Kleinberg, R.L., Schlumberger-Doll Research Chow, E.Y., Schlumberger-Doll Research Plona, T.J., Schlumberger-Doll Plona, T.J., Schlumberger-Doll Research Orton, M., Schlumberger-Doll Research Canady, W.J., Schlumberger-Doll Research Summary Currently deployed devices for detecting fractures in borehole environments often fail to detect fractures, give an indication of a fracture when none is present, or both. This work describes a systematic laboratory and signal-processing study of the effects of unfavorable formation and borehole conditions on the effectiveness of two circumferential acoustic fracture detection techniques. The conditions investigated include lithological variation, surface roughness, shallow cracks, transducer standoff, blind spots, nonradial fractures, variable fracture closure, and variations in acoustic path length. Introduction Various techniques have been used to search for fractures intersecting boreholes. These include analysis of amplitudes in the sonic log, the fracture identification log (FIL), which relies on electrical measurements, and a borehole televiewer. Devices using sound waves propagating circumferentially around the borehole include the circumferential microsonic tool (CMS/CMT) and the circumferential acoustic device (CAD). A number of studies have indicated that a very careful analysis of some of these logs can be useful in locating fractures. However, others have found currently available tools to be lacking in reliability. In this paper we present the results of an investigation of the sensitivity and reliability of two fracture detection methods based on measurements of acoustic waves propagating circumferentially in a borehole. A series of laboratory experiments that simulated measurements in various borehole environments was conducted. The analysis of the measurements provided insights into the performance of these fracture detection provided insights into the performance of these fracture detection techniques. Under Experiments we describe the general experimental conditions under which measurements were made. Then the acoustic fracture-detection techniques under study are described. We then discuss sensitivity of the fracture detection methods to effects such as depth of fracture, lithology changes, and surface roughness. Experiments The majority of measurements were made in an Indiana limestone formation of 16% porosity. The geometry of the block is shown in Fig. 1. The formation has a smooth borehole 20 cm [8 in.] in diameter and can be separated into two pieces by means of the radial sawcuts that intersect the center of the borehole along a diameter. These radial sawcuts act as smooth fractures. The two nonradial sawcuts extend only halfway through the block. A radially intersecting, "natural" vertical fracture is also shown in Fig. 1. The entire formation was submerged in a tank of water. Measurements were made with the blocks connected to form a complete borehole. The bulk slowness in the limestone measured 216 mu sec/m [66 mu sec/ft] for the compressional wave, 383 mu sec/m [117 mu sec/ft] for the shearwave, and 410 mu sec/m [125 mu sec/ft] for the planar pseudo-Rayleigh wave. The density of the saturated limestone was 2.5 g/cm3 pseudo-Rayleigh wave. The density of the saturated limestone was 2.5 g/cm3 [156 lbm/cu ft], and the fluid speed measured 666 mu sec/m [203 mu sec/ft]. The transducers used in this study were commercially available ultrasonic hydrophones with radially symmetric radiation patterns. Direct sonic paths between transmitter and receiver were blocked by fitting the transducers with reflective baffles. The typical measurement involved moving a source and receiver pair around the borehole while keeping the angle between them constant. The waveforms transmitted from the source to the receiver were recorded for analysis. The transducers could be adjusted in height such that in a complete rotation around the borehole, they would cross either three or five vertical fractures. The three fractures included the two sawcuts and the "natural" fracture. The extra two fractures were the nonradial sawcuts. Center frequencies varied between 50 kHz and 130 kHz [50,000 and 130,000 cycles/sec]. For ease of interpretation the graphs of fracture response vs. angle are coded (with wavy lines) to indicate sectors in which fracture detection would be anticipated. The marked sectors differ from graph to graph because of variations in the transducer spacing, the path along which the transducers were moved, and the waveform analysis, either single (energy) or double (semblance). JPT p. 657
Read full abstract