Abstract

The optimization of unconventional oil and gas assets requires simultaneous consideration of fracturing and porous media flow phenomena. Many recent works have demonstrated that a fully coupled fracturing and flow modeling workflow can be used for a variety of applications (McClure et al., 2023; Fowler et al.,2023, Pearson et al., 2022). By coupling fracturing and production physics, solutions implicitly reconcilethe impact of fracturing designs on well productivity and resource recovery.

In order to optimize fracturing design for a pad of upcoming wells in their Bakken development, LibertyResources constructed a coupled model which was calibrated and used in an optimization workflow and documented in URTeC paper 4043244 (Pearson et al, 2024). The presented dataset was calibrated against a five well development program in a 2×1 mile spacing unit – one parent well followed by 4 infill wells drilled after 3-/1/2 years of production.

Single-variate analysis of a non-optimized dataset was shown to be significantly inferior to a multi-variate optimization process to develop a “cloud” or “onion” of datapoints allowing the operator to select an economic optimum that either focused on Discounted Profitability Index (return of capital) or PresentValue created.

The scope of this paper is to take the multivariate datasets presented in URTeC 4043244 and to dissect them for the individual single variate components. A genetic-algorithm multi-variate optimization procedure was used on the history matched dataset. The first generation used a space-filling design, sampling all regions of the design parameter space (where each completion/stimulation design variable is a parameter).Subsequent generations goal-seek the optimal objective value based on the results of the prior generation.The overall result of the process is a cloud or “onion” of optimization points as a function of the economic parameter being considered. This paper presents the optimized results as sliced datasets which immediately allows the completions engineer to visualize the dependence of the multivariate optimization on the individual components in the completion design.