An Efficient EOR Screening Approach with Statistical Pattern Recognition: Impact of Rock/Fluid Feature Selection and Extraction

Abstract

Application of secondary and tertiary recovery methods to hydrocarbon reservoirs for enhanced oil and gas extraction plays a pivotal role in petroleum industry. Therefore designing and selection of practical enhanced recovery methods become increasingly important in planning field development scenarios. EOR screening techniques are widely utilized to provide engineers with suitable recovery techniques corresponding to reservoir and fluid characteristics. A proper screening algorithm enables critical decision making on potential enhanced oil recovery strategies with the limited rock/fluid information. In this paper we employ statistical pattern recognition along with feature selection/extraction algorithms to efficiently decide on the appropriate EOR method. We implement and test novel screening techniques through the means of machine learning and pattern recognition techniques, which are well-established in computer science literature, in order to discriminate between various EOR methods. A comprehensive study is conducted utilizing various classification rules to solve EOR screening problem. Also, the effects of various rock/fluid features in EOR screening performance and outcome are studied as well. Well-known feature selection methods like data whitening, feature ranking and dimensionality reduction have been integrated to the classification process to improve the performance of the classifiers. In order to perform feature selection, some important fluid and reservoir characteristics such as permeability, depth, API, temperature, oil saturation and viscosity are taken into account. The proposed data-driven screening algorithm is a high-performance tool to select an appropriate EOR method such as steam injection, combustion, miscible injection of CO2 and N2. In this innovative approach, we integrate both theoretical screening principles such as Taber criteria and successful field EOR practices worldwide. The proposed algorithm makes it possible to integrate different types of data, eliminate arbitrary approach in making decisions, and provide accuracy and efficient computation. The suitability of the proposed method is demonstrated by different synthetic and real field EOR cases. We showed that the proposed EOR screening algorithm is able to predict the appropriate EOR method correctly in more than 90% of cases. We also ranked the proposed screening algorithms based on their screening performance. Similarly we found the most effective set of rock/fluid properties in selecting best EOR method and also reduced the number of features without losing screening performance. Moreover the proposed feature selection/extraction approaches enhanced the performance of the classifier by reducing overfitting, complexity and run time, as well as simplifying the decision rule. The proposed EOR screening algorithms are automated and intelligent tools that offer the appropriate EOR methods along with their associated probability of success. Employing rock/fluid feature selection and extraction further improves the screening performance by decreasing model complexity and also helps to better understand the underlying physical process.