This article reviews that the US Department of Energy’s Pacific Northwest National Laboratory in Richland, WA, is developing the self-diagnostic monitoring system (SDMS) with the aim of improving the situation for plant operators. The SDMS differs from current monitoring systems in two ways: its wireless and it has predictive capability. SDMS uses multiple sensors that are housed on radio-frequency tags, each with its own power source. The system monitors current conditions and by focusing on the root causes of degradation, the sensors identify problems and help predict failure of components. The SDMS as well as decision support for operations management (DSOM) systems include protocols in which the sensors can self-check, to make sure that data sets are logical. Potential data corruption from electromagnetic interference in the plant can be dealt with by data encryption as well as by other means, such as using more than one sensor or running current or voltage checks to ensure that sensors are functioning correctly.
Unscheduled shutdowns, maintenance headaches, and inefficient operation of a nuclear power plant can often be traced to the culprits of vibration, biofouling, erosion, and corrosion in the plant's service water system. Available monitoring schemes, even with trending, may provide data on the current status of the system, but they won't tell the operator that a particular pump is likely to fail six months from now, resulting in a forced shutdown while it is being replaced.
The U.S. Department of Energy's Pacific Northwest National Laboratory in Richland, Wash., is developing the Self-Diagnostic Monitoring System with the aim of improving the situation for plant operators. The SDMS differs from current monitoring systems in two ways: It's wireless and it has predictive capability, according to Leonard Bond, senior staff scientist at the Pacific Northwest lab.
SDMS uses multiple sensors that are housed on radio-frequency tags, each with its own power source. The system monitors current conditions and, by focusing on the root causes of degradation, the sensors identify problems and help predict failure of components, according to Bond, who presented a paper on the SDMS system to the American Nuclear Society late last year.
The service water system presents a useful test bed for a prototype wireless smart multisensor scheme. The service water system comprises pumps, reverse-osmosis filters, heat exchangers, and other components that are subject to harmful vibration, fouling, corrosion, and related actions that can result in eventual failure. It also operates at temperatures, pressures, and flows that lend themselves to lab-scale study.
For the purposes of testing the monitoring scheme, researchers built a one-tenth-scale service water system at the laboratory. The essential service water system that is being simulated in the lab is safety critical, because it supplies water to emergency diesels and safety-related heat exchangers throughout the plant.
Many currently operating nuclear power plants were designed in the 1960s, and they rely on human operators and centralized control rooms. Unlike the present control setups, which are wired with all the cables coming back to a central controller, the SDMS system is designed to provide distributed process monitoring and measurement. "What we are doing is using a wireless data network with encryption," Bond said. "Then you can distribute the intelligence, so you have a distributed processing hierarchy." The encryption is a set of software protocols and hardware that was developed at the lab to ensure the integrity of the data against problems of electromagnetic interference or security breaches.
At the overall system level, there will still be a control room, said Bond. But at the subsystem level, computing and monitoring capabilities can be localized, reducing the amount of data that must be passed back and forth to the central control room, he said. "If you lost your central control room, there would still be enough intelligence distributed to be able to take that subloop or system component into intrinsically safe conditions." The distributed processing began with an earlier set of operations and maintenance tools developed at the lab, known as Decision Support for Operations Management, or DSOM, and was developed further with the SDMS project.
Distributed intelligence could be applied right down to the level of components, such as pumps, Bond added. Such a distributed network could improve safety as well as operational efficiency, he said.
The prototype monitoring system is based on wireless RF tags and distributed sensors that provide real-tin1e monitoring of components. The RF tag developed at the laboratory is basically a small electronic circuit board that contains an antenna or coil. The monitoring system uses active tags, which include batteries and can both send and receive data. The active tag is capable of initiating communications and can conU11Unicate over distances in the plant. RF tags are also available in simpler versions, without their own power sources and more limited in communication abilities.
