A Foundation fieldbus-based process control system proves easy to install and cost effective for a kraft pulping application at Bowater in Gatineau, Que.

Bowater Pioneers Foundation Fieldbus Automation at Gatineau Newsprint Mill

Behind the pioneering fieldbus application is a long evaluation conducted by Bowater. Early in the 1990s, the company concluded that the future of process automation would lie in fully-distributed control having two key components: an industry-standard device level network, and intelligent and interoperable field devices. Such advancements would be crucial to future economical automation projects and efficient mill operations.

No industry-standard device-level communications protocol was available in the early 1990s that could eliminate the expensive, cumbersome, and slow gateways that competing proprietary protocols needed to talk to one another. Further, none of the proprietary protocols had the power to rise above the others and become the de facto standard. Vendor alliances promoting particular proprietary protocols as a way to achieve de facto status were not the answer either; the ground beneath these alliances shifted too easily. Bowater believed that proprietary protocols, for all practical purposes, would always lock an end-user into a restricted choice of devices and technologies.

WHY FIELDBUS? The fieldbus protocol allows any Foundation-registered device to communicate with any other Foundation device. Bowater believes Foundation fieldbus is the best conceived device-level protocol ever developed, offering multiple speeds, function block programming, synchronized real-time control, and protection for vendors’ proprietary advances in technology while still assuring interoperability. Fieldbus also shows signs of becoming as popular as 4-20 mA is today. Bowater has long advocated Fieldbus Foundation efforts toward a standard protocol created with inputs from both end-users and vendors that has the following features:

• Inexpensive • Bi-directional

• Digital • Multi-drop

• High speed • Deterministic

• Bus powered • Robust

• Industry-wide

A Bowater engineer inspects the field cabinet housing the scalable process management system for the Gatineau mill’s kraft pulping operation. The system supports both fieldbus and 4-20 mA field devices.

SPECIALTY PULP TO THE TEST. Anxious to try the new technology in a real-world environment, Bowater embarked on a project to automate kraft pulping. The area, which blends dried kraft pulp sheets in a slurry with water, made an ideal first fieldbus candidate because of its relatively small size and an ability to store finished material as a protection against unanticipated shutdowns during commissioning. The kraft pulp is added to machine stocks in portions up to 10% to increase sheet strength in lighter newsprint grades.

The area’s single loop controllers and mimic panels were replaced with a Fisher-Rosemount DeltaV scalable process management system consisting of a modular controller containing a power supply, CPU, fieldbus communications module, two conventional analog I/O modules, and six conventional discrete I/O modules. The components are mounted in a cabinet in an electrical equipment room that is near the equipment controlled by the process management system.

To compare fieldbus with analog technologies, the system’s field devices include both fieldbus and conventional 4-20 mA instruments. PID logic is variously located in fieldbus transmitters, fieldbus valve controllers, and the DeltaV controller. Some loops are all fieldbus; others are hybrid fieldbus/4-20 mA configurations.

Completing the architecture is a DeltaV PC workstation linked to the controller via 1,200 ft of fiber optic IEEE 802.3 (10baseT) Ethernet. The workstation serves as a combination operator /engineering terminal.

A Bowater operator uses a PC workstation, located in the mill control room, to monitor the pulping area. The workstation is linked to the control cabinet (Photo 1) by 1,200 feet of fiber optic IEEE 802.3 (10baseT) Ethernet.

TWO FIELDBUS SEGMENTS. Two fieldbus H1 (31.25 kbps) segments were installed at Bowater’s Gatineau mill: one 200-ft long to the pulper, the other 700-ft long to kraft storage chests. The segments utilize the mill’s standard instrumentation cable of four No. 18 AWG twisted and shielded pairs. One pair is used for fieldbus, the remaining three pairs for the 4-20 mA signals. Noise levels on the H1 segments are generally below 40 mV; the noise rejection standard for the instrumentation selected is 75 mV.

Residing on the pulper’s H1 segment are five Foundation fieldbus registered digital instruments:

• Temperature transmitter (Rosem-ount 3244)—monitors white water used in dilution. This measurement is read by two sensors in one probe, with a deviation alarm built into the instrument’s logic. Redundant sensing was easily accomplished with fieldbus.

• Level transmitter (Rosemount 3051L) and a valve controller (Fisher DVC-5000)—mounted on a globe valve, this controls pulper level. PID logic is located in the valve controller.

• Valve controller (Fisher DVC-5000) —mounted on a rotary V-ball valve, this controls the supply of white water to the pulper.

