As long as there have been hydrocarbon processing industry (HPI) facilities to process crude oil and intermediates, there have been instruments in place to assist plant operators in measuring, recording and controling pressures, flows, levels, temperatures and other process variables. Initially, these were (by todays standards, at least) crude Rube Goldberg-like instruments utilizing ingenious mechanical and/or pneumatic mechanisms.
In the early petroleum refineries and petrochemical plants, many of the control concepts conceived during the Industrial Revolution of the 18th and 19th centuries were further developed, refined and proven. It would be exceedingly difficult, if not impossible, to operate a typical present-day refinery or petrochemical plant without good, closed-loop process control. The continuous, generally steady-state nature of processing liquid petroleum feedstocks and intermediates lends itself to closed-loop feedback control, using a standard set of measurement, control and final control/actuation instruments. However, with so many process variables, interactions and nonlinearities involved, the process can overwhelm the human mind. Additional challenges include more complex processes and plants, less uniform feedstocks, increasing use of unconventional feedstocks, variable energy costs, and an increasingly difficult regulatory environment. These further increase the reliance on automation and help explain the universal acceptance of sophisticated process control and automation systems in todays plants.
Anatomy of control
While it hasnt always been the case, process control isfor the most partan exacting science made possible by continuous advancements in control theory, processing technologies and process-control instrumentation and systems. These developments enable not just individual control loops, but entire units, plants, and even integrated petrochemical complexes to be operated in a close-to-optimum manner (with optimum determined by product cost, quality, yields, throughput and so on). Rather than those quaint, if ingenious, mechanical and pneumatic instruments utilizing basic feedback control, current process measurement and control in HPI plants are performed by networked field instrumentation with more onboard intelligence than early mainframe computers and computer-based process automation systems that are many times more powerful and capable than the NASA control centers that sent the first men to the moon. This article will briefly trace how we got from point A to point B.
The early years
Modern process control instrumentation evolved from the basic instruments and devices developed to control prime movers, such as James Watts steam engine, during the Industrial Revolution. In the 1850s, following the revelation that crude oil could be refined via distillation into kerosine for lighting purposes, which ended the whale-oil industry, the first petroleum refineries were constructed in Europe and the US. Sensing the opportunity, a handful of instrumentation companies (Honeywell, Fisher, Foxboro, Bailey, Bristol, Taylor, Brown, etc.) began to adapt their temperature, level and pressure gauges, and pen-based, mechanical circular chart recorders to meet the basic measurement and control needs of the early refineries. During this industrial period, control was purely manual, with field operators monitoring the gauges, taking notes and making any needed process adjustments by manually opening or throttling valves. Often, this meant the operator had to move around quite a bit, including climbing up to places that are restricted under current OSHA rules.
In the late 1890s and early 1900s, refineries began to implement automated feedback control using a combination of direct-connected, pneumatically operated instrumentation utilizing ingenious combinations of nozzles, flapper valves, bellows, springs and other mechanismsall powered by compressed air. This provided a reasonable degree of on-off control. Separate indicators and chart recorders were often used to provide the human interface and record-keeping functionalities. By combining basic mechanical pressure, level, flow and temperature measurement instrumentation with field-mounted pneumatic controllers and actuator-driven valves, closed-loop feedback control became possible. Fisher Controls (now part of Emerson Process Management) introduced its field-mounted Wizard pneumatic controller in 1930, as shown in Fig. 1A.
Fig. 1A. Indicating/recording pressure gauge
circa 1908. Photo courtesy of Invensys
Operations Management. 1B. Fisher
introduced its field-mounted Wizard 1
pneumatic controller in 1930. Photo courtesy
of Emerson Process Management.
More sophisticated, large-case field-mounted pneumatic instruments incorporating control, indicating and recording functions began to appear on the scene around 1915. Initially, these just provided on/off and/or proportional control capabilities. Foxboro (now part of Invensys Operations Management) introduced the Model 40, the first proportional-plus-integral controller, in 19341935. In 1941, Taylor Instruments (now part of ABB) introduced the Fulscope 100, the first controller to provide full proportional/integral/derivative (PID) control capability in a single unit. At present, PID methods remain the workhorse of process control in refineries, petrochemical plants and other process plants around the world.