The Prototype System
The monitoring system in prototype uses three tags, each of which can house up to 16 sensors. Three tags were sufficient for the proof-of-concept purposes of the test, although they could be expanded to many times that number in an actual full-scale system, Bond said. Information is sent over a wireless local area network, as well as over a wired LAN for checking purposes. Bond noted that a wired network would likely be used in certain safety-critical applications in an actual nuclear plant. Information from the sensors is fed back to the central control unit via RF tags.
The prototype system includes a programmable logic controller, which also sends information to the central control unit, for checking purposes. Although PLCs would not provide enough intelligence for prognostics capability, they could still have a role in a full-scale system in an actual plant, where they could serve as an interface between the sensors and distributed processing, said Bond. "You still need some control functions, and there is flexibility in how you interface between PLCs, sensors, and distributed processing."
Bond describes the tags as "next generation." He said, "We have added much more intelligence to the tag in terms of doing the process." They are used with a suite of sensors of different types. The tags are capable of performing monitoring and control functions, operating on a two way protocol that can send information both ways. For exan1ple, the tags can be used to convey a signal to change a pump's speed or to alter valve settings. Control functions could come from a central control room or be done at the local level, where pumps could be set to an intrinsically safe operating.
The other capability that sets this monitoring scheme apart achieves a level of diagnostic and prognostic capability by focusing on certain "stressors," or root causes of failures, Bond said. For example, cavitation in a pump will result in erosion. Control of cavitation, perhaps by changing suction pressures, will extend the life of the pump, resulting in better asset management and avoiding component failures and unexpected shutdowns.
The focus is on the actual cause of the degradation mechanism rather than on symptoms, he said. Monitoring the stressor agent-misalignment, or suspended solids, for example-rather than the effects of the degradation mechanism itself, gives a much earlier diagnosis of the process that is causing the system to degrade.
Focusing on the stressor provides "an early diagnosis of a process that is happening that will cause degradation, rather than waiting until you get between a rock and a hard place, where you have already eroded or corroded something or fouled something," Bond said. "By monitoring stressors, one can improve the potential for early diagnosis and also improve asset management."
In the water service system in the SDMS project, the centrifugal pump has degradation mechanisms of cavitation, vibration, erosion, and corrosion, and stressors of flow throttling, misalignment, suspended solids, acidity, and chlorine. The reverse osmosis filter has fouling as a degradation mechanism, and stressors are suspended solids and chlorine. The heat exchanger, which was not included in this year's project, has fouling, corrosion, and erosion as degradation mechanisms and suspended solids, chlorine, and highflow velocity as stressors.
Wireless communication could eliminate the need to have "major looms" of cable, said Bond. Instead, "You will have a little chip with multiple sensors sitting on an individual pump," he said.
Diagnosis and monitoring are pushed down to the component level, and combined health monitoring data, instead of raw data, can be sent back to the central control room to make the information easier for the operator to manage. "It's easier to look at 20 dials than to look at 200," Bond said. "But if something goes out of operating range, he can have his attention focused on what is going on."
Present and Future Aims
The SDMS project is focusing on future nuclear power plants-what are referred to in the industry as generation three and four plants-that could come online after 2020, Bond said. Yet the system is also flexible enough to be retrofitted on present nuclear power plants and can be applied to other types of industrial applications, including conventional power plants and manufacturing process lines.
The SDMS project stems from Decision Support for Operations Maintenance. DSOM is a set of computerized operations and management tools based on plant design information that can be used to improve plant performance and safety. The DSOM system has already been applied to the U.S. Marine Corps' Twenty nine Palms central heating plant in California, its Parris Island cogeneration plant, and a heating plant site of the New York City Housing Authority.
The SDMS project builds on the original Decision Support software to build prognostic capability and integrate it into wireless distributive networks, together with process monitoring and control, Bond said. The Nuclear Energy Research Initiative, a project of the Department of Energy that is focused on next-generation nuclear power plants, funded the project.
Bond expects that SDMS will simplify the tasks of installing instrumentation in new power plants. According to Bond, if the system proves itself on the service water system, it may eventually be applied to all parts of the plant, including reactor core operations. He added that reactor core application is very futuristic.
One benefit of the project is that it should avoid unwanted shutdowns due to component failures, said Bond. It could also expand the intