Using Production Data to Constrain Resource Volumes and Recovery Efficiency in the Marcellus Play of West Virginia

Abstract

Abstract Recovery efficiency (RE) is a poorly constrained parameter with respect to unconventional resource development. However, the increasing volume and quality of data related to Marcellus development in West Virginia provide an opportunity to assess this parameter. In this report, we develop a map of estimated ultimate recovery (EUR) per square mile (mi2) for the Marcellus Play in West Virginia and compare that map to published estimates of both technically recoverable resources (TRR) and original gas-in-place resources (GIP). These comparisons reveal that Marcellus wells appear capable of producing significantly more gas than previously estimated and, throughout large areas of the State, more gas than has been assessed to be in place. While explanations for this observation likely include both potential overestimation of EUR and underestimation of GIP, our review indicates that a primary source of the discrepancy is overly-conservative GIP calculations. New GIP calculations have been conducted for a range of wells that appear to account for the bulk of the "excess gas" by including GIP volumes from bounding formations determined likely to be part of the larger Marcellus Reservoir Unit (MRU). In northeastern West Virginia, the MRU includes both the Marcellus Formation and the lower 300 ft of the Mahantango Formation. In northwestern West Virginia, units up to the Cashaqua Shale Member of the Sonyea Formation are included in the MRU. These new GIP/mi2 values, when compared with our estimates of TRR/mi2, indicate that ultimate RE for ongoing development generally ranges between 20 to 60%. Our approach to estimate TRR/mi2 is based on identification of 166 Marcellus "development sites" in northern West Virginia. Each development site was selected to consist of multiple horizontal wells that were drilled as a coordinated activity at a common spacing over a short period of time by a single operator. Sites and wells also were selected to maximize both the relevance (post 2012 vintage wells where possible) and reliability (more than 18 months production history) of the EUR forecasts. Site area was determined as the product of cumulative lateral length and well spacing. The cumulative EURs for all wells in each site were estimated (using two independent sources for EUR estimation) and those sums were divided by the area to obtain a single point estimate of TRR/mi2. Our review indicates that Marcellus Play resource assessments tend to be overly conservative. This analysis shows the value of focusing on the most relevant production data and using those data as a ground truth check on resource assessments.