Sie sind auf Seite 1von 5

71

st
EAGE Conference & Exhibition Amsterdam, The Netherlands, 8 - 11 June 2009


5908
Gas Hydrate in the Krishna-Godavari Basin, India
P. Dewangan (National Institute of Oceanography), P. Jaiswal* (Rice
University), T. Ramprasad (National Institute of Oceanography), S.
Gullapallis (National Institute of Oceanography), C.A. Zelt (Rice
University), M.V. Ramana (National Institute of Oceanography), M.V. Lall
(Directorate General of Hydrocarbons), B.J.P. Kumar (Directorate General
of Hydrocarbons) & A.V. Sathe (Oil and Natural Gas Corporation)
SUMMARY
Large quantities of gas hydrate were discovered while drilling in the Krishna-Godavari (K-G) basin along
the Indian east coast in 2006. Drilling also showed significant inhomogenity in hydrate distribution. Using
the unified imaging approach of Jaiswal and Zelt (2008) we develop a velocity model and a depth image
that are consistent with the well observations. We use the velocity model for estimating hydrate
concentration with effective medium modeling proposed by Dvorkin et. al. (1999). We realize that current
rock physics schemes may fall short in hydrate quantification in fracture dominated systems such as the K-
G basin.

71
st
EAGE Conference & Exhibition Amsterdam, The Netherlands, 8 - 11 June 2009
Introduction
The offshore extension of the Krishna-Godavari (KG) basin along the eastern coastline of
India (Figure 1) has gained significant attention after the discovery of large quantities of
hydrates during the drilling/coring operations in May-August 2006 with research vehicle
JOIDES Resolution. Drilling also showed significant inhomogenity in hydrate distribution. In
this paper, with the help of a high-resolution (dominant frequency ~80Hz) seismic line and a
well located close by (<50m), we present a case study of gas hydrates quantification through
rock-physics modeling using a P-wave velocity model developed using a combination of
depth migration and traveltime inversion.

Study Area

Figure 1. Study
Area. Seismic
line is indicated
with black line
with CDPs
labeled in
multiples of 200.
Location of the
Well W is
indicated by solid
dot. Bathymetry
is labeled every
50m. Location of
the study area
with respect to
India is shown in
inset with rivers
Krishna and
Godavari labeled.
Bathymetry is
labeled in km.


The KG Basin is a peri-cratonic passive margin basin on the east coast of India (inset; Figure
1). The basins characteristic feature is its enechelon horst and graben system which is filled
with a thick pile of sediments of Permian-to-Recent age. The onland part of the K-G basin
covers 28,000 km
2
and is mostly alluvium covered. Drained by the rivers Krishna and
Godavari, the offshore basinal area also consists of predominantly fine grained alluvial
sediments in the shallow (< 200m) subsurface. Previous studies in the K-G basin using
industry style MCS data indicates the presence of large scale growth faults in the continental
shelf and upper slope regions, shale ridges and diapirs in the slope region, and imbricate
thrust faults on the lower slope and uppermost rise. Studies by Dewangan et al. (2008) using
chirp sonar, sub-bottom profiler (SBP) and multibeam swath bathymetry in the slope region
strongly suggests that gas hydrate occurrence and distribution in the basin are linked with
shale tectonics.

Data and processing
The seismic line used for this study was acquired in the slope setting in water depth of 1000m
in 2004 as a planning tool for JOIDES resolution (Figure 1). The variation in seafloor
topography along the seismic line is ~180 m. The data were acquired in an end-on acquisition
style with a source and receiver spacing of 12.5 m and a nominal fold of 60. The seismic line
is ~8 km long and comprises 576 shots fired using an airgun bubble source. The minimum

71
st
EAGE Conference & Exhibition Amsterdam, The Netherlands, 8 - 11 June 2009
and maximum offsets were 70m and 1570m respectively. The data were recorded with a
sampling interval of 1ms. Overall the data have a high S/N ratio and a broad frequency
bandwidth of 10600 Hz (the background noise has ~30 dB lower power than the signal).
The seismic data was processed for enhancing the S/N ratio and imaging only the
broad features of the geology. The first step in processing is trace editing for removal of the
noisy traces. As a result of this step, 10% reduction took place in data volume. Following this,
10-15-80-160 Hz minimum phase Ormsby bandpass filter is applied to the data. An additional
requirement of predictive deconvolution with operator length of 16ms is recognized for
suppressing the bubble pulse. The processed data was then used for velocity analysis.

