A New Method for Fracture Identification Using Array Sonic Tools

Abstract

Summary. The development of array sonic tools has enabled the recording of an array of sonic waveforms at each depth location with several receivers. In this paper, we develop a new method to study the sonic amplitude/energy by taking advantage of the redundant information in these waveforms. First, we identify the wave components of interest. Second, we apply windowing procedures to extract the desired wave components. Third, we calculate the energy of the wave component at each receiver depth location. Finally, the calculated energies are displayed as an energy log or a variable-density image as a function of depth. Open fractures are detected when the method is applied to the low-frequency Stoneley wave. Reductions of the Stoneley wave energy caused by open fractures appear as a characteristic pattern on the output. The vertical extent of the pattern provides an estimate of the fracture dip angle, and the magnitude of the energy reduction is related to the hydraulic conductivity of the fractures. We show examples of fracture identification using data from the Conoco Test Well 33–1 located in Oklahoma. Complementary information regarding the location and nature of the fractures is available from full-core and borehole televiewer (BHTV) logs. The Stoneley wave response to different fractures is discussed. Combined with borehole images, this method provides a valuable way for distinguishing between open and closed provides a valuable way for distinguishing between open and closed fractures. Introduction Modern digital sonic tools are capable of recording down-hole sonic waveforms with an array of receivers. The recorded waveforms contain information about the in-situ rock properties, such as compressional and shear slownesses, and attenuation. In this paper, we propose a new method to estimate wave amplitude/energy from recorded array sonic waveforms and to display the information as a well log. Three important contributions are emphasized. First, we discuss ways to extract the sonic amplitude/energy information from array sonic waveforms. Second, we display the information in a meaningful way, taking full advantage of the data redundancy in array sonic waveforms. Third, and most important, we apply the method to formation evaluation, where we have found that the method has great potential for describing the heterogeneous properties of fractured rocks. This is important because the location and evaluation of fractures is critical for oil and gas production in low-permeability formations. Method The acoustic wavefield recorded in the borehole is extremely complex because of the presence of the overlapping wave components: compressional and shear headwaves, trapped-fluid modes, and surface waves. One way to separate the components in the recorded waveforms is to use a sonic tool with a long transmitter-to-receiver (TR) spacing. In this way, wave components that propagate with different velocities can be isolated with a time window. Besides exhibiting a time separation, the wave components may also have harmonic separation in frequency. The frequency content of compressional and shear waves, for example, is generally higher than that of the Stoneley wave. This allows us to apply low-pass filtering to extract the Stoneley wave from the waveforms. Fig. 1 shows an array of sonic waveforms recorded in hard rock with a digital array sonic tool and shows that the different arrivals can be isolated from one another by application of a time-windowing technique. Fig. 2 shows the amplitude spectra of the time-windowed compressional, shear, and Stoneley waves. It is clear that the low-frequency Stoneley wave (2,000 to 5,000 cycles/sec [2 to 5 kHz]) is the strongest event and is separated from the headwaves (approximately 6,000 to 15,000 cycles/sec [6 to 15 kHz]). The filtered output (2,000 to 4,000 cycles/sec [2 to 4 kHz]) of the waveforms is shown in Fig. 3. Note that the filtered waveforms contain only the low-frequency Stoneley wave and not the headwaves. Wave components of interest may be extracted with time windowing or frequency windowing, or with a combination of the procedures. The choice of the windowing procedure and its implementation will depend on the procedure and its implementation will depend on the application. For example, one might be interested in the compressional headwave in the 10,000- to 15.000- cycle/sec [10- to 15-kHz] band. JPT P. 677