Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/260593111

Sand Fraction and Pore Fraction Inversion through Acoustic Impedance Data:
Another Alternative Tool for Qualitative Thin-bed Reservoir Characterization
and Net Pay Volumetric Estim...

Conference Paper · June 2013

CITATIONS READS

0 139

2 authors, including:

Ee Juan Boon
Universiti Teknologi PETRONAS
2 PUBLICATIONS   0 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Ee Juan Boon on 08 March 2014.

The user has requested enhancement of the downloaded file.


Sand Fraction and Pore Fraction Inversion through Acoustic Impedance Data: Another
Alternative Tool for Qualitative Thin-bed Reservoir Characterization and Net Pay
Volumetric Estimation

GAN Wei Di1 and BOON Ee Juan2


Faculty of Geosciences and Petroleum Engineering
Universiti Teknologi PETRONAS, Tronoh, 31750, Perak, Malaysia
1
Email: maxor.gwd@gmail.com; 2wisejuan13@gmail.com

KEYWORDS: Acoustic Impedance Inversion, Thin Bed, Reservoir Characterization, Pore


Fraction, Sand Fraction

ABSTRACT
This paper presents a case study on inversion and characterization of thin-bed reservoirs which is below
seismic resolution. The ultimate aim of this study is to deterministically estimate the volume of thin pay
constrained by inverted acoustic impedance (AI) data with minimal well controls during the exploration stage.
Since most of the thin-beds are not imaged by the seismic, we propose the use of Sand Fraction (i.e. sands
content) in the seismic scale to quantify thin-beds and heterogeneity in the reservoirs. Combine with the
quantification of total porosity (Pore Fraction), they could provide us with flexibility in incorporating various
shale distribution models in our volumetric estimation. Firstly, seismic is inverted to AI. Then, both Sand
Fraction and Total Pore Fraction are further inverted with a function of inverted AI. This paper also discusses
various strategies for volumetric estimation based on various shale distribution models.

INTRODUCTION
This study is conducted on a gas field in the Malay Basin during the exploration stage. Data used in this study
includes 3D seismic data (near, far, full stack) with two wells 4km apart. The reservoirs consist of packages of
thin-bed fluvio-lacustrine clastic sediments, which are dominated by sand, shale and coal interbeds (Figure 1).
The goal of this study is to invert Sand Fraction and Pore Fraction through impedance data from seismic for
volumetric estimation. The conventional formula for GIIP (Gas Initially In Place) is as follow:

where

GIIP = Gas Initially In Place

GBV = Gross Bulk Volume of Reservoir Body

NTG = Net Pay to Gross (Fraction)

⌀ = Porosity (Fraction)

= Gas Saturation (Fraction)

FVF = Formation Volume Factor of Gas (Fraction)


Gan and Boon

In this case study, we had identified several limitations which we wish to solve:

1. Thinly laminated lithology such as sands, shales and coals, which are below seismic resolution, are
difficult to be identified. In our case study, gas sands and coals are undistinguished in inverted seismic.
The limit of seismic resolution or tuning thickness is 20m.

2. Heterogeneity especially gradual change of reservoir properties (i.e. shale volume, porosity, etc.) such
as graded bedding makes classification of facies very subjective. For example, setting the cutoffs to
classify shales, sandy shales, very sandy shales, very shaley sands, shaley sands and clean sands are
very subjective.

3. Various distribution modes of shale in reservoir could also affect porosity (Thomas & Stieber, 1975).
Figure 10 shows the effect of different shale distribution modes on the resulting effective porosity
( ) which could directly influence volumetric calculation.

In this study, we would discuss the methodology we use to characterize thin beddings which are not imaged by
the seismic, by introducing Sand Fraction and Pore Fraction. This is an extended characterization study through
acoustic impedance inversion. The motivations behind our approach are:

1. To address the limitations of seismic thin-bed during the exploration stage with limited well controls, it
is proposed that empirical relationship of Sand Fraction and Pore Fraction as a function of Acoustic
Impedance (AI) are exploited from wells. Packages of thin-bed sand, coal and shale could then be
represented in the seismic scale by averaging Sand Fraction (1-Vshale) in impedance domain.

2. To address the heterogeneity of reservoir in the seismic scale, it is proposed impedance data from
seismic inversion are used as a constraint.