Velocity model building and depth imaging
Velocity model building and depth imaging was performed using the Unified Imaging scheme
proposed by Jaiswal and Zelt (2008). This scheme begins with conventional stacking velocity
analysis. The interval velocity model from stacking velocities is used for pre-stack depth
migration (PSDM). The stacked data and the PSDM image are interpreted for common
horizons (three in this study; Figure 2) and the corresponding wide-aperture reflections are
identified in the shot gathers. Using the interval velocity model the stack interpretations are
inverted as zero-offset reflections for constraining the corresponding interfaces in depth; the
interval velocity model is maintained stationary. The velocity model is updated by inverting
traveltime picks from shot gathers till the interpreted and inverted horizons are consistent to
within the interpretational uncertainties (< 5m). The Unified Imaging is then said to have
converged yielding the final velocity model and depth image (Figure 2).
The traveltime inversion part in Unified Imaging updates both velocities and horizon
depths. It defines an objective function (E; Equation 1) and minimizes it until the normalized
data misfit is close to unity.
1 ]} )[ 1 ( ] [ { ) , (
1
,
1
,
1
,
1
z C z s C s s s C s d C d z s E
h z
T
v s
T
z h s
T
d
T
o o | o o o o | o o

+ + + =
In equation (1), od is the difference between the observed and predicted traveltimes, os is the
slowness (inverse of velocity) perturbation vector; s
0
is the starting slowness vector, C
d
is the
data covariance matrix; covariance matrices C
s,h
, and C
s,v
measure horizontal and vertical
roughness of the slowness perturbation, respectively, is the trade-off parameter, and s
z
determines the relative importance of maintaining vertical versus horizontal model
smoothness. oz is the interface depth perturbation vector being solved for and C
z,h
is the
covariance matrix that measures the interface roughness. | determines the relative weights of
slowness and interface regularizations. For updating only the interfaces, | is set as zero.
For PSDM, the data are sorted from common-shot (for processing) to common-offset
domain and are binned according to their shot-receiver offsets. The resulting common-offset
sections are padded to ensure that each bin has at least one trace per receiver station. The
migration aperture is set to ~2 km and a frequency up to 80 Hz is migrated.

Effective Medium Modeling (EMM)
EMM (Dvorkin et. al., 1999) accounts for the unconsolidated nature of gas-hydrate bearing
sediments. For porosity less than critical porosity (|
c
- porosity at which solid appears to be
suspended in fluid), the bulk (K) and shear (G) moduli of the dry frame are given by:
3 / 1
2 2
2 2 2
3 / 1
2 2
2 2 2
1
1
) 1 ( 2
) 1 ( 3
) 2 ( 5
4 5
,
) 1 ( 18
) 1 (
2
2
8 9
6
;
/ 1 /
3
4
3 / 4
/ 1
3 / 4
/
(
(

=
(
(

=
|
|
.
|

\
|
+
+
=
(

+
+
=

+
+
=

P
G n
G P
G n
K
G K
G K G
Z Z
Z G Z G
G
G
G K G K
K
c
HM
c
HM
HM HM
HM HM HM c
HM
c
dry
HM
HM
c
HM HM
c
dry
v t
|
v
v
v t
|
| | | |
| | | |



71
st
EAGE Conference & Exhibition Amsterdam, The Netherlands, 8 - 11 June 2009


Figure 2. Combined PSDM image and velocity model. BSR is indicated with solid arrows.
The resistivity log from the well W is overlaid. A zone of diminished reflectivity is indicated
with dashed ellipse. 1, 2, and 3, are horizons used for inversion; 2 is the BSR, and 1 and 3 are
indicated by dashed grey lines.

