Rx for Instrumentation
Plant productivity depends heavily on the condition of process instrumentation, from sensors and transmitters for flow, level, pressure, or temperature to analytical instruments and valves. What's needed to keep field instrumentation healthy and operating properly without breaking the bank? Here, experts in the field suggest some actions to take and discuss the relative merits of EDDL vs FDT/DT...
Plant productivity depends heavily on the condition of process instrumentation, from sensors and transmitters for flow, level, pressure, or temperature to analytical instruments and valves. What's needed to keep field instrumentation healthy and operating properly without breaking the bank? Here, experts in the field suggest some actions to take and discuss the relative merits of EDDL vs FDT/DTM (see sidebar) for presenting valve parameters.
Rx#1. Make the right choice
The first step in maintaining process instrumentation is to choose the right devices in the first place—right meaning appropriate for the application.
For example, don't buy sophisticated intelligent instruments to do a low-priority job.
For critical processes, however, choose intelligent instruments with F OUNDATION fieldbus, Profibus, or HART communications. These types of instruments are especially important when accuracy is crucial, says Tom Wallace, global marketing manager, PlantWeb Emerson Process Management, because 4-20 mA systems involve a digital-to-analog converter in the transmitter and an analog-to-digital converter in the control system. Both are subject to errors and drift. And, he adds, digital transmitters provide a great deal of information.
Use isolated transmitters; even though most DCS and PLC inputs are isolated, there is often enough electrical noise from drives and other sources to introduce errors. Isolated transmitters can help minimize the effects.
Don't depend on switches alone to control pressure, temperature, or level, because they can fail undetected. Use them as backup to an analog signal, which produces a continuous output that can be monitored.
Rx#2. Install equipment correctly
The best instrument in the world will do a bad job or fail prematurely if installed incorrectly. Pay attention to:
Device placement. For example, don't install a pressure transmitter on top of steam pipe or the condensate will continuously drain out, and the sensor will be exposed to live steam 100% of the time.
Pipe runs. Make sure there is enough straight run, or a flow conditioner, upstream of a flowmeter (see illustration).
Mounting. Mount instruments in the right orientation: up for gas, down for liquid.
Routing. Route impulse tubes carefully (see illustration). If practical, eliminate them.
Power input. Make sure power is reliable and free of electrical noise, surges, and spikes.
Grounding. Some instruments—magmeters, for example—must be grounded properly. Ensure shielding is grounded at one point only.
Ambient temperature. Operating a sensor or transmitter outside its rated temperature range will shorten its life.
Corrosive atmosphere. According to Scott Saunders, vice president of sales and marketing at Moore Industries-International Inc., some people in the chemical industry will "take an RTD [resistance temperature detection] measurement to a transmitter; they'll install it—a brand new application—and everything is working great. They'll come back to their preventive maintenance schedule six months later, remove the RTD leads, stick in their RTD source simulator, dial in a particular temperature, and call back to the control room to make sure it's reading 4, 12, and 20 going through the scale. Once it checks out, they install the RTD right back onto the leads of the transmitter. Now they've re-introduced the biggest error of the whole measuring system: corrosion of the 3-wire RTD leads." A 3-wire RTD assumes the lead impedance will not change. In a corrosive atmosphere, the lead resistance can increase over time. A 4-wire installation minimizes this problem.
Class I Div. 1 installations (see illustration): "It's amazing to me the number of Class I Div. 1 housings—explosion-proof housings—that are not sealed all the way down to all their threads," says Saunders. "I've even seen some of our transmitters in Class I Div. 1 housings where, probably during start-up, they were calibrating the instrument, wiring the instrument, and they set the Class I Div. 1 glass top, but didn't screw it down tight all the way to the gasket. Those threads are there for a reason: to cool that gas before it hits the atmosphere."
Once an instrument is installed, it is vital to verify proper setup and installation. Use device diagnostics to check for installation errors. If a device has a simulate mode, use it to check and make sure everything works, adds Emerson's Wallace.
Rx#3: Leave it alone
Once an instrument is installed and working properly, it is often best to leave it alone unless it sends an alert. "Top-tier devices are so accurate and so stable, you don't calibrate them when you get them in the plant," says Wallace. "You take the factory calibration, then zero them once they're installed in proper orientation and have had process fluid applied to fill impulse legs. Then you re-zero them and go away for a decade." Recalibration, unless needed for re-ranging, he points out, is often counterproductive. Check the manufacturer's documentation. "Factory calibration will almost always be better than calibration at the plant or in the field," he says.
