Measure More...Without Wires
Salt air slowly corrodes the internal cooling passages of a U.S. nuclear plant's 2,500-hp water-circulation pumps. Corrosion leads to clogging and inefficient heat transfer, and, if undetected, will ultimately result in equipment failure. As the temperatures of water-circulation pump stator-coils increase, efficiencies and effective operation decrease.
Salt air slowly corrodes the internal cooling passages of a U.S. nuclear plant's 2,500-hp water-circulation pumps. Corrosion leads to clogging and inefficient heat transfer, and, if undetected, will ultimately result in equipment failure. As the temperatures of water-circulation pump stator-coils increase, efficiencies and effective operation decrease. Those temperatures are indicators of impending failure that can result in costs and lost revenues exceeding $400,000 per incident.
Wireless sensors from Sensicast take real-time temperature measurements from each pump's stator coil and, by hopping frequencies around a mesh network, deliver the information hundreds of feet even in a harsh, metal-laden environment full of radio frequency interference. Thus proven, wireless sensors were installed to monitor gas emissions and serial information from equipment in other remote sections of the plant.
Scenarios like this are real and are progressing rapidly. Observe how the Institute of Electrical and Electronic Engineers' (IEEE) 802.11 standards (aka WiFi) have been widely adopted—seemingly overnight—by consumers. While industrial processes generally are less fault-tolerant than the average consumer, industrial implementation of wireless sensors seems to be poised on a cusp, much like WiFi was not that long ago. 'In 10 years 70-90% of new plants should be wireless,' predicts Honeywell Inc.'s Tom Phinney, senior fellow, Integrated Security Technology, ACS Research Labs.
'Wireless, for the last decade, has been an easy sell in non-critical applications,' comments Graham Moss, Australia-based Elpro Technologies' general manager. The company literally got its fingernails dirty in Australia's mines, developing its first- through third-generation wireless devices. He notes that wireless has been a more challenging sell in mission-critical situations.
Early on Moss saw the need for a network approach to wireless communications as more antennae sprouted from control room rooftops to accommodate RF (radio frequency) from different monitoring and control devices. Elpro already provided data transmission interfaces to/from DCS and PLCs, increasing the utility of Profibus, Modbus, DeviceNet, and others.
HART wireless specification—due in early 2006—will serve as a catalyst to drive more expansive adoption of wireless, Moss suggests. 'Big vendors such as Emerson and Honeywell—by embedding $20 wireless HART chips in all their products—will make HART the defacto standard.' He also sees wireless mesh networking as a system driver in providing semi-self-configuration capabilities, similar to a PC's 'plug and play' facility.
However, with evermore information needed at the enterprise and operations levels, there is an increasingly strong need for sufficiently robust and reliable devices to negate the criticisms of the doubters.
Mike Robinson, MicroStrain Inc. vice president of sales and marketing, says he's seen, 'a wide range of industry that's very interested in wireless sensors, including aerospace, automotive, civil engineering, biomedical and such. I've seen growth rate projections of 200% year-over-year, but I can't see that as reasonable [maintainable].' Nevertheless Robinson expects that exponential growth of the market is likely. Indeed, Moxa Technologies Inc. cites West Tech Research's projection that, by 2009, the low-power wireless sensor market will exceed 550 million units.
What will likely drive that growth, says Robinson, is that many in industry intuitively recognize that 'the cost of a wired system is a lot more when the installation is considered with cable, conduit, and the like. It can run triple the sensor and cable cost. Moreover ease of installation [with wireless] can be remarkable. We had a job in the U.S. Capitol for 18 sensors. With wireless, it took a half day; if had been wired, it would have taken a week.'
Necessity is sometimes the motivator as Phinney observes: 'Shell has had a wireless safety loop since the government wouldn't allow a wire to be strung across a freeway.'
'Motivation,' says Adalet marketing and sales manager, Matthew Piecuch, 'includes lower cost, faster and easier installation as well as maintenance compared to hardwired systems. Different architectures are available—we typically use 900MHz systems in the unlicensed instrument/medical/scientific (ISM) band.'
