Making the Right Choice With Smart Wireless Sensors
Smart sensors have been around for a long time, but when combined with wireless capabilities, the sky’s the limit.
If you are anticipating a wireless future, you can easily spend $6,000 or more for an intelligent sensor capable of preventing and withstanding an explosion, but you probably don’t need to spend that much. Instead you can choose from self-contained wireless sensors/transmitters or use wireless RTU (remote terminal unit) and I/O units with a cluster of wired sensors that will tie into the plant network through a gateway device. To have a successful and upgradeable wireless installation, however, you need to know what you want to accomplish, as well as how much your needs will grow.
With the ISA100.11a specifications complete, processors can choose from new ISA100-compatible sensors and gateways or several already existing WirelessHART devices on the market-from low-cost, “hockey-puck” sensors right on up to the $6,000-plus explosion-proof sensor. In addition, several alternative proprietary solutions are available with price tags ranging from a couple hundred dollars for a temperature node to about $2,000 a point for some RTU and remote I/O-based units that let users cluster wired sensors together for transmitting to a remote data collection unit or Ethernet gateway. The new ISA100.11a specification allows for WirelessHART and other protocols (e.g., Modbus TCP) to be “tunneled” through its system components, giving users the option of using these with ISA100-compatible equipment.
If you are planning an intelligent, wireless sensor network, there are some important considerations above and beyond radio frequency (RF) communications theory. This article looks at some of the practical issues in choosing a wireless sensor network. While cellular technology is a long-haul technology supported by several wireless sensor suppliers, it is not covered in this article. Should you need the longer coverage cellular systems provide without an investment in licensed-radio technologies, be sure to ask your sensor/automation supplier about this option.
Cases for wireless
In some ways, it might be counterintuitive to think wireless sensor networks could provide better solutions than wired networks when interference is an issue. For instance, Craig McIntyre, Endress+Hauser chemical industry manager, relates a story about a wired network at a brewery near Heathrow Airport. As is the case with shielded cables, this Profibus-based wired network could not be grounded at all locations because of ground-loop, induced noise. When the new sensor network was brought online, sensor signals were skewed if not totally unusable. Upon further investigation, sensor network engineers found the radar system at Heathrow was interfering with low-level sensor signals. Some additional network cable grounding procedures solved the problem. While a wireless network is not completely free of interference, in this case, it might have been a better solution.
Wireless sensors can overcome the limitations of wired devices in rotating machinery. Kaitlin Carter, Banner Engineering marketing analyst for wireless products, tells the story of a bottling plant that tried using a level sensor with a signal that passed through slip rings on a rotating filler. Getting a usable signal proved to be a maintenance headache, so the bottler’s engineers installed a wireless node on the rotating machine, and now the system receives a noise-free signal from the level sensor.
Several wireless opportunities exist in a plant or warehouse. For example, if it’s too long a distance to be practical for a cable, or if infrastructure has to be torn up and replaced to route a cable to a sensor on a tank or silo, a single wireless sensor used in conjunction with a wireless gateway attached to the plant network may offer a complete solution. However, if the distance between the two is a little too long, or if there are obstructions in the path of the signal, there are a couple of options. First, mesh networks use multiple sensors, each with a radio transmit-receive capability and the ability to relay signals from one to the other. Second, point-to-point configurations using a single or multiple sensors can have their range extended with the use of repeaters-devices dedicated to retransmitting the signal.
Mesh sensor networks
If a manufacturing plant needs to measure additional process variable points that are located between the first wireless sensor and gateway, further wireless sensors that support a mesh network could be employed. According to Gareth Johnston, ABB’s global wireless product manager, WirelessHART sensors have built-in mesh networking and can support the range and robustness expected for process applications. In a mesh network, each sensor can act as an intelligent repeater, so if the first path isn’t reliable enough to pass the signal, the network will reroute the signal to the gateway using the optimal number of wireless sensors in between as repeaters. In a mesh configuration, three or more wireless sensors will get the signal to the gateway.
Processors with Emerson Process HART-based wired sensors in place and close enough to form a mesh network can convert the sensors to wireless operation for the incremental cost of an add-on WirelessHART adapter. This is helpful when there is nothing wrong with the sensor, but the wiring has seen better days. Rather than pull or change old wiring, the maintenance crew can attach Emerson’s SmartWireless THUM adapter and put the sensors on a WirelessHart mesh network. This device also lets processors gain access to advanced diagnostics for remotely managing devices and monitoring their health.
The THUM technology was recently added to Rosemount’s tank gauging systems, and in the first field installation of these systems, all are operating successfully, says Mikael Inglund, Rosemount technical product manager. “All those tanks had power cabling but no proper signal wiring, so the customers saw significant cost savings by not having to do new cabling work.”
Mesh sensor networks are suitable for applications where redundancy is more important than transmission speed and data rates, says Dan Dudici, Invensys Operations Management business development manager. When speed and data rates are more important, he recommends choosing a point-to-point solution such as the Invensys WiModPak wireless I/O solution, which provides low latency (about 25 ms) and data rates of 250 kbps using proprietary, optimized technology.