This practical approach works for two reasons, says Moore Industries' Saunders: First, "these are reliable devices. They aren't very expensive. If there is a major problem, you can just throw it away and get another one." Second, digital instruments have internal diagnostics. Monitoring their status will warn if something goes wrong.
Some people may be reluctant to abandon periodic recalibration. The way to be convinced it is unnecessary is to build a database. A documenting calibrator can help here, but it's the database that tells the tale. Some people, says Saunders, will invest in "newer technologies like F OUNDATION fieldbus or Profibus, but they don't put the appropriate middle layer in, like an asset management optimization package to take advantage of digital data such as status bits, calibration, and drifting."
Such a set-and-forget philosophy doesn't apply to validated process devices in safety service and mission-critical measurements, or to those with components that can wear or foul or are in severe service. Analytical devices and valves fall into this category. Devices installed in corrosive, abrasive, severe vibration, or extreme-temperature environments, also need extra care. And, the concept doesn't apply to valves with moving parts that change their behavior with time and use.
Rx#4: Listen for signs
One way to tell if an instrument is healthy is to check it for process noise. A reading that has not changed in weeks may be the mark of a stable process, or it may indicate a failed instrument. "I'm reminded of an analytical installation where a technician said, 'this pH has been rock-solid on 7.2 for 18 months,'" says Emerson's Wallace. "That's because for the past 18 months the probe has been totally fouled and hasn't actually measured a process variable in that time. A little process noise is probably an indication that your device isn't dead."
Some transmitters include a several-second damping function that can obscure noise that indicates it's working properly. However, if you have a device using F OUNDATION fieldbus or Profibus, noise should show up in the digital data, even if it might not be apparent in the process variable. It may also be useful to track the standard deviation of the process noise. An increase, says Wallace, "may indicate entrained air in a process stream, slugged flow, or plugged impulse lines." In oil and gas applications, it may reflect sand in the flow.
Rx#5. Pay attention to secondary variables
Many modern instruments gather data on more than one variable, even if they don't normally report it. Many flow and pressure instruments measure temperature as well—sometimes process temperature, sometimes the instrument's temperature—and can keep a record of the reading. "Electronics have certain ambient conditions they like to be around," says Paul Schmeling, product marketing manager for pressure products at Emerson Process Management. For example, "a PRT [platinum resistance thermometer] that sits on the circuitry can tell whether the electronics are being abused more than they should be."
An instrument can also be set to remember pressure excursions it has experienced. These abnormal conditions can shorten instrument life or cause actions such as zero shifts. A device can be set to signal when it's time to run a check or to indicate such circumstances as heat tracing left on when it should be off, or vice versa.
Rx#6. Be careful what you ask for
The drawback to watching all available data from all digital devices in a system is a so-called alarm flood. A process upset may create alarms from so many sources that operators are overwhelmed and unable to respond quickly, with potentially dire consequences. Perhaps the best-known U.S. incident of this kind was the near-meltdown at Three Mile Island nuclear plant in 1979, but there have been others, both in the United States and other countries.
The best way to prevent such a situation is to put in an alarm management system (six alarms per hour is one industry benchmark), but some judicious programming in a DCS can help a lot. If there is no emergency, it's often best to ignore minor errors that have little potential to cause problems.
"A lot of these devices have both warning limits and alarm limits so that they can offer a philosophy of 'kind of a high' and a 'high-high' rating for how urgent the problem is," says Charlie Piper, product manager of fieldbus programs at Invensys Foxboro. "They might have 20 or 30 warning and alarm limits in these devices," he adds, "so the system should be able to look at those limits and trigger a work order in some cases."
Rx#7. Convert data to information
All the data available from intelligent sensors and transmitters do no good without correlation. Some users, says Moore Industries' Saunders, install "all the tools to get all the information, but they lack the reporting process or the software level that can actually provide the predictive maintenance needed to get a return on investment or proper calibration." An asset management system can save considerable maintenance costs because it can create information enabling change from preventive maintenance to predictive maintenance.
