COALINLAS, a software for detecting coal beds in well-logs

作     者：Yusefi, Amir Ramazi, Hamidreza 

作者机构：Amirkabir Univ Technol Tehran Polytech Dept Min & Met Engn Tehran Iran 

出 版 物：《EARTH SCIENCE INFORMATICS》 (地球科学信息学)

年 卷 期：2019年第12卷第1期

页      面：129-142页

核心收录：

学科分类：07[理学] 0708[理学-地球物理学] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

主　　题：Coal Well logging Software Detection algorithm Probability density function Generalized extreme value function 

摘      要：In this paper, the algorithm and working method of the software are described, which allows the detection of depth and thickness of coal beds by digital information in well-logs. The software known as COALINLAS is designed and developed in Visual Studio using C# by the authors. In this software, a boundary value of each log is defined for detection of coal and non-coal layers, by importing the data of a borehole, called the reference borehole. Inputs of the software are the reference borehole data including digital files of values of well logs in .las format, and core-sampling data;and target functions for accepting a layer as coal. The main engine of software is an algorithm called the detection algorithm. Importing the well-logging and core-sampling data, this algorithm calculates the boundary values to separate the coal bed from the other by the frequency distribution function for values of well logs near coal beds. The frequency distribution of well logs follows the generalized extreme value (GEV) function. The location of distribution depends on the mode and the scale depends on the standard deviation and the software calculates three boundary values in conditions where the cumulative density function (CDF) is equal to 50%, 70 and 90%. The case study used to test the performance of software shows that the boundary limit calculated for CDF of 70% separates the layers more precisely. In this case study, it is concluded that the software has the ability to detect all coal beds in the boring path using well-logs data. Moreover, COALINLAS can identify fine-scale changes in the characteristics of layers and detect dispersed thin layers neglected in core-sampling.