Geoscience Reference
In-Depth Information
I
Phase II (drainage)
Phase III (postdrainage)
5
2.5
0
-2.5
-5
0
20
40
60
80
100
120
140
160
180
200
Times (s)
Figure 5.41
High-pass filtered self-potential data showing the level of electrical noise produced during drainage. The full 195 s
electrogram is shown. In phase II, the electrogram exhibits large negative voltage spikes with decaying tails. The negative voltage
spikes typically involve only few channels, implying that they are located not too far from these electrodes.
removed with a high-pass filter. It can be seen clearly in
Figure 5.41 that in phase II, the
“
noise
”
clearly begins at a
much lower level and increases dramatically with mostly
negative excursions (Figure 5.41), which dominate as
time increases (Figure 5.41). The high-pass filter causes
a differentiation-like effect converting negative-only
pulses into pulses with both positive and negative excur-
sions. Phase II is dominated by a series of different types
of electrical events, but most prominent are the negative-
going voltage leading edge and a long (up to 1 s) tail.
Negative-going voltage spikes are consistent with water
impulsively descending during drainage. Most of the
negative voltage spikes do not have a coincident signal
offset; however, some of them show various types and
magnitudes of offsets (Figure 5.40). The cause of the vari-
ety of these offsets has not been determined, but could be
suggestive of a localized process of consolidation of the
sand near the lesser compacted surface or within the vol-
ume, altering the resistivity of the local sand volume.
Video recordings show movement of sand grains near
the surface of the sand during drainage, lending support
to the consolidation suggestion as an explanation of the
observed signal offsets.
Similar data have been obtained for an imbibition
experiment (not shown here) using the same test
apparatus. However, the amplitude and the number of
electrical bursts were observed to be much smaller than
for drainage. During imbibition, the most important
events are positive excursions in the self-potential signals.
These observations are consistent with Haines jumps
during the imbibition process described earlier and the
upward movement of water within the sand volume.
As explained earlier, the sudden movement of
the meniscus during drainage experiments has been
documented in a number of prior studies (Aker et al.,
2000). In these prior works, the invading fluid (air in
the present case) was found to suddenly invade a larger
region of the pore space. Those studies demonstrated that
the observed bursts of fluid flow (Haines jumps) were
characterized by large fluctuations in the pore water
pressure and AE (Sethna et al., 2001; DiCarlo et al.,
2003). In our drainage experiment, we did not measure
AE, but a crackling AE was easily discerned by human
ears during drainage and was heard at the same time that
the electrical impulses occurred in the data. However, the
acoustic and electrical noise phenomena have not yet
been correlated with a simultaneous data acquisition.
We attribute Haines jumps as the source of the signals
we observed.
We studied the nature of the imbibition and drainage
processes using a statistical analysis of the occurrences of
impulsive events in different data sets involving the two
different processes. As described earlier, the impulses from
each data set were tabulated for a set of voltage thresholds
and displayed in histogram form (Figure 5.42). The differ-
ence in voltage magnitude distributions for drainage and
imbibition is a result of the difference in the types of
pore-filling processes. Note that the magnitude of the
self-potential events is much larger on drainage than on
imbibition (x-axis on both panels in Figure 5.42). During
drainage, the magnitude of the electric events shows a
power law distribution with an exponent of
−
1.7 ± 0.1 fit-
ting the data well over the complete range of event mag-
nitudes. These results are in agreement with the seismic
data analysis of DiCarlo et al. (2003) who also found a
power law exponent of
−
1.7 ± 0.2.
During imbibition, no large events observed, and the
histogram of event sizes drops off much steeper with a