To avoid tampering by operators who were worried about keeping their jobs in the face of all this automation, instrument suppliers had to start putting locks on the cases to keep employees out. While this practice certainly helped secure the instruments integrity, it created many problems when the instrument had to be adjusted or repaired, especially when no one could find the key!
While relatively primitive compared to todays digital controllers, these early mechanical/pneumatic instruments did a surprisingly good job of controlling process variables. They were so reliable that a number of them are still operating in some older refineries and petrochemical plants.
Move from stand-alone instrumentation to control rooms
A major breakthrough in process control occurred around 1938 with the introduction of pneumatic transmitters and large-case instruments modified to accept pneumatically transmitted signals from field-mounted transmitters and then sending pneumatic control signals back to valve actuators. For the first time, this made it possible to physically separate the process-measurement instrumentation from the recording/indicating/controlling instrumentation. This led to the appearance of local control rooms in refineries and other process plants, as shown in Fig. 2.
Fig. 2. Large-case pneumatic instruments in a
control room at Gulf Oil Co.s Port Arthur,
Texas, refinery. Photo courtesy of Petroleum
Refiner, March 1949.
In some cases, these local control rooms were located up to several hundred feet away from the processing units (but no further, due to the distance limitations of the pneumatic signals). With this instrumentation, control room operators could remotely monitor process variables, setpoints and valve outputs, and switch between automatic and manual control. To ensure that different suppliers instrumentation would function properly together, the industry soon established the 3-psi to 15-psi standard signal range for pneumatic transmission, which remains in effect today.
Since control room space in HPI plants is usually limited and always expensive to build, following World War II (WWII), instrumentation suppliers focused on reducing the size of the instruments mounted in the control room. The resulting miniaturized controllers typically measured approximately 6-in. by 6-in. on the front faceplate, complete with a built-in indicator. With these smaller instruments, it now became practical to embed the indicators, controllers and recorders in appropriate locations on wall-sized graphical diagrams, as shown in Fig. 3. These diagrams illustrated the process unit, providing operators with a more intuitive sense of how the instrumentation related to the process. While these graphic panels helped reduce training requirements and enabled operators to monitor process operations more effectively, they still required fairly large control rooms. This led to the development of semi-graphic panels. These graphic displays used less space and still provided much of the intuitiveness of full graphic panels.
Fig. 3. View of the graphic instrumentation
board of Rock Island Refining Co.s FCC unit.
Photo courtesy of Petroleum Refiner, March
On the sensor side, suppliers began introducing a number of measurement products during the invigorating post-WWII years that would see wide applicability in the HPI. These included the first pneumatic-differential pressure transmitter, introduced by Foxboro in 1948. In conjunction with a simple flange-mounted orifice plate, this provided a practical and low-cost method to obtain accurate and repeatable fluid flow measurements. In 1956, Beckman Instruments introduced the first gas chromatograph for chemical analysis based on earlier research by A. T. James and A. J. P. Martin.
At around this same time, we started seeing more analog electronic instruments appearing in refinery control rooms. They were often interfaced to existing pneumatic instruments using current-to-pressure (I/P) and pressure-to-current (P/I) converters. In 1951, the Swartwout Co. introduced its AutroniC, the first electronic controller to use vacuum tubes. At the 1958 Instrument Society of America (ISA) show in Philadelphia, Pennsylvania, Foxboro, Taylor Instruments, Honeywell, and Leeds & Northrup (now part of Honeywell) all demonstrated electronic controllers. In 1959, Bailey Controls (now part of ABB) introduced the first fully solid-state electronic controller, followed shortly by several other instrumentation suppliers. During these years, we also began to see the shift from single-loop to multi-loop electronic controllers. In 1952, several engineers at Shell Development also presented the feasibility of direct digital control (DDC) in the Transactions of ASME (American Society of Mechanical Engineers).
