09 April 2001
by John Rezabek
Engineers want to leave 4-20 mA behind; justifying digital is The Challenge.
With the emergence of certified field devices and control systems hardware to support them, Foundation fieldbus (FF) has become worthy of serious consideration for new projects and existing plants.
Of course there are challenges involved in implementing this technology in a new chemical plant. These include justifying it with the manufacturing, commercial, and project management people. There are design issues that are unique to the technology. There are also cost and construction issues.
And what are the actual benefits of this new bus technology?
Gang way, fieldbus
In the middle of 1998, the company that was then BP Chemicals was in the early phases of detailed design for a new 70-kiloton/year 1,4-butanediol (BDO) plant. When the slated TDC-3000 system became unavailable due to the sale and continued operation of neighboring Lima Refinery, the project needed to select a distributed control system (DCS).
The TDC is not an FF system; rather, it used Honeywell's proprietary DE protocol for digital integration of field devices. While only Honeywell transmitters were capable of this form of communication, we had been happy with the transmitters and impressed with the benefits of digital communication.
Having experienced the benefits of digital integration of field devices with this DCS, we couldn't envision ourselves commissioning a plant in 2000 that would shackle us to legacy 4-20 mA technology.
How could we convince conservative project, manufacturing, and commercial management that a technology as new as fieldbus was a good choice for our project?
We needed to work on management because BP had never before implemented FF technology, some science projects in lablike environments notwithstanding. We had to ask, "Where do we want to be in 2001?"
We were early adopters, and we had to go further up the ladder than ever imagined when we made this decision.
Align with corporate strategies
Users are fortunate if they have some preexisting corporate strategies or directives that an FF application would further. These would include concepts such as manufacturing excellence, quality improvement, empowering operators, and process optimization.
Any of these or similar concepts help justify a fieldbus application. Typically, management will have bought into these at some level. This can lessen the degree to which any justification must be rigorously commercial. FF, in particular, offers a great many features, such as diagnostics, that directly serve such ends.
Our vision for the BDO's fieldbus implementation included the following:
- Have operators ascertain data validity, zero transmitters, and diagnose and repair instruments.
- Have operators perform routine maintenance of instruments, thus saving one technician.
- Have operators use predictive maintenance to maintain valve health. Valve problems are innately self-revealing. This lets engineering plan outages and minimizes on-site spare parts.
- Have maintenance pull and service only the right valves and all of the right valves for four consecutive turnarounds.
- Use transmitter diagnostics to predict sensor failures, impulse line plugging, and problems with upstream pumps.
- Use online diagnostics to avert five spurious process interruptions.
- Use online diagnostics to avert serious damage to equipment or catalyst.
- Use instruments and valves that are self-documenting.
- Have clerks handle many process safety management and International Organization for Standardization documentation chores.
By showing the degree to which these properties served our corporate manufacturing vision, we effectively got our foot in the door and gained management's ear.
Operators sell technology
One aspect that helped sell fieldbus was our operators' enthusiasm for it. We invited them to preview the various technologies with us. Fieldbus offered features that impressed them.
Options such as real-time feedback of valve position, improved measurement diagnostics, the ability to zero transmitters, and secondary variable availability that helps avoid freezing or cooking transmitters made the technology attractive to most who looked at it.
These individuals made their feelings known to their supervisor, which helped convince him it was a worthwhile choice. In our case, the endorsement of one or two experienced operators is far more persuasive than control engineers' pontification.
Commercial justification, savings
The potential commercial incentives for a fieldbus application have been extolled for many years. What we found was that in reality, the economic incentives for an application such as ours were not phenomenal.
Instead, they were more in line with those identified in a 1999 study at Syncrude Canada by Ian Verhappen (see "Syncrude totes up the numbers on fieldbus installation," September 1999 InTech,)
Overall, we claimed about $200,000 in savings for the project. What we finally realized in the execution of the job was probably less than might be realized because of the following:
- Because we used the "chicken foot" topology, our installation was nearly identical to a traditional point-to-point installation from the junction box (JB) to the field devices.
- Fieldbus devices can cost up to 15% more than traditional smart devices.
- DCS factory acceptance testing is complicated by fieldbus. We lacked any kind of simulator to prove our configuration, with the applicable portion fully downloaded to devices.
- Our engineering contractor made JB diagrams as well as segment diagrams, so few savings were realized there.
- Many engineering contractors are optimized for point-to-point wiring, each with their own (frequently homegrown) in-house database for populating loops and JB drawings with terminations. Fieldbus doesn't fit neatly into these point-to-point schemes, with many devices sharing effectively the same terminations and pair or wires.
- Our job was a mix of fieldbus and point-to-point, which added complexity.
- For optimum integrity and fault tolerance of the fieldbus segments, we chose termination hardware, which was significantly more expensive than traditional terminations. For example, to minimize the chance of a segment becoming "shorted" when a device was removed for maintenance, we used InterlinkBT connectors at each device.
Group I/O functionally
Our engineering contractor was enthusiastic about the fieldbus technology and helped support the decision. We did nothing special to show fieldbus applications on the piping and instrumentation diagram (P&ID). Instead, we chose to view fieldbus as a control application nuance similar to I/O card loading.
Engineers obsess to varying degrees about the precision and detail to show on a P&ID, but I can't imagine anyone shows I/O card loading. Controls specialists are nonetheless judicious about what terminates where, applying some process knowledge to minimize common-mode failures, optimize fault tolerance, and group I/O functionally with respect to the process.
