3 Approaches to Process Plant Wireless
One of the words circulating in industrial wireless discussions is infrastructure. Just as there is wired networking infrastructure, any situation where there will be more than a simple point to point device installed will require additional equipment to support the application. For someone considering where and how best to use wireless networking for industrial plant floor solutions, the topic...
One of the words circulating in industrial wireless discussions is infrastructure. Just as there is wired networking infrastructure, any situation where there will be more than a simple point to point device installed will require additional equipment to support the application. For someone considering where and how best to use wireless networking for industrial plant floor solutions, the topic of infrastructure deserves careful analysis.
This article looks at three major wireless providers or alliances: Emerson Process Management and Dust Networks; Honeywell Process Solutions; and Apprion and Invensys.
WirelessHART: Shortest path to interoperability?
They were not chosen with the belief that they are necessarily the best; many vendors in this space offer similar platforms or overlap at strategic points. They were chosen for the following reasons:
Each has clearly articulated its strategy and approach over a relatively long period;
Each has the resources and enough recognition in their areas to influence potential buyers; and
Each represents a different approach with specific benefits and weaknesses.
Comparing and contrasting these architectures can help you examine your own needs and then choose one or examine other offerings.
Connectivity without infrastructure
In October 2006, Emerson Process Management launched its Smart Wireless instrumentation system. The company later identified Dust Networks as a partner providing the embedded wireless module.
TSMP communication sends information from one device to another until it reaches the gateway. Here, information from the device on the right make two extra “hops.”
This introduction reflects Emerson’s assertion that wireless does not have to be complicated. With the right equipment and software, systems effectively form themselves. “Nobody says they want to go out and buy wireless infrastructure,” says Steve Toteda, vice president of marketing for Dust Networks.
“There’s a lot of excitement today around wireless,” adds Bob Karschnia, vice president of technology for the Rosemount division of Emerson Process Management. “We introduced our Smart Pack as a simple way for people to try it easily and with a limited number of devices. We have other customers that are ordering equipment and installing it as fast as they can. That’s really a big statement, because people in oil refineries and chemical plants and power plants are all busy today, but these people are taking time out to get these installed and are actually using them.”
Emerson’s wireless devices function as transmitters, receivers, and repeaters. They form communication links with each other, support their network partners as needed, and report to a larger system when there are disturbances in the links.
All instrumentation in this scenario is battery powered, and only one gateway point needs external power and a wired network connection. Dust Networks says its TSMP (time synchronized mesh protocol) is very efficient with power consumption, allowing battery life of several years and longer even in adverse conditions. Consequently, there is effectively no need to add wiring to the communication devices.
Everything in a Dust/Emerson platform is optimized to maximize battery life, or at least make it predictable. Individual devices minimize power consumption and communicate data in bursts of 10 ms. Instruments not practical for battery powering will require mains or some other method. Some instruments create data that is difficult to send in short bursts, such as vibration analyzers. These problems can usually be solved, but consider the application, and the functions of required instrumentation.
Under Emerson’s current application of the SmartMesh XT protocol, the shortest reporting interval for an instrument is 15 seconds. However, individual devices can be called upon by neighbors to relay data even when they have no data of their own to send. Devices closest to the gateway see the most relaying activity, however the network can be optimized to ensure that no single devices have to bear more than their share. This spreads activity through the network and stabilizes battery life. Emerson predicts that with effective optimization, any device, regardless of its position in the network, should operate at least 5 years.
Of the protocols examined here, the Dust / Emerson solution with its multi-hop mesh networking has the highest potential for latency in communication since data may have to jump from device to device before reaching the gateway. This does increase transit time, although the practical effect is debatable in the context of an instrument’s data reporting interval.
Under this approach, each hop adds 31.2 ms, so if an instrument is 10 hops from the gateway (which could be the case for some on the fringes of a huge network) latency is 312.5 ms. More typically, a network will require three or four hops, so latency will be about 100-135 ms. In most installations, up to 30% of the devices in a network talk directly to the gateway, so those have no extra hops. If an application is critical enough that this type of latency is an issue, discuss it with any system provider. There are strategies to mitigate latency, but some can reduce battery life.
