A Machine-Studying-Based mostly Fuel Raise Optimization Workflow for Unconventional Fields

arXiv:2607.25885v1 Announce Kind: cross
Summary: On this paper, we current an automatic data-driven workflow utilizing Machine Studying (ML) for gasoline carry optimization in unconventional fields. This workflow integrates a ML mannequin that precisely forecasts the Fuel Raise Efficiency Curve, and a Bayesian Optimization Framework to resolve for the optimum gasoline injection charges beneath the constraints of facility capability. The ML mannequin leverages the historic manufacturing time collection knowledge with out requiring downhole gauges or multi-rate effectively assessments. We piloted this workflow on 30 wells throughout 5 effectively pads in Bakken and obtained >5% manufacturing uplift on common. With the success of the pilot, we now have now fully-deployed this workflow in Bakken throughout 200+ gasoline carry and plunger-assisted gasoline carry (PAGL) wells. Furthermore, the ML-based gasoline carry optimization workflow introduced on this paper is an efficient and financial answer for different belongings the place downhole knowledge or multi-rate testing will not be out there/possible because of value or facility constraints.



Source link