Role of the Cluster Analysis in Logfacies and Depositional Environments Recognition from Well Log Response for Mishrif Formation in Southeast Iraq

Text-only Preview

International Journal of Engineering Research & Science (IJOER) ISSN: [2395-6992] [Vol-3, Issue-12, December- 2017]
Page | 35
Role of the Cluster Analysis in Logfacies and Depositional
Environments Recognition from Well Log Response for Mishrif
Formation in Southeast Iraq.
Jawad K. Radhy AlBahadily1*, Medhat E. Nasser2
Department of geology, college of science, Baghdad University, Baghdad, Iraq.
*Corresponding Author Email: [email protected]
AbstractThe recognition of depositional environments from well logs is based on the principle that well log responses are
related to changes in thickness, texture, grain size and litholog y along the well path. The variations in sedimentary rock
response are due to reservoir heterogeneity. A similar set of log responses that characterizes a specific rock type and allows
it to be distinguished from others, have defined an electrofacies.Cluster analysis includes a broa d suite of techniques
designed to find groups of similar items within a data set. Partitioning methods divide the data set in to a number of groups
predesignated by the user. Hierarchical cluster methods produce a hierarchy of clusters from small clusters of very similar
items to large clusters that include more dissimilar items. Hierarchical methods usually produce a graphical output kno wn
as a dendrogram or tree that shows this hierarchical clustering structure.The method has tested on a 398.5 m thick interval
of Carbonate deposits in a vertical well from Amara field, located in southeast Iraq. Modal data have collected from b oth
core and cutting samples. Cluster analysis and electrofacies classification have per formed using advan ce interpretation in
Interactive Petrop hysics software version 3.6. Carbonate microfacies and marine depositional environments studied for
Mishrif Formation depended on the available thin sections though they were not enough to cover all the depositional
environments of Mishrif Formation. Therefore, previous studies and well logs were also depended in this study.Correlation
of determined logfacies with those defined from cores and cuttings is fundamental to ch eck the reliability of used met hods
and to define a meaningful cut off level for wells from which no cores or cuttings are available.
KeywordsCluster analysis, Mishrif Formation, Logfacies, Well Logs response,
I. INTRODUCTION
The identification of depositional environments from well logs is based on the principle that well log responses are related to
changes in thickness, texture, grain size a nd lithology along the well path. The variations in sedimentar y rock response ar e
due to reservoir heterogeneity. The description of a rock in terms of its type, origin, and depositional environment is called a
Lithofacies description. This can be done by direct o bservation of the rocks or inferred from analysis and interpretation of
well log data. Determining lithofacies from well logs requires calibration to known rocks (cores, samples, or outcrops).
Understanding the rock facies is the only way to reconstruct the paleogeography of a lithologic succession, rock sequence,
which in turn provides clues as to a potential reservoir quality and lateral extent(Crain, d. g.).
The purpose of well log cluster analysis is to look for similarities/dissimilarities between data points in the multivariate space
of logs, in order to group them into classes also called electrofacies(Euzen, T., Delamaide, E., Feuchtwanger, T., &
Kingsmith, 2010).
An electrofacies is defined by a similar set of log responses that characterizes a specific rock type and allows it to be
distinguished from others. Electrofacies are obviously influenced by geology and often can be assigned to one or another
lithofacies, although the correspondence is not universal(Doveton, J. H., & Prensky, 1992).
International Journal of Engineering Research & Science (IJOER) ISSN: [2395-6992] [Vol-3, Issue-12, December- 2017]
Page | 36
Electrofacies are primarily observational in nature, and the classification procedure is based on three steps: principal
component analysis, cluster analysis, and discriminant analysis. Principal co mponent analysis is used to summarize the data
and to reduce the dimensionality of the data without any significant loss of information. The method displays the data as a
function of new variables, called principal components, that are simple linear combination of the well logs. The aim of
cluster analysis is to classify the well-log data into groups that are internally homogeneous and externally isolated on the
basis of a measure of similarity or dissimilarity between t he groups. The clusters define elec trofacies on the basis of the
unique c haracteristics of well-log measurements reflecting minerals and lithofacies within the loggedinterval.Once the
electrofacies are identified, we can use discrimi- nant analysis, a multivariate statistical method, to assign an individual
observation vector to one of the predefined electrofacies(Perez, Datta-Gupta, & Mishra, 2005).
