Cutting the wires of communications

By Walt Boyes, Editor in Chief

ONE OF THE most discussed topics in the past year has been the use of wireless communications in process automation. No other topic since the heyday of the Fieldbus Wars has generated so much interest, comment, information, disinformation, FUD (fear, uncertainty and doubt), and sheer noise.

We're going to try to cut through the noise for you, and help you separate what’s real from what’s just hype. We'll take a tour through the wireless landscape, show you the applications, talk about the technology, tell you who the players are, and show you how to write your own scorecard. After all, it’s going to be you end users who get stuck with this technology, and get to make it work in your plant.

We've written about wireless before, both in Control magazine (September 2005, November 2005) and in Sound Off!! (the Control editor's blog), and our sister magazines have covered it as well. We've also discussed wireless several times on ControlGlobal.com's Process Automation Radio Network.

What End Users Want
Herman Storey, senior automation consultant for Shell Global Solutions, and co-chair of ISA's SP100.11 committee says, "Frankly, I don't think there are as many wireless applications as some vendors believe. Those applications that can be done with wires will be done with wires."

Jim Sprague, a member of Control's editorial advisory board, and an automation engineer with a major Middle Eastern oil company adds, "What I want is for the wireless vendors and SP100 to produce standards."

Storey agrees. "Shell will not be installing more than the minimum wireless applications until there is a standard we can work with."

Despite these hurdles, many other end users have told Control that they believe wireless automation systems can be made robust enough and reliable enough to be used, even for loop control. Polls conducted by B&B Electronics ("Wireless Winning Wider Acceptance in Automation"), and by the SP100 committee itself (see "Tell SP100 What You Think" sidebar below) indicate that there’s widespread interest in having the ability to use wireless sensors for process monitoring, for asset management, and even for loop control in what they describe as "non-critical" applications. A minority of end users even believe that wireless communications can be made reliable enough to be used in critical, closed-loop control and even in safety instrumented systems (SISs).

As we’ve reported before, the bottom line is that end users want a standard or a set of standards that they can design to and procure from before they’ll use wireless in their plants other than in demonstration-type applications, or where they absolutely have no other choice.

Such "no other choice" applications abound, at least apocryphally. That is, everybody has a story, but nobody will identify where the application actually was located. There’s the steel plant that uses wireless temperature sensors to protect the water bath around the electric arc furnace. There’s the tank farm with all wireless level sensors. There are lots of others, and many vendors and even some end users will talk about them, but won't name them.

So, it's hard for end users to figure out what's FUD and what's not.

The Wireless Landscape
As we reported nearly a year ago ("Users Want and Industrial Wireless Standard," wireless isn’t like wired communications. "It’s not enough to say, as if we were Rodney King, ‘Can't we all just get along?’” says Dr. Peter Fuhr, of Apprion Corp., and chair of SP100’s interoperability committee. The wireless landscape is tricky, and littered with standards and applications that have little or nothing to do with process automation, with the exception of one glaring fact—they’re all in use in process plants from refineries to biopharma installations. As you can see in Figure 1 below, there are a lot of wireless technologies.

Many standards can mean tough decisions of process control engineers.

In fact, as a senior ABB Inc. official told me recently, the use of wireless in discrete automation and robotics is far advanced when compared to the use of wireless in continuous process or batch process automation. "We're even delivering power to the machine wirelessly," he said.

Many plants, whether discrete, process, or hybrid are already using wireless technologies such as radio frequency identification (RFID) and WiFi (802.11b or 802.11g) both in offices and on the plant floor.

Even though there are more than 100 industrial communications buses, as long as they stay in the wired world, everything is okay, and all you have to have is interoperability between devices on the same bus. So, you can have Profibus, HART, Foundation fieldbus, Modbus, and many more devices in the same plant without worrying that adding a device or a network will cause the rest of the communications buses in the plant to go dead.

However, once you go wireless, you’re in a different world. Interoperability of devices on the same network is still an issue, but it pales in comparison to the problem of [ital]coexistence[end ital] between wireless devices on the same or adjacent networks. Now you have to decide if adding a device or another network node will work, work part of the time, or simply bring down the entire network. Wireless communications are such that you may not even know the answer until you turn the new device on.

