Earth Resources Laboratory :: Abstracts

Seismic Velocity and Attenuation from Full Waveform Acoustic Logs

Mark Elliot Willis

Submitted to the Department of Earth, Atmospheric, and Planetary Sciences in April 1983 in partial fulfillment of the requirements for the degree of Doctor of Philosophy

Abstract

Automatic models of determining P and S wave velocity and attenuation from Full waveform acoustic logs are studied and compared. The suggested P wave velocity method is an event detector which is based on the threshold detection in a window near the previous picks and fine adjustment by a semblance correlation. The moveouts found by the correlation process are used to find the common source P velocities as well as effective borehole compensated P velocities. We call the most reliable velocity method for S waves the P correlated S method. It consists of correlating the P waveform with the rest of the record to find the S arrival. The S waveform is then correlated with the next record to determine the S velocity. The method does not necessitate a distinct S wave arrival and can utilize the existence of the reflected conical wave (or normal mode) whose phase velocity is controlled by the S velocity of the formation. When these methods are applied to synthetic borehole seismograms they reproduce the model velocities to within less than 0.5%.

In order to increase the depth resolution of the slowness (or velocities) determined, and remove the effects of borehole radii changes, we propose a method to formally invert the travel times and moveouts determined above. On synthetic travel time data the method obtains slownesses which are very well resolved. In the presence of noise the estimates display a larger variance but are locally unbiased. The standard borehole compensated method remains biased around bed boundaries and over small layers but produces estimates which appear fairly insensitive to the noise.

For the determination of intrinsic rock attenuation, the short receiver separations available make the estimates quite susceptible to noise. On synthetic data Qp estimates appear obtainable using fixed length time window around the P headwave. On actual data the estimates obtained using the slopes of the spectral ratios varied substantially even though the maximum likelihood spectral estimation was utilized to reduce the effect of noise. More stable estimates were obtained when the ratio at the peak of the near receiver amplitude spectrum was used. For S attenuation, there appears to be a complicated interaction between the shear headwave and the reflected control wave which masks the formation shear attenuation.

Finally, we investigated the sensitivity of the full waveform logs to borehole asymmetries and tool positioning. Our results show that the waveforms are indeed sensitive to these effects. We noted the following trends due to the above effects: 1) a reduction in the size of the P headway packet, 2) a reduction in the central portion of the reflected conical wavetrain, 3) the Stoneley wave is basically in affected, and 4) that the higher frequencies should be affected more than the low frequencies.