Available online at https://ijcpe.edu.iq and www.iasj.net

Iraqi Journal of Chemical and Petroleum
Engineering

Vol.19 No.2 (June 2018) 9 – 13
ISSN: 1997-4884

Corresponding Authors: Wafa Al Kattan , Email: NA, Sameer N. AL Jawad, Email: d.sameer@yahoo.com, Haider Ashour Jomaah, Email:
haiderashour@yahoo.com
IJCPE is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License

Cluster Analysis Approach to Identify Rock Type in Tertiary

Reservoir of Khabaz Oil Field Case Study

Wafa Al Kattan
a
, Sameer N. AL Jawad

b
and Haider Ashour Jomaah

a
Petroleum Engineering Department/ College of Engineering/ University of Baghdad

b
Ministry of Oil/ Department of Reservoir and Field Development

c College of Engineering/ University of Baghdad

Abstract

Rock type identification is very important task in Reservoir characterization in order to constrict robust reservoir

models. There are several approaches have been introduced to define the rock type in reservoirs and each approach

should relate the geological and petrophysical properties, such that each rock type is proportional to a unique hydraulic

flow unit. A hydraulic flow unit is a reservoir zone that is laterally and vertically has similar flow and bedding

characteristics. According to effect of rock type in reservoir performance, many empirical and statistical approaches

introduced. In this paper Cluster Analysis technique is used to identify the rock groups in tertiary reservoir for Khabaz

oil field by analyses variation of petrophysical properties data that predicted by analysis of well log measurements. In

tertiary reservoir four groups identified by cluster analysis technique, were each group was internally similar in

petrophysical properties and different from others groups.

Keywords: Rock identification, Khabaz oil field

Accepted on 23/5/2016

1- Introduction

Cluster analysis is a multivariate approaches which aims

to distribute a sample of subjects of a set variable

measured into a different number of groups where similar

subjects are placed in the same group ‎[1]. Well log cluster

analysis is process aim to look for similarities and

dissimilarities between data points in the multivariate

space of logs, in order to distribute them into classes

called electrofacies Suzan et al 2010. An electrofacies is a

unique set of log responses that characterizes the rocks

physical properties and fluids contained in the volume

that investigated by the logging tools ‎[2]. There are

different methods that can be used to make a cluster

analysis; these methods can be classified as follows ‎[1]

1.1. Hierarchical Approach

In this approach there are different methods which

clusters should be joined at each stage. The main methods

are summarized as:

a. Nearest neighbor method in this method the distance
between two clusters between the two closest

members, or neighbors.

b. Furthest neighbor method In this case the distance
between two clusters is defined to be the maximum

distance between members.

c. Average method the distance between two clusters is
calculated as the average distance between all pairs of

subjects in the two clusters.

The distance between two subjects can be measured by

Euclidean distance as following ‎[1].

√∑

(1)

1.2. Non-Hierarchical or K-means Clustering Methods

In these methods desired in advance the number of

clusters to specify and the best solution is chosen. When

large data sets are involved, Non-hierarchical cluster

analysis tends to be used it allows subjects to move from

one cluster to another so, sometimes preferred because

this isn't possible in hierarchical cluster analysis. There

are two disadvantages of k-mean cluster analysis first

know how many clusters likely to have often difficult and

therefore the analysis may have to be repeated several

times and second it very sensitive to the choice of initial

cluster ‎[1].

The clustering proses based on two stages. Firstly, the

well log data is classified into manageable data clusters so

that the number of clusters should be enough to cover all

the different data ranges that can be detect on the logs

W. Al Kattan, S. N. AL Jawad

and H. A. Jomaah / Iraqi Journal of Chemical and Petroleum Engineering91,2 (2018) 9-

data .the reasonable number of clusters for most data sets

are between 15 to 20 clusters.

The second step based on takes these 15 to 20 clusters

and group them into a manageable number of rocks types

and reducing the data to 4 to 5 homogenous groups ‎[3].

