IPA14-SE-118 PROCEEDINGS, INDONESIAN PETROLEUM ASSOCIATION Thirty-Eighth Annual Convention & Exhibition, May 2014 FORMULATION OF ROCK TYPE PREDICTION IN CORED WELL USING FUZZY SUBTRACTIVE CLUSTERING ALGORITHM Farizal Hakiki* Aris Tristianto Wibowo* ABSTRACT Knowledge of reservoir rock porosity and permeability are essential elements of competetent reservoir simulation and management. Vertical and horizontal heterogeneities are critical components of reservoir characterization and are among the key input parameters into three-dimensional geological and flow simulation models. A previous study based upon Keshtkar (2010) has been carried out to predict permability and rocktype of a well in a gas field in Iran using fuzzy C-means clustering method (FCM). A critical review of that paper will be presented in this paper according to the use of fuzzy C-means clustering method. By using another method of fuzzy clustering, fuzzy subtractive clustering algorithm (FS), the authors attempt to present a formula to predict carbonate rock type. The result will be compared to the results of applying fuzzy C-means clustering method. FS is used when there is no clear idea how many clusters there should be for a given data set. This method, however, strictly depends on the value of accepted ratio, rejected ratio, influence range and squash. The fuzzy subtractive method, theoretically, has the effect of reducing the number of computations significantly, making it linearly proportional to the number of input data instead of being exponentially proportional to its dimension. The result of rock type determination using this method resulted in circle clustering. Hence, this method would be suitable for heterogeneity grouping rather than equation-based trend of porosity-permeability grouping. Keywords: Rock type, Fuzzy subtractive algorithm, Carbonate rock INTRODUCTION In this paper, the authors emphasized the role of artificial intelligence for reservoir characterization. * Institute of Technology Bandung Reservoir characterization plays a crucial role in modern reservoir management. Pore distribution, a reflection of permeability distribution and environmental deposition, is the main point of discussion of this paper. Porosity and permeability values acquired form 579 cores of carbonate rock were input to the clustering program using fuzzy subtractive algorithm (Jang et al. 1997). Fuzzy clustering is designed to identify natural groupings of data wherein each data point belongs to a cluster to some degree that is specified by a membership grade. Each data point is specified by member degrees of fuzzy sets hence each data point has a possibility to be grouped into every cluster. It means the data is not absolute and can belong to more than one group. Data with most member grades shows the highest possibility to be substituted into a particular group. Fuzzy C-means clustering was devised by Dunn in 1973 then further developed by Bezdek in 1981. In 1994 it was redeveloped with a new name, Fuzzy Subtractive Clustering by Chiu. The purpose of this study is to classify types of carbonate rocks and critize the use of Fuzzy C- Means for classifying rock type and permeability prediction in a paper written by Keshtkar. Shortly, it is called rock typing. Rock typing is a process of subdividing a reservoir into groups; each has specific characteristics geologically as well as petrophysically (Archie, 1950 in Permadi et al. 2011). This study was evaluated to learn the applicability of fuzzy clustering in grouping petrophysical properties of rocks. METHODOLOGY According to Chiu, the initial clustering step with Fuzzy Subtractive Clustering begins with determining data that have higher potential than surrounding data. For instance, there are n elements © IPA, 2014 – 38th Annual Convention Proceedings, 2014