Direct digital control and the dawn of CIM
Exciting news came in March 1959, with the announcement thatfollowing almost two and a half years of effortTexaco and the Thompson Ramo Wooldridge (TRW) Co. installed the first direct digital control computer online in a refinery, as shown in Fig. 4. This heralded what would later become known as the computer-integrated manufacturing (CIM) era for the HPI. An excellent article entitled, Texaco closes the loop, which appeared in Business Week, April 4, 1959, chronicled the drama: Shortly before 11 a.m. on March. 12, a veteran Texas Co. process operator named Marvin Voight flipped the switch ... The action closed the loop in the first fully automatic, computer-controlled industrial process. Moments later, the most vital parts of the 1,800-bpd polymerization unit at Texacos Port Arthur (Texas) refinery were under the unblinking eye and almost instantaneous control of a Thompson Ramo Wooldridge Corp. RW-300, a desk-size digital computer designed for just such control jobs as this. Texaco hopes the computer will raise the plants efficiency by a healthy 6% to 10%.
Fig. 4. The Thompson Ramo Wooldridge
(TWR) RW-300 direct digital control process
control computer was installed at Texaco Inc.s
new No. 1 poly unit at the Port Arthur, Texas,
refinery. Photo courtesy of Motiva Enterprises.
In addition to TRW, which contributed the computer, the Bristol Co. (now part of Emerson Process Management) redesigned its recording controllers to interface with the computer. Leeds & Northrup supplied onstream analyzers to chart the chemical content of the raw material and product streams.
The description of the computers function provided by Charles Richker, Texacos chief process engineer at the time, doesnt sound all that different than that for a present-day optimization project. The computer ... gets an analysis of incoming gas and outgoing gas; it senses and measures pressure, flows and temperatures; it calculates catalyst activity; then it weighs all these together and decides what the processing unit should do to get the most product for the least cost. Finally, it sets the controls and rechecks its figuring.
According to Business Week, the computer cost was $98,000 (in 1959 dollars); the custom I/O required to convert analog measurement signals to digital language cost $36,000; andnot surprisinglythe expense for engineering and extra instrumentation was more than double that of the capital cost for the computer and I/O hardware. So how did Texaco cost-justify this major (for 1959) $300,000 science project? To begin with, apparently, the company would have spent at least one-third of that on new instrumentation for the polymerization plant anyway. In hard terms, the company anticipated that the new computer would boost conversion efficiency from the 85%87% considered the maximum for the most skilled operators using automatic controllers to 93%, while saving up to $75,000/yr by prolonging catalyst life. Based on this information, Texaco expected an early payout on its investment. In soft terms, according to a Texaco executive, the company also expected to gain invaluable knowledge and experience from full-scale operation.
Not surprisingly, the familiar question of whether all this automation would make the human operator obsolete frequently came up during the project. But, obviously, thats not the case. While the computer does the dull repetitive work of reading, calculating and resetting, if something should go amiss, it would sound an alarm to which a human operator would have to respond to handle the situation.
Following this initial direct digital control implementation at Texacos Port Arthur refinery, TRW installed an RW-300 DDC computer at Monsantos new Chocolate Bayou, Texas, petrochemical plant in 1960. During the same approximate time period, IBM installed its first special-purpose computer for process control, the IBM 1700, at an American Oil refinery in Indiana, as shown in Fig. 5, and at a Standard Oil of California refinery, and (in 1962) at a DuPont chemical plant. In 1961, IBM announced its first standard computer for process control, the 1710 model. In the 1960s, Foxboro introduced several digital systems, including the M9700 process computer and its Digital Equipment Corp. (DEC) PCP-88-based DDC system, which was installed at the Esso Aruba Refinery. The PCP-88 incorporated dual DEC PDP-8s with a shared disk drive.
Fig. 5. IBM process computer installed in the
control room at American Oil Co.s Whiting,
Indiana, refinery. Photo courtesy of
Hydrocarbon Processing and Petroleum
While much of the supplier activity during this time focused on process control computers, automation suppliers also had to figure out how to interface the installed base of largely pneumatic field transmitters and actuators with their new-fangled electronic controllers. In 1959, Honeywell introduced the 4-mA to 20-mA analog signal, which, in conjunction with P/I converters mounted in the control room, provided the interface between the companys pneumatic field instrumentation and electronic controllers. Ultimately, 4 mA to 20 mA won out over Foxboros proposed 10 mA to 50 mA signal as an industry standard (ISA SP-50) for analog field communications.
In 1965, DEC introduced its first minicomputer, the PDP-8. Eventually, the company supplanted this with the PDP-11, which was used widely for real-time process control applications. In 19681969, Honeywell introduced the Series 16 DDC, with a modular hardware/software package. In the 1970s, Bailey Controls and Taylor Instruments introduced their own DDC systems. Rather than trying to build the computers themselves, these early process control systems were based on DEC, MODCOMP, Data General, and other companies minicomputers using purpose-built software and I/O. In 1971, Foxboro introduced its Fox 1, the first in a popular series of process control computers and, in 1972, its SPEC 200 split-architecture analog electronic controllers and INTERSPEC digital data highway. Also in 1972, Fisher Controls introduced its Series 1000 split-architecture system, with separate controllers and faceplates. In 1973, Taylor introduced real-time programming to the control industry with the companys process-oriented language (POL), an adaptation of BASIC programming, first used on the Taylor 1010 and MOD 3000 control systems. Also, in this appropriate time frame, process control engineers starting experimenting with reusable control-block structures, which arguably formed the basis for todays ubiquitous object-oriented programming techniques.
The DCS era
Thanks to continuing improvements in solid-state microprocessors and digital communications, automation suppliers were able to squeeze ever-more-powerful functionality into their electronic devices and systems. This led to the development of the distributed control system (DCS). While some might challenge this point, its generally accepted that Honeywell coined the phrase and introduced the first DCS, the total distributed control (TDC) system, in 1975. At just about the same time, Yokogawa introduced the companys CENTUM DCS.
Despite their high cost, TDC 2000 and CENTUM received strong acceptance within the HPI, particularly in North America and Japan. Within the next several years, several other companies, including Bailey Controls, Fisher Controls, Fischer & Porter (now part of ABB), Taylor Instruments, and Foxboro introduced their own DCSs. The Foxboro SPECTRUM DCS, began to show up in refineries and petrochemical plants around the world, providing strong competition for Honeywell and Yokogawa. Yamatake, which shared some intellectual property with Honeywell and manufactured many of the TDC 2000 components, also marketed the system in Japan.
Unlike the monolithic DDC systems which it replaced, the DCS distributes much of the functionality across multiple processors, helping to minimize the impact of failures on the ability of the plant to produce product. In theory, at least, the DCS architecture also moved some of the control functionality closer to the process to minimize latencies. The microprocessor-based, multi-loop controllers were connected to supervisory computers, floppy disk drives, CRT-based operator displays and push-button-equipped workstations, and line printers, now often located in a central (rather than local) control room via a proprietary data highway. In practice, however, the harsh environmental conditions in HPI facilities required that both the process controllers and I/O had to be mounted in air-conditioned rack rooms, often located fairly close to, if not immediately adjacent to, the central control room.
Fig. 6. Analog electronic instrumentation
installed in the main process control center at
Marathon Oil Co.s Garyville, Louisiana,
refinery. Photo courtesy of Hydrocarbon
While DCSs offered far more control and real-time information handling capabilities and other functionalities than previously available, they were not without their obvious flaws. For example, while the CRT-based operator displays provided control-room operators with a remote view of one or more process units while seated in front of a workstation in the control room, the computer displays lacked the intuitiveness of the full- and semi-graphic panel boards that they supplanted. This increased training requirements and often led to the operator switching from automatic to manual control because they just didnt trust the computer. Also, as operators in process plants know all too well, since its relatively easy and inexpensive to configure process alarms in software-based DCSs (compared to hard-wired annunciators), there was also a tendency to configure unnecessary and often confusing alarms, leading to often-terrifying alarm storms.