2- Main Section

Khabaz oil field is one of Iraqi field with multiple pay

zones similar to most of the northern Iraqi carbonate oil

fields. It's located to North West of Kirkuk city and far

away about 12 km from Kirkuk city center ‎[4].

This paper is developed by depending on data from

eleven wells was selected for this study. Interactive

petrophysics program 3.5 used to apply cluster analysis in

order identify rock type for tertiary reservoir in Khabaz

field.

Sonic (DT), bulk density (RHOB), water saturation

(Sw), and effective porosity (PHIE) and predicted

permeability(K) logs for the eleven studied wells, were

used as input data for cluster analysis, 20 clusters assume

to cover all data variation.

K-mean statistical technique used to seed input data in

to given clusters by assume initial guess mean value for

each cluster for each input loge data and then try to

minimize sum of squares deference within cluster

between data points and cluster mean value.

The 20 cluster consolidated by hierarchal technique

which based on compute distance between clusters and

merger two closest clusters in distance then return

compute the distance between new clusters and re-merge

the two new closest clusters .the processes done until all

clusters merged in one cluster.

Consolidating of the clusters into a known number of

rock types is easily indicated by Cluster Randomness Plot

done by plot random thickness index vs. number of

clusters.

The randomness is performed on original logs data by

calculating the average number of depth levels per cluster

which represent the average cluster thickness layer.

Then the theoretical average random thickness is

determined by assuming the clusters to be assigned

randomly at each depth level.

The randomness index represents the ratio of average

cluster thickness to average random cluster thickness ‎[3].

Av. Thickness = Number of depth levels / Number of cluster layers (2)

– (3)

Where, pi is the proportion of depth levels assigned to the

i-th cluster.

Randomness index = Av. Thickness / Random Thickness (4)

Randomness plot for tertiary reservoir shows four groups

can be depend as rocks types by collecting the number

heights peaks as shown in the Fig. 1.

Fig. 1. Cluster group’s randomness for tertiary reservoir

in Khabaz field

The hierarchal technique shows merging process of rock

type in groups distinguished by different colors explained

in tree- diagram as shown in Fig. 2.

Fig. 2. Cluster grouping tree diagram for Tertiary

Reservoir in Khabaz Field

Table 1. Cluster analysis results for each Rock type

CLUSTERS PHIE SW DT K RHOBC

cluster Groups Points Mean Std. Dev. Mean Std. Dev. Mean Std. Dev. Mean Std. Dev. Mean Std. Dev.

