Completion Optimization Utilizing Multivariate Analysis in the Bakken and Three Forks Formations

Abstract

Abstract Completion optimization has been a key factor for ensuring an economic well in the Williston Basin. In North Dakota, most completion information is public after three to six months unless the wells are confidential. This enables operators to closely monitor offset completion and production performances. The goal was to optimize well economics through analyzing completion parameters. The area of interest for this analysis was confined to the Fort Berthold Indian Reservation and surrounding acreage in North Dakota. Public and internal completion, reservoir and geologic data was gathered, interpreted, and used as inputs for the multivariate statistical analyses. These analyses were designed to understand the impact that both reservoir quality and completion design parameters have on production. The multivariate tool, coupled with quality checked and carefully interpreted data, has aided in understanding how sensitive well performance is to these parameters, as well as how these variables behave. This tool provided an analytic method to determine what parameters and their ranges are most critical to good production. This paper will elaborate on the following high sensitivity completion and reservoir parameters and their behaviors: lateral length, proppant volumes, stage lengths (plug to plug or sleeve to sleeve), proppant type, treatment rate and water cut. The completion parameters were easily controlled by frac design, but additional analysis and review was needed to better understand the causes of high water cut. An understanding of completed reservoir (Middle Bakken or Three Forks) water saturations, job size, and geologic anomalies was critical in understanding and predicting this negative reservoir parameter. This information was combined to predict the performance of various completion designs without having to invest the capital in large field tests. Economic analysis was then done using the cost of the respective completion design and expected performance. The optimized completion design was derived, in part, from the multivariate analyses and is currently being field tested.