Chemical profiles of produced water submitted to the Pennsylvania Department of Environmental Protection (PA DEP) in residual waste reports often remain unused after submission. However, these data are valuable and can potentially be used to recognize or predict changes in reservoir properties, cross-formational flow, mineral scale formation, and permeability. In this study, we use computer science (data mining and analytics) techniques to extract data from produced water chemical assay reports submitted to the PA DEP. Geochemical modeling techniques are applied to create a high resolution (well-pad scale) spatio-temporal snapshot of changes in mineral saturation indices of produced water within the northeast and southwest production regions of the Marcellus Shale in Pennsylvania over a 5-year timespan (2012-2017). This approach combines multidisciplinary faculties in computer science and geoscience to generate valuable insight for stakeholders from publicly available, though disparate data sources.


The Marcellus Formation is a prolific natural gas field that lies underneath significant portions of Pennsylvania, West Virginia, New York, and Ohio. In Pennsylvania, natural gas is produced from the shale members of the Marcellus in a number of counties. However, the majority of production (by volume) occurs in northeast (NE) and southwest (SW) sections of the state. The formation is a black, organic-rich shale with a varying carbonate composition (de Witt et al., 1993; Werne et al. 2002; Lash, 2008). Organic material and carbonate content, along with other geologic characteristics, vary spatially between NE and SW regions of Pennsylvania (Carter et al., 2013). These regional geologic characteristics as well as differing post-depositional histories result in varying produced water characteristics (Barbot et al., 2013).

Heterogeneity in production water composition could exert a spatial control on production challenges experienced by operators. Specifically, regional variations in analyte concentrations could increase the frequency of mineral scale occurrence at certain well pad locations. Additionally, temporal variations in produced water chemistry could suggest different water-rock reactions occurring in the subsurface. These can include reactions related to well completion or stimulation techniques. For example, temporal variations in produced water composition could be the result of a fracture extending outside of its intended target zone. Another potential issue is cross-formational flow, which can happen after periods of extended production lower pressure in the targeted reservoir. Dilution effects observed in the produced water of a nearby pre-existing well could also suggest a parent-child relationship, where newer wells negatively impact production in existing wells. Likewise, production water reuse can pose issues as well when these fluids are combined with other incompatible fluids, such as produced water from another region or oxygenated surface waters. These phenomena combined with pressure changes within the reservoir are often related to mineral scale occurrence that can impede production from a well.

This content is only available via PDF.
You can access this article if you purchase or spend a download.