In 1977, Honeywell introduced the first redundancy scheme for a process controller, and, in the early 1980s, following several years of development and an investment of approximately $80 million, Honeywell introduced the companys second-generation DCS, the TDC 3000. This system offered more powerful controllers, new workstations, enhanced information management and other important features. According to some sources, Esso installed the first TDC 3000 system at the companys Cold Lake Refinery in Alberta, Canada. Throughout the 1980s, all DCS suppliers continued to enhance their systems with new control and information management capabilities, making the DCS the de facto platform for process control.
Honeywell introduced the first smart pressure transmitter, the ST3000, in 1983, followed shortly by Foxboro, Yokogawa, Rosemount and others. When combined with the respective suppliers proprietary digital field communications scheme, these smart pressure, temperature and flow transmitters improved performance over analog transmitters by transmitting the process variable(s) and often secondary measurements (such as ambient temperature, which is vital in colder climates) in a precise digital format; allowed the transmitters to be re-ranged remotely; and enabled operators and maintenance technicians remote (if relatively crude) access to transmitter status and diagnostics. This eliminated many unnecessary trips to the field.
In 1989, 30 years after the first DDC computer went online at Texacos Port Arthur refinery, the Purdue Reference Model for Computer Integrated Manufacturing was published. This evolved into todays ISA 95 architectural model and schema for plant-to-enterprise integration.
Fig. 7. DCS-equipped control room with CRTs
and projected display. Photo courtesy of ABB.
APC pushes the boundaries of economics
The improved visibility into the process and robust PID and advanced regulatory control capabilities provided by many DDC and DCS platforms helped operators and control engineers in HPI plants to stabilize control loops to a considerable degree and also solve other problems. Recognizing the opportunity that advanced regulatory control offered to help stabilize some of their trickier, more interactive control loops, process control engineers began to take greater advantage of these embedded capabilities, often experimenting on their own to further expand the envelope.
Model predictive control (MPC) was pioneered largely by dedicated groups of control engineers at Shell (including both Charlie Cutler and Steve Treiber) and other energy companies beginning in the early 1970s. In the late 1980s, Shell Research engineers in France developed the Shell Multivariable Optimizing Controller (SMOC), a significant advancement. A handful of small specialist companies, such as DMCC, Setpoint and Treiber Controls (all three subsequently acquired by AspenTech), plus Predictive Control in the UK and Profimatics also began to develop, refine and license MPC technology. Not to be outdone, control gurus at the major DCS suppliers (Honeywell, Foxboro, Yokogawa, etc.) also either began to develop their own MPC solutions or the companies acquired and further developed licensed technology from third parties.
The resulting breakthroughs in MPC helped solve the previously daunting multivariable constraint problems encountered in many HPI processes. Advanced process control (APC) software systems such as these, which typically ran in separate supervisory computers, provided the DCS controllers with the precise setpoints needed to further stabilize the process, reduce variability, and safely operate processes closer to physical constraints. Assuming that the plant process control operators trusted the APC enough to keep it turned on (which was not always the case), this typically provided owner-operators with significant economic benefit.
Open control and real-time information systems
While they represented a step change in process control technology over the all-in-one DDC systems and stand-alone analog electronic controllers, DCSs were handicapped by their closed, proprietary nature. This tended to speed obsolescence and make it difficult and costly to integrate the DCSs with other plant- or enterprise-level systems. Seeking to gain every possible competitive advantage, DCS suppliers were loath to share their proprietary communication technologies with other suppliers, or to even open up their software codes to their customers. This was particularly troublesome in HPI enterprises, where lots of data and information need to flow back and forth between the plant-level systems used to produce products (DCS) and the enterprise-level planning and scheduling systems.
However, during the 1980s, IBM, DEC, Microsoft, AT&T, and other high-technology companies were investing huge sums of money and dedicating their impressive brain trusts to advancing and reducing the cost of the general-purpose information technology (IT). This laid the foundation for the Internet-enabled Information Age. These technologies included open, standards-based operating systems (such as UNIX) and graphical user interfaces, Ethernet networking, TCP/IP communication protocols, object-based programming approaches, and many others that we take for granted today. Unlike the DCS, these technologies were based on open standards and many were available commercially, almost literally right off the shelf.
