Development of Hybrid Intelligent System for Virtual Flow Metering in Production Wells

Abstract

Abstract The demands placed on operators to effectively manage production operations become more complex and challenging with increasing numbers of high recoverable oil and gas fields in frontier environments (remote onshore, deepwater and ultra- deepwater). Predicting flow rates of wells located in these frontier environments accurately is of great importance to enable asset performance monitoring, field surveillance, production accounting, production optimization, reservoir management decision, volumetric input to reservoir simulators and reservoir estimate tracking, though out the field life cycle. Although several studies have focused on real-time well rate estimation using physical multiphase flow meters, theoretical approach (mass, momentum and energy coupled with real-time field data) and artificial neural network models, little attention has been given to continuous flow rate surveillance based on hybrid intelligent systems. Hybrid intelligent systems combine intelligent techniques synergistic architecture in order to provide solution for complex problems. These systems utilize at least two of the three techniques: fuzzy logic, genetic algorithm and artificial neural networks. The goal of their combination is to amplify their strengths and compliment their weaknesses. This paper presents a novel approach of using hybrid intelligent modeling technique, available time series field data and well configuration information to develop a virtual flow meter for production wells. The simulation results from the hybrid intelligent based virtual flow rate meter are compared to those obtained from real life field data. The validated model is used to predict future performance of existing wells. The effects of various parameters are performed to determine their impacts on the predictive accuracy of the new approach.