Production Analysis of Marcellus Shale Well Populations Using Public Data: What Can We Learn from The Past?

Abstract

The scope of this paper is to use publicly sourced data as a starting point for production analysis of two distinct well populations. The well populations that will be examined are horizontal Marcellus Shale wells (1) frac'd earlier in time with a defined plug-to-plug stage spacing and (2) frac'd at the same time or later in time with 1.5 to 3 times the number of stages, a production acceleration technique known as Reduced Cluster Spacing (RCS). Public production databases were used to create snap shots of completion activity over time. To speed work flow, a more robust and regularly updated commercially available database that contains both completion and production data was used. Two distinct populations of wells for analysis were identified. A conservative decline curve analysis (DCA) was made to forecast estimated ultimate recovery (EUR) for all wells using only public data. For selected wells, reciprocal production rate versus square root of time plots were generated. This paper will attempt to answer the following questions: (1) were statistically meaningful populations of both stimulation techniques found to exist; (2) how much production uplift did RCS achieve both early on and later time; (3) was there a real increase in EUR; (4) was there a difference in stimulation efficiency; and, (5) what can we learn by plotting P10, P50, and P90 values and EUR distribution curves for specific well populations. To the authors' knowledge this would be the first SPE paper to explore this topic over such large Marcellus Shale well populations in Pennsylvania.

Application of Machine Learning Method of Data-Driven Deep Learning Model to Predict Well Production Rate in the Shale Gas Reservoirs

“…Since unconventional reservoirs have nano-Darcy permeability, horizontal drilling with multi-stage hydraulic fracturing is used to expand the hydrocarbon drainage to maximize production [1]. Reservoir modeling for productivity prediction in such systems is an extremely complicated task, owing to the need to simulate fluid flow in a network of induced natural fractures coupled to geomechanical effects and other processes such as non-Darcy flow, adsorption-desorption and water blocking [2][3][4]. Various numerical simulation methods that simulate multi-stage hydraulic fracturing behavior in horizontal wells are continuously being studied [5][6][7][8][9].…”