Initially, at least, automation suppliers either ignored, or tried their best to ignore, these goings on, convincing themselves that their industrial customers would never accept using commercial off-the-shelf (COTS) technologies in their plants. Foxboro, with the introduction of its I/A Series system in 1987, was the first mainstream automation supplier to incorporate UNIX, Ethernet and other commercial-type technologies into a system designed to manage and control mission-critical industrial processes. Foxboro also spent millions of dollars developing the worlds first real-time object manager. Since there were still performance and availability concerns about Ethernet at the time, Foxboro developed a redundant/fault-tolerant scheme for its Ethernet-based process control network, which the company intentionally referred to as a serial backplane, rather than a network, because company officials were concerned that industrial users wouldnt be able to get their heads around the idea of an Ethernet-based system.
Fig. 8. Recently renovated central control room
in a refinery at Major Global Energy Co. Photo
courtesy of Emerson Process Management.
Unfortunately for Foxboro, serious software and manufacturing issues with the I/A Series system, which took several years to fully resolve, prevented the company from capitalizing on this innovative technology, with more conventional and field-proven systems, such as Honeywells TDC 2000/3000 and Yokogawas CENTUM DCSs continuing to gain market share in the HPI plants in North America and Japan, respectively. In Europe, ABB began to make inroads into the HPI with its MOD 300 DCS.
The development of the DCS over the past 30 years has closely mirrored that of the overall process automation business, moving from proprietary technologies and closed systems to COTS components, industry-standard field networks, and Microsoft Windows operating systems. Today, the DCS has moved from a system-centric architecture to one that is more focused on supporting collaborative business processes and helping owner-operators achieve operational excellence in their process plants.
The drive toward openness in the 1980s gained momentum through the 1990s with the increased adoption of COTS components and IT standards. Probably the biggest transition undertaken during this time was the move from the UNIX operating system to the Windows environment, particularly for human-machine interface (HMI) and data analysis and presentation applications.
The invasion of Microsoft at the desktop and server layers resulted in the development of technologies such as OLE for process control (OPC), which is now a de facto industry connectivity standard. Internet-based technology also began to make its mark in industrial automation and the DCS world.
The impact of COTS was most pronounced at the hardware layer. Standard computer components from manufacturers such as Intel, Motorola, IBM, Sun Microsystems and Cisco Systems made it cost prohibitive for DCS suppliers to continue making many of their own servers, workstations and networking hardware (although most DCS suppliers still assemble their own process controllers and I/O modules, albeit using many COTS components). COTS not only resulted in lower manufacturing costs for the supplier, but also in steadily decreasing prices for the end users, who were also becoming increasingly vocal over what they perceived to be unduly high hardware costs. Some suppliers that were previously stronger in the programmable logic control (PLC) business, such as Rockwell Automation and Siemens, have been able to leverage their expertise in manufacturing control hardware to enter the DCS marketplace with competitive offerings.
The current state of most process automation system offerings available on the market today relies heavily on the incorporation of international standards, a common control and configuration environment, a common hardware platform, and a common information infrastructure that is designed to accommodate a wide range of applications from multiple suppliers. Although the DCS of today has come a long way from the almost totally proprietary world of the 80s, there is still considerable progress to be made in the quest for full standards adoption.
In ARCs latest global DCS market outlook study, published in 2011, Honeywell retained its dominant position in the global refining market, followed by Yokogawa, Invensys, Emerson Process Management, ABB and Siemens. In chemicals, Yokogawa has the leading position globally, followed by Siemens, Honeywell, ABB, Emerson Process Management, Invensys and Yamatake.