3. Since Sand Fraction and Pore Fraction are continuous properties, volumetric estimation no longer
needs to rely on discretized blocky model of litho facies, which would be subjective in condition where
boundary between facies is not clear.

THEORY

According to Widess (1973), seismic is ideally capable to resolve rock thickness down to the limit of of the

dominant frequency. However, due to many factors like signal to noise ratio, seismic could practically resolve

rock thickness down to the limit of . Likewise, Sheriff and Geldhart (1995) propose that we can detect the

present of thin layers till the finest thickness of . Beyond that, it is undetectable in seismic data.

2
Gan and Boon

METHODOLOGY
Rock Property Analysis

Figure 2 shows initial rock property analysis that is conducted on the reservoirs by using well data. The study
revealed that Acoustic Impedance (AI) alone is capable of discriminating gas sand from background lithology in
well resolution (0-125Hz). However, after applying a low cut filter and resample the same data to 2ms as with
the seismic resolution (0-85Hz), AI could no longer separate gas sand from coal effectively. It is also observed
that gas sand and shale starts to overlaps partially in the AI domain. This suggests that maximum resolution of
seismic data (85Hz, 2ms) would be insufficient to resolve the response of thin coals from gas sand, and thin gas
from shale. Hence, the inversion of seismic data into AI with the same resolution would not be able to
distinguish gas sand from background lithology. Therefore using cutoffs in AI ranges for pay quantification
would likely results in underestimation as most thin beds are seismically invisible.

Coal Characterization

Some of the coal beds could be characterized in the seismic volume through AVO difference. AVO difference is
an attribute derived by subtracting the envelope of near seismic amplitude from far seismic amplitude. As
illustrated in Figure 3. The extreme negative values (colored in dark green) show the dimming response of coal
with offset. The results are consistent with well lithology log. The purpose of coal characterization is to allow
qualitative interpretation of reservoir body to exclude thicker coal from being quantified as Sand Fraction.

Sand Fraction and Pore Fraction Characterization: An Alternative Tool

Instead of capturing blocky model of litho facies, Sand Fraction is proposed as an alternative tool for a more
representative quantification of thin beds in the seismic scale. From well data (Figure 4), we invert ranges of AI
into averaged fraction of shales (Vshale) where averaged Sand Fraction can be obtained by 1 - Shale Fraction.
Sand Fraction here refers to the sand content, not sandstones. The idea is to weigh sand contents according to AI
"trend" as observed in the well at a specific reservoir, so that fractions of "seismically invisible" thin sands are
taken into accounts. As for coals, they are being discounted by clipping them as non-producing shales (100%
Vshale). We assume that thin sands would not be seen in the seismic data but collectively they could be detected
as a package of homogenous gross sands. Together with inverted Pore Fraction from AI (Figure 5) and also gas
saturation data, we could calculate various volumetric (i.e. GIIP) based on various shale distribution models in
the reservoirs. This step would be explained further in the discussion section.

Acoustic Impedance Inversion

Figure 6 illustrates the workflow of seismic inversion to obtain absolute AI. First, seismic is tied to the well and
wavelet is extracted from full-stack seismic around the well. Then, frameworks of interpreted surfaces (i.e.

3
Gan and Boon

horizons and faults) are used to construct a structural model, which is used to guide the interpolation of well AI
trend (0 – 6Hz) to build a Low Frequency Trend Model. Together with the wavelet, Constraint Sparse Spike
Inversion (CSSI) is employed to invert full-stack seismic data into absolute AI. In the inversion process,
inverted AI had been successfully calibrated to the well. Several Quality Controls (QCs) showed that Inverted
AI are able to match the well AI while inverted synthetics shows a good match to seismic Figure 7. The
distribution of lithology in inverted AI also remains similar to well.

