Sadegh Baziar; Mohammad Mobin Gafoori; Seyed Mehdi Mohaimenian Pour; Majid Nabi Bidhendi; Reza Hajiani
Abstract
Klinkenberg permeability is an important parameter in tight gas reservoirs. There are conventional methods for determining it, but these methods depend on core permeability. Cores are few in number, but well logs are usually accessible for all wells and provide continuous information. In this regard, ...
Read More
Klinkenberg permeability is an important parameter in tight gas reservoirs. There are conventional methods for determining it, but these methods depend on core permeability. Cores are few in number, but well logs are usually accessible for all wells and provide continuous information. In this regard, regression methods have been used to achieve reliable relations between log readings and Klinkenberg permeability. In this work, multiple linear regression, tree boost, general regression neural network, and support vector machines have been used to predict the Klinkenberg permeability of Mesaverde tight gas sandstones located in Washakie basin. The results show that all the four methods have the acceptable capability to predict Klinkenberg permeability, but support vector machine models exhibit better results. The errors of models were measured by calculating three error indexes, namely the correlation coefficient, the average absolute error, and the standard error of the mean. The analyses of errors show that support vector machine models perform better than the other models, but there are some exceptions. Support vector machine is a relatively new intelligence method with great capabilities in regression and classification tasks. Herein, support vector machine was used to predict the Klinkenberg permeability of a tight gas reservoir and the performances of four regression techniques were compared.
Majid Bagheri; Mohammad Ali Riahi
Abstract
Seismic facies analysis (SFA) aims to classify similar seismic traces based on amplitude, phase, frequency, and other seismic attributes. SFA has proven useful in interpreting seismic data, allowing significant information on subsurface geological structures to be extracted. While facies analysis has ...
Read More
Seismic facies analysis (SFA) aims to classify similar seismic traces based on amplitude, phase, frequency, and other seismic attributes. SFA has proven useful in interpreting seismic data, allowing significant information on subsurface geological structures to be extracted. While facies analysis has been widely investigated through unsupervised-classification-based studies, there are few cases associated with supervised classification methods. In this study, we follow supervised classification scheme under classifiers, the support vector classifier (SVC), and multilayer perceptrons (MLP) to provide an opportunity for directly assessing the feasibility of different classifiers. Before choosing classifier, we evaluate extracted seismic attributes using forward feature selection (FFS) and backward feature selection (BFS) methods for logical SFA. The analyses are examined with data from an oil field in Iran, and the results are discussed in detail. The numerical relative errors associated with these two classifiers as a proxy for the robustness of SFA confirm reliable interpretations. The higher performance of SVC comparing to MLP classifier for SFA is proved in two validation steps. The results also demonstrate the power and flexibility of SVC compared with MLP for SFA.
Morteza Nouri Taleghani; Sadegh Saffarzadeh; Mina Karimi Khaledi; Ghasem Zargar
Abstract
Porosity is one of the fundamental petrophysical properties that should be evaluated for hydrocarbon bearing reservoirs. It is a vital factor in precise understanding of reservoir quality in a hydrocarbon field. Log data are exceedingly crucial information in petroleum industries, for many of hydrocarbon ...
Read More
Porosity is one of the fundamental petrophysical properties that should be evaluated for hydrocarbon bearing reservoirs. It is a vital factor in precise understanding of reservoir quality in a hydrocarbon field. Log data are exceedingly crucial information in petroleum industries, for many of hydrocarbon parameters are obtained by virtue of petrophysical data. There are three main petrophysical logging tools for the determination of porosity, namely neutron, density, and sonic well logs. Porosity can be determined by the use of each of these tools; however, a precise analysis requires a complete set of these tools. Log sets are commonly either incomplete or unreliable for many reasons (i.e. incomplete logging, measurement errors, and loss of data owing to unsuitable data storage). To overcome this drawback, in this study several intelligent systems such as fuzzy logic (FL), neural network (NN), and support vector machine are used to predict synthesized petrophysical logs including neutron, density, and sonic. To accomplish this, the petrophysical well logs data were collected from a real reservoir in one of Iran southwest oil fields. The corresponding correlation was obtained through the comparison of synthesized log values with real log values. The results showed that all intelligent systems were capable of synthesizing petrophysical well logs, but SVM had better accuracy and could be used as the most reliable method compared to the other techniques.
Mohsen Karimian; Nader Fathianpour; Jamshid Moghaddasi
Abstract
Porosity is considered as an important petrophysical parameter in characterizing reservoirs, calculating in-situ oil reserves, and production evaluation. Nowadays, using intelligent techniques has become a popular method for porosity estimation. Support vector machine (SVM) a new intelligent method with ...
Read More
Porosity is considered as an important petrophysical parameter in characterizing reservoirs, calculating in-situ oil reserves, and production evaluation. Nowadays, using intelligent techniques has become a popular method for porosity estimation. Support vector machine (SVM) a new intelligent method with a great generalization potential of modeling non-linear relationships has been introduced for both regression (support vector regression (SVR)) and classification (support vector classification (SVC)) problems. In the current study, to estimate the porosity of a carbonate reservoir in one of Iran south oil fields from well log data, the SVR model is firstly constructed; then the performance achieved is compared to that of an artificial neural network (ANN) model with a multilayer perceptron (MLP) architecture as a well-known method to account for the reliability of SVR or the possible improvement made by SVR over ANN models. The results of this study show that by considering correlation coefficient and some statistical errors the performance of the SVR model slightly improves the ANN porosity predictions.