[URTeC 2024] Enhancing Production Forecasting Accuracy: A Machine Learning Approach in the Permian Basin

Enhancing Production Forecasting Accuracy: A Machine Learning Approach in the Permian Basin

Technical Paper Details::
Technical Presentation: Tuesday, June 18th, 2024
Theme 7: Production Forecasting II
Topic: Enhancing Production Forecasting Accuracy: A Machine Learning Approach in the Permian Basin

Authors: K. Sathaye*1, S. Iceton1, T. Cross1. 1. Novi Labs.

Abstract:

Half of lower-48 production comes from wells under 18 months of age. These wells are not effectively modeled by traditional decline curves due to sparseness and noise in the data, especially when considering monthly volumes from state agencies with different reporting requirements. So, despite the importance of these wells for NPV calculations or supply projections, we do not have an accurate method of forecasting these wells with traditional decline curves. In this study, we compare a machine learning method to traditional curve fitting to determine the potential for accuracy improvements on forecasting young horizontal wells. This machine learning model is an autoregressive tree-based model, trained on blend of proprietary production data and public reporting sources such as FracFocus, Texas Railroad Commission, and the New Mexico EMNRD. Utilizing this approach, the forecast for young wells can be informed by the decline curves of surrounding similar wells, in addition to the well’s own short production history. In aggregate we find that the machine learning model reported Mean Absolute Percent Error of 8.5% and 9.0% for the Midland and Delaware Basins, respectively. Traditional Arps decline curves showed errors of 12.4% and 12.3% for those same basins for a 6-months ahead forecast. These types of machine learning methods, which can combine production history with external well data, can provide a basis for better decision making in asset evaluation and resource allocation for future shale development and field management. 

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.