Emerson chose this approach based on reliability, says Karschnia. “We tested everything, but none of it worked the way customers expect it to. We need something that’s robust and can tolerate the huge electromagnetic disturbances in a plant.”
Of the three architectures discussed here, the Dust / Emerson solution has no capability beyond instrumentation—for now. “Emerson made a strategic choice to start with simplicity and introduce wireless mesh solutions at the instrumentation level,” explains Jane Lansing, vice president of marketing. “Field sensor networks can be easily installed and deliver significant value without the need for investing in a plant-wide wireless infrastructure. Customers have been taking advantage of this since we started shipping our field products in December 2006. As we have deployed these field networks, we believe the market is now ready to consider higher performance wireless networks and the applications that run on them.”
At present there is no timeline; however, Emerson’s Global User’s Exchange, in September, has frequently been a venue for new product launches.
Lansing said Emerson wants to “make wireless easy and standards based. We want to make wireless solutions accessible to all customers and not lock them into single vendor proprietary technologies since we believe customers will prefer to purchase products for non-automation applications from already established vendors in this space. Therefore our approach is to partner with the best in class supplier of the customer’s choice for this complementary equipment.”
Honeywell’s approach meshes at the multi-node level rather than between instruments.
Infrastructure can add versatility
Honeywell Process Solutions introduced its OneWireless platform in 2006, promising a combination of simplicity and versatility. Honeywell’s earlier introduction, the XYR 5000, available since 2003, is a single purpose sensor network without additional functionality. The OneWireless approach is more involved than Emerson’s, but Honeywell contends a small amount of infrastructure provides greater versatility.
In its early efforts, Honeywell learned “that there’s a very finite amount of air space available at these plants, and a lot of them today are already using other wireless applications,” says Jeff Becker, Honeywell’s director of global wireless business. “Because wireless is a shared medium, you start to hit interference limits. You can only run so many applications at the same time before they start to step on top of each other.”
Honeywell decided to coordinate wireless use rather than have many point solutions that could interfere with each other. “There’s a lot of management overhead that comes into it,” says Becker. “A wireless network is easy if you have one point solution, but when you have two point solutions and you have to manage between the two, it gets more complicated, and it keeps growing. Customers don’t want to have to manage it, so OneWireless made a whole lot of sense.”
Honeywell’s OneWireless uses externally powered multi-nodes scattered throughout the coverage area. These communicate with each other and provide a backhaul to bring information to a gateway where it passes into the larger wired network. Individual instrumentation devices do not mesh with each other or act as repeaters. They simply broadcast data to the multi-nodes, which take it to the gateway. The meshing action takes place between nodes, not at the instruments. For safety, at least two multi-nodes receive data from each instrument, creating a redundant path.
Honeywell has created a network with low latency and very deterministic communication which are critical for closed loop control functions. Alex Chernoguzov, Honeywell engineering system architect explains, “While we know initial applications will predominantly include slower-speed monitoring and alerting, we intentionally architected the system so that customers can, in the future, add control and higher speed data communication using existing wireless infrastructure. If you design a system to monitor only, it is hard to upgrade it to accommodate control. If you design the system for control, it is easy to do monitoring. So, we intentionally designed it with high reliability, high throughput and a high degree of determinism and latency control to support multiple-use applications.
“In addition, customers asked us to support fast alerting. Many people don’t understand the difference between monitoring and alerting. Monitoring means you send a value back periodically no matter what. Alerting means that you send a value back immediately if a threshold is reached. A good example is a level sensor for a tank. For monitoring purposes, you might only need the value updated every 30 seconds.” During an overflow, “you don’t want to wait 30 seconds, you want to know right now. That’s alerting, and it looks a lot like control because it requires minimal latency, guaranteed delivery, etc.”
Honeywell chose a variant of SP-100 (an ISA standard still in progress) for its instrumentation, but recognizes that others may prefer an 802.15.4 platform. Honeywell developed SP-100 based devices believing it more robust and efficient in bandwidth use. However, it has prototyped a multi-node for 802.15.4 radios as well. An end user can use both approaches if both types of multi-nodes are deployed.