• Valve controller (Fisher DVC-5000)—mounted on a rotary V-ball valve. A 4-20 mA consistency transmitter provides the process measurement for this control loop. The hybrid loop’s PID control algorithm is located in the DeltaV controller.

Seven fieldbus instruments reside on the kraft storage area’s H1 segment:

• Two level transmitters (Rosemount 3051L)—one for each of two storage chests.

• pH transmitter (Rosemount Analyti-cal 4081)

• Pressure transmitter (Rosemount 3051) and a valve controller (Fisher DVC-5000)—mounted on a globe valve, this is used for dilution water pressure. PID logic is located in the transmitter.

• Two valve controllers (Fisher DVC-5000)—mounted on Neles-Jamesbury double-acting valves. A 4-20 mA consistency transmitter provides the process measurement for this control loop. The PID algorithm is also set up to select the valve to be used in the control loop, depending on which kraft storage tank is in use. This hybrid loop’s PID is also located in the DeltaV controller.

Fisher-Rosemount engineers in Austin, Texas, stage scalable process system modules and fieldbus instruments during development of the new Bowater control system. The bright yellow fieldbus cable is readily visible.

THE PAST AS A GUIDE. Unlike some end users evaluating device-level control, feasibility was not a concern when Bowater began the pulper automation project. The company has had good experience since 1988 with a PLC system that uses a twisted pair to link distributed I/O blocks. In other words, the company knew device-level control could work.

One of the first questions encountered with the new automation project was how to run fieldbus wiring. Reservations have been expressed in industry about having too much plant control dependent on a single non-redundant pair. Bowater, however, has relied on non-redundant PLC network pairs in liquid-tight, flexible armored conduit for 10 years and has never had a failure. The fieldbus cable was installed in the same fashion.

A second question was how much spare channel capacity and cabinet space to allocate. Bowater’s rule of thumb has been to install 25% more I/O than initially required, 50% more cabinet space, and a 100% larger rack room. The plant did the same for the new automation—minus the rack room—because it saw no need to alter a successful strategy. Overall, the scalable automation’s space requirements are much lower than a conventional DCS.

Ports on the Foundation fieldbus connector block at the Gatineau mill link the main fieldbus cable to spur cables serving particular instruments.

AVOIDING UNWANTED COMPLEXITY. Unlike conventional DCS architecture, where intelligence is concentrated in centralized controllers, fieldbus permits field devices themselves to perform PID and other sophisticated tasks. It also enables the devices to store and transmit a wealth of information, and to talk to one another “behind the backs” of their controllers.

This meant that Bowater had to evaluate the potential for unwittingly building complexity, confusion, or even chaos into the control system. Little chance of that actually existed on such a small project, but problems could possibly occur on larger projects in the future. The company therefore took careful account of questions such as which devices should control locally and which should not, and whether distributed control should be placed in the transmitter or valve. Other questions included how fieldbus devices should coordinate with other fieldbus devices and with devices on the controller I/O, when and what data should fieldbus devices report to one another and back to the controller, and what should be done with that data.

Revising mill standards. Concerns about complexity and confusion also prompted Bowater to begin an ongoing process of revising mill standards for project definition, control, and instrumentation system design, control system configuration, design of operator displays, etc., for Foundation fieldbus-based control systems. The purpose was not only to assist development and configuration of the kraft pulp application, which the company performed in-house to gain experience, but also to guide the company and outside integrators on future projects.

Bowater is now asking questions and writing specifications about integrating PLCs, drives, chromatographs, etc., with fieldbus. Questions abound. How is PLC data requested and sent? Does each coil and register have a tag? How will commands be called? What packet size? What will a variable frequency drive’s device description look like? How will an analyzer broadcast synchronous data from an asynchronous request? What about valve output tracks, bumpless transfers, ramping, output rate limits, dead time compensation, and the like?

CONSULTANTS/CONTRACTORS AFFECTED. A consultant without experience in configuring a fieldbus control system is likely to underestimate development and configuration costs. Instrumentation systems must now be designed as part of a total network, not just an individual loop. Much more data is available from a fieldbus device than a conventional device. Control strategies will be set up differently, with a great deal of attention being paid to what control strategies should operate in the field devices and which should operate in the controller.

