Laboratory Experiments in Fracture Propagation

This paper describes fracturing experiments in dry blocks of various rock materials. The results have application to evaluation of hydraulic fracturing theories. The block dimensions were 3 in.×4 in.×12 in. [7.6 cm×10.2 cm×30.5 cm] with metal plates epoxied to the 3-in.×12-in. [7.6-cm×30.5-cm] faces. Remaining faces were coated with soft epoxy to provide an impermeable jacket. The blocks were loaded in a pressure cell with an upper movable piston bearing on the 3-in.×4-in. [7.6-cm×10.2-cm] faces. A servo-controlled press applied constant stress to these faces higher than a lateral confining stress applied by oil pressure. Fractures were initiated by injection of various fluids into a small notch located on a center plane parallel to the 4-in.×12-in. [10.2-cm×30.5-cm] faces. Fracture growth along the same plane was assured by the stress conditions. Use of these experiments to test theories of fracture propagation required measurement of three variables, fracture width bi, and propagation pressure pi at the notch entrance, and fracture length, L. bi was determined by a capacitance method, and pi was measured directly by a pressure transducer. L was measured by two methods - either ultrasonic signals or pressure pulses generated in miniature cavities. The ultrasonic method confirmed the existence of a Barenblatt liquid-free crack ahead of the liquid front whose relative length decreased with confining stress. The metal plates bonded to the 3-in.×4-in. [7.6-cm×10.2-cm] faces prevented slip at the top and bottom of the fracture, giving a three-dimensional (3D) crack of constant height. However, the bi, pi, and L data followed trends predicted by two-dimensional (2D) (plane strain) elastic theory reasonably well. Fracture closure measurements after shut-in showed an initial period of leakoff-controlled closure and a final period of creep-controlled closure. A pi slope change at the transition is identified with the instantaneous shut-in pressure (ISIP) in field records and is higher than the true confining stress. Introduction Methods of predicting crack dimensions during fracturing operations are essential to proper design of field treatments. Many fracture-propagation theories have been advanced. Contributions have been made by Barenblatt,1 Khristianovitch and Zheltov,4,5 Howard and Fast,6 Perkins and Kern,7 LeTirant and Dupuy,8 Nordgren,9 Geertsma and de Klerk,10 Daneshy,11 and Cleary12,13 among others. However, practical methods of evaluating the theoretical work have been few. Mostly they have been. limited to indirect and generally inconclusive field evaluations. The Sandia mineback experiments14–16 have provided more direct evaluations. However, even here important fracturing parameters are uncontrolled or unknown. This paper describes laboratory-scale hydraulic fracturing experiments that provide critical data for evaluating crack propagation theories. In these experiments we measured the fundamental variables of crack growth under controlled conditions with known fracturing parameters. Experimental Methods All fracturing experiments were carried out in dry blocks 3 in.×4 in.×12 in. [7.6 cm×10.2 cm×30.5 cm] in size. Mesa Verde sandstone and Carthage and Lueders limestone were used as sample materials. Scaling considerations were important. It was necessary to scale down injection rate and leakoff to be consistent with fracture dimensions. The scaling factor of importance was taken to be fluid efficiency, the ratio of crack volume to injected volume. This factor was controlled through appropriate combinations of sample permeability and fracturing fluid viscosity. As fracturing fluids we used thick grease, hydraulic oils of various viscosities, and gelled kerosene (Dowell's YFGO™). Fluid efficiencies ranged from 3 to 70%. Most experiments were conducted at efficiencies between 30 and 50 %, a range typical of most field treatments. Fig. 1 shows the experimental arrangement. Shaped aluminum plates were bonded with Hysol clear epoxy to the 3-in.×12-in. [7.6-cm×30.5-cm] faces of the sample block as shown. The remaining faces were coated with a thin layer of the same epoxy to provide an impermeable jacket for confining pressure. One of the aluminum plates contained an injection port communicating with a 1.4-in. [0.64-cm] borehole as illustrated. A pair of brass plates with faces 0.2 in.×0.5 in. [0.5 cm×1.3 cm] was epoxied into the borehole at its center. These plates, separated by a gap of 0.01 in. [0.025 cm] served as a parallel plate capacitor. They were connected to a capacitance bridge that detected changes in gap width through changes in capacitance. This provided a direct, continuous measurement of fracture width at the borehole.