AAPG Annual Convention and Exhibition

Datapages, Inc.Print this page

Modeling the Marcellus: An Integrative Approach to Creating a Facies Model


Exploitation efforts in the Marcellus Formation have generated widely varied well production results over relatively short distances. We hypothesize this is in part controlled by changes in reservoir quality as dictated by lithofacies stacking patterns. This necessitates an enhanced understanding of the lateral and vertical distribution of lithofacies as influenced by the depositional system. Initially to differentiate facies within the Marcellus Shale, 21 unique lithofacies were defined in 27 cores, stacking patterns were quantified and mapped across the basin. In an attempt to fill significant spatial gaps, and improve our quantitative definition of facies for modeling purposes our goal is to develop a core-calibrated petrophysical model from 114 pilot holes in northern PA. The result is an improved dataset that allows us to more precisely predict the lateral continuity and vertical thickness of facies by linking high resolution core descriptions to broader, regional-scale seismic attributes and well performance data. The petrophysical model's output results in 8 discrete facies, with 3 facies (HiFive, HiGR1, HiGR2) particularly prominent in the Union Springs Member of the Marcellus Formation. We suggest maximizing the length of wellbore that intersects HiFive: a high porosity, high total organic carbon (TOC) petrofacies. The HiFive facies has a 4.01%-by-weight mean TOC, and both HiFive and the combined HiGR facies have >3.2x TOC than any other facies. Net-to-gross thickness maps within the Union Springs Member show a first order positive correlation with well performance for HiFive, and no correlation with the combined HiGR facies. Additionally, HiFive thickness correlates well with seismic attribute values picked along a surface. For example, inverted gamma ray data cross-plotted with HiFive facies show a linear correlation R2 = 0.78. In southern Bradford County, our model suggests over 75% of the Union Springs Member comprises HiFive facies, correlating with high well performance in the area. In contrast, Lycoming and Clinton Counties to the southwest have only 30% HiFive facies in the Union Springs Member, and likewise, surrounding well performance drops >69% from Bradford County. Our results demonstrate the need to understand petrophysical variability in terms of facies distributions in unconventional reservoirs rather than interpret reservoir properties from bulk porosity and permeability measurements to adequately describe the reservoir system.