Because of the latency of wireless systems, food processors use industrial wireless equipment primarily for monitoring and data gathering and, in most cases, aren’t ready to trust a wireless system to run any controls-at least critical ones. The lack of determinism in a mesh network does not lend itself to control when the network may spend time reorganizing and rerouting signals, says Russ Graybill, Yokogawa general manager of network solutions. In many cases, mesh topology is overkill and adds more complexity than necessary. When sensors aren’t being moved from one location to another, it makes more sense to use point-to-point routing, adds Graybill. To anticipate the needs of its users, he says Yokogawa lets customers turn off mesh protocols and use fixed routing in its mesh products.
The redundancy of a mesh network might be what users value most-so much that some mesh network users think network redundancy and reliability are better than their wired networks, says McIntyre. With the built-in redundancy of WirelessHART and ISA100 protocols, users not only get the process variable from the sensor, but also information about the network’s health-like the signal indicator on a cell phone. But when a wire breaks on a sensor network, users may have little information to know where the break is and how the network is affected, he adds.
If the cost of mesh-capable wireless sensors is an issue, and more than one or several process variables need to be measured at a single site, a good, less-expensive alternative might be to aggregate all the sensor data into a compact, local, programmable automation controller (PAC), I/O system or other device with the capability to communicate over Wi-Fi, says Ben Orchard, Opto 22 application engineer. There are two advantages to this method. First, wireless Ethernet equipment usually has better range than discrete wireless sensors, and second, wired sensors provide users with a wider range of devices-and are more affordable as well.
Should the Wi-Fi option become a little challenging because of distance, processors can add repeaters between endpoints-such as data collectors and loggers and gateways. “Nominal range between our [wireless] loggers and data collectors is about 500 ft., line of sight,” says Steve Knuth, T & D Corp. factory representative. Because coverage can be affected by intervening structures, he recommends the use of a wireless repeater to extend the range, or daisy-chaining wireless repeaters to extend the distance even farther.
Temperature monitoring is critical in the regulated industries, and often a point-to-point wireless system can do the job inexpensively and in a timely manner. Ashish Desai, Omega Engineering application engineer, has seen wireless RTD-based (resistant temperature detector) sensors used in some pharma company freezers. In this application, temperature data is passed to a central alarm system that alerts the maintenance crew as soon as there is a trend in the wrong direction. Other point-to-point applications include tank level monitoring and temperature monitoring inside greenhouses and seed storage units. Users can start out with an industrially packaged sensor (thermocouple [T/C], RTD, RH, pH, process voltage/current) and a receiver (IP68 packaging if needed) capable of working with 48 wireless sensors. The receivers come with software supporting Oracle databases and Excel spreadsheets, and can connect to a PLC or data acquisition interface.
One automated HACCP solution provider in the UK makes use of wireless temperature monitoring, providing time-stamped temperature data logging for small food processors, freezers and retailers. The records satisfy regulators that correct temperatures have been met and held. The Kelsius HACCP solution is paperless and records air and product temperatures in all refrigerated areas. “In dealing with food safety, the system has to be paperless, more efficient and less labor-intensive,” says John Ireland, fresh food manager for the Gala Group. According to Ireland, the system also eliminates the risk of human error during the food preparation process.
It’s often difficult to track where sensors are being used, but Knuth points to popular applications for wireless sensor networks including monitoring the storage condition of certain areas in refrigerated/frozen warehouses where specific areas (microclimates) present varying temperature requirements. Loading dock temperatures are also important, and these wireless systems provide electronic records with time stamps that meet 21 CFR Part 11 requirements. Other key applications include a deli/meats processor that monitors the cooking temperatures of prepared meals and a cheese manufacturer that continuously monitors its process temperatures.
One food processor in the Northeastern US that wanted to monitor several batches decided to use wireless sensors whose data wound up at an electronic chart recorder. Because sensors were always being moved as changes were made to the process, the manufacturer opted to use wireless T/Cs, which transmitted to a wireless gateway connected to the chart recorder through the plant network, says Graybill. With all the location changes of the sensors, putting in fixed, wired sensors wasn’t practical.
Though several suppliers sell wireless hardware (Opto 22 included), most customers are not yet comfortable with doing control wirelessly, says Orchard. For industrial control in food processing applications, wireless control isn’t fast or deterministic enough to satisfy the needs of packaging, slicing, bottling and other high-speed applications, though it may be fine for longer-running batch process operations where sensor data is transmitted at the seconds rate (as opposed to micro- and milliseconds).
Faster sensor update rates will always have a higher price tag, says Dudici, so the practicality of using wireless for both sensing and control will most likely be out of the budget compared to wired systems. But prices will come down as new technologies are developed, standards are in place, and competition intensifies.