And the level of sophistication is expanding. Robinson says that, less than a year ago, a one-channel strain-gage was on the market; now three channel strain-gages are available. Moreover, seeming insurmountable wireless situations are being overcome: wireless communication was applied successfully in a refrigeration plant by installing a 'sniffer antenna' inside the unit to pick up signals, he says.
Regarding the concerns about frequency saturation or congestion, Adalet's Walker uses an Internet metaphor: 'at one time people were wringing their hands that the number of available IP addresses would be exhausted. That hasn't happened, and there're no more 'chicken littles' running around today.' He's confident that, 'new technology will address such issues, as it traditionally has.'
Providing unique capabilities
Additionally, wireless sensors can provide unique capabilities, such as monitoring the stress/strain or temperature of 'dummy parts' in the manufacturing process. This provides real-time in-process data, rather than data measured along the periphery of the parts production process.
MicroStrain offers an RFID-type sensor with signal conditioning to monitor production in process. However, Robinson laments the semantics—and the image of RFID being inherently cheap. He advises that any RFID-type sensor will cost well over $1 per unit.
Indeed, Adalet Wireless sees an increasing trend with wireless sensor hardware to interface with pre-existing thermocouples and process pressure sensors. Additionally, the radio is becoming more tightly integrated to the sensor.
Some limits do exist to ongoing physical downsizing, such as terminal blocks. And the sensor side is becoming more affordable. Adalet Wireless' Steve Walker comments that wireless sensors may become something of a commodity with postage-stamp size sensors that can be affixed to a tank to transmit temperature levels over a short range.
Unsurprisingly, in Adalet Wireless' experience, the current adopters of wireless sensing technology tend to be in the sectors with the money—pharmaceuticals and energy, especially petrochemicals. The wide spectrum of industry embracing wireless sensors today also includes municipalities and chemical companies, and system integrators. Well-known entities include Pfizer, Merck, BP, and New York City.
A typical approach sees a facility trying wireless sensors on one application and, after getting comfortable with it, expanding to other functions within the plant. One example is tank monitoring, where levels tend to change slowly and update frequencies are lower. Another example is discrete signaling; a switch input can be easily replicated in many locations with little or no wiring.
Adalet works with the customer to address specific needs or concerns. Piecuch said that, in using 900 MHz, there're likely few other competing devices at a plant, such an employee walkie-talkie.
Walker described some creative applications. One included a Texas hospital that, as required—by National Electric Code (NEC) 700.7, 701.8, 702.7—to have notification via beacon or strobe when primary power is lost and generators kick in; this to make alternate arrangements for critical care patients. Another was a ski area where snowmaking equipment, such as water pumps, are managed wirelessly.
In sharp contrast to skiing, Adalet has a long history in underground coal mining, having manufactured so-called explosion-proof boxes (to contain electrical relays and the like) over decades. Recent coalmine disasters have highlighted the potential of wireless communications to provide beacons for miners. A 'relay net' concept could employ 900 MHz signals.
Indeed, hazardous locations are seen by Adalet as a forte of wireless sensors. In Class I Division I situations, explosion-proof boxes allow much higher RF output compared to low-power intrinsically safe radio at 1W output. Walker says, 'It's about 50-times more powerful than intrinsically safe units—and the transmission range is proportional to the square root of power density—so you're looking at seven times the range.'
Issue of energy
Robinson, of MicroStrain, notes that the biggest energy drain is the transmitter, and that's the reason that multi-hop radios are such power hogs—the radio is on almost continuously. So it's unsurprising that one of the biggest concerns is power. It, in turn, tends to be driven by device accessibility—either to a primary power supply or physically. Robinson says a year's battery life is not unreasonable. The obvious energy-conservation approach would see low sampling rates or data stored onboard and periodically transmitted. With a low sampling rate, several years' battery life is possible, he suggests. Others see up to five year's life depending on data update frequency.
