Petrophysics
PetrophysicsSome Causes for Bad Sonic Logs and Some Editing Options
Introduction
The objectives of this report are modest - to help you to identify bad sonic logs and to present methods of "cleaning-up" sonic data for input into Synthetics programs.
Poor quality sonic logs make creation of conventional synthetic seismograms difficult or impossible. Cycle-skipping is the most obvious symptom of bad sonic logs. Less obvious, but more widespread is the problem of acoustic signal attenuation resulting in anomalously high sonic values (low apparent velocity).
This report presents;
- Some reasons for poor quality sonic logs,
- Criteria for recognizing bad data, and,
- Some methods that can be used to "clean up" the original data.
Two methods are presented for calculating velocity from resistivity logs.
Causes of Bad Sonic Logs
Conventional wisdom holds that the sonic log is relatively unaffected by poor hole conditions. Most log analysis programs default to using the sonic for porosity determination where hole conditions are bad. In rugose hole, the sonic log may appear to be satisfactory with perhaps some cycle-skips. However, these first impressions of sonic log accuracy may be misleading.
Poor sonic logs can be caused by low transmitter strength, "road noise" and attenuation of compressional sound waves.
Low sonic transmitter strength will result in less than optimal receiver signal amplitudes. Under extreme conditions this will result in cycle-skipping. Records of transmitter signal strength are not kept, so it is not possible to determine how important this effect is in causing poor quality sonic logs.
Road noise is caused by tool movement along the borehole that generates a high frequency noise component that is superimposed onto the normal acoustic signal. These noise spikes travel along the tool body and can result in early stopping of the sonde timing clock. Road noise results in random spikes of varying amplitude on the sonic log. The far sonic detectors are more affected by road noise than the near detectors because of the reduced signal amplitude with increasing travel time.
Attenuation (decreased amplitude) of the compressional acoustic wave is probably the major cause of poor sonic logs. Attenuation results in the signal at the receiver crossing the threshold amplitude later than for a stronger signal. This results in "dt-stretch" where the apparent formation velocity is less than the true velocity. The magnitude of the "dt-stretch" can be up to 6 micro-seconds per foot (Tittman, 1986). Put in the perspective of porosity, over-estimation of transit time by 6 usec/ft in sand corresponds to porosity estimates that are about 4.5 porosity units too high.
Severely attenuated sonic signals may have compressional wave amplitudes lower than the detector threshold value. This results in the first compressional arrival not being detected. Instead, later, higher amplitude arrivals such as the Shear arrival are detected. This appears on the sonic log as the familiar and easily recognised cycle-skips where the transit time is obviously much too high and often has a "spiky" appearance.
Sonic "strech" is more common than cycle-skipping but is rarely recognized or documented.
Some causes of compressional signal attenuation are;
1) Low formation velocity. The sonic wave travels from the transmitter to the borehole wall through the mud. Some energy is refracted vertically and travels along the borehole wall. Energy is continually refracted back to the borehole where eventually some is detected at the receivers. Energy loss is therefore a function of formation velocity with slower (longer travel time) formations causing more signal attenuation.
2) High porosity. High porosity formations have poor grain-to-grain coupling with the result that acoustic enery is transmitted to the much slower fluid. The energy arriving at the receiver has diminished amplitude. This effect is most common at shallow depths.
3) Shale content. In laboratory experiments, Gardner et. al. (1968) noted a substantial decrease in signal amplitude with the addition of small amounts of clay to the formation. The reason for this attenuation has not been adequately explained but may explain why very dense shales have anomalously high transit time values.
4) Thin beds. Reflection and refraction of acoustic energy occurs at boundaries of beds with different velocity. This results in a reduction of signal amplitude. Attenuation is a function of both the velocity contrast between beds and the number of beds in the travel path.
5) Alteration of formation near the borehole. Drilling fluid often causes alteration of minerals (especially clay) near the borehole. The effect of this is to create a zone with lower velocity than in the virgin formation. Since the diffracted sonic waves travel close to the borehole wall, acoustic energy is attenuated by factors 1 and 2 above. Energy is also lost by refraction at the interface between the altered and virgin formation. Mud solids (clay) may be introduced into permeable beds with resulting signal attenuation by factor 3 above.
6) Eccentering of the sonic tool. Sonic receiver signal amplitude falls rapidly as the sonde moves away from the borehole axis (Tittman, 1986). This results because wave fronts travelling different paths undergo destructive interference.
7) Transmitter-receiver spacing. Sonic signal strength falls with increasing distance between the transmitter and receiver. This is an argument for not using long spaced sonic tools.
8) Borehole rugosity. In rugose hole it is not possible to ensure that the sonde is always centered in the borehole. The result is severe attenuation of the acoustic signal as in 6 above. Energy is also lost by diffraction at "angles" in the rugose hole. The transmitter-receiver distance is effectively much longer in rugose hole than in smooth hole. Since hole rugosity is often a result of alteration of the formation by drilling mud, several of the above mentioned factors contribute to signal attenuation.
