Authors: Oscar Molina*, Camilo Mejia, Jerry Webb, and Rebecca Nye, Enovate Upstream



This work presents two cases for the application of artificial intelligence (AI) techniques as complementary tools to classic physics-based modeling. On the one hand, we discuss the benefits from the implementation of an AI-assisted non-linear solver, especially designed for analytical simulation of multi-fractured horizontal wells, that led to obtaining fast simulation results, this way opening an avenue for the use of reservoir simulations for type-well analysis in unconventional reservoirs. On the other hand, we introduce the application of AI for drilling, more specifically, optimization of the rate of penetration (ROP). In both cases, we used data from offset wells to better understand the characteristics of the target formation, pinpoint opportunities for ROP optimization, and identify key production drivers to determine their impact on short and long-term production. In this manner, we are able to generate physically meaningful production forecasts. In this paper, we examine the results from the application of the proposed AI methods to a field case study for a type-well study in the Lower Eagle Ford formation. Analysis of results show the importance of accounting for static reservoir properties and completion properties to generate accurate production forecasts. Similarly, we observe that the introduction of artificial neural networks (ANN) as a means for ROP optimization, using offset wells data, allowed to generate optimized drilling schedules to be used for drilling infill wells in the area of interest.