PID 174 A Mapping of Shale Volume
PID 174 A Mapping of Shale Volume
PID 174 A Mapping of Shale Volume
Keywords
𝐺𝑅 𝑙𝑜𝑔 −𝐺𝑅𝑚 𝑖𝑛
𝑉𝑠ℎ = (1)
𝐺𝑅 𝑚𝑎𝑥 −𝐺𝑅𝑚𝑖𝑛
Where,
Post-stack seismic
inversion (AI) Volume of
Figure 1: Showing distribution of oil fields and well loactions in
study area of Upper Assam Basin. Shale (Vsh)
Mathematical
Methodology transformation
Using empirical
Post-stack seismic inversion is carried out and results Equations
used to derive SI, ρ and ∅ using transform on AI
volume. The seismic attributes AI, SI, ρ and ∅ are Seismic Attributes: MLFN
used as input in computation of Vsh value in MLFN AI, Density,
model. This methodology is applied to three wells Porosity, SI
(M1, M2 & M3) and 2D post stack seismic section Predicted Vsh section
from parts of Upper Assam basin (Russel and
Hampson, 1991). Two wells (M2 and M3) are used
as training well and M1 well is used to validate the
result. Trained neural network from two of the wells Figure 2: Flowchart showing methodology for prediction of
is applied to the 2-D seismic section to generate Vsh volume of shale using MLFN model.
distribution and validate the results using the third
well. Finally the trained network is used to generate
subsurface Vsh distribution map on the seismic Using acoustic impedance (AI) inversion from post-
section within a time interval of 1550–2250ms. The stack data the seismic amplitude is transformed to its
low values of training error and validation make layer properties (Figure 4). AI, SI, ρ and ∅ sections
reliable Vsh prediction. have been used to train the MLFN model for
mapping Vsh obtained from well logs in part of Upper
Figure 2 is showing the procedure for estimation of Assam basin. SI, density and porosity sections are
Vsh from post-stack seismic section (Figure 3) generated by mathematical transformation from
crossing three wells. Actual values of shale volume at inverted AI section. An empirical relation is carried
the three well locations have been evaluated from out between SI and AI for well M1 of Upper Assam
Mapping of Shale Volume using Neural Network Modelling
basin as shown in Figure 5(a). The transformation Porosity and AI (∅𝐷 = -0.00005AI +0.51) is
equation is obtained as SI=0.75AI-1936 with a established (Figure 5.b). Using this relation, porosity
coefficient of determination (R2) =91%. Using this section is created by transforming inverted AI section
transformation, SI section is generated from inverted (Figure 6). Density section is derived from the
AI section. Similarly, an empirical equation of inverted AI and P-wave velocity (Vp) sections.
Figure 4: Inverted Acoustic Impedance section of the 2-D post-stack Seismic data.
Figure 5: Empirical relation between (a) SI and AI & (b) Porosity and AI for well M 1 of Upper Assam Basin.
Mapping of Shale Volume using Neural Network Modelling
In Figure 9, a high Vsh zone above 1500 ms low values of training and validation error make
corresponds to the Girujan clay formation. The same reliable estimation of Vsh in this study area. The
zone is clearly indicated by the low impedance (in estimated MLFN predicted Vsh model follows the
Figure 4) and high porosity (in Figure 6) above general geological trend of Upper Assam basin. Since
1500ms. The Tipam series includes sandstones that presence of shale impact the reservoir quality, shale
are somewhat coarser, mottled clays, and a few volume estimation is important parameter in log
conglomerates . The Surma series consists of analysis. So it is important to determine the shaly
alternating beds of shale , sandy shale , shaly sand analysis in reservoir characterization studies.
sandstone, sandstone, and conglomerate (Corps, Better estimations of hydrocarbon production can be
1949). The zone between 1550-1750 ms shows achieved by understanding the shale distribution and
alternate layers of shale and sand (Figure 9), that knowing the volume of shale in the reservoir.
corresponds to part of Tipam and Surma group.
Comparatively low Vsh zone with shaly sand features
is exhibited from 1800 to 2200ms (Figure 9), which
is the Barail group of Oligocene age. The same
Figure 9: MLFN predicted volume of shale (Vsh) distribution for the seismic section of Upper Assam basin.
This new approach to shale volume prediction from Das, B. and Chatterjee, R., 2016, Porosity Mapping
post-stack seismic section using neural network from Inversion of Post-Stack Seismic Data;
proved to be effective in Upper Assam basin. The Georesursy, 18, 306-313.
Mapping of Shale Volume using Neural Network Modelling
Kumar, R., Das, B., Chatterjee, R. and Sain, K., Singha, D.K., Chatterjee, R., Sen, M.K. and Sain, K.,
2016, A methodology of porosity estimation from 2014, Pore pressure prediction in gas-hydrate bearing
inversion of post stack seismic data; Journal of sediments of Krishna– Godavari basin, India, Marine
Natural Gas Science and Engineering, 28, 356-364. Geology, 357, 1-11.