[URTeC 2020] Evaluating the Impact of Precision Targeting on Production in the Midland Basin using Machine Learning Algorithms (ID 3062)

With the right dataset and right machine learning model, you can get high-resolution answers on fine-tuning your target selection — potentially increasing production by over 10% at zero cost.

We studied the impact of target in the Midland Basin, and found that the sweet spots vary by county and by zone. GOR and WOR can also see dramatic changes (and can provide insights as to what’s driving the production variance). This approach provides a useful complement to traditional, petrophysics-based methods.

talk details::

Wednesday, 11:30 AM. Theme 3: Reservoir Monitoring and Well Spacing I

Shapley values for position in zone across the main three targeted formations in the Midland. Highest “target impact” is seen in the upper Wolfcamp A.

Abstract::

One of the least expensive ways to improve the production of a horizontal well is to target the best rock. Though operators have developed robust petrophysical methods for selecting targets, regional-scale evaluations of the impact of that target selection could be improved. This type of information could expand capabilities to objectively identify analogs that were not previously obvious. This larger analog list could then be used to improve production, benchmark offset operators, and customize spacing/stacking configurations. Here, we quantify the impact of target on production from the Midland Basin. We apply a decision trees-based machine learning algorithm on a regional Midland dataset composed of production, geologic grids, completions header, and detailed geologic targets, totaling over 7,000 wells. The target data contains the location of the well within the zone, reflected as a fraction from 0 (top of zone) to 1 (bottom of zone). We leverage SHAP values (Shapley Additive exPlanations) to interpret the impact of target on expected production. The impact of position in zone varies from 13% for the Wolfcamp A to 5% for the Wolfcamp B and 7% for the Lower Spraberry Shale. This represents approximately 20,000 barrels two year cumulative oil for Wolfcamp A wells, for an net present value uplift potentially over $100,000/well. The best target within the Wolfcamp A and Wolfcamp B are usually near the top of the zone, though some location and target combinations, such as Irion County Wolfcamp B, show best production near the middle of the zone. We also show that with some targets, increased oil production is associated with higher water production, and in others, with lower water production. Selecting the best target and geosteering to keep in that zone are among the most cost- effective means to increase production, though changes to landing zone must also take into consideration the impact of spacing degradation and sharing of production from other zones. For instance, a Wolfcamp B well may see highest production if landed near the top of the zone, though that may increase spacing interference with Wolfcamp A neighbors. This study provides guidelines to inform landing, geosteering, and spacing/stacking planning decisions.

Enter your name and email below. We will email the paper to you right away.

First name(Required)
Consent