The SP100 committee has developed a taxonomy to describe the industrial wireless landscape. The committee has defined five different classes of wireless automation, ranging in complexity from Class 5 to Class 0.

Class 5: Monitoring without immediate operational consequences
The ISA SP100 Call for Proposal states, "This class includes items without strong timeliness requirements. Some, like sequence-of-events logs, require high reliability; others, like reports of slowly changing information of low economic value, need not be so reliable, since loss of a few consecutive samples may be unimportant."

Many experts, like Harris (Hesh) Kagan, director of technology for new business at Invensys Process Systems, and president of the Wireless Industrial Networking Alliance (WINA), believe this will be the largest category of wireless applications in the near future. "The biggest bang for the buck, the low hanging fruit," he says, "is probably going to be in the area of maintenance and condition monitoring, followed very quickly by an opportunity for incremental process measurements."

These applications include all of the condition-monitoring applications that are being used to add incrementally to asset management, as well as all of the process measurements Kagan is talking about.

In the old days, the only process variable that was measured was the one that was being used to control the loop: flow, temperature, level, pressure, density, pH, conductivity, and so forth.

The advent of huge relational databases optimized for process control, such as Invensys' Wonderware historian, Honeywell's PHD, or OSIsoft's PI, make it possible to actually use information from process sensors not directly involved in loop control. The problem is getting those sensors to connect to the historian's server. Wiring up a bunch of extra sensors just so you can see process optimization trends is simply not going to get past the funding gate. Wire is expensive for sensors not directly needed to make the process go. Wireless sensors, on the other hand, have a great deal to offer these applications.

Class 4: Monitoring with short-term operational consequences
"This class includes high-limit and low-limit alarms and other information that might instigate further checking or dispatch of a maintenance technician," reads Call for Proposals. "Timeliness for this class of information is typically low (slow), and measured in minutes or even hours."

This class of applications includes monitoring the inventory level of tanks in tank farms, as well as alarms that are designed only to actuate if the process goes significantly out of whack, but that don’t instantly require a response. Alarms, however, are a special case. They can be of any class (see “Alarms” sidebar below).

Class 3: Open-loop control
This class includes actions where an operator, rather than a machine, 'closes the loop' between input and output. Such actions could include taking a unit offline when conditions so indicate. Timeliness for this class of action is human scale, measured in seconds to minutes.

Class 2: Closed-loop, supervisory control
"This class of closed-loop control usually has long time constants, with timeliness of communications measured in seconds to minutes," adds Call for Proposals. "Examples are batch unit and equipment selection." The quintessential examples of closed-loop, supervisory control are oilfield pipeline and water/wastewater SCADA systems. These have been wireless for two decades, already.

Class 1: Closed-loop, regulatory control
"This class includes motor and axis control as well as primary flow and pressure control," says SP100. "It is a matter of semantics," adds Kagan.

"We get a lot of head-nodding," he continues, "when we ask end users if they want to use wireless for closed-loop control. However, when you peel the layers of the onion away, what we find out is that they’re really talking about using wireless for incremental process measurement. They’re not looking for closed-loop control. There’s no one I’m aware of, an end user, who’s interested in wireless, at this point in their lives, for closed-loop control."

Class 0: Emergency action
This class includes safety-related actions that are critical to both personnel and plant. Most safety functions are, and will be, performed through dedicated wired networks to limit both failure modes and susceptibility to external events or attack. Examples are safety-interlock, emergency-shutdown, and fire control. However, Angela Summers, Ph.D., president of SIS-TECH consultants and chair of ISA's TR84.02 and TR84.04 committee says, "Wireless sensors have not demonstrated the reliability necessary for SIS applications in the petrochemical and refining industry.

