Quantifying the Probability of Success of Stimulation Treatments When Information is Limited

Hoeink, Tobias (Baker Hughes, a GE company) | Cotrell, David (Baker Hughes, a GE company) | Odusina, Elijah (Baker Hughes, a GE company) | Ghorpade, Sachin (Baker Hughes, a GE company)

OnePetro 

Abstract

A paradigm shift in dealing with subsurface uncertainty in hydraulic fracturing treatments is introduced. The mathematically rigorous application of uncertainty and sensitivity analyses for a proposed stimulation of a lateral well within an unconventional reservoir in the Marcellus with limited formation data delivers the ability to identify the optimum treatment parameters and to quantify its probability of success. Selection of the optimum reservoir stimulation treatment is achieved by systematically investigating thousands of hydraulic fracture simulations over a large parameter space covering formation properties with inherent uncertainties (e.g., stress gradients, leak-off coefficients) and tunable treatment parameters (e.g. pumping rates, fluid and proppant properties, perforation spacing), and computing an objective function. Operators commonly select objectives based on technical (e.g., propped fracture length, fracture height containment), operational and investment considerations. Here, the average fracture conductivity at closure is selected as the primary technical objective to be maximized. A subsequent uncertainty analysis of the optimum treatment plan that expressly includes the limits of formation property knowledge quantifies the probability of success. Production forecasts of specific cases illustrate the range of possible outcomes. Results from more than 12,000 hydraulic stimulation simulations demonstrate a wide distribution of results in terms of average fracture conductivity. Surprisingly, only a small, isolated fraction (< 5%) of the design space returns clearly superior results compared to the majority of investigated scenarios. The optimum treatment designs in this study are associated with relatively low volumes of a gel treatment pumped at relatively high rates. Production simulations illustrate that the best 10% of cases significantly outperform production over the first two years by approximately 50%. Collectively, the approach presented here illustrates the application of uncertainty and sensitivity analyses on several thousand simulations that cover a large, realistic parameter space. Embracing uncertainty, this approach enables identification of the best treatment plan and quantification of the probability of success given limited formation data. In addition, this methodology offers input for risk assessment and return on investment decisions.