Examination of a Cored Hydraulic Fracture in a Deep Gas Well (includes associated papers 26302 and 26946 )

Abstract

Abstract A hydraulic fracture stimulation conducted during 1983–1984 in non-marine, deltaic, Mesaverde strataat a depth of 7100 ft (2164 m) was cored in a deviated well in 1990. The observed fracture consists of two fracture intervals, both containing multiple fracture strands (30 and 8, respectively).while the core had separated across many of the fracture strands during coring, the rock remained intact across 20 of the strands, preserving materials within the fractures. Nine of the remaining intact strands were split open, revealing abundant gel residue on the surfaces of every fracture examined. Of 7 strands associated with major bedding planes, 4 displayed offsets of 1–3 mmat the planes and 3 strands had their growth terminated at the planes, showing the importance of bedding (petrophysical heterogeneities) on fracture propagation. Implications of all these findings for propagation. Implications of all these findings for hydraulic fracture design and analysis are also addressed. Introduction One of the principal hindrances to an accurate description of the hydraulic fracturing process has been the inaccessibility of the created fracture. As a result, the conventional view of a hydraulic fracture has been driven primarily by the idealized version developed by modelers to predict fracture geometry and behavior. Only a few mineback experiments, occasional TV logs, lab tests, and inferences from diagnostics have provided amore realistic view of the hydraulic fracturing process. process. Current fracture models assume that the fracture is a single plane with fluid frictional effects proportional to the fluid resistance in smooth proportional to the fluid resistance in smooth parallel plates or ellipses. Variable-width parallel plates or ellipses. Variable-width cross sections are typically handled by integrating one of the simple fluid resistance laws across the cross-sectional area. This procedure yields small pressure drops throughout the majority of the crack, particularly when large tip pressure drops are introduced. The effects of more complex fracture behavior, such as multiple strands, offsets, and waviness are seldom considered in the modeling process, although they have been used to interpret field results. One reason for ignoring complex fracturing is the difficulty in determining the size of the effect; a second reason is the lack of field data (other than shallow minebacks) to support such behavior. This simplistic, model-driven paradigm for hydraulic fracturing has biased our view of many other aspects of the process as well. Height growth, for example is treated as an elastic process, involving equilibrium models, time-constant approximations, or fully 3-D finite element models. Little attention has been given to the inefficiencies of fracture growth across bedding and non-elastic behavior, factors which may be as important as the easily analyzable elastic components. Fluid rheology, leakoff and damage are generally considered to be known quantities, based on laboratory data taken at conditions that seldomresemble the in situ state. Of particular significance is the neglect given to gel residues, which can cause conductivity damage to the hydraulic fracture and to the natural fractures which are intercepted. P. 597