What goes around, comes around
While the HPI may be very conservative in some respects, traditionally, this industry has been quick to accept new technologies that offered clear potential to help companies operate and maintain their complex assets better and more efficiently. Process computers, direct digital control systems, DCS systems, APC, simulation and plant-wide historians, are just a few examples of the new technology adopted by the HPI. This has also been the case for fieldbus, the technology that provides a digital link between intelligent, microprocessor-based field instrumentation and the host DCS.
Unlike the 4-mA to 20-mA analog electronic standard for communications between field instruments and the control system (and the 3-psi to 15-psi pneumatic standard that preceded it), which required point-to-point wiring (or pneumatic hoses) for each device, digital fieldbus technology enables multiple field devices to communicate with the host system on the same wire. While fieldbus segment sizing, topology, and hazardous area-related decisions can add engineering complexity and cost compared to point-to-point analog field wiring, the wiring savings alone can reduce fieldbus installation costs to a significant degree.
More importantly, fieldbus provides bidirectional digital communications between the field devices and the host system. Thus, in addition to communicating one or more process variable measurements for monitoring and/or control, the field devices can communicate secondary measurements and important device status and asset management-related information to the host system. This eliminates the tedious and time-consuming effort previously required to ring out and verify potentially thousands of different field terminations during system commissioning; reduces ongoing maintenance costs and effort by eliminating unnecessary trips to the field; and in conjunction with appropriate softwareenables HPI plants to implement highly effective condition-based plant asset management strategies to help improve equipment availability, while minimizing unnecessary maintenance. (As many owner-operators have learned, too much work during planned turnarounds is done based on habit, rather than on actual need; while needed work sometimes goes unattended.)
That is the good news. As automation users in HPI plants know all too well, due to the snail-like pace of standardization efforts and significant politicking among national standards bodies and automation suppliers, its taken far too long for fieldbus standards and technology to arrive at its current state. Initially, DCS suppliers offered proprietary digital communications that provided many of the benefits of todays standard fieldbus technology, albeit in a single-vendor environment. In other words, each suppliers smart transmitters could only communicate digitally with its own control system. Not an optimum situation, particularly for end users.
Pressed by their customers, national standards bodies in Europe and North America, working in conjunction with automation suppliers, initiated a number of different, competing fieldbus standardization efforts. In 1999, in a creative, if not terribly helpful, attempt to break the stalemate and end these fieldbus wars, the International Electrotechnical Commission (IEC) came up with a compromise standard, IEC 61158, that recognized eight different fieldbus approaches (including FOUNDATION fieldbus, ControlNet/Ethernet IP, PROFIBUS, WorldFIP and INTERBUS), grouping these into different types, but creating common physical, data link and applications layers.
While waiting for these fieldbus wars to sort themselves out, many users simply avoided the issue altogether by installing transmitters with HART communications capability. While HART-enabled transmitters dont offer multi-drop capability and still commonly communicate the process control variable in analog, 4 mA to 20 mA (rather than the available digital) format, they do offer the potential to access transmitter status diagnostics and interact with the field device remotely from the control room, maintenance shop or plant reliability center. In the past, many users found this particularly useful when commissioning the instruments; far fewer used this capability for ongoing asset management. However, this situation appears to be changing as automation and other suppliers have introduced plant asset management toolkits that fully exploit the potential of HART.
Ultimately, in Europe, PROFIBUS emerged as an industry fieldbus standard for both process and discrete applications, gaining wide acceptance among suppliers and users alike on that continent. In North America, FOUNDATION fieldbus emerged as the digital fieldbus for process plants.
Not surprisingly, Shell and other leading HPI companies were among the first to experiment with and implement FOUNDATION fieldbus. They did so cautiously at first, with the initial implementations in pilot plants and for other smaller-scale projects, and ultimately, in virtually all their new plants and/or major expansion projects.
FOUNDATION fieldbus-enabled control valves and transmitters include standard process control blocks, so thatonce againfor simple process control loops, at least, both measurement and control can be performed in the fieldjust like in the early days of process control! Whats more, as with the early single-loop controllers, this can help limit the negative impact of instrument or other faults compared to DCS controllers, which handle dozens, or even hundreds, of control loops (albeit, normally in redundant configurations designed to enhance fault tolerance).
