Coring adjacent to a hydraulically fractured horizontal well in Eagle Ford shale by Conoco-Phillips has revealed several closely spaced parallel hydraulic fractures (separated only by few inches) propagating in the direction perpendicular to the wellbore axis. The number of observed hydraulic fractures greatly exceed the number of clusters according to the recent paper titled "Sampling a Stimulated Rock Volume: An Eagle Ford Example". The observed behavior is contrary to the conventional practice of hydraulic fracture modeling where often a single fracture from each perforation cluster. This assumption stems from a simplified concept of the stress shadow that inhibit the growth of multiple parallel fractures under very tight spacing. In this study we show that correct modeling can in fact capture the field observed fracture clusters or swarms of closely-spaced fractures.
Numerical model based on displacement discontinuity method is used to simulate non-planar hydraulic fracture propagation. Fracture deformation, fluid flow and perforation friction are fully coupled. Fracture propagation from a single cluster consisting of 20 perforations under 1800 phasing spanning 5 ft is considered. The effect of controlling parameters such as far-field stress contrast, perforation properties, and fracture toughness on multiple hydraulic fracture growth from a cluster of perforations is studied.
The results show that closely spaced fracture clusters or swarms can occur for a certain range of conditions and operational parameters. The in-situ stress contrast, perforations conditions, and injection rates exert a significant influence. Under the right conditions, closely-spaced fractures can extend to distances exceeding tens of feet from the wellbore. Early termination and/or coalescence of closely spaced fractures can also occur.
To our knowledge, our modeling results are the only ones that can explain the data from the Conoco-Phillips field observations regarding the occurrence of fracture swarms. The resuts show that the assumption of a single fracture per cluster does not hold true under all conditions. Moreover, such assumption would significantly underestimate stimulated rock volume near the wellbore. Finally, our results capture the injection pressure data which can be used as a diagnostic tool to infer the perforation effectiveness (i.e., the number of perforations that are in contact with fluid flow).