Volumetric Estimation through Sand Fraction and Pore Fraction

Figure 8 illustrates the workflow of volumetric estimation by using Sand Fraction and Pore Fraction. To
proceed with volumetric estimation, geometry of reservoir gross bodies is delineated by using interpreted
surfaces and Gas Water Contacts (GWCs). As discussed earlier, inverted AI in each reservoir’s geometry is later
converted into Sand Fraction by using empirical relationship of Vshale versus AI in wells. Likewise, inverted AI
is converted into Pore Fraction by using relationship of total porosity versus AI in wells. Now the reservoirs are
characterized by Sand Fraction and Pore fraction, and are directly calibrated to the well. A quick QC shows that
areas of high Sand Fraction and High Porosity correspond to sandy interval in the well log (Figure 9). Effective
Pore Fraction is then calculated by multiplying Sand Fraction with Pore Fraction. This is to discount shale and
its pore from being calculated in net pay quantification. Our model assumes distribution of shale is laminar
where micro-pore in shale is not producible. Next, Effective Pore Fraction is further multiplied with pore-
weighted-average gas saturation from the well to calculate net pay. Since the reservoir geometry is in time
domain, the final step of the workflow is to convert the whole geometry to volume in depth (billion cubic feet)
domain by using P-Velocity data. The P-Velocity data is inverted from absolute AI using empirical relationship
from the well.

RESULTS

As with conventional approach, Sand Fraction and Pore Fraction in addition with gas saturation are able to
estimate these reservoir parameters: gross bulk volume, net sand volume, total pore volume, effective pore
volume, and finally net pay volume. GIIP are then computed by multiplication of Formation Volume Factor
(FVF) with net pay volume. Table 1 shows Volumetric Estimation (i.e. GIIP) in Billion Cubic Feet.

DISCUSSION
The inversion of seismic to Sand Fraction and Total Pore Fraction has allowed us great flexibility in applying
various shale distribution modals in our shaley sands reservoirs’ volumetric (i.e. GIIP) estimation (Figure 10).
The models that will be discussed in this paper are the modification of Thomas and Stieber (1975). For laminar
shale model, partial porosity and partial grainosity of sands matrix are replaced by shale’s micro-porosity and
grainosity. As demonstrated in the methodology, if assumption of micro-porosity in shales is not gas productive,

4
Gan and Boon

effective porosity is computed by discounting shale matrix and its micro-porosity by multiplying Sand Fraction
with Total Pore Fraction. Then, the volumetric equation will be as follow:

GIIP = Sand Fraction x Total Pore Fraction x Sgas x FVF

For structural shale model, the shale matrix is not occupying any sand’s porosity and not having any micro-
porosity. Hence, effective porosity is equal to total porosity. The volumetric equation will be as follow:

GIIP = Total Pore Fraction x Sgas x FVF

For dispersed shale model, the shale matrix occupies pore space of a sands matrix supported rock. These shales
do not have any micro-porosity. Thus, effective porosity can be computed by deducting Shale fraction from
Total Pore Fraction. Then, the volumetric equation will be as follow:

GIIP = (Total Pore Fraction - (Shale Fraction x Total Pore Fraction)) x Sgas x FVF

This method does not delineate any specific gas sands body (i.e. net pay). Sand Fraction and Pore Fraction of
the whole reservoirs will be used to estimate volumetric (i.e. GIIP). In this case study, the less productive parts
of the reservoirs will have low Sand Fraction and low Total Pore Fraction. Thus, the multiplication of these two
fractions will result in low Effective Pore Fraction and making their contribution to volumetric calculation
insignificance. In the opposite, the more productive parts of the reservoirs will contribute significantly to
volumetric calculation.

FUTURE WORK
Rigorous testing and more case studies need to be done on this methodology to assess its effectiveness in
estimating volumetric. Besides, Coal Fraction could be inverted from AVO Difference (i.e. Envelop Far –
Envelop Near) to discount thin coal from volumetric calculation. This is particularly useful in quantifying pore
in shaly sands or possibly fracking jobs on highly consolidated shales for tight gas.

CONCLUSION
With minimal well control in this 2msec 3D seismic dataset, it is shown that minimum gross sands interval of
20m with corresponding Sand Fraction is able to provide a suitable average to represent thin-bed lithology in the
seismic scale.

Sand Fraction and Pore Fraction are calibrated directly to well, easy to derive, yet provide the flexibility to
adjust volumetric calculation based on different shale distribution models in reservoirs, objective, interpretations
and assumptions.

5
Gan and Boon

ACKNOWLEDGEMENTS
The authors would like to thank Petroleum Management Unit (PMU), PETRONAS for providing the data and
giving permission to present and publish this paper. The authors are also grateful to Geosciences Department of
Universiti Teknologi PETRONAS for providing the necessary facilities to carry out this work.

