A New Approach For Improving Permeability Estimation by Integrating Sequential Gaussian Simulation (SGS) With Co-Active Neuro Fuzzy Inference System (CANFIS) Network

Abstract

This paper shows that geostatistical modeling integrated with Artificial Neural Network (ANN) can be effectively used to estimate permeability when limited amount of core and log data is available. The goal of this study is to create a good calibrated data set, and then designing a proper network for permeability estimation in one of Iranian oil reservoir. In this reservoir coring have been done only in one well. Therefore due to widely scattered and limited number of data, conventional methods of permeability estimation such as empirical equations and multi linear regressions gave poor prediction. New approach was developed in which, geostatistical modeling is performed to create comprehensive data set. After that this data set will be used to design and train neural network for prediction of permeability. In this work, at first the reservoir has been classified in to two different flow units to overcome the extreme heterogeneity of the reservoir. Then Sequential Gaussian Simulation (SGS) was performed to obtain very fine permeability distribution versus depth and log data in cored intervals of each flow unit of the reservoir. After that uncertainty analysis has been done to create data set by selecting the best points with highest probability and lowest value of standard deviation. These data were input to design and train the Co-Active Neuro Fuzzy Inference System (CANFIS) Network. This network is based on Fuzzy Logic algorithm that its parameters obtained from learning rules of neural network. Then core data are used to test the network. The results show the acceptable estimation of permeability. Finally network can effectively be used to estimate permeability in uncored but logged wells. Introduction The degree of success of many oil and gas reservoir management depends on the accuracy of the models used in a reservoir description. Permeability is an important parameter in a heterogeneous reservoir characterization. It controls the strategies involving well location, number of wells, completion and reservoir management. Permeability, therefore, is a key parameter in any reservoir characterization study that governs in great extension its handling and development. Permeability is usually measured in laboratory on core samples. However, most drilled wells are not cored. As a result, models are needed to estimate permeability in un-cored but logged wells. Estimation of permeability in un-cored but logged well is a generic problem common to all reservoirs. Many researchers applied rules of thumb developed over the years for given fields and formations to estimate rock permeability using log data in un-cored well. These rules basically stated that a relationship between porosity and permeability might be established. Petroleum engineering concepts also inspired many empirical models to estimate formation permeability from well log responses. These models established the existence of a relationship between permeability, porosity and fluid saturation. This normally requires a calibration data set that is represented by one or more key wells where comprehensive information is available in terms of core and log data. This data which is not always available in real cases, are used to build the predictor and to test the reliability of the results. Preparing good data set is one the fundamental challenges of reservoir engineers ever since the beginning of oil industry.

Performance improvement in vertical heterogeneous handoff methodology using CANFIS classification approaches

“…Figure shows the classification stages in CANFIS approach, which consists of five layers . The extracted features from normal BS and malicious affected BS are given to X and Y node of this CANFIS classifier.…”

“…Figure 4 shows the classification stages in CANFIS approach, 18 which consists of five layers. 19 The extracted features from normal BS and malicious affected BS are given to X and Y node of this CANFIS classifier. In this paper, "mamdani" fuzzy rules with "Triangular" membership functions (MF) are used for CANFIS classification design.…”

Production-System Optimization of Gas Fields Using Hybrid Fuzzy/Genetic Approach

“…Recently, fuzzy-logic systems have been used to solve a number of different problems in a variety of areas, including physical and chemical systems, production planning and scheduling, location and transportation problems, resource allocation in financial systems, and engineering design. In the petroleum industry, fuzzy logic has been put to work in many different areas since the early 1990s, including reservoir characterization (Hambalek and Gonzalez 2003;Lim and Kim 2004;Rafiei et al 2009;Shokir 2006;Soto et al 2001;Taghavi 2005;Zhou et al 1993;Cuddy 2000;Kedzierski and Mallet 2006;Finol et al 2002;Hajizadeh 2007a), optimal well operations (Alimonti and Falcone 2004; Lababidi 2003, 2005;Rivera 1994;Dumans 1995;Xiong et al 2001), stimulation treatment (Xiong and Holditch 1995;Nitters et al 2000), and economic analysis (Zolotukhin 2000;de Salvo Castro and Fereira Filho 2001;Agbon and Araque 2003;Chang et al 2006). An extensive list of such applications is presented in Table 1.…”