At the same time, a consultant will probably overestimate the costs of routine configuration and database building because of some fieldbus control systems’ Windows NT and IEC 61131.3 graphical format and simplified global database building. From experience thus far, Bowater can better detect unrealistic or excessive contract engineering estimates; this alone could be a major benefit from the first fieldbus application. In a similar vein, the company is also developing specifications for electrical contractors.

SOLID CONTROL IS DEMONSTRATED. Since implementation in June 1998, this first foundation fieldbus control system has operated reliably except for a valve communications failure, which was caused by a poor cable connection and was quickly remedied. Other than that, communications between fieldbus devices and the host controller are proving remarkably robust. No differences have been detected in the quality of loop control or response, whether PID logic resides in the fieldbus transmitter, the fieldbus valve controller, or the DeltaV controller. The location of PID logic is probably best left to other process control design factors. Delta V scan times are running 450 mS on the pulper’s fieldbus segment and 650 mS on the storage area’s segment, which are well within process requirements.

The mill saved approximately 50% of the wiring costs normally associated with new instrumentation. Instead of pulling three four-pair instrumentation cables into each area, only one was required. Modular wiring components with quick-connects also helped cut costs and prevented accidental short circuits and incorrect wiring. Wiring savings were further realized when a second pH transmitter was added to the system in December 1998. With a conventional 4-20 mA scheme, another cable would have had to be pulled to serve this transmitter. With fieldbus, the instrument was simply connected to the nearby H1 segment with a short length of cable.

As a beta test, a Foundation fieldbus magnetic flow transmitter has recently been installed in the stock line between the pulper and storage. A major reason for selecting a fieldbus meter is to compare it to analog magmeters in the mill. Although accuracy and stability differences are expected to be insignificant in the application, differences do exist between the two instrument types in measurement technology. Even though interoperable, they might not be interchangeable.

MANY HOURS SAVED. The mill estimates that 80% of the man-hours normally required to commission a control system were saved as well. Much less time was spent tracing wires and diagnosing problems. Although commissioning improvements could not boost production (the pulper doesn’t run full time), the mill believes that significant benefits will accrue in the future when installing or upgrading fieldbus based automation on a continuously running process, such as a paper machine. Typically, instrumentation and controls are the last items finished during a paper machine rebuild because mechanical work must be completed first. By reducing commissioning time, production might resume sooner. A day saved in commissioning a paper machine can increase mill revenue by C$250,000 to C$360,000.

Before startup, Bowater had formulated a backup plan should the new automation not work. A small army of company engineers and technicians were on call and enough HART parts were on hand to establish conventional control within hours if necessary. For the first days after startup, everyone tensely waited to see if the system would crash. But after a few months without a hitch, and after checking diagnostics for signs of hidden problems, everyone became confident the right automation choice had been made.

Judging from its hands-on experience in automating the kraft process, Bowater foresees project engineering costs for initial Foundation fieldbus projects being slightly higher than for conventional scalable automation. The reason is the additional engineering hours required to design the unfamiliar fieldbus segments. As experience is gained, however, engineering costs should fall. In any case, the mill expects any higher initial costs to be paid back many times over by faster checkout, earlier startup, and lower operating costs. Greater mill output and more consistent quality pulp and paper products should also result.

USING THE INFORMATION. An expected benefit now being studied is how to make use of information so easily gathered by the fieldbus devices and accessed by the workstation. Built-in device diagnostics is one. For example, the pulper’s pH transmitters indicate when their probes should be cleaned. Control valves can provide information about packing friction, number of cycles, and total travel. Such information could allow the plant to predict when valves should be serviced or replaced. It would be ideal to schedule service so that a process need never be shut down outside of a planned outage.

In the future, instrument vendors will probably add features such as vibration analysis and temperature sensing to detect abusive excursions, frequency analysis to indicate if a valve is fully shut, and so forth. Training maintenance crews to make use of these and other tools is very important because traditional 4-20 mA device troubleshooting and calibration methods can no longer be used.

The mill expects process upsets and unplanned downtime will also be reduced by making use of available diagnostic, calibration, servicing, and historic information for process control equipment asset management. Much of the mill’s fieldbus efforts today are concentrated on training technicians and on expanding this first fieldbus application with other vendors’ devices to check interoperability.

All in all, the mill is finding itself gradually shifting its thinking from the DCS mindset of the past 20 years to the new world of fully distributed process automation.

Trung Phung, P.E., is a process control engineer for Bowater Pulp and Paper Canada Inc., Gatineau, Que., and Wade Stewart, P.E., is a project engineer for Fisher-Rosemount Performance Solu-tions, Austin, Texas.