REFERENCES

Sheriff, R. E., & Geldart, L. P. (1995). Exploration Seismology (2 ed.). Cambridge University Press.

Thomas, E. C., & Stieber, S. J. (1975). The distribution of shale in sandstones and its effect upon
porosity. 16th Annual Logging Symposium. SPWLA.

Widess, M. B. (1973, December). How Thin Is A Thin Bed. Geophysics, 38(6), 1176-1180.

APPENDIX
Table

VOLUMETRIC ESTIMATION - Billion Cubic Feet


GROSS BULK SAND TOTAL PORE EFFECTIVE PORE GAS PAY Formation GIIP
VOLUME VOLUME VOLUME VOLUME SATURATION VOLUME Volume Factor

Reservoir-TOP1 21.98 10.57 4.32 2.14 0.39 0.84 127.00 106.60


Reservoir-TOP2 10.34 5.03 1.90 1.02 0.46 0.46 127.00 58.83
Reservoir-TOP3 16.15 7.18 2.96 1.34 0.50 0.67 127.00 84.74
TOTAL 48.47 22.78 9.19 4.50 1.97 250.17
Table 1: Volumetric Estimation – Gas Initially In Place (GIIP) in Billion Cubic Feet.

6
Gan and Boon

Figures

Figure 1: Well logs view of the reservoir intervals in time domain showing (a) Seismic data, (b) P-Impedance (i.e. AI)
data in seismic resolution and (c) lithology log in well resolution. Note that the reservoirs consist of thinly laminated
sand and shale sequences, and if compared to the length of seismic wavelet, the beddings are essentially too small for
detection.

7
Gan and Boon

Figure 2: Cross-plot of well Density verses P-Impedance (i.e. AI) on the left (a) shows coals (black) are able to be
distinguished from gas sands (i.e. pay in red) by AI in well resolution, while the very same cross-plot on the right (b)
shows coals and are no longer distinguishable by AI after filtered into seismic resolution (0-85Hz). Also note that the
pay would be partially overlapped with wet sand (yellow) and Shale (grey) in seismic resolution compared to
resolution in well (a).

8
Gan and Boon

Figure 3: Section view of (a) Inverted Absolute P-Impedance data (i.e. AI) and its corresponding AVO Envelope
Difference (b), overlaid with well lithology log and density data. On the left (a) coal and pay share the same range of
AI. However in AVO Envelope Difference (b), Coal could be identified. The extreme negative value (colored in dark
green) shows the dimming response of coal in AVO Envelope Difference. Spikes in well density data indicate presence
of coal, which are color coded as black in the lithology log.

9
Gan and Boon

Figure 4: Cross-plot of Shale Fraction (i.e. Vshale) versus P-Impedance (i.e. AI) shows empirical relationship from
well data. For every 1000 range of AI bands, all the sampling points’ Vshale are summed and divided by the total
number of points to calculate the Arithmetic Averaged Vshale of the respected AI band.

10
Gan and Boon

Figure 5: Cross-plot of Total Pore Fraction versus P-Impedance (i.e. AI) shows an almost linear relationship from
well data (red line = best fit). Best fit’s weights are given to the red data point, which is the pay.

11
Gan and Boon

Figure 6: Schematics showing the workflow of Absolute P-Impedance (i.e. AI) Inversion through Constraint Sparse
Spike Inversion (CSSI) algorithm.

12
Gan and Boon

Figure 7: Section data showing inverted Absolute AI from seismic data QC with well AI (a), and synthetics from
inversion QC with original seismic data (b). It shows a good calibration of seismic to well, where inversion result is
consistent with well AI.

13
Gan and Boon

Figure 8: Workflow of inversion from Absolute P-Impedance (i.e. AI) to Sand Fraction and Pore Fraction through
empirical relationships. Later, these fractions could be used in volumetric estimation, depending on presumed shale
distribution models as suggested by Thomas and Stieber (1975).

14
Gan and Boon

Figure 9: Section view of Sand Fraction (a) and Pore Fraction (b) QC with well lithology. Area of high Sand Fraction
and Pore Fraction correspond to main reservoir intervals as shown in well log lithology.

15
Gan and Boon

Figure 10: Illustrations of various distribution modes of shale in reservoirs and their effect upon porosity. Modified
from Thomas & Stieber (1975).

16

View publication stats

You might also like