Page 117 - Well Logging and Formation Evaluation

Integration with Seismic 107

Big increases in AI(t) (“hard kicks”) will result in a large peak. In the
SEG (Society of Exploration Geophysicists) convention, this is displayed
as a black loop (to the right). Similarly, decreases (“soft kicks”) will
display as white loops (to the left). A similar convention is used for R(t).
The depths corresponding to major changes in lithology, and hard/soft
kicks, which are usually also formation tops, will have been converted
from depth to time along with the AI log. These may then be overlain on
the seismic to see if they correspond to seismic events. It is at this stage
that the shape of the synthetic is important. Because of problems with
“statics” (seismic shifts due to shallow events) or gas effects, events may
have become shifted on the seismic so that they do not match the syn-
thetic. By comparing the seismic and synthetic and matching up each loop,
it may be possible to apply a static shift to the synthetic so that the two
match up. Then the position on the seismic log of formation tops as seen
in the well may directly tie to the seismic section.
Whether or not this works will depend on how big the AI contrasts
resulting from lithology changes are, and the quality of the log data and
seismic analysis. I once spent two years exclusively tying synthetics to
seismic sections from logs from all around the world. In my experience
the method worked very well where there were major boundaries, good-
quality seismic data (usually from offshore), and reasonable log data. A
near-perfect match was obtained in about 5% of cases. In about 50% it was
possible to tie at least one event with confidence. It is worth remembering
that given the typical frequency content of seismic (say, 70Hz) and the
formation velocities (say, 5000m/s), one would only expect to be able to
resolve events having a minimum thickness of half a wavelength, i.e., 0.5
*5000/70 = 36m. Below this thickness, the logs will start interfering with
each other and the situation becomes much more complicated to interpret.
It should also be remembered that even if two random traces are com-
pared with one another, they can be expected to match up about half the
time. I have seen a supposedly good match proclaimed for well and
seismic log data that were afterward found to have come from different
countries!
When modeling a particular seismic anomaly (e.g., the pinch-out of a
GOC [gas/oil contact] against the top of a structure), it is sometimes nec-
essary to create artificial logs by successively removing a particular
section of log or changing the thickness of a sand body. This can be done
quite simply in the time or depth domain and the resulting traces filtered
as before to create an artificial seismic section in which the effect can be
modeled.