Siemens' product manager Jeff Raimo says while energy harvesting costs a few dollars more than battery powered versions today, they should be comparable in cost in two to three years. Today's cost premium is typically below what it costs to swap the battery one time—battery cost, labor, and so on. So, over an expected lifetime of say 15 years, there would be significant savings from a self-powered sensor.
All technology improvements notwithstanding, one fact remains: batteries are problematic for large-scale wireless applications. They have to be monitored for charge, useful lives vary depending on the operating environment, and there are labor, stocking and replacement costs to consider. Furthermore, devices must be accessible for battery replacement, and batteries are fast being categorized as a hazardous waste.
Energy harvesting provides continuous, renewable and ample energy—meaning one can monitor sensors and transmit more frequently than what is typically permitted by battery-powered devices. Potential energy sources include light, vibration, thermal gradients, pressure differential, motion, and piezoelectric (from manual depression of a push-button switch)—present in the environment. Until fairly recently, the amount of energy generated via harvesting was insufficient to send a wireless signal any practical distance. However, improvements to harvesting techniques, combined with very low-energy electronics, means independent battery-less technologies are a commercially viable reality.
Energy-harvesting radios' output is up to 10 mW compared to the 1 mW typical for battery-powered radios, so the range is longer. With existing technologies engineers can now expect transmission distances of up to 30 m indoors, with signal strengths sufficient to reliably transmit through walls and other structural elements.
Key enabling features of these devices are their highly efficient power management technologies and extremely short signal duration. When the devices are not transmitting, they revert to an ultra-low-energy sleep mode. However, when signaled, the devices quickly wake up, transmit a burst of data, and then go back to sleep—all in &0.001 sec.
Currently, solar-powered temperature sensors are among most common application specified today. Wall-mounted piezoelectric switches that operate from energy harvested when a person pushes the switch is another device gaining acceptance. Other applications include solar-powered magnetic contacts in security systems.
Due to the necessity to keep the power draw of a given wireless device extremely low, data must be transmitted in short bursts. This is not an issue for applications such as a temperature sensor that intermittently transmits a small packet of data. But for high-bandwidth applications, such as streaming control or system monitoring data over extended periods of time, energy harvesting just can't generate enough power to handle the job—yet.
Another concern is reliability, because wireless signal can only be transmitted one-way; there is no acknowledgement that the signal was properly received by the intended device. This, of course, is unlikely to be an issue for applications where the user can visually confirm transmission, but for more sophisticated control systems, device status is often a critical aspect of overall system or network integrity.
Beyond the power supply considerations cited, Tendril Networks' CEO, Tim Enwall, observes that adopting companies should anticipate two further software challenges in deploying wireless sensors:
Months may be required to program even basic wireless-sensor networks using the software that end-user companies typically use today; and
There is equal difficulty in adding company-desired control capabilities to such sensor networks.
Indeed, Enwall says that most companies that Tendril works with have cited the foregoing deployment factors and are seeking solutions to get over the software speed bump. He added that the sensor hardware has gotten fairly robust, but system software to manage these networks—a wireless network OS if you will—has been lacking.
Because of these software limitations, companies have needed to employ specialized programming teams with hard-to-find skills to write applications for these deployments. But even with their expertise the programming work has focused primarily on just getting useful packets back and forth between devices rather than developing higher-level enterprise functions.
Enwall says software-related challenges have made wireless sensor network deployments time consuming and costly to date, but new software is becoming available that addresses these issues. For example, Tendril's Service Broker promises to cut deployments from months to days and allow companies to add even complex control capabilities as a simple step in the process.
Physics, a hindrance?
The physical magnitude of a plant may be formidable, limiting the effectiveness of wireless sensors. ExxonMobil's Baytown (Texas) refinery, for example, covers some 40 mi.2—not including the adjacent chemical plant. 'A natural approach [there] would be adoption of a cellular structure overlaying a plant, with plant compartmentalization along the boundaries of plant upgrades,' says Phinney.
