Comment

01 August 2004

Open bus integration

By Craig Schiro and Donald Lum

Justifications for closed vendor proprietary systems have vanished.

Here is a viable, proven, and applied approach to constructing an integrated petrochemical enterprise using commercially available off-the-shelf technology.

Profound changes in the process control industry are underway, and economies of scale are the drivers. Wide availability of low-cost commodity hardware and software technologies abounds. Companies are growing up around niche areas, providing network connectivity at low cost.

The nucleus of change is as follows:

• TCP/IP
• Microsoft Windows technology
• Industrial Ethernet
• OLE for process control (OPC)
• Open fieldbus technologies

Differentiate hardware

Over the past twenty-plus years, the evolution of controls from mainframe-based to total distributed control has been fast-paced. Before the explosion of networked computing, control vendors relied on custom proprietary hardware and software to meet the needs of demanding real-time control environments.

The past justifications for closed vendor proprietary systems have vanished! Control manufacturers can no longer rely on selling proprietary hardware and software systems, because the life-cycle costs are astronomical.

As an example, one large name in process control uses a 6-megabit-per-second token-passing control network to form its totally distributed control system. To support only the network part of this system required an electrical engineering staff, an embedded software engineering staff, an applications software engineering staff, a staff of systems software and test engineers, and fabrication facilities.

This engineering overhead was required for very low volumes of high-priced network interface boards and systems. Today, because of economies of scale and the explosion in PC technologies, the function of this obsolete, slow network can be replaced for less than $19 an interface card.

Distributed control system (DCS) and programmable logic controller manufacturers are under greater cost pressures to differentiate their control hardware. This accelerates the race to transform their companies into intellectual property businesses, which takes place by incorporating their knowledge in software. Control vendors are moving to commodity hardware and software operating systems such as Windows 2000 Professional to replace the low-volume, high-cost systems of the past.

Sharing the cost to design modern control systems is part of the "open" thrust.

Hardware is becoming more scalable due to open interoperable technologies, the most significant of which are the following:

• Actuator Sensor Interface (AS-i) is a serial bus technology optimized for cost-effective integration of discrete I/O-wide multivendor interchangeable modules (www.as-interface.com).
• Profibus DP/PA addresses medium speed and lower speed loop-powered fieldbus communications (www.profibus.org).
• Foundation fieldbus is a single, open, interoperable fieldbus developed with input from more than 140 companies, including process and manufacturing automation companies, and end users worldwide (www.fieldbus.org)
• OPC is a nonproprietary technical specification that defines a standard set of interfaces based on Microsoft Component Object Model (COM) technology. OPC standards provide vendors a neutral way to implement control domain interconnections (www.opcfoundation.org).
• Industrial Ethernet is Ethernet networking addressing the needs of the industrial marketplace.

These open technologies provide the path to the future for users and integrators to break the monopolies of the control vendors. The technologies are real and are solving problems today. Let's take a look at a typical oil refinery and how OPC technology can form the basic building blocks for integration. The business domain, referred to as level 4 in the Purdue Model of Process Control, is fast becoming a tightly integrated adjunct to the control domain referred to as level 3 in the Purdue model.

In the past, implementing an open information solution for a refinery or chemical plant computing system was not a choice. Integration came about through expensive gateways where the user paid significant sums for repeatable specials. The high prices discouraged information engineering and information integration. Evaluation of the vendor with the best proprietary solution was the best option. The gateway served to bring all information and external DCS interfaces to and from the DCS. Custom interfaces then connected the DCS to the business domain. The result was data isolation and difficulty in managing data manipulation, reporting, and analysis tools. This situation is changing for the better. Interestingly, close adherence to designing an enterprise based on open technologies yields an asset not readily visible.

For example, if the initial vendor you select for the OPC historian falls from your favor, you can seek a better offering and replace the old OPC historian. This is true for almost all OPC software and hardware devices designed to the dominant fieldbus technologies and open standards. All of this happens without significant disruption to your enterprise. This is fast becoming the norm and not the exception.

Take a close look at the common business functions segregated between level 3 and level 4 for a typical petrochemical facility. Companies identified the need to segregate business functions to accomplish computerized integrated management (CIM) in the 1980s, so that is not new.

A team of industry leaders and university professors led by Professor Theodore Williams of Purdue University in West Lafayette, Ind., recognized this as part of their work on CIM.

Segregation of the control domain from the business domain makes economic sense. The business objectives of the two domains are significantly different. High availability of designs for the control domain as well as the need for maximum security between these designs justifies this segregation. Process control requires a deterministic environment that further supports segregation.

The gateway that bridges vendor propriety hardware and software systems to the enterprise is fast giving way to OPC technologies.

OPC has a critical role in integrating the enterprise with best-in-class technologies. The OPC Foundation manages specifications for open vendor interchangeable technologies. As common needs arise for new COM objects, the OPC Foundation generates new OPC specifications for the needed data interfaces. OPC software is available directly from the hardware vendors. Today it is not uncommon to outsource OPC driver development to other companies who specialize in OPC software technology.

Matrikon of Edmonton, Canada, a charter member of the OPC Foundation, specializes in OPC software driver technologies. It has OPC software for more than 400 devices. For example, the company wrote the Triconex (an Invensys company) OPC server used to communicate with the Triconex Network Communica-tions Module. A mutual software licensing agreement allows Triconex to focus on its core technologies and leverages the OPC software strength of an outside expert—Matrikon.