Interoperability versus Coexistence
It isn't enough, however, to limit your thinking to just these classes because they don't define all of the wireless applications you'll find in your plant. If you'll go back to Figure 1, you can see a lot of different standards, and each standard is designed for one or more distinct applications. For example, RFID is becoming more and more prevalent, even in process plants. RFID is being used today for worker-location systems, inventory management, and even batch-management applications. So, when you’re producing specifications for your process plant’s wireless systems, you need to take into account the effects of the RFID transmitters in your wireless control system, and vice versa. Perimeter security is another significant wireless application, and workers in your plant, probably you too, are already using walkie-talkies and cell phones, as well as wireless laptops, PDAs, and other wireless-enabled tools (see “Wireless While We Work” sidebar below). So, what will happen when somebody keys a microphone right next to your wireless loop controller?

"There are lots of pieces of equipment in the 'wireless cloud' that surrounds a plant," says Dave Kaufman, Honeywell’s new business development director (See Figure 2 below). "So, you have to make sure that not only are all your pieces of wireless equipment compatible with each other, but also that you have some way of assuring that the next piece of wireless equipment you install won't bring down the existing networks."

There are lots of pieces of equipment in the "wireless cloud" that surrounds a plant.

Kaufman continues, "When you have wired networks, you’re most interested in making sure that every device on the wire is 'interoperable' with every other device, and that there are few if any incompatibilities. This is magnified in the wireless world."

What happens when you try to install too many wireless sensors in what Emerson Process Management's director of technology, Bob Karshnia, and John Berra, Emerson's president, call, "canyons of steel?”

According to Graham Moss, president of pioneering Australian wireless company, Elpro Technologies, what you get is increasing latency. Latency is the time lag between the time the message is sent, and the time it’s received at the address to which it was sent. Currently, as congestion or interference increases, latency also increases. So, if you want your signal to be received quickly (a critical alarm, perhaps), you may be in trouble. "This means that just installing a single, huge network with 10,000 sensors isn't likely to work," says Moss. He predicts "multiple networks, arranged in layers."

This is why the SP100 committee included coexistence and quality of service as two of its main goals. Some of the questions each responder to SP100's Call for Proposals must answer include:

• How does the system deal with poor quality RF links due to high noise, low signal due to long path lengths or obstacles, multipath fading, or interference?
• How does the system adapt to changing RF environments, and loss of key components such as routing nodes, gateways, and network managers? What components can be redundant?
• How does this system coexist with other systems operating in the same band, and how does it coexist with other systems within the facility?
• What is the throughput of the system? With many sensors? How scalable is the system? 
• How does the system deal with quality of service (latency)? How does adding other services affect the latency of existing services in the network?

A Hacking We Will Go
And then there’s the problem of security. The only way somebody can enter bad data into a wired network is to hack that network. Assuming a competent firewall, this is hard to do from outside the plant. In addition, most process control systems aren’t directly connected to the Internet, making it even more difficult to intrude. However, a wireless network theoretically can be hacked by anybody who can pick up the signal.

At my house, I can pick up anywhere from 7 to 12 wireless networks, just in my neighborhood. Most of them aren't even encrypted with WEP (a basic security precaution). There’s more than one reported case of wireless tampering with process systems in water and wastewater plants. Any industrial wireless standard will have to have a significant security component. In addition, those wireless components will have to have some sort of certification that they won't keel over and die the minute somebody tries to attack them.

HART Wireless Update: Draft Spec by End of 2006?
For many reasons, Emerson's announcement in February (see the Process Automation Radio Network’s podcast with Karshnia, Gabe Sierra, Emerson’s product manager, and others, that it’s producing a system for Rosemount to market that’s intended to meet the HART Communications Foundation's draft specification threw its HART Wireless Working Group into confusion. Based on work done at its recent meeting in Minneapolis, however, most observers believe, along with WINA’s Kagan, that "HART Wireless is once again on track, and may very well meet your (Control editor-in-chief Walt Boyes') challenge to have a draft specification by the end of 2006."

Rumor has it that there might even be an interoperability demonstration of a HART Wireless specification at ISA Expo 2006 this coming October.

With approximately 20 million devices already installed, and the specification expected to be backward compatible with every device with HART installed, this is good news for people who want to start using their HART-enabled devices for more than just smart calibration.