In equation 2, n is the number of contacts per grain in a sphere packed at |
c
(8.5 in this
study), P is the effective pressure, and K, G and v are the bulk and shear modulus and
Poissons ration of the solid phase. The critical porosity in the region, from multiple core
analysis, is determined to be approximately 62%. In the reference well, the variation of
porosity with depth was estimated using Athys Law,
) / (
0
) (

| |
z
e z

= ; |
0
is the initial
porosity and is a free parameter. The porosity values were found to be less than critical
porosity justifying use of equations 2. Elastic constants of the solid phase in the reference well
were computed assuming 100% clay mineralogy (JOIDES onboard observation). The
saturated bulk and shear moduli were calculated from their dry counterparts using
Gassmanns equation and 100% water saturation. The sonic velocity profile using the
saturated bulk and shear moduli were calculated using the equation:
; / ) 3 / 4 (
Sat Sat P
G K V + = K
sat
, and G
sat
, are the bulk and shear moduli of water saturated
sediments, respectively, and is the bulk density. The sonic profile thus generated was
compared with the sonic log in the reference well (Figure 3a).
A satisfactory comparison encourages us to apply the same scheme to well W. While
computing the dry frame bulk and shear moduli for well W, we assume that the hydrates are
part of the fame of the sediments. The key to estimating hydrate concentration is thus to find
out how much of the frame is to be constituted by the hydrates so that the velocity profile
estimated by inversion (Figure 2) can be reasonably predicted. In a trial and error manner, the
hydrate component in the frame is then varied from 0-60%. For every concentration value,
Hills average formula is first used to calculate the composite moduli of the dry frame from
individual components - clay and hydrate. The saturated moduli and the sonic velocity are
then computed in turn. The process is repeated for all hydrate saturation till the overall
mismatch between the predicted and the inverted velocity profile is found to be less than 10%.
The resulting hydrate saturation profile is presented in Figure 3b (red curve).

71
st
EAGE Conference & Exhibition Amsterdam, The Netherlands, 8 - 11 June 2009
Figure 3. Rock physics modeling.
a) Velocity comparison in the
reference well. The match suggests
that the adopted rock physics model
is reasonable. b) The concentration
profile of hydrates required to
match the inverted velocities along
the well W (in red) compared to
concentration estimated from
resistivity log (in black). The
resistivity log is displayed in Figure
2. The mismatch could be indicative
that including hydrates as part of
dry sediment frame may not be a
sufficient assumption for modeling
hydrate concentration.





Results and discussion
The final velocity model and the final depth image are combined in Figure 2. The model is
characterized by alternating increasing and decreasing velocity zones from the northwest to
the southeast end. In the combined velocity model and depth image, parts of the image with
BSR appear to be associated with overlying increase in velocity. The region between CMPs
830-930 and depth 1125-1223m show diminished reflectivity (blanking?) and coincides with
an enhancement in velocities - another indication for presence of hydrates. The resistivity log
from W (Figure 1) is overlaid on the depth image (Figure 2). The BSR in the resistivity log (~
160 mbsf) matches fairly well with the BSR in the depth image (solid arrows; Figure 2)
indicating the reliability of the depth image and in turn that the velocity model is
kinematically correct.
Using the bulk and shear moduli for clay from the reference well, the estimated
hydrate saturation for W is shown in Figure 3b and compared with the saturation obtained
using the resistivity log. The match appears to be better in parts with relatively low hydrate
concentration (90-120m depth). In other parts, the velocity predicted by EMM falls short.
Helgerud et. al. (1999) show that maximum enhancement in velocity occurs when hydrates
are included in the frame (as opposed to pore-filling fluid) ruling out the possibility of using
hydrates in any other state. The JOIDES coring recovered many intervals of massive hydrates
in features resembling fractures e.g., between depths 20 and 80m (Figure 3b). An open ended
question (currently under investigation) thus remains in the end if modeling hydrate
accumulations in presence of fractures requires a new type of rock physics modeling.

References
Dewangan, P., Ramprasad, T. Ramana, M.V., Mazumdar, A., Desa M., and Badesab, F.K.
[2008] Tectonics in the Krishna-Godavari Basin: Implication in Gas-Hydrate Exploration.
Earth and Planetary Science Letters, communicated.
Dvorkin, J., Prasad, M., Sakai, A., and Lavoie, D. [1999] Elasticity of marine sediments:
Rock physics modeling. Geophysical Research Letters, 26, 12, 1781-1784.
Helgerud, M.B., Dvorkin, J., Nur, A., Sakai, A., and Collett, T. [1999] Elastic-wave velocity
in marine sediments with gas hydrates: Effective medium modeling. Geophysical Research
Letters, 26, 13, 2021-2024.
Jaiswal, P. and Zelt, C.A. [2008] Unified imaging of multichannel seismic data: combining
traveltime inversion and prestack depth migration, Geophysics, 73, doi:10.1190/1.2957761.

Das könnte Ihnen auch gefallen