Evaluation of multi-stage hydraulic fracturing techniques to optimize production in naturally fractured reservoirs using a DFN-based numerical technique

Abstract

Over the past decade, significant improvements have been made in hydraulic fracturing design and strategy to incorporate the unique characteristics and complexities in naturally fractured shale reservoirs. Several hydraulic fracturing techniques including zipper fracturing and alternating fracturing have been developed to mitigate the impacts of in situ stress alteration compromising the desired fracture network due to the propagation of fractures in unfavorable orientations. The traditional approaches used to evaluate the fracturing techniques were typically based on flow simulation using a synthetic fracture network with a geometry pre-determined at a fixed location. Considering fracture propagation parameters such as the interactions between pre- existing fracture and the approaching hydraulic fracture, the impacts of rock properties on fracture growth, the stress shadow due to in situ stress alteration, and the fluid and proppant transport can enhance accurate and realistic representation of the simulations for production forecasting in high formation complexity shale reservoirs. In this research study, a new approach using an integrated complex discrete fracture network (DFN) and fluid flow model has been proposed to accurately evaluate hydraulic fracturing techniques using multiple-horizontal-well layout in naturally fractured reservoir by including the four aforementioned significant factors. A 3D DFN model has been developed for a particular reservoir using the properties and natural fracture characterization specific to the reservoir. Production forecast as a result of the fluid flow simulation has been handled utilizing unstructured grid blocks exclusively created for the specific DFN. A case study elaborating these processes has been conducted based on the data from the Eagle Ford shale. The complex DFN model, the integrated DFN and fluid flow model, and the input dataset have been successfully validated against microseismic fracture mapping and commingled production data obtained from a well pad located in the Eagle Ford oil window. By utilizing this integrated model, the proposed approach has been used to evaluate the hydraulic fracturing patterns in various aspects in this research study. According to the production results simulated based on the approach and strategy described in the research study, the well pad with an alternating fracturing technique produces 5.7% more oil than a pad with zipper fracturing during a 20-year production period. Both pads experience a significant drop in production rate within the first year. The initial oil rate in the alternating fracturing case is greater than the zipper fracturing for about 40 months after which zipper fracturing stays higher. Under the constant FBHP condition, the production rate of each well in the same pad is not equal to the others at any given time and zipper fracturing case presents larger difference over the entire production period. The ratio of the production contributed by each well to the commingled production changes with time in an unpredictable pattern during the initial production. However, the trend becomes predictable after the initial production period and it is significantly influenced by the well placement, the formation heterogeneity, and the complex fracture network geometry and properties, considered for a well in comparative relation to the others. Impacts of these factors are evidenced by the variation in reservoir pressure, distributing over the reservoir area at a time. As a result of the time-variant “ratio of the production contribution” introduced in this study, production optimization in hydraulic fracturing techniques using multiple-horizontal-well system has become challenging. Assuming the constant “ratio of production contribution” over the entire production period or determining the production of each well at a time based on normalized rate from the commingled production could lead to the errors in production forecasting. While the comparative relation of well spacing and the formation characteristics to the other wells can be properly addressed in traditional model with synthetic (pre-defined) fracture geometry, the regional fracture network geometry and its behavior handled by the realistic representation capability developed with the proposed approach introduced in this study help to provide a more accurate solution.