<p>Acoustic transmission data obtained in laboratory experiment were used to estimate main stages of hydraulic fracture onset, growth and filling by fracturing fluid. Laboratory setup consists of two horizontal disks with a diameter of 750 mm, and a sidewall with an internal diameter of 430 mm. The disks and the sidewall form a pressure chamber with a diameter of 430 mm at a height of 70 mm. There are a number of holes in the disks and the sidewall that are used for mounting ultrasonic transducers, pressure sensors, as well as for fluid injections. As a model material, a mixture of gypsum with cement was used, which was poured into the chamber. The sample was saturated with water gypsum solution and loaded with vertical and two horizontal stresses using special chambers. The fracture was created by viscous fluid (mineral oil with viscosity 0.1 Pa*s) injection with a constant rate 0.2 cm<sup>3</sup>/s through a cased borehole (diameter 12 mm) with a horizontal slot, which was preliminary located in the center of the sample. Hydraulic fracturing monitoring was carried out by recording of ultrasonic pulses passing through the sample during fracturing. To separate the ultrasonic pulses, the frequency of their sending was used. After that, the envelope of each record fragment was constructed using the Hilbert transformation and its maximum was found. Comparison of the ultrasonic pulse amplitude variations and injection pressure led to the following observations. Initial decrease in the pulse amplitudes began before the maximum pressure was reached, which may indicate the hydraulic fracturing onset at a pressure less than the maximum. The amplitude decline occurs smoothly, so it is difficult to identify any characteristic point on these curves and, accordingly, it is difficult to establish an accurate time of the fracturing onset and the fracture rate. The fracture rate was estimated by different methods previously as &#8776;130 mm/s. After the decline, the pulse amplitudes started to increase, that was related with the injection fluid front propagation in the fracture. In contrast to the decline, the beginning of the amplitude growth was clearly detected. Taking into account the spatial locations of the ultrasonic pulse source, receivers, and fracture, it is possible to estimate the propagation velocity of the fracturing fluid front as &#8776;35 mm/s. After the increase, the ultrasonic pulse amplitudes started to decrease significantly (up to 3 times), which is probably due to the further expansion of the fracture aperture. On the transducers located closer to the well, this decline is maximum. When the injection is stopped, the ultrasonic pulse amplitudes began to grow again, which indicates the fracture closure as the injection pressure decrease. In the experiments on the fracture re-opening under various stress applied to the sample, a linear relationship between the fracture re-opening pressure and applied vertical stress was found. This type of relationship should be expected, but values of the relation parameters declined from the values suggested in theoretical research, which was explained by taken into account back-stresses and non-linear behavior of the sample material.</p>
Research on support vector machine method for comprehensive evaluation after compression
