Advanced acoustic data analysis

Processing acoustic data downhole as well as at the surface is necessary to transform the raw acoustic signals recorded by modern logging instruments into data suitable for interpretation and analysis. Data processing takes place:

• During acquisition
• In the logging tool itself
• In the surface acquisition unit
• In post-acquisition processing at computing centers

There are a variety of sources of noise in the downhole environment that contaminate the recorded acoustic signal:

• Tool ("road") noise
• Measurement error
• Reflection and scattering from rough borehole or bed boundaries
• Mode conversion
• Interferences that occur in the downhole environment

The goal of acoustic-data processing is to minimize the data noise while maximizing the petrophysical information. Data preprocessing reduces the influences of these sources, thus allowing extraction of the true formation signal.

Following the rapid theoretical advances in acoustic-wave propagation made during the 1980s and 1990s, significant advances in data processing provided improved quality in slowness measurements and enabled a number of new applications using Stoneley and dipole-shear wave in open and cased holes. The combined interpretation of Stoneley and dipole-shear acoustic measurements with NMR and borehole imaging enhances formation evaluation.

Slowness analysis

One of the primary goals in borehole acoustic logging is to obtain formation slowness from array measurements. Accurate slowness analysis is vital to most petrophysical, geophysical, and seismic applications. A variety of techniques are used for computing slowness from array tools (Table 1).

• 

  Table 1

Semblace and Nth-root stacking

The two most commonly used techniques to determine slowness from borehole acoustic array are:

• Semblance
• Nth-root stacking

Both are cross-correlation, coherency techniques that compare signatures in an acoustic array and find the similarities that correspond to coherent wave types. Semblance has a direct physical interpretation, whereas Nth-root stacking is a purely mathematical solution. Although the semblance method is faster, the Nth-root stacking method is more tolerable to noise and ignores amplitude variations across the array and, in general, provides better results. Collecting the array data, either as receiver or transmitter (source) arrays (Fig. 1), enhances the slowness output from both of these techniques.

• 

  Fig. 1 – Diagram showing the gather of array acoustic data into either common-source subarrays or common-receiver subarrays. These subarrays cover the same depth interval. Data grouping is in the vertical direction for the source gather, and in the horizontal direction for the receiver gather (courtesy of SPWLA).

The primary application of the receiver- and transmitter-array-derived slowness curves is to provide compensation (DDBHC) to the monopole compressional and shear slowness for borehole irregularities such as washouts or cave-ins. Borehole compensation is achieved by averaging the slowness estimates from the receiver and transmitter arrays. The dipole-shear wave does not require borehole compensation because the flexural-shear wave is not as sensitive to the borehole geometry although it is often used to improve results in rough boreholes.

The objective of the semblance technique is to find the slowness that maximizes the coherence (time domain) among the wave power spectra over the receiver array. The Nth-root stacking technique is very similar to the semblance technique, except that the waveform amplitudes are modified in a different way to produce a so-called "pseudo" power spectrum instead of the true power spectrum.

Although the algorithms used in semblance and Nth-root stacking differ, the processing procedure is similar. A time window of fixed length is set up on each signature in the array. The windows are offset by a specified time interval on the successive signatures and a correlation is computed over these windows. The windows on all but the first signature are then stepped a certain distance out in time and another correlation value is computed. The process continues for the range of anticipated slowness in the well. This process is repeated as the window is progressively stepped on the first signature.

The group of correlation values obtained is known as a correlogram, in which the waveform coherence or correlation values either from semblance or Nth root method are displayed as a function of slowness and arrival time creating a 3-D surface. The correlation values in a range of arrival times are further combined to form a "combined correlogram," which is a projection of the 3-D surface on the slowness axis. The correlogram peak at each measured-depth level is used to obtain the Δt slowness for the wave type. These individual slowness values are then displayed as a continuous log curve. Filtering for the compressional wave or for the dispersive shear wave is normally required before using either method. As a quality check, the slowness results of semblance or Nth-root stack processing can also be plotted in the combined correlogram together with a "computed travel time" which is plotted against the waveform. The computed travel time represents the sum of the transit-time measurements from the transmitter to the receiver along with the mud travel time. Dipole shear-slowness processing must also take into account dispersion of the flexural wave. Depending on the frequency of the flexural mode, a correction may be required to obtain shear-wave slowness.

Resolution enhancement

The aperture (vertical resolution) of array tools is typically 3.5 ft. This means that a bed must be at least that thickness to measure true acoustic slowness, although such tools can detect (resolve) beds down to 2-ft thickness. To meet the need for the higher resolution necessary for thin-bed evaluation, waveform-matching (multishot) processing techniques use the redundant information contained in overlapping receiver subarrays to improve vertical resolution (Fig. 1). Recently introduced processing techniques reduce the aperture to 0.5 ft and thus achieve a true vertical resolution of 1.0 ft (Fig. 2). However, reducing the aperture makes the measurements more susceptible to noise.

• 

  Fig. 2 – Compressional-wave-slowness curves obtained for various configurations (apertures) of possible overlapping subarrays based on an array tool having four transmitters and eight receivers. Note the increasing resolution of the curves with decreasing subarray aperture. Track 9 is a consistency check obtained by averaging the curves to 3.5-ft aperture and overlaying the results (courtesy of SPWLA).

Attenuation analysis

Acoustic-wave attenuation correlates with a variety of petrophysical parameters, including:

• Formation lithology and pore-fluid permeability
• Degree of fluid saturation
• Fractures type

A recent study suggests that the combination of compressional- and shear-wave attenuation logs may provide a potential formation-evaluation tool (Figs. 8 and 9).

• 

  Fig. 8 – Comparison between shear-wave attenuation logs and core permeability in a permeable oil zone. The two attenuation curves agree in high-permeability zones (courtesy of SEG).

• 

  Fig. 9 – Correlation between compressional- and shear-wave attenuation and fracture location. The interval of 2920 to 2930 m in which P- and S-wave attenuation curves overlay corresponds to the interval of high acoustic reflectivity (Track 4) and fractures identified by the down-going Stonely-wave reflectivity (Track 5) (courtesy of SEG).

Stoneley-wave analysis

The advanced use of Stoneley-wave analysis is best demonstrated through the following examples:

• Permeability estimation with Stoneley waves
• Gas detection

Noteworthy papers in OnePetro

Use this section to list papers in OnePetro that a reader who wants to learn more should definitely read

External links

Use this section to provide links to relevant material on websites other than PetroWiki and OnePetro

See also

Acoustic logging

Acoustic logging tools

Anisotropy analysis

Acoustic logging while drilling

PEH:Acoustic_Logging