Trials are underway to prove a wireless network can support PID control, says Johnston. WirelessHart has been designed with PID control in mind and adds a time stamp to process data, which can be used in a modified algorithm to account for message delays, he states.
Whether or not controls can be connected wirelessly, maintaining a separation strategy has its advantages, says McIntyre. “A field device usually is first connected to a control system via a fieldbus or traditional 4-20 mA dc I/O. The control system then is connected to the IT system through some type of middleware such as a database server. However, for some smart field devices, particularly those involved in process monitoring as opposed to process control, direct connection between the field device and an IT technology platform may be a better solution.
“After all, the smart field device already has available, fully defined engineering values so further scaling is redundant,” adds McIntyre. This approach avoids the deterministic delivery, data management, safety and security issues associated with more-demanding control system environments. Connecting directly from the field device to the IT system also cuts costs substantially-for three reasons. First, a direct connection eliminates the need for traditional hardware components and software systems. Second, the field device in question may not need to be part of the control system’s overall validation and maintenance program. Third, field device access can be controlled through existing IT security systems, according to McIntyre.
“Many environmental applications such as EPA reporting and enterprise applications such as inventory management don’t require the millisecond update speeds or deterministic behavior inherent to most control systems; in fact, there may be prudent reasons to keep environmental reporting infrastructure separate from the control system,” says McIntyre. Some facilities have found up to 70 percent of their field measurement devices don’t have any associated real-time control functions or critical time delivery, he adds.
After the recent Stuxnet virus attacks, most people are more aware of the need to keep their data safe and away from prying eyes. “If someone wants to get in, it doesn’t matter if it’s wired or wireless,” says McIntyre. “That’s [another] good reason to keep the controls platform separate from the monitoring system.” A good way to keep intruders out, he adds, is to have as few remote servicing points as possible.
Fortunately, with wireless systems that use relatively low power, range is not that significant-compared to a licensed, high-powered system. With the limited range, an intruder would almost have to be on site to “steal” the signal. “It would be nearly impossible to break into the system (i.e., pick up one of our packets),” says Carter, “and even if you could, what would be the value of the data? The data would be a Modbus register, so first the intruder would have to know something about Modbus, and second know the context in which it is used.” Breaking into this data stream is not the same as stealing a neighbor’s Wi-Fi, and of what importance would a tank level be to the intruder?
Some suppliers are building in additional security schemes to their wireless systems. Besides using a proprietary communication protocol, T& D designed an optical registration system into its products, says Knuth. To have a product join the network, a user must have physical control of the device and register it to the data collector through a photo-array that records the new device’s serial number. Without this information in the system, the network will not “see” an unregistered device.
Though many wireless sensors can be powered by AC power or DC systems, most have the ability to use internal battery power. The key to long battery life is using the least amount of power possible to minimize battery changes. This is important when wireless sensors are in hard-to-reach or dangerous places. Depending on transmit intervals, some WirelessHART devices can work anywhere from four to 10 years on a single battery.
There are, however, other sources of power, and some are more obvious than others. A simple source involves using solar cells for outdoor applications when possible; in this case, the solar cells are used primarily to recharge or hold up a sensor’s battery. Another source involves harvesting power from vibration-especially if vibration is the process variable to be measured. However, if it’s a new vibration monitoring application (meaning all new equipment), be sure there is enough vibration to power the sensor.
Enough energy to power a wireless sensor can also be harvested from a temperature gradient. “We can use a small (3°C) differential in temperature to create electricity to power sensors,” says Jim O’Callaghan, EnOcean Inc. president. The heat can be harvested from machinery parts, radiators, smoke stacks or even the human body, he adds. The most important aspects of any wireless sensor that harvests energy are that users don’t have to change the battery, the sensor can be completely sealed, and it can work in any food and beverage environment.
With the finalization of wireless sensor standards and research in finding ways to minimize sensor power requirements and harvest power from the surrounding environment, wireless sensors are filling varied roles in monitoring process applications. In addition, processors have the option to cluster several wired sensors in an application and transmit all their data to a wireless Ethernet gateway through wireless RTUs and data collectors. In terms of cost and variety of possible solutions, it would seem the sky’s the limit.
For more information:
Craig McIntyre, Endress+Hauser, 317-535-1368, firstname.lastname@example.org
Kaitlin Carter, Banner Engineering, 763-512-3414, email@example.com
Gareth Johnston, ABB, +44-1480-475312, firstname.lastname@example.org
Ben Orchard, Opto 22, 951-695-3028, email@example.com
Steve Knuth, T & D Corp., 518-669-9227, firstname.lastname@example.org
Dan Dudici, Invensys Operations Management, 847-404-1288, email@example.com
Russ Graybill, Yokogawa, 770-254-0400 (ext. 4467), firstname.lastname@example.org
Ashish Desai, Omega Engineering, 203-359-1660, ext. 2326, email@example.com
Jim O’Callaghan, EnOcean, 925-275-6601
Mikael Inglund, Emerson Process (Rosemount), 800-722-2865