SP100 Standard—Where Do We Stand?
ISA SP100 has split into several working groups. The most important of these groups are the SP100.11 and SP100.14 committees. SP100.11 is concentrating on Classes 1 through 5, with consideration for Class 0 applications. SP100.14 is concentrating on Classes 4 and 5, with consideration for the other application classes. There’s some overlap, but fundamentally SP100.11 is working with process measurement and control networks, while SP100.14 is working with long-term monitoring and asset management applications primarily.

Throughout this summer, SP100 member are working on the Calls for Proposal issued on July 14. A proposer conference is planned for early September, and final proposals for the standard are due in early December. Selection of the final proposal is scheduled by the end of January 2007, and draft specifications are expected to be written by August 2007. A period of comment and voting will take place with a standard recommended out of the committee by the end of June 2008.

Assuming an end to the infighting that has dogged ISA standards committees since the initial meetings of the fieldbus standard committee, ISA should approve an S100 standard by October of 2008. Tom Phinney, of Honeywell, who is the convenor of the equivalent IEC standard committee, estimates a similar time scale for IEC approval. Recently, the signs have been good that the infighting will end. Emerson and Honeywell have publicly pledged to turn over any patents and other proprietary intellectual property to the HART Communications Foundation and SP100 committee. "We believe in standards," say Emerson's Berra and Honeywell Process Systems’ CEO, Jack Bolick, in nearly the same words. Peace is being made.

Where's All This Going?
What does all this mean for you, the end user? It means that you'll be able to confidently purchase wireless networking gear for your HART-enabled field devices by early 2007, and know that they’re being built to a standard. By sometime in 2007, some vendors may be producing "draft S100 standard" compatible products, and by 2008, there should be a single standard for industrial wireless in process automation.

Figuring Out Your Own Use Cases
Between now and the time the HART Wireless Working Group and SP100 committee report their standards, there are still a lot of things end users can do to get ready.

Tell SP100 What You Think

WHAT DO you want to use wireless for? Where would you like to deploy wireless? What are the benefits you expect to get in using wireless?

For once, it’s easy for end users to pipe up and tell the industry what they think. The SP100 committee continues to solicit end user opinions on these questions. We encourage you, as end users, to take the end user poll.

Alarms Go Jingle, Jangle, Jingle

ALARMS CAN be of any of the classes in which SP100 has divided wireless applications. For example, Class 0 alarms could include radiation monitoring or toxic gas sensors, many of whose alarms have a completely automated response, like automatic containment processes. Class 1 alarms would be those alarms that would cause immediate shutdown of the process, what the SP100 committee calls, "high-impact process condition." Class 2 alarms would include any normal automated response to a normal change in process conditions, like a flow diversion or a low level alarm causing the control system to begin changing from one storage tank to another. A Class 3 alarm would describe a process condition where the intervention of a human operator is required, such as deciding whether to divert from Tank 1 to Tank 2, or Reactor 1 to Reactor 2. Class 4 alarms would describe an equipment condition, generally, with a short-term response requirement, such as sending a maintenance technician to find out why the low pressure alarm is going off, but the process seems to be operating within nominal parameters. A Class 5 alarm might be to order spare parts to repair an analyzer that isn't critical to the process.

Wireless While We Work

AT THE Harry Tracy Water Treatment Plant in Daly City, Calif., San Francisco Water Department plant supervisor Leland Fong's operators have cut the cord and become wireless workers, using tablet PCs from Wonderware.

"My operators are more efficient," Fong says. "Before, if they had an alarm, they had to go to the control room to take care of it. With the tablet PCs, they can work with the alarm wherever they are, if they’re in contact with the WiFi network."

Fong says that many jobs that used to be two-man jobs are now one-man jobs, which really leverages the technology. "For example," he continues, "an operator can log in that he’s working on the chlorine residual analyzer, so the operator in the control room will see that he shouldn’t worry if the chlorine residual alarm goes off. Calibration of our field instruments has become very easy with the tablet PCs."

Fong's operators are on the City of San Francisco’s WiFi LAN. So, if they have the appropriate permissions and are in range (the Tracy plant has some physical-location issues with hills and tunnels), then they can operate the plant from wherever they are.