Robust Facies Model by Combining Geostatistics and Multi-Attribute Analysis: A Case Study from

Middle Eocene Pay of Nawagam Field, Cambay Basin

Sankhadip Bhattacharya*1, Pratip Sengupta 2, Ravendra Kumar 1, T. R Joshi 2, R. K. Thakur2

Oil and Natural Gas Corporation Limited, India

(1 GEOPIC, Dehradun ONGC Ltd; 2 E&D Directorate, Dehradun ONGC Ltd)

bhattacharya_sankhadip@ongc.co.in

Keywords

Nawagam, Sequential Indicator Simulation (SIS), Seismic attributes, Geostatistics.

Summary

Geologically sound facies model is of prime

importance in building any static model. The

classical variogram based property propagation

methodology has limitations when it comes to model

facies distribution in a complex geological

environment. These techniques usually fail to capture

spatial as well as temporal heterogeneity of sinuous

fluvial channels. In order to have a good facies model

which captures heterogeneities and therefore

movement of reservoir fluids, it is necessary to

integrate several sources of information in a proper

manner. This paper discusses a useful methodology

devised to build a geologically sound facies model of

a Middle Eocene pay of fluvial origin in the

Nawagam field of Cambay Basin, through blending

of seismic attributes using classical geostatistical

approach. Facies propagation was performed using

Sequential Indicator Simulation (SIS) and seismic

attribute trend was used as an external guide to

improve channel facies definition away from well

data points. Vertical proportion curves derived from

well data captured temporal heterogeneity in the

model. Results of several realizations were used to

estimate facies model, which helped to predict

suitable stratigraphic entrapment locales based on

complex facies distribution pattern. The study was

validated by recently drilled wells in the field.

Introduction

Nawagam field is located in the central Ahmedabad-

Tarapur-Broach tectonic block of Cambay basin, one

of the most explored petroleum provinces of India.

The field was discovered in 1963. Geologically, the

area comprises two prominent intra-basinal highs,

NW–SE trending Miroli–Nawagam high and NE-SW

trending Nandej–Wasna high. Both of them are

merged to EW trending Nawagam-Wasna Ridge

which divides northern Jetalpur Low from southern

Tarapur Depression. Regional tectonic map of the

Nawagam field is shown below (Fig. - 1).

Figure 1: Regional Tectonic setting of Nawagam field,

Cambay Basin (Courtesy: ONGC Bulletin)

Nawagam

Field

Not to

scale

Robust Facies Model by Combining Geostatistics And Multi-Attribute Analysis

So far, two major hydrocarbon bearing heterogeneous

reservoirs have been established in the field, namely

Palaeocene multi-layered Lower pays of Olpad

Formation consisting mainly of trap wash, formed

during the syn-rift phase and Middle to Late Eocene

arenaceous Upper Pays of Kalol Formation,

deposited mainly in the fluvio-deltaic setting during

the post-rift phase. Recently, commercial

hydrocarbon has been discovered in Middle pay

within argillaceous Cambay Shale Formation of Late

Palaeocene to Early Eocene. Exploring for fluvio-

deltaic pays of Middle Eocene in Nawagam is really

challenging because of its complex facies distribution

pattern. The same pay sand is producing in nearby

fields like- Sanand, Jhalora, Kalol, Wadu etc. The

present paper mainly deals with an attempt to build a

robust geologically sound Facies Model of the

shallower part of the Middle Eocene Pay within

Kalol Formation combining Seismic Attributes and

Geostatistics, since variogram based classical

stochastic analysis alone cannot capture the

complexities of a fluvio-deltaic reservoir 1.

Methodology

The structural framework was built using the fault

framework, and seismic guided horizons, in depth

domain. Thereafter, a 3D geological grid with 32

layers (cell dimension: 50*50*1) within the zone of

interest was constructed using the structural

framework, which acts as the basis for geostatistical

propagation of properties, like facies, porosity,

saturation etc. (Fig. - 2).

Figure 2: Structural framework of the study area

The workflow adopted here is shown in Chart – 1.

The first step was to mark four different types of

facies, namely sandstone, siltstone, shale and coal in

all the 45 wells, among which siltstone is the

dominant reservoir facies. The log motif of a Middle

Eocene section is shown in Fig. - 3.

Figure 3: Log motif of Middle Eocene Pay

The discrete facies logs were then up-scaled using the

most suitable arithmetic average method. The process

was iteratively done till we achieved a low value of

standard deviation and variance by comparing

histograms of input raw logs and output up-scaled

logs (Fig. - 4).