The nodes also provide connectivity using 802.11 a, b, g, communication to bring wider bandwidth WiFi applications to manufacturing operations. This supports other functions such as:
Wireless to standard laptop computers;
Walk-around HMI and control room interface;
Employee locating and tracking; and
Plant floor voice and video communication.
To support wider adoption, Honeywell expects to make its platform available to other instrumentation vendors. “This isn’t designed to be proprietary,” says Becker. “We want the interface to be standard and so we encourage other vendors to work with us or work with others to develop these radios.”
At this point, no other providers have publicly promised participation, but market forces will drive manufacturers to satisfy demand as it emerges. Honeywell has been broadening its offering of instrumentation, undoubtedly with this in mind, to dispel any concerns from potential users about limited availability from a proprietary offering.
Honeywell Process Solutions considers itself the most experienced of this group within the wireless arena. If all Honeywell divisions are counted, it claims to have deployed 35 million wireless devices (many related to its building management product businesses). In addition its XYR 5000 point-to-point systems have received broad adoption (relatively speaking) with more than 500 user plants cited.
Infrastructure first, then instrumentation
Apprion’s network claims the widest ability to interface with a variety of devices and manufacturers.
The third wireless technology is driven by Apprion and its industrial partner, Invensys. Their position begins with two key assumptions.
First, wireless has huge potential but must be approached with a plan. A plant-wide assessment is a critical first step, and there has to be appropriate network management throughout. Both Apprion and Invensys cite studies that say wireless networks require more management than comparable wired counterparts. Users that build haphazardly will ultimately find a mess of conflicting communication.
Second, they say, no single vendor technology (as characterized today at least) is sufficient to exploit all that wireless has to offer. Therefore, infrastructure has to work with a variety of approaches and vendors.
Ian McPherson, vice president of products for Apprion says, “The needs of the application are going to drive people to one of three propositions: you can have low latency, high bandwidth, or long battery life. You can’t reconcile all of those in one technology. Our ability to be technology agnostic doesn’t mean you should have anything and everything, but you can have choice of multiple vendors to find the best for any application or networking challenge.”
This being the case, McPherson expects to see plants with multiple sensor networks operating side-by-side depending on the nature of the tasks they perform. The Apprion infrastructure will keep them all working together cooperatively, he says.
Moreover, a well managed wireless network can perform any wired functions and more. Apprion describes its infrastructure as a cloud that covers a plant and can do virtually anything. Invensys supplies applications that operate within this cloud, and when fully deployed, it can integrate with any wireless instrumentation platform, and handle walk around HMIs, personnel tracking, condition monitoring, voice and video communication, etc. An end user can select devices from any supplier, using point-to-point or mesh networking. Invensys characterizes this as “both vendor-independent and standards-agnostic.”
Apprion’s hardware approach is to deploy externally powered nodes, called Ionizers, around the plant. Like Honeywell’s OneWireless, these have multiple radios to communicate with each other and a variety of devices. Ultimately, all information travels via a WiMax backhaul that connects all the Ionizers. However, the Ionizers will not necessarily be able to communicate directly with all individual wireless devices. For example, a group of Emerson Smart Wireless instruments may still use the Emerson gateway as the access point. If that gateway can be wired to the larger plant network, it might not communicate with the Apprion system at all except to share security and network management functions.
Designing a wireless network begins with a site survey to determine all relevant end user requirements and the plant layout. From that survey Apprion can recommend the quantity and locations of Ionizers. Apprion also recommends and provides ongoing network management and maintenance as a service.
This approach will have the least appeal to a company who wants to approach wireless incrementally. The infrastructure has to be built before applications can be implemented, and that represents a major expense compared to a buying just a few points. The infrastructure is scalable but not really designed for small installations. But once the system is in place, applications are simple to add since most possibilities should have been considered in the site survey.