An asset management system is also very useful with valve positioners, says Foxboro's Piper. "You can track how many times a positioner has reversed direction. Frequently, you can also track the total amount of stem travel." And you can tell when something is going wrong, he explains. "There are parameters in most valve positioners, such as excessive deviation, which would tell you they are developing higher friction. You're commanding the valve to move to a certain position, and it's not getting there." By looking at the 20 or 30 variables that some of these devices provide, it is possible to trigger work orders as needed, he says.
"If you have the proper software and asset management system in place," adds Saunders, "you can test every so often on valve characterization. You can look at component drift. You can do reports that push out predictive maintenance (PM) schedules." He mentions a user who, by counting how often a particular valve went to a seated position, discovered that it did so infrequently enough to not need a PM schedule. What's interesting here, he says, is that the determination was made not with an asset management system, but with a simple HART interface module that triggered an input to a DCS whenever the stem position was less than 1%.
Data gathered and correlated by an asset management system make it possible to develop strategies for predictive maintenance, but in certain cases old-fashioned PM or even run-to-failure are better options. For example, losing a level transmitter on a tank that takes a week to fill or empty may not be an emergency.
Overall, if the newest technology is installed or if it's a non-critical element, often the wisest course of action is to do less instead of more. "If a device or final element has parts that can wear or foul, and the measurement has a safety, environmental, or significant economic impact, preventive or predictive maintenance is in order," says Wallace. "In most cases, change the mindset of preventive or predictive maintenance from 'go out there and do something' to 'monitor and, if necessary, initiate automated diagnostics.' One of the greatest causes of device problems is going out there and doing something. If the consequence is small, ignore it until it fails, then replace it."
Presenting valve parameters: EDDL or FDT/DTM?
The best way to communicate between a field device and a control system has been the subject of considerable debate. "Fieldbus wars" have seen supporters of F OUNDATION fieldbus, Profibus, and various other protocols struggle for dominance. Disagreement has also arisen about the best way to handle settable parameters in a field device and present them to the user. On one side are supporters of Device Description Language (DDL), which has become Electronic Device Description Language (EDDL). On the other are those who prefer Field Device Tool/Device Type Manager (FDT/DTM) technology.
With the first and older technology, manufacturers create Device Description (DD) files containing essential data about each device. The host then takes these data and presents them to the user. "The field device manufacturer describes the association of parameters and how they ought to be grouped and shown on the display, but the precise format is up to the host," says Martin Zielinski, director of technology for HART and fieldbus at Emerson Process Management. FDT/DTM technology was designed as a standardized environment for field device maintenance and configuration. A DTM is a device maintenance and diagnostic application that plugs into the host system's engineering and asset management system. Unlike DD language, a device vendor's DTM plug-in includes a user interface to manage and maintain the device. It lets the device vendor create the host application, and control how the interface should be navigated and what information should be shown.
Recent changes have blurred the distinctions between the two technologies in some ways. On the EDDL side, enhancements have added a graphical capability, "where you could draw charts and graphs, and [also] a persistent data capability with ability to store information on the system hard drive and then retrieve it in the future," says Zielinski.
The enhancements, says Foxboro's Piper, "start to organize the configuration screen. Instead of having just a flat list of parameters, [users] can start to organize them into different screens and groupings and tell you which ones should be presented as a pull-down choice. This improves the appearance of the screen, from a fill-in-the-blank kind to a more form-like format."
Despite enhancements, DD has limitations from the host manufacturer's perspective, says Piper. "We can't build a user interface for every valve because each is different," he says. "The valve manufacturer has to do it." FDT, he maintains, is the best technology for doing that. (See diagram.)
Emerson's Wallace, on the other hand, believes that "EDDL has significantly more potential to deliver maximum gain with minimum sacrifice." As evidence, he cites support levels. "There are about 1,200 device types that currently use EDDL. There are around 110 that use DTMs." And, he adds, "EDDL is a global standard supported by all host vendors."
Piper disagrees. "Right now, every system vendor except one has declared support for FDT technology," he adds.
Wallace does suggest an arrangement that gives some credence to each side. "My recommendation is to require EDDL and use it as the standard technology. When EDDL, and application packages or 'snap-ins/snap-ons' that use EDDL, are not available for a needed function, consider DTMs."