9) Fractures. When an acoustic wave reaches a fluid filled fracture, part of it reflects back into the rock and part changes to a fluid wave in the fracture. When the fluid wave reaches the opposite fracture wall, there is further reflection loss and conversion back into compressional, shear and Stonely waves. (Schlumberger, 1987). Both reflection and mode conversion contribute to signal attenuation.
10) Hydrocarbons. Gas in high porosity formations attenuates compressional sonic waves and result in anomalously high transit times (Gardner et.al. 1968).
Recognition of Bad Sonic Logs
Velocity gain and offset errors resulting from incorrect Sonic tool operation are annoying but are not critical for generating accurate synthetics. Log analysis techniques in frontier areas often use baselining techniques that partly normalise-out the effects of errors in log readings. In more developed areas, logs from offset wells should be used to recognize and correct bad log data. In general, sonic logs should be assumed to be incorrect in rugose hole. Often, log analysis results (porosity and water saturation) using sonic data show a significant increase in "noise" in rugose hole. In many cases this is probably the result of optimistic porosity estimates from "stretched" sonic data. In rugose hole it is often difficult to baseline sonic and resistivity logs for source rock analysis. These intervals often appear to have high TOC values that are not supported by laboratory data and where resistivity values do not indicate significant organic matter.
Editing of Sonic Logs
Some of the editing options for sonic logs are:
1) Use a graphical editing facility such as that available in GEOLOG. This method is excellent for removing obvious cycle-skips over short intervals but is tedious for long sections. Graphical editing is not effective for correcting intervals affected by "sonic stretch".
2) Delete sections of bad data and replace with realistic values or interpolate between the top and bottom of the deleted interval. This may facilitate creation of synthetics, but valuable information may be lost.
3) Filter the log using a moving average filter. This method smooths out noise spikes but retains "relics" of bad data such as cycle skips and negative values. Also, vertical resolution of the log is degraded.
4) Use a median filter to eliminate questionable data. This method reduces curve variance in the filter window and eliminates unrealistic values. However, valid data is "clipped" from peaks and troughs in thinly bedded formations.
5) Use the GEOLOG "Despike" program. This program removes data that meets the criteria of having more than a specified amplitude range (spike amplitude) in a given depth interval (spike width). In practice this program is hard to use because both spike amplitude and width are unlikely to be constant.
6) Use the GEOLOG filtering option to create a turning point filter to remove unwanted high-frequency cycles from the log.
7) Replace sections of bad sonic with estimates from other logs such as the resistivity.
In some cases minor cycle skipping caused by hole caving does not significantly affect integrated time because high and low sonic values above and below the cave are "averaged out" in the integration.
Creating Sonic Logs From Resistivity Data
Sometimes sonic log data is not available, has missing sections or is very poor quality. In the last case it is futile to create integrated transit time and synthetics from the original data.
Often there is a positive correlation between velocity and resistivity. If so, synthetic sonic logs can be created from resistivity logs using either the Faust method derived from empirical relationships or from petrophysical relationships between resistivity and porosity and porosity and sonic transit times.
Faust (1953) developed the following empirical relationship;
In practice we have found that K is not constant but can be described either as a simple function of depth or as a zoned constant.
The Archie equation that describes the relationship between porosity and resistivity and the Wyllie time-average equation relating porosity to changes in transit time can be used to derive a relationship between resistivity and transit time. Assuming that there are no hydrocarbons;
Well 1: The Archie-Wyllie relationship was used to estimate transit time from the resistity log in this well.
Figures 1 and 2 show separation between measured and predicted DT in rugose hole. Good agreement between the measured and predicted values above and below the rugose section indicates that the resistivity derived transit time is probaly more reliable than the recorded sonic log. Figure 2 shows cycle-skips in extremely rugose hole and evidence of "sonic stretch" in adjacent sections.
Well 2: This well has extremely poor logs. The hole is very rugose and the sonic log through most of the logged interval is badly affected by cycle skips. The resistivity log is also very noisy. The GEOLOG filter program was used to create a turning-point filter to remove high frequency signal components from the resistivity log. Available sonic and resistivity data was used to generate a curve for the Faust "constant" K;
The GEOLOG graphical edit facility was used to define major trends in the K log. This edited K was used in the Faust equation to calculate a transit time curve from the resistivity log. The Faust sonic matches the measured sonic where hole conditions are reasonable. In intervals with cycle-skips, the Faust sonic curve is realistic. The synthetic created from the Faust sonic has far fewer reflectors than were generated from the original sonic log but character match of major reflectors with seismic data events is fairly good.
References
Dewan, John, Essentials of Modern Open-Hole Log Interpretation", 1983.
Faust, "A Velocity Function Including Lithologic Variation": Geophysics, v. 18, pp. 271-297, 1953.
Gardner, G. H. F., Harris, M. H., Velocity and Attenuation of Elastic Waves in Sands, SPWLA, 1968.
Kokesh, F. P., Blizard, R. B., Geometrical Factors In Sonic Logging, Geophysics, Feb. 1959.
Schlumberger Technical Review, v. 35, no. 1, 1987.
Tittman, Jay, Geophysical Well Logging", Academic Press, 1986.
|
|