System design and management play to Apprion and Invensys’ strengths, since neither have wireless process instrumentation. This is in contrast to Emerson and Honeywell, which each have much hardware available and more on the way. Hesh Kagan, strategic technology director at Invensys Process Systems, explains, “We’re developing wireless instrumentation, but many of our systems today are being used with competitive instrumentation in the wireless world, just as it is in the wired world. We’re about working with customers to develop an ongoing strategy. Wireless is just an enabling technology. Our whole message is the InFusion model of integrating the whole enterprise.”
Both companies stress the capabilities of the network to support large scale integration from plant floor to enterprise. Just as Invensys calls InFusion the “world’s first enterprise control system,” the intention is for its wireless platform to support similar thinking.
One interesting element of this discussion is the convergence of technologies. All three are moving in the direction of offering full scale, managed wireless infrastructure solutions. Given the potential for interconnectedness, does that make the choice of initial vendor all the more critical or much less important?
“Some day there may be software defined or cognitive radios that could run different protocols on a single radio, but that’s many generations from now,” says Apprion’s McPherson. Until that day, approaching wireless technology, like any technology, should begin with understanding your applications.
WirelessHART: Shortest path to interoperability? While there has been much discussion about wireless standards at the instrumentation level, an organization that has a reputation for creating interoperability is blazing a path that shows promise. The HART Communication Foundation, known for its HART protocol for wired instrumentation, has made much progress in drafting a wireless protocol that it hopes will be just as widely applied. (Bear in mind that the protocol is not currently complete and has not been formally accepted, so technical statements here may change.)
Given the current discussions in the marketplace among competing vendors, this protocol has potential to be misunderstood if not studied carefully. It is intended to cover only the instrumentation level and does not extend into more extensive wireless applications.
WirelessHART covers all aspects of instrumentation mesh networking, to the extent that any devices made by any manufacturer according to this protocol will interoperate fully with each other and appropriate gateways. HART, known for forward and backward compatability with its wired protocol, aims to replicate that benefit and other functionality in WirelessHART applications.
“Users have expectations of what they want,” says Ron Helson, HART executive director. “In the wireless world, we felt it was essential to preserve the same user experience, to ease the transition. So interoperability was also important to ensure that devices from different suppliers will work in the same wireless network. Buy your field devices from company‘A’ or ‘B’ and a gateway from ‘C.’ It’s a HART device, it just doesn’t have any wires.”
“There’s general excitement about this,” says Robert Shear, director of market development for Dust Networks. “The trump cards of HART are the installed base, and existing tool chain. People understand HART. They rely on it to be simple and practical, which will allow HART to do well in the market. WirelessHART and [ISA] SP-100 do not have to be incompatible, but people are still trying to figure out how the two groups can work together.”
Here are some key facts:
Like Emerson’s SmartWireless, multiple-hop communication in a WirelessHART network has potential to add latency. “WirelessHART has 100 TDMA slots per second,” says Wally Pratt, chief engineer for HART. “This means each hop costs 10 ms, so a 4 hop deep network would add 40ms to devices generating 1 second measurements. A 4 hop network (i.e.ocating the access point near the place where fast updates are required.”
How many devices can be used in a network communicating with one gateway? Pratt answers, “This is application specific. What we recommend is that you schedule no more than 25-30% of the available bandwidth. This leaves room for ad hoc communications and retries, etc. Consequently, the number of devices you can have in a network depends directly on combined scheduled process data rates. For example, if the average update rate is one minute, there could be 1,200 to 1,800 devices per access point. An average update rate of one second supports 25 to 30 devices per access point.”
With the installed base of HART and vendor support, this protocol has significant potential to shape the direction of wireless instrumentation. Emerson, for example, expects to adopt it, even though its current Smart Wireless offerings are not compatible. Will a migration path be available? “Our customer base is pretty diverse,” says Karschnia. “We have a slew of options on how to migrate from what they have today to WirelessHART. As an industry leader, we want ensure that our customers recognize that we’re here to support them with the existing products they have today and with migration paths.”
Peter Welander is process industries editor. Reach him at PWelander@cfemedia.com .