Keywords: sensors; instruments and actuators; control systems; control strategies
In early January 2004, NASA's Mars rovers Spirit and Opportunity arrived at Mars — some 70 million km away at closest approach — after a journey of over half a year, landing under fully autonomous control. The world marvelled at the automation of their flight and landing control, a new milestone for process automation. Clearly the technology matters not only to daily life but to the exploration of space. This article reviews the development of process automation and its trends.
What is process automation? Standard DL/T 701-1999 (Thermal automation terminology for fossil-fired power plants) defines it concisely: the use of detection and control systems to carry out production in place of direct manual operation.
Like automation elsewhere, process automation hardware consists mainly of three parts: sensors and instruments (including display), control devices or systems, and actuators (actuating plus regulating mechanisms). Notably, process automation usually works in harsh, hazardous plants where interruptions are costly, so it demands high anti-interference, reliability and availability — clearly unlike building automation.
Sensors/instruments and actuators are the foundation; sensors have evolved through mechanical, electronic and microprocessor stages. With rapid advances in microelectronics, MEMS, intelligence and integration, plus new materials and processes, sensors are developing toward miniaturisation, digitalisation, intelligence, networking and virtualisation. Endress+Hauser's ceramic-capacitance pressure sensor, for example, is a dry sensor without transfer fluid, measuring 0–60 MPa with stable performance, annual drift below 0.1% F.S and overload capacity of hundreds of times full scale.
If sensors are an automation system's sense organs, actuators are its limbs: they receive control signals and change the manipulated variable so the process runs as required. With automation, electronics and computing advancing, actuators too are becoming intelligent — more and more carry communication and smart-control functions. Emerson, Smar and others offer smart valve positioners with highly integrated microprocessors, replacing force balance with digital balance, converting electric commands into pneumatic positioning increments, configurable for dead band, direct/reverse action and travel range, supporting split-range, equal-percentage and quick-opening corrections, with self-tuning and self-diagnosis.
Control devices are the nerve centre. From the early field-mounted controller (transmitter-indicator-controller in one), through pneumatic and electric modular instruments and direct digital control (DDC), to today's ubiquitous DCS and PLC: DCS passed through infancy (1975–80), maturity (1980–85) and expansion (after 1985), advancing in reliability and maintainability, richness of control algorithms, information-processing power and speed, configuration software usability, networking and openness — becoming the mainstream of process control. PLCs, compact, focused, fast, reliable and cheap, spread just as quickly as the other mainstream. PLC-based DCS is developing fast; PLC and DCS interpenetrate, merge and compete — the current trend. With improving PC reliability and price-performance and unmatched openness, integration, product/support and market share, PC-based industrial control is thriving. And the newer fieldbus control system (FCS) shows strong vitality through excellent interoperability and function distribution, more powerful system functions (multivariable I/O per device, networked device management, wider diagnostics, richer status), easy installation and configuration, higher measurement/control accuracy, lower engineering and maintenance cost, and scalability.
Recently the big vendors have been launching larger, higher-level total automation architectures — Emerson's PlantWeb, Honeywell's TotalPlant, Siemens' TIA — collectively called Process Automation Solutions (PAS). These are no longer mere integration of hardware and software (OS, middleware, applications) but include services and even information integration from field to enterprise. In short, control systems are heading toward networking, intelligence, integration, distribution, informatisation and openness.
Control methods and strategies are the soul of process automation. Since the late 20th century the main direction has been applying artificial intelligence. Methods have grown from classical control (PID, ratio, cascade, feedforward) to optimal, adaptive and self-tuning, self-learning, nonlinear, hierarchical intelligent, expert, fuzzy-logic, neural-network, human-imitating, pattern-recognition-based, multi-mode variable-structure, chaos, robust, extenics-based intelligent control, H∞ control and μ-synthesis. Large turbo-generators, for instance, are very complex control objects with long dead time, strong coupling, essential nonlinearity and much unknown disturbance, making boiler combustion control, mill control and main/reheat steam temperature control over wide load ranges hard for classical strategies. Deep research has followed, and advanced theories are entering practice: ABB and Sulzer built state-observer SCO models for main and reheat steam temperature; Siemens created controlled condensate throttling (COT) models and fuzzy NUC (new coordinated unit control), producing optimised schemes for different units and conditions already showing clear results in domestic plants. Germany's Krupp Hoesch Westfaien steel works improved its models with neural networks for notable gains — product dimensional deviations down 12%. Vendors also ship commercial intelligent controllers, e.g. CyboSoft's model-free adaptive (MFA) controllers Cybocon and Cybocon CE with algorithms (standard, inverse-delay, nonlinear MFA, robust MFA) per process, considerably improving control. As global competition intensifies, advanced control strategies will spread ever wider; combining methods in service of control objectives is the effective path to advanced control systems.
In summary, the mainstream trends of process automation are intelligent detection and control, digitalised measurement information, and integrated control-management. New process control systems must keep improving performance to strengthen enterprises' competitiveness and create good social and economic returns.