II. BASIC CONCEPTS
Cluster analysis includes a broad suite of techniques designed to find groups of similar items within a data set. Partitioning
methods divide the data set into a number of groups predesignated by the user. Hierarchical cluster methods produce a
hierarchy of clusters from small clu sters of very similar items to large clusters that include more dissimilar items.
Hierarchical methods u sually produce a graphical output known a s a dendro gram or tree that shows this hierarchical
clustering structure. Some hierarchical methods are divisive, that progressively divide the one large cluster comprising all of
the data into two smaller clusters and repeat this process until all clusters have been divided. Other hierarchical methods are
agglomerative and work in the opposite direction by first finding th e clusters of the most similar items and progressively
adding less similar items until all items have been included into a single large cluster. Cluster analysis can be run in the Q-
mode in which clusters of samples are sought or in the R-mode, where clusters of variables are desired(Holland, 2006)
The clusters define electrofacies on the basis of the unique characteristics of well -log measurements reflecting minerals and
lithofacies within the logged interval. Once the e lectrofacies are identified, we can use discriminant analysis, a multivariate
statistical method, to assign an individual observation vector to one of the predefined electrofacies (Perez et al., 2005).
The Cluster Analysis module uses standar d statistical routines to allow the user to cluster the data into groups to produce a n
electrical facies log. This log can then hopefully be used to correlate to geological facies (Senergy, 2008) .
III. METHODOLOGY
The theory of Cluster Analysis is the module works in two stages. Firstly, the data is divided up into manageabl e data
clusters. The number of clusters should be enough to cover all the different data ranges seen on the logs.15 to 20 clusters
would appear to be a reasonable number for most data sets. The second step, which is more manual, is to take these 15 to 20
clusters and group them into a manageable number of geological facies. This may involve reducing the data to 4 to 5 clusters.
(Stage-1 K-mean clustering) The first stage of Facie s Clustering uses the K-mean statistical technique to cluster the data into
a known entered number of clusters. For this to work an initial guess has to be made of the mean value of each cluster for
each input log. The initial guess can affect the results and in order to get good results the initial values should cover the total
range of the logs. K-mean clustering works by assigning each input data point to a cluster. The routine tries to minimize the
within-cluster sums of squares of the difference between the data point and the cluster mean value. The routine works by
calculating the sum of the squares difference for a data point and each cluster mean and assigning the point to the cluster w ith
the minimum difference. Once all the data points have been assigned to the clusters the new mean values in each cluster are
calculated. Using the new mean values the routines starts again re-assigning the data to the clusters. This loop continues until
the mean values do not change between loops. These then become the results.
International Journal of Engineering Research & Science (IJOER) ISSN: [2395-6992] [Vol-3, Issue-12, December- 2017]
Page | 37
FIGURE (1): CROSSPLOTS AND HISTOGRAMS BETWEEN (RHOB, NPHI, DT, GR, LLD AND MSFL)AS
GENERATED BY K-MEANS CLUSTER ANALYSIS FOR GROUPS OF MISHRIF FORMATION
(Stage-2 Cluster Consolidation) Cluster consolidation can be done completely manually by using the crossplot and log plot
output to group data, or a hierarchical cluster technique can be used to group the data. Hierarchical clustering works by
computing the distances between all clusters and then merging the two clusters closest together. The new cluster distance to
all other clusters is then recomputed and the two closest clusters merged again. This process continues until you have only
one cluster. The results can be plotted as a dendrogram, which IP displays. The dendrogra m shows how the clusters were
merged and the order they were merged (Senergy, 2010).
FIGURE (2): CLUSTER GROUPING DENDROGRAM OF MISHRIF FORMATION
Cluster grouping dendrogram