efficiency: The Impact of Precision Targeting in the Midland Basin

Authors: S. Christian*1, T. Cross1, J. Garzon1, 1. Novi Labs.

Abstract:

Over a decade into unconventional developments, the Midland Basin has seen a number of technological advancements reshape the economics of oil extraction, including the adoption of precision targeting techniques for optimizing production efficiency. With over 20,000 horizontal wells in the basin, machine learning has a wealth of examples to analyze, providing a data-driven complement to traditional geological methods for selecting landing points. This study seeks to leverage machine learning models to predict three-year cumulative oil volumes using high resolution target information alongside geology, completion and spacing parameters as model inputs. 

Using a basin-wide causal inference model, we predicted oil volumes at 30-day increments for the first three years of a well’s production. For training features, we employed completion parameters, geological variables, and spacing parameters. Most importantly, we included a “position in zone” variable which represents the position of the well’s midpoint with respect to the top of its formation. We took an input data set of approximately 18,000 wells with sufficient data from across the Midland Basin. We tested the capability of the “position in zone” feature to capture geologic variability by running the model with the Lower Spraberry bench divided into its subzones and without.  We also analyzed the impact that “position in zone” had in the Wolfcamp A and Wolfcamp B formations. Model results were interpreted using SHAP Values (SHapley Additive exPlanations), a powerful tool for machine learning model explanations.

It is demonstrated that the feature “position in zone” is able to capture geologic variability. When the model was run without Lower Spraberry subzones, the “position in zone” feature rose in the rank of the model’s most predictive features and the impact that the feature had on predicted 3-year cumulative oil was delineated by the varying geology in the Lower Spraberry bench. When the model was run with subzones, we observed that wells targeting the top half of the Jo Mill had a positive impact on production while wells targeting the bottom half of the Jo Mill had a negative impact. This trend did not exist in the Lower Spraberry Shale and Lower Spraberry Sand. In the Wolfcamp A, wells targeting the top half of the zone correlated with an uplift to production of up to 5% while wells targeting the bottom half negatively impacted production by up to 10%. In the Wolfcamp B, we observed a slight negative impact to production for wells targeting the lower half of the zone. 

Precision targeting not only facilitates the identification of high-potential drilling locations but also enhances well productivity. Inflation, as a determinant factor influencing capital investments, operational expenses, and overall project viability, necessitates a meticulous assessment of ways to tip the scale towards favorable returns. With clear trends exhibited in each formation and little to no additional cost involved, operators would benefit from optimizing landing position in their future development.

Please fill out the form below, and we’ll email you the paper as soon as it becomes available.

First name(Required)
This field is for validation purposes and should be left unchanged.

Meet Novi Labs' knowledgeable experts.

Invite to Speak

Our team of experts is passionate about sharing valuable insights on US energy. Please provide us some information about your event, and we will get back to you as soon as we can

Date
This field is for validation purposes and should be left unchanged.