Experience with Using Data Analysis Technologies in Identification of Lost Production Zones

Abstract

Abstract Recent reserves depletion and increasing complexity of methods of development efficiency improvement and production stimulation requires solving the problem of "lost sections" of oil-saturated reservoirs. One of the approaches of lost production section search is based on comprehensive analysis of well logging and core data and automation of multi-well analysis of this data. In this case the process of search of lost sections includes both stages of well logging and core data interpretation (including initial work) and analysis of data interpretation results and rating of candidate wells for geological and engineering operations. During initial work and well logging interpretation, specialist has to perform routine tasks that take significant time. For example, such tasks include: initial unification of well logging and core data, depth matching of well logs (single- an multi-type), calibration and splicing of well logs, depth matching of well logging and core data, self-consistent with well logging and core data adaptation of parameters of petrophysical functions and interpretation models, determination of stable correlation "core-log" relationship for complex reservoirs, etc. Most of the tasks listed above involve data analysis. This paper reviews how to automate performance of routine tasks arising during lost section search. Widely used technologies of data analysis (Data Mining), that are first of all related to tasks of optimization, relationship determination, pattern identification and classification, are to some extent applied during geological modelling. For instance, artificial neural networks, decision trees, Bayesian networks, genetic algorithms, etc. can be used for analysis of geological and geophysical information. Statistical methods such as correlation and regression analysis are often referred to data analysis methods and are applied as well to geological and geophysical information analysis. However, specific research area set its own limits and requirements on the use of tools related to Data Mining. These are the examples of problems causing unsatisfactory outcome when direct using of Data Mining tool and technologies: –the data does not correlate with each other (in terms of depth), i.e. it is necessary to perform depth matching before data analysis;–training set is inconsistent;–inverse or optimization problem does not have unique solution. The article discusses the problem of application of data analysis methods and tools to geological and geophysical tasks under presence of uncertainties. Methods of preparing and pre-processing of well data at various stages are presented. There are introduced the results of using algorithms of data processing and methods of data analysis developed and adapted for particular application tasks.