Figure 4: Raw vs Upscaled facies log histogram

A flexible pixel-based stochastic modeling technique,

called Sequential Indicator Simulation (SIS) was

used for propagating 4 types of upscaled discrete

facies in the reservoir layers of the 3D geological

grid. Vertical Proportion curves were also generated

from well data to control the vertical distribution of

facies in a particular zone and seismic attribute trends

of Spectral Decomposition maps tuned for particular

Middle Eocene pays at selected frequencies (22 Hz)

were used as spatial trends, which guided the

propagation of reservoir facies away from the well

data points, within the zone (Fig. - 5). Spectral

A A’

Middle

Eocene

Pay

Fault

Model

Horizon

Model

A-A’ Sectional View of the

layer model with faults

NG-X Well

Middle Eocene

Pay top

Robust Facies Model by Combining Geostatistics And Multi-Attribute Analysis

decomposition is chosen over other attributes since

subtle stratigraphic features get tuned at particular

frequencies and can be identified easily 2.

Thus, away from well data, conditional distribution

follows the values read from the seismic trends and

close to well data, the continuity expressed through

the variogram and kriging become more dominant.

After several iterations, we observed that 50:50

weightage distributions between variogram and

seismic trend give the optimum result in this case

(Fig. - 6).

Validation of Facies Model

In the present study, Stochastic facies modeling was

adopted over deterministic approach, since it has the

advantage of generating multiple realizations based

on the input data which will all be equally probable

and can assist in better understanding of the

associated uncertainties.

A detailed, geologically sound facies model for input

to rock property modeling was thus generated after

considering several realizations, which was subjected

to extensive quality checking, like uncertainty

analysis in variogram ranges, histogram analysis,

arbitrary profiles through wells etc. and the resultant

facies maps were correlated with well data. Even to

check the quality of facies model, a few blind wells

were left and after modeling actual lithology were

checked with the modeled lithology at recently

drilled well locations (like well NG-Y) and a good

match was observed (Fig. - 7). The well has

produced commercial oil on conventional testing.

Figure 5: Input trends for facies modeling (A- Vertical

proportion curve; B- Seismic trend; see text for explanation)

Figure 6: Average reservoir facies occurrence map of

Middle Eocene Pay with producer wells (NG-Y in circle)

Figure 7: Facies Model validation

Model

Extracted Saturation

Average reservoir facies occurrences within zone

NG-Y

Well

N

GR Facies

Actual

Facies

Upscaled

Model

Extracted

Facies

Model

Extracted

Porosity

Middle

Eocene Pay

Top

A B

Robust Facies Model by Combining Geostatistics And Multi-Attribute Analysis

Chart-1: Facies modeling flowchart

Figure 8: N-S Sectional view of Facies Model

Conclusions

Heterogeneity in facies distribution influences the

flow pattern of hydrocarbon in a reservoir. 3D

geological models that account for facies variations

are more reliable since basic depositional trends are

taken into consideration. Modeling fluvial reservoir

facies is itself a challenging task using conventional

geostatistical techniques of facies modeling. The

methodology described herein helps to capture both

spatial as well as temporal heterogeneity in the

complex facies distribution pattern of fluvio-deltaic

Middle Eocene Kalol pay sand by integrating

different types of data, like- well data, seismic data,

petrophysics and geostatistics (Fig. - 8). Such

geologically sound facies model provides more

confidence in the facies biased petrophysical property

propagations. The model has already been tested by a

number of recently drilled wells as well as some

missed opportunities were identified based on

complex facies architecture of the pay zone. It

ultimately facilitated in achieving a reliable

estimation of the in-place volume of hydrocarbon,

aiding in more comprehensive and suitable

management strategies.

References

1. Ronny Meza, et. al., 2015, Combining

Geostatistics with Seismic Attributes to

Improve Reservoir Management Strategies: A

Case Study from the Faja Petrolifera del

Orinoco; WHOC15- 327, 1-14.

2. Chopra, S., 2011, Extracting meaningful

information from Seismic Attributes; p- 8,

CSEG Distinguished Lecture, Canada, 2011

Acknowledgments

Authors would like to thank Sri. A. K. Dwivedi,

Director (Exploration) of ONGC Ltd. for giving the

permission to publish this paper in 12th Biennial

International Conference and Exposition of SPG,

2017. Authors would like to express their sincere

gratitude to Sri Ashutosh Bhardwaj, ED- HOI

GEOPIC for allowing the work to publish. Authors

greatly acknowledge Sri Anand Sahu, Retired ED-

Chief E&D Directorate for continuous support and

motivation during the entire course of the study.

Authors would like to thank the respective teams

from Asset and Block for providing all the necessary

G&G inputs. Sincere thanks to Mr. K. Vasudevan,

GM (Geology) for critically reviewing the paper.

Thanks are due to all the colleagues who are directly

or indirectly involved and contributed in various

means in this study. The views expressed herein are

solely of the authors and do not necessarily reflect the

views of the organization.

Data QC and

Loading

Seismic

Attributes

Structural

Framework

Upscaling

SIS

Gridding

Facies Model Model QC

N S

NG-A

NG-B

NG-C NG-D

NG-E

NG-F NG-G