We established some engineering guidelines for loading fieldbus segments.
The first step was to identify critical valves . . . and later on, critical measurements. A critical valve is generally one whose failure has a high likelihood of leading to a costly process shutdown. This would include the fuel-throttling valve to a process heater or boiler, for instance. We identified three valve classes:
- Level 1, valves whose failure immediately shuts down the process.
- Level 2, valves whose failure requires very rapid operator intervention to avoid a process shutdown.
- Level 3, valves whose failure requires operator intervention and impacts rates or quality but would not cause a shutdown.
We determined that no Level 1 valve would share a segment or H1 card with any other Level 1 valve or Level 2 valve. Restrictions on Level 2 valves were fewer, and there were few restrictions on Level 3 valves. In some cases, we chose to design segments with three or more Level 3 valves.
In general, we aimed to include the primary process variable on the same segment as its associated valve and to implement the proportional, integral, derivative (PID) algorithm in the valve positioner. With the application of backup link active scheduler, we determined that having PID in the field was more fault tolerant than performing all PID in the central control room- based controllers.
A shorted segment or segment power supply failure would send valves to their failure positions, regardless of where the system solved the PID. PID in the field means control would continue with only loss of view, in the case of an H1 interface card failure.
We chose to keep primary and secondary components of cascade loops on separate segments, reasoning that at least one variable would remain visible in the event of a communications interruption or failure.
When using conventional instruments, the assignment of I/O can occur independent of the piping, instrument location, and electrical design effort. However with fieldbus, one may need to locate field instruments well before detailed piping design is even 50% complete and before any instrument JBs are located.
In other words, the allocation of devices to segments must happen with knowledge of where they are physically located. Instrument designers accustomed to relying on completion of piping design may be unprepared for or uncomfortable with trying to prelocate instruments.
For butanediol, we went through a best-guess process based on major equipment location. This method worked well. Except for a few devices that ended up across a roadway, there were few serious problems with this approach.
As to transmitter calibration, we chose to perform no calibration in the field. Instead, we paid an extra $25 per transmitter for our supplier to certify each device and provide a record of calibration. We have seen no negative consequences from this strategy.
As to valve calibration, we have not performed any bench checking of control valves. Each valve is autocalibrated in place, in line. There have been no negative consequences from having taken this path.
As to loop checking for fieldbus devices, we performed only a communications check. We disconnected the transmitter once at the device to ensure that the correct tag number was in its correct location. We did the same thing at the junction box to ensure that the instrument tagging was correct. This allows future maintenance to disconnect the instrument at the box, confident they are dealing with an accurately tagged cable.
As to device commissioning and downloads, unlike traditional smart instruments, fieldbus instruments aren't preloaded at the factory with their configuration. A control scheme can have blocks residing in several different devices, as well as the controller. It's likely to be impractical to specify all this information when the transmitters are ordered. Indeed, it's costly and impractical for the manufacturer to install this configuration.
Performing device configuration as part of DCS configuration and then downloading in the field at the time of commissioning can work. However there are certain scheduling anomalies that occur and that one should account for in advance through refined planning.
Minor issues only
While the fieldbus control system is still early in the commissioning and precommissioning phases, we are pleased with its reliability and performance. But there are certain comforts from the old days and ways that we miss.
Inexpensive local indicators, which many process industry cultures are used to having near control valve bypasses, are a problem. Due to the lack of any easily applied and reasonably priced local indicators, we ran a significant amount of conventional wiring to drive 4-20 mA local liquid crystal display meters.
Using separate, conventional analog junction boxes and conduit systems is one of the ugly and expensive penalties of being an early adopter. Nonetheless we're happy to have the infrastructure in place and are ready when such appliances become available.
Compact, neatly packaged, redundant segment power: We're used to having highly reliable I/O power. With the products available at the time we designed BDO, we limited ourselves to applying redundant power only for critical segments, or about 20% of all segments.
The bulk 24 volts DC for all segments is redundant and diode auctioned. But the fieldbus power conditioners, which allow the segment to float and eliminate cross talk, contain active devices and fuses and constitute a single point of failure for the segment.
Our solution for critical segments, provided by our DCS vendor, is reliable and does the job but has a large footprint.
Portable tools for configuration and simulation: There are no handhelds. In our case, the DCS was our handheld. There were also no simulation tools.
If H2 becomes widespread, we can get away from Modbus. Modbus is used for multiplexers, emergency shutdown communications, and assorted other tasks and easily consumed half or more of our configuration time. Modbus works and is reliable, but there is much work to coordinate scaling, addressing, alarming, and "unpacking" of discrete data.
On/Off or digital valves came to the market too late for us to use. Our rack room footprint would be smaller by one third if this type of product had been available when we had to choose.
FF is ready for prime time. BP Amoco will seriously consider FF for future projects. IT
Figures and Graphics
- A twisted-pair or home run cable connects the control room with devices. A terminator (T) at each end of the fieldbus cable allows the twisted pair to carry digital signals. A power conditioner (C) separates conventional power from fieldbus wiring.
- Common ground
- Daisy-chain wiring connects devices along the home run cable with spurs (S).
John Rezabek is the lead control engineer at BP Amoco in Lima, Ohio. He is an ISA member.