Completion Design Optimization for Unconventional Wells Using Large Scale Computational Science

Abstract

In the current state of the oil and gas industry, unconventional resources are a significant source of the total production output. Unconventional wells remain profitable at various price points, because initial stimulation treatments can be tailored to changing market conditions, reflecting completion costs and (estimated) hydrocarbon prices. The same holds true for re-stimulation of already producing wells. Stimulation treatment "opens" up the subsurface to ultimately allow for better drainage of the reservoir hydrocarbons. The primary stimulation treatment currently in use is hydraulic fracturing, in which the wellbore is broken up into multiple stages, and highly pressurized fluid (oftentimes water) is pumped into each stage of the wellbore. This causes fractures to propagate away from the wellbore, which in turn enhances the local reservoir permeability and allows for economical production. Historically, the number of stages, and clusters per stage, for hydraulic stimulation has been based on wellbore horizontal length (e.g., 200 ft or 400 ft), or much valued previous experience in the same or similar area, as well as other investment considerations. Over time, a strong tendency has developed to place stages and clusters closer together to improve production. However, it is reasonable to assume that there will be a point beyond which adding another stage becomes more expensive than what is gained by increased production revenue from the greater stage count (i.e., less profitable depending on the time of investment). This scenario frames a classic optimization problem which is solved using Monte Carlo methods. Results show that optimal stimulation treatment configurations are robust for many objective functions related to the fracturing process (e.g., propped length and propped height). However, we find that objective functions related to production, production revenue, and profit often provide different optimum treatment configurations, and that those optima shift with respect to the considered timeframe. Because business decisions will ultimately be based on profit decisions over a given time span, we propose utilizing the appropriate objective function together with an integrated modeling approach such as presented here.