Users have the choice of hardware and OPC software technologies. This allows them to select the best components to build hierarchical information architectures. OPC standards are constantly emerging as business needs become clear.

Some important standards widely in use today are

• data access for reading and writing data between an application and a process control device flexibly and efficiently,
• event and alarm mechanisms that notify OPC clients of specified events and alarm conditions, and
• historical data access for reading, processing, and editing historian engine data.

OPC connects devices to applications using standard data access technology developed by Microsoft called COM. All OPC-compliant applications (interface using OPC client) can interface with any OPC server.

A case of tank farm automation

A major expense of a tank farm upgrade involves the cost of a new wiring infrastructure. This cost makes wireless technology extremely enticing. Wireless technologies are in their infancy, and reports of failures for those on the bleeding edge of this technology are common. There are some emerging radio technologies that show promise, but more development is still necessary.

As tank farm operators look to upgrade their tank field infrastructure, techniques to maximize the return on this investment are an important consideration. In the past, automating a tank farm required a separate cabling infrastructure to accommodate control of pumps, motor-operated valves, miscellaneous I/O, and automatic tank gauging.

Today's use of Industrial Ethernet as the off-site backbone is critical to providing an obsolescence-resistant tank farm automation infrastructure. As wireless devices mature, the fiber-optic network can facilitate reliable deployment of radio handheld devices as well as wireless supervisory control and data acquisition.

Past systems did not provide for power distribution. It is not uncommon for tank farm operators to incrementally expand and patch old gauging systems with later technology, yielding mixed-up wiring systems. Common to all modern tank gauging systems is the need for a reliable power distribution system. The incremental patching of the system commonly involves connecting to field lighting panels for power. This random method of sourcing power significantly affects the system reliability. Electrical noise produces failures in the gauge electronics. Someone trying to turn lights on or off produces an unexplained loss of view with mysterious return-to-service events before maintenance personnel can identify the outage cause.

Most gauging systems are more than thirty years old. The best path forward if an infrastructure upgrade is planned is to use the technologies listed below to construct an open vendor-independent system.

• Fiber-optic Industrial Ethernet: Certified for Class I, Division II groups A, B, C, and D. The end of all fieldbus wars is Ethernet.
• Profibus DP and PA: Use four-wire radar tank gauging for custody transfer. Use two-wire loop-powered radar for inventory measurement. Use microcontrollers for ancillary control loops, using Profibus PA to measure and control the following tank farm processes: control valves, pressure, temperature, flow, and miscellaneous analog inputs and outputs.
• AS-i: Automation discrete devices such as pump automation, motor-operated valves control, tank mixers automation, and independent high-level alarms.
• OPC-DA: Integrate tank farm controls to the DCS and concurrently provide direct interface to the plant information system.
• Web-based human-machine interface
• Redundant: 480 volts alternating current (VAC) to 120 VAC power distribution system to reliably power the new automation infrastructure.

Older tank gauging systems do not include power distribution infrastructures. All modern tank gauging systems require reliable power to function. A redundant 480 VAC power system provides secondary selective power distribution as a key element of the infrastructure. It supplies reliable power to the instrument subsystems for high system availability and overall maintainability.

As of today, most of the two-wire tank gauges manufactured by tank gauging vendors support Profibus PA. Foundation fieldbus (FF) is available for some, but it is not as prevalent as Profibus PA. Most vendors have plans for two-wire FF gauges, but PA is still dominant today. High-power radar devices (four wire) such as those manufactured by Saab Rosemount can communicate on Profibus DP.

In a recent study conducted for a major Gulf Coast oil refinery, the task consisted of investigating open technologies to replace an aged tank farm gauging system and a separate wiring system dedicated to off-site valve and pump control. The valve and pump control system used an old two-wire wiring system developed by Powell Industries Inc. in Houston, Texas. The C-2 system has a dedicated pair of wires that run from the tank field to the control house. Some of the wire pairs also served to wire standard 4–20 mA readings. The tank gauging, valve and pump, and miscellaneous I/O system are more than thirty years old.

The recommended system, based on criteria-based evaluation techniques, uses three open bus technologies and Industrial Ethernet to construct a vendor-independent system. It was clear after a detailed investigation of many vendors' offerings that a combination of Profibus PA, DP, AS-i bus, and Industrial Ethernet technology provides the most resilient and long-lasting system.

Resilience from technology obsolescence rested on these three critical points:

1. There is little doubt that IP-based Ethernet systems will prevail long term in the industrial controls market (all fieldbus wars end at Ethernet).
2. Profibus has the support of many vendors, probably even more than Foundation fieldbus, although Profibus is more prevalent in Europe than in the U.S.
3. At least three different vendors supplied AS-i technology that meets the requirements for the tank field valve and pump control. There are hundreds of AS-i manufacturers to choose from.

It is not uncommon for tank farm wiring infrastructures to be past or approaching the end of their useful lives. A 3 June 2000 Houston Chronicle article highlighted that aging wiring is rapidly becoming a national issue. Major cable manufacturers' data indicates the design life of wire insulation is forty years. Many of the nation's tank farms have wiring and tank gauging systems that are more than thirty years old.

Behind the byline

Craig Schiro (Craig.Schiro@infotech-eng.com ) is an ISA member and has an electrical engineering degree. He is president of InfoTech Engineering with more than twenty-four years in the petrochemical industry. Donald Lum (don.lum@infotech-eng.com) has a B.S. and a PhD in chemical engineering. He is vice president of advanced process control technologies at InfoTech.