A number of new technologies introduced into the oil field over the last couple of decades now provide the hardware basis for continuous field-wide optimization (CFO). CFO will require computer integration of field hardware (e.g., downhole sensors, remotely activated completions, surface facilities) for continuous decision-making in a feedback fashion (data acquisition, data processing, actuation). We introduce a hierarchy of oil field operations that identifies various levels of detail and time-scales for decision-making processes. We propose to use this hierarchy in a multi-level / multi-scale approach to CFO. An important element in that approach is the availability of predictive models that can be used at various levels of the hierarchy, so that optimal actions can be continuously selected through optimization over a moving horizon.. In this context, artificial neural networks (ANN) are a tool that has been used to build data-driven models. This paper is structured in two parts:elements of CFO, andbrief review of ANN basics, known ANN applications in the petroleum industry, and a critical view of ANN capabilities.
Background
Oil field efficiencies from integration of diverse new technologies
After almost two decades of intensive efforts to increase its competitiveness, the oil industry has shifted its focus from depending on oil prices for its viability to using advanced technology for more efficient exploration, drilling, and production. Innovations in seismic data processing, high-angle drilling, complex well architectures, and downhole monitoring and control instruments have revolutionized the industry and are drastically lowering finding and lifting costs. Realized or projected benefits stemming from the use of new technologies include improved overall recovery; increased/accelerated production; reduced well construction costs; reduced frequency and cost of well intervention (the largest single expense to occur during the life of most producing wells); and reduced surface facilities (Drakeley and Douglas, 2002).
Spurred by the successful application of individual well and surface innovations such as the above, the oil industry is now seeking the next technological level to be realized by bringing these disparate tools under a single completion scenario known as smart wells, or intelligent completions. In fact, there is a vision of computer integrated operation and continuous optimization &control of entire fields, succinctly described by the term intelligent fields. Continuous optimization and control will require careful orchestration of hardware, software, and humans. It can capitalize on the key elements of a feedback loop which may impart intelligence to a field and which are already individually available in the oil industry, namely (a) real-time measurements; (b) downhole flow-control valves; and (c) computing/communication power and algorithms for data processing and decision making. There is also good potential to benefit from related technologies developed in other industries, such as real-time optimization and plantwide control of oil refineries and chemical plants, or completely paperless aircraft design.
Currently, data processing and decision making (item c in last paragraph) is a key area needing development (Airlie, 2002). Huge volumes of field data are being gathered, but due to lack of appropriate data management, well/completion models, and skilled people, much of this important data is not being used to the best possible effect. Appropriate data-driven modeling of systems, coupled with optimization and control - either fully automated or manual - is essential to closing the feedback loop and realizing the full potential of intelligent fields.
Oil field efficiencies from integration of diverse new technologies
After almost two decades of intensive efforts to increase its competitiveness, the oil industry has shifted its focus from depending on oil prices for its viability to using advanced technology for more efficient exploration, drilling, and production. Innovations in seismic data processing, high-angle drilling, complex well architectures, and downhole monitoring and control instruments have revolutionized the industry and are drastically lowering finding and lifting costs. Realized or projected benefits stemming from the use of new technologies include improved overall recovery; increased/accelerated production; reduced well construction costs; reduced frequency and cost of well intervention (the largest single expense to occur during the life of most producing wells); and reduced surface facilities (Drakeley and Douglas, 2002).
Spurred by the successful application of individual well and surface innovations such as the above, the oil industry is now seeking the next technological level to be realized by bringing these disparate tools under a single completion scenario known as smart wells, or intelligent completions. In fact, there is a vision of computer integrated operation and continuous optimization &control of entire fields, succinctly described by the term intelligent fields. Continuous optimization and control will require careful orchestration of hardware, software, and humans. It can capitalize on the key elements of a feedback loop which may impart intelligence to a field and which are already individually available in the oil industry, namely (a) real-time measurements; (b) downhole flow-control valves; and (c) computing/communication power and algorithms for data processing and decision making. There is also good potential to benefit from related technologies developed in other industries, such as real-time optimization and plantwide control of oil refineries and chemical plants, or completely paperless aircraft design.
Currently, data processing and decision making (item c in last paragraph) is a key area needing development (Airlie, 2002). Huge volumes of field data are being gathered, but due to lack of appropriate data management, well/completion models, and skilled people, much of this important data is not being used to the best possible effect. Appropriate data-driven modeling of systems, coupled with optimization and control - either fully automated or manual - is essential to closing the feedback loop and realizing the full potential of intelligent fields.