Radio propagation can be highly variable due to physics, says Phinney. Movement of a crane or other large equipment can create reflections, fades, and such. Moreover, the favored 2.4 GHz frequency, which is usable in most countries, is the same as microwave ovens. That means any leaky in-plant microwave oven becomes a 1-kW jammer when up against the nanowatt- and picowatt-level signals that sensors are trying to send or receive. And, because about a third of a typical plant's outlets have reversed wiring, such microwave sources jam both phases of the power system. Outside the walls of the plant, there even have been instances where nearby local coffee shops have created massive interference by setting up WiFi hotspots.
Phinney says, 'Clearly, in the longer term, government must designate spectrum for critical infrastructure that has no other legal interferers.' ISA's SP100 standard is taking a strategic approach to coexistence—recognizing that multigenerational instrument radios and instrument mesh networks will coexist simultaneously in different regions of a plant. If a small part of a plant is upgraded, existing technology—from inventory—will be employed. But if say 20% of a plant is upgraded, the then-latest technology will be deployed.
In 10 years Phinney believes that deliberate jamming is likely to be fully avoidable due to technology such as low-power software-defined radios. Because a perpetrator can't know when such a radio is on, high-power jamming won't be feasible without continuous jamming, which indirectly identifies the source. At that point, 90% penetration of industrial markets by wireless will be achievable, contends Phinney.
Despite challenges, industrial applications use wireless sensors to provide performance unparalleled by hardwired systems. The technology exists and the solutions are being successfully deployed. Are you ready for wireless?
Wireless application checklist
With battery-less devices, Siemens cites several considerations:
Is the application non-mission critical (where reduced reliability is acceptable, such as in cases where an interrupted signal’s impact would be minimal)?
Is it a low-bandwidth application where only small packets of data are transmitted intermittently?
Are there a large number of wireless devices using batteries where battery maintenance is a significant issue and/or the location of the devices is such that battery exchange is difficult?
Is the application in a hazardous environment where measures to counter the risk of spark discharge require elaborate and costly measures? The low operating voltages of these devices and lack of cabling may make more sense and provide a simpler solution.
Is the amount of radio frequency (RF) interference in your facility a concern? Because of the ultra low energy of the radio signal used (a result of the extremely short transmission time) the amount of RF “pollution” added to the environment is considerably less than compared to battery-powered wireless devices.
Is the application in an area where the devices need to be sealed and watertight? This may be easier to accomplish with a device that never needs to be periodically opened up for battery replacement.
Tough, real-world applications
Out of the frying pan into the fire. There are few industrial locales more inhospitable to sensors than a steel plant. Extreme temperatures and harsh chemical, metallurgical, and mechanical processes make monitoring a challenge. Yet process observation is critical to maintaining steel plant profitability—any unexpected maintenance shutdown directly hits the bottom line.
In steelmaking, electrical current sometimes arcs to the sides of the furnace, superheating the furnace wall, sometimes overloading its temperature controls. If the cooling system then fails, molten steel could burn through the furnace walls in a matter of minutes, allowing molten metal to pour out onto the shop floor, endangering workers’ lives and millions of dollars worth of equipment, shutting down the plant for a significant period of time.
For one major steel manufacturer, it was critical to find an accurate and cost effective way to monitor and record temperature fluctuations within furnace water jackets. After much research, the company chose a wireless mesh-network of temperature sensors.
With around-the-clock production, there was almost no downtime to install or replace the sensors. They had to be placed or replaced quickly during a brief cool period in the melting cycle. Even during these cool periods the temperature at the sensor site exceeds 125 °F. The need for a quick installation, the high heat, and difficult positioning made installing wired sensors impossible. Moreover, molten steel splashing around would vaporize any wires it touched.
Simple traditional wireless monitoring techniques were out, too. Strong magnetic and electrical fields, plus large masses of steel would interrupt RF transmission, causing communication outages. In an application where a minute’s delay can mean disaster, this was not acceptable.
Placed between the inner and outer walls of the furnace, Sensicast H900 RTD nodes were able to withstand the heat from melting steel, wide temperature swings, powerful magnetic fields, water spray, and vibration.
Smart mesh networking routes data around any temporary trouble spots that might occur. Frequency hopping let the H900 nodes search out the best channel for clear communication, assuring a constant and accurate flow of information to the plant’s maintenance and management teams.