Since DCS process controllers are often physically located at a significant distance from the process, fieldbus-enabled control in the field can sometimes reduce the time latencies involved when a measurement signal has to be transmitted from the field devices to a DCS controller and the control signal transmitted from the controller back to the final control device in the field. While a study commissioned by the Fieldbus Foundation revealed that this can help improve performance, particularly in fast-acting control loops, owner-operators have been slow to accept fieldbus-enabled control in the field to date.
ARC believes that this is probably because users are already very comfortable and generally satisfied, with their DCS controllers, and because control in the field doesnt add value for the types of interactive control loops found in many HPI processes.
Fig. 9. Collaboration wall decision support for
a hypothetical integrated downstream energy
enterprise. Photo courtesy of Invensys
While the state-of-the-art in process automation systems has only advanced incrementally in recent years, ARC believes that well soon see some major advancements emerge in industrial automation, andif the past is any lessonowner-operators in the HPI will likely to be among the first to implement many of these advancements. Some of these advancements include:
- Even smarter field devices that are capable of conveying their health in absolute terms
- Automation systems with increased functional distribution
- Platform-independent, software-based function blocks supporting intelligent autonomous agents
- Increased use of cloud computing to serve data to authorized users, anywhere, at any time
- Increased use of wireless field devices, including wireless measurements for process monitoring and control
- Increased use of tablets, smartphones and other mobility devices by plant operators, maintenance technicians, engineers and others
- Increased use of advanced analytics for real-time decision support, fueled by the big data currently buried in many plant historians.
ARC analysts will aim to keep HP readers informed about these and other trends in our monthly Integration Strategies columns. HP
Paul Miller is a senior editor/analyst at ARC Advisory Group and has 25 years of experience in the industrial automation industry. He has published numerous articles in industry trade publications. Mr. Miller follows both the terminal automation and water/wastewater sectors for ARC.
Dick Hill is vice president of ARC Advisory Group, Dedham, Massachusetts, responsible for developing the strategic direction for ARC products, services and geographical expansion. He is responsible for covering advanced software business worldwide. In addition, he provides leadership for support of ARCs automation team and clients. Mr. Hill has over 30 years of experience in manufacturing and automation. He has broad international experience with The Foxboro Co. Prior to Foxboro, Mr. Hill was a senior process control engineer with BP Oil, developing and implementing advanced process control applications. Prior to joining ARC, he was the US general manager of Walsh Automation, a major engineering consulting firm and supplier of CIM solutions to the pulp and paper, petrochemicals, pharmaceutical, and other process and manufacturing industries. He is a graduate of the Lowell Technological Institute with a BS degree in chemical engineering.
Dave Woll is vice president of the consulting services at ARC Advisory Group where he provides high-level consulting services for ARC clients. He has been with ARC since 1997 and has been defining and applying process automation for over 35 years. This includes the marketing and application of control, safety, SCADA, measurement systems and business Integration. Prior to ARC, Mr. Woll held numerous positions at both The Foxboro Co. and Bristol Babcock. He holds a BS degree in electrical engineering from the University of Connecticut.
While certainly functional, the only way those early pneumatic controllers could be tuned to provide the desired control response was through tedious and wasteful trial and error, which didnt always work either. During WWII, two engineers at Taylor Instruments, John Ziegler and Nathaniel Nichols, spent a lot of time tinkering around with PID simulations on the companys Fulscope 100 controller until they came up with a satisfactory solution. In 1942, they published their now-famous paper, Optimum settings for automatic controllers. This established clear rules for tuning PID controllers in refineries, petrochemical facilities and other process plants. This came in very handy during the ensuing war years, when now-well-tuned controllers helped chemical plants produce synthetic rubber for tires and other wartime necessities, helped refineries produce massive quantities of gasoline and diesel to fuel jeeps, trucks, tanks and heavy equipment; and produce newly developed, high-octane aviation fuel for fighter planes, strategic bombers and other aircraft. HP
Ziegler and Nichols develop a method for tuning closed-loop controller response