Geology Reference
In-Depth Information
a)
Net pay
Well K = 4m
Well L = 13m
Well L
Tuning curve using wavelet derived
from nearby well and scaled to the
maximum amplitudes
b)
-3000
-2000
Well L
-1000
Well K
0
0
2.32ms
1.96m
3.76ms
3.2m
5
10
15
20
25
11.5ms
9.7m
Time thickness peak to trough (msecs)
Figure 10.13
Net pay prediction using simple amplitude scaling; (a) composite amplitude map (reds are high amplitudes) with
two-way-time structure and showing locations of wells L and K; (b) tuning plot from seismic together with tuning curve scaled to the
maximum amplitudes. Courtesy RPA Ltd.
thickness is known in the model the variation of the
scaler with apparent thickness is straightforward to
derive. Plotting the inverse of the scaler on the seismic
tuning plot gives a
there are wells available to generate the coloured
inversion operator but there are no pay values to
guide the fitting, the scaler would be calibrated such
that the calibration curve would be positioned above
the data points on the average bandlimited impedance
versus seismic thickness crossplot (Connolly et al.,
2002
). The positioning of the curve would reflect the
general understanding of N:G.
A useful simplification of Connolly
which provides
a useful graphical appreciation of the calculation (i.e.
net pay
'
calibration curve
'
¼
(ABLI/calibration curve value)*apparent
thickness).
To illustrate how the net pay prediction technique
works, a model was generated by using data based on
a real field example (
Fig. 10.15
), including both thin
and thick layering situations. The model comprises
gas sands and shale, in which the acoustic properties
are held constant between the wells, and there are no
hydrocarbon contacts. For each of the model wells,
apparent thickness and average bandlimited imped-
ance were calculated. The scaler function was then
calibrated as described above.
Figure 10.15d
demon-
strates the viability of the technique. In practice, if
s(
2007
) tech-
nique is to assume that the scaler varies linearly with
apparent thickness (Simm,
2009
). This may be appro-
priate for example when there are no wells and the
effective bandpass wavelet is unknown. A linear scaler
assumption leads to the following:
'
ABLI × apparent thicknes
2
Net pay
¼ð
Þ
constant
:
Fig 10.16
illustrates that the simplification works
quite well with the model dataset.
232
