Structuring an Integrative Approach for Field Development Planning Using Artificial Intelligence and its Application to an Offshore Oilfield

Abstract

In early stages of reservoir depletion, it is often a challenging task to accurately determine reservoir properties that are representative of the actual field. Due to different scales of data obtained from various sources like seismic data, well logs, cores, and production data, there is a lot of uncertainty in solving the inverse problem of estimating formation rock and fluid properties from the field data. Hard-computing protocols like reservoir simulation are time and labor intensive. The objective of the current study is to develop a reservoir characterization tool using a novel approach of correlating seismic attributes with well logs and production data using artificial intelligence approach. The tool will enable construction of spatial oil maps at different times revealing sweet spots and aid in optimized field development planning. A workflow is developed for devising a comprehensive reservoir characterization tool based on artificial expert systems. A case study of an offshore deep-water asset is used in demonstrating the tenets of the workflow. The reservoir under consideration is highly heterogeneous in terms of property distribution and is believed to be highly channelized. The ANN based tool assists in identifying sweet spots by predicting optimal well location/completion parameters and production profiles. The multilayer feedforward back-propagation based neural network tool developed is able to capture the correlations that exist amongst seismic data, well logs, completion data, and production data. Well logs are correlated to seismic attributes and geometric location of wells with an average testing (blind test) error of less than 20%. Having correlated seismic data with well logs, synthetic well logs are generated for the entire area of seismic coverage. Synthetic well logs combined with seismic data are able to correlate well with the production within 21% error. The tool developed enables users to predict entire well log suites for even a directional well of user defined configuration through a graphic user interface in a short period of time (typically less than a minute). This methodology uses a unique way of computing seismic attributes following a horizontal well path and correlating them with the suite of well logs. Incorporation of interference effect from neighboring producers and injectors, schedule of production and functional links based on geographic location has made the production performance module robust and reliable. The workflow enables generation of oil production forecast maps through production performance network. NPV (net present value) calculations integrated with production forecasts is used in identifying the potential infill well locations. The results discussed in the paper showcase the robust nature of the methodology.

Artificial Intelligence Applications in Reservoir Engineering: A Status Check

“…A high-resolution production potential surface map can be generated to visualize the sweet spot of gas recovery, which guides the selection of infill drilling locations. Later, Bansal et al [31], Ketineni et al [32] and Ozdemir [33] have extended the workflow to characterize onshore and offshore oil reservoirs. Along with seismic survey data, various types of well log data are also utilized as the input of the AI model to predict the well production performance.…”

Practical CO2—WAG Field Operational Designs Using Hybrid Numerical-Machine-Learning Approaches

“…The machine-learning technologies exhibit strong competences to solve a large spectrum of petroleum engineering problems, including sweet spot identification [15], history matching [16], fluid property characterization [17], and field development strategy optimization [18]. In the field of reservoir simulation, the machine-learning models comprehend the fluid transportation dynamics in porous media via learning the data structure presented by a knowledge base instead of solving the partial differential equations using numerical and analytical methods [19].…”

Acceleration of the NVT Flash Calculation for Multicomponent Mixtures Using Deep Neural Network Models

“…Since 2012, due to the breakthrough of AlexNet [22], deep learning, a subset of machine learning, made a revolutionary progress in image recognition, language processing, artificial intelligence, and so forth. The great potential of machine learning has drawn close attention from the petroleum industry, and remarkable efforts have been made in many aspects, such as reservoir characterization [23], development planning optimization [21] and permeability estimation [1]. Furthermore, a number of researches have been devoted to speed up both phase stability test and phase splitting calculation using the machine learning method.…”