Impressed by the reliable operation of the initial monitors, the plant has since installed a system to deliver data directly to its control system—allowing more efficient preventive maintenance and eliminating unscheduled downtime.
Cleanliness counts. Siemens’ product manager Jeff Raimo says a Canadian manufacturer’s flexible manufacturing system employs clean rooms that have to be moved periodically. A wireless monitoring system saved front-end costs as well as the time and expense required for hard-wired-systems in periodically redeploying the rooms. And he noted that an unrelated U.S. Department -of Defense project employed a wireless network as backup since it couldn’t afford any downtime.
It’s apparent that any combination of necessity and/or plant management focus on applying wireless technology is key to successful deployment of the technology. As successful implementations are seen by their peers, the exponential growth is likely to continue.
Surprisingly, some in industry hold that security is not a major consideration with wireless sensors because the typical range is a 70 m radius and encryption is an option for further comfort. Honeywell’s Tom Phinney, however, maintains that security is an issue. Two principal aspects are:
Secrecy of information; and
Forgery of messages.
Phinney says protection against external actions has no payback, but it must be provided. He observes, “Some say that the information security is not that important, yet today’s neural network capabilities greatly enable cyber thieves. Today’s neural network capabilities allow cyber thieves to simulate a plant, learning its capacity, cost of operation, inventory levels, trade-secret process parameters and recipes. Reportedly, Detroit Edison wanted to put remote meter readers in General Motors Corp.’s plants but was rebuffed for that very reason.
Forgery, on the other hand, is integral to authentication and message integrity issues. Hurricane Katrina showed the vulnerability of the infrastructure such as refineries. Such an attack on infrastructure could be carried out from abroad. Moreover attacks in parallel are feasible against facilities such as refineries, where similar equipment is in use, tied to different control rooms. And one of the comparatively lower threats comes from entities such as the Russian mafia that can use such access and capabilities to extort money from a plant.
Malevolent insider actions can result from employees being suborned or disgruntled. A network IDS (intrusion detection system) is necessary. Deep inspection of messages to assure legitimacy, where messages are interpreted in sequence—not in isolation, is a useful technique, but requires access to unencrypted message content. This puts insiders at risk of discovery—a good deterrent for white-collar perpetrators, but less so for blue-collar sources. A key is avoiding use of uninspectable information conduits. A colleague of Phinney’s from BSI—Germany’s version of the U.S.’ National Security Agency—has the saying, “all trust is limited.” It boils down to: compartmentalization, defense in depth, verification, and periodic testing. Every five years there is a major background investigation of personnel who hold U.S. government security-clearances. There is never a final yes or no answer—only a qualified yes or no.
'Limitless' energy to be harvested
As the technology evolves, fueled by increasing amounts of vendor-supplied development dollars and demand, more complex wireless devices capable of sending larger amounts of data can be expected to go battery-less. The number of applications are equally likely to grow as companies figure out new ways to apply the technology in situations where the absence of wires can generate cost savings and other enterprise-wide facility management benefits.
Siemens' product manager Jeff Raimo cites a Siemens' spin-off company, EnOcean, as among the first to commercialize energy harvesting. He says that, as sensors on a chip (RF and power management electronics) are developed, size and power requirements will shrink, and multiple parameters—such as CO and CO 2 —will be able to be sensed, enhancing wireless sensors' attractiveness.
In that area, Siemens' work on its MEMS (microelectronic mechanical systems) program is aimed at integrating 15 chemical and bio-sensors and micro-security cameras employing patented field-effect-device or FET technology. The future of this expertise is trending toward nanotechnology.
Small solar panels have been employed in very remote applications. MicroStrain's energy harvesting captures power from machine vibration or strain (in a structure, such as a bridge).
In another vein, Honeywell says fuel cells offer 3-5 times the energy density of conventional batteries, but they do have a current limit, and cannot give burst output. Comparatively, the cells also offer a very long shelf life and are well suited to 5-10-sec scan cycles.