1 1 343 0.24 0.03 0.35 0.21 74.30 5.17 296.64 61.22 2.29 0.08

2 1 421 0.22 0.02 0.16 0.09 77.19 4.11 182.19 34.16 2.31 0.04

3 1 640 0.20 0.03 0.36 0.16 69.13 2.88 147.90 31.49 2.40 0.04

4 1 677 0.18 0.02 0.20 0.10 72.66 3.62 85.00 23.52 2.39 0.05

5 2 696 0.15 0.03 0.22 0.10 68.25 2.86 13.98 11.98 2.45 0.03

6 3 194 0.19 0.04 0.86 0.14 74.13 4.43 45.79 41.95 2.44 0.06

7 2 586 0.08 0.02 0.40 0.14 65.55 2.80 2.28 4.01 2.52 0.04

8 2 678 0.17 0.04 0.51 0.10 66.05 2.59 22.18 20.60 2.49 0.04

9 2 437 0.14 0.04 0.22 0.11 58.88 3.46 23.20 20.54 2.46 0.06

10 3 585 0.16 0.03 0.88 0.11 67.15 2.86 9.44 9.61 2.45 0.04

11 3 517 0.07 0.03 0.95 0.09 68.69 3.14 0.49 1.09 2.55 0.05

12 2 705 0.12 0.02 0.50 0.14 60.38 2.16 7.82 9.43 2.56 0.03

13 2 692 0.06 0.02 0.42 0.15 55.64 2.74 0.80 3.07 2.60 0.05

14 3 940 0.12 0.02 0.95 0.08 63.97 2.28 4.21 5.22 2.53 0.03

15 4 1134 0.08 0.02 0.98 0.06 61.45 1.61 0.38 1.20 2.60 0.03

16 4 615 0.02 0.02 0.99 0.06 62.04 2.43 0.01 0.01 2.63 0.05

17 4 810 0.08 0.02 0.95 0.09 56.82 1.92 0.48 2.19 2.58 0.04

18 4 1478 0.05 0.02 0.99 0.05 57.92 1.58 0.03 0.07 2.65 0.03

19 4 816 0.02 0.02 0.98 0.07 53.06 1.97 0.02 0.07 2.68 0.04

20 4 207 0.00 0.02 1.00 0.00 54.65 3.28 0.01 0.01 2.83 0.05

K-Mean Clusters Results

W. Al Kattan, S. N. AL Jawad

and H. A. Jomaah / Iraqi Journal of Chemical and Petroleum Engineering91,2 (2018) 9-

The quality of four rocks types identified from k-mean

values of petrophysical properties which are used as input

parameter for cluster analysis that tabulated in the Table

According to k-mean values for each cluster within rock

type groups each group given grad as:-

Best quality rock type
Good quality rock type

Moderate quality rock type

Bad quality rock type

Final graphical of cluster analysis for selected wells in

this paper explained in the Fig. 3

Fig. 3. The final graphical result of clustering analysis

W. Al Kattan, S. N. AL Jawad

and H. A. Jomaah / Iraqi Journal of Chemical and Petroleum Engineering91,2 (2018) 9-

Fig. 4. Rock type for Kz-2 by cluster analysis method

Fig. 5. Rock type for Kz-3 by cluster analysis method

KZ-3Scale : 1 : 1300

DEPTH (2230.05M - 2460.M) 04/08/2015 18:28DB : IP (3)

Depth

DEPTH

(M)

Tops

Saturation

SW ((%))
1. 0.

Porosity

PHI ((%))
0.5 0.

Rock

Rock Type 1

Rock Type 2

Rock Type3

Rock Type 4

Matrix

PHI ((%))
1. 0.

Clay

Porosity

Matrix

2250

2275

2300

2325

2350

2375

2400

2425

2450

J
e
r
ib

e
U

n
it

U
n
it

"
U

n
it

U
n
it

E
U

n
it

W. Al Kattan, S. N. AL Jawad

and H. A. Jomaah / Iraqi Journal of Chemical and Petroleum Engineering91,2 (2018) 9-

3- Conclusion

Cluster analysis is a simple method can be used to

identify rock type fore reservoir depending on log data.

1- Cluster analysis of log data for wells that penetrated
tertiary reservoir shows that the tertiary reservoir can

be divided in four groups as shown in roundness

plot.

2- Plotting rock type in continuous form in selected
wells shows that the unit B is the most interested

zone in tertiary reservoir for Khabaz oil field.

References

[1] Rosie Cornish, Statistics Cluster Analysis,
Mathematic Learning Support Center, 2007.

[2] T. Euzen, Well log cluster analyses and electrofacies
classification: probabilistic approach for integration

log with mineralogical data. GeoConvention 2012,

[3] Interactive Petrophysics IP- V3.5 User Manual 2008.
[4] Integrated Reservoir Study for Tertiary Reservoir in

Khabaz field 1989, North Oil Company – Ministry of

Oil.

http://www.searchanddiscovery.com/pdfz/documents/2014/41277euzen/ndx_euzen.pdf.html
http://www.searchanddiscovery.com/pdfz/documents/2014/41277euzen/ndx_euzen.pdf.html
http://www.searchanddiscovery.com/pdfz/documents/2014/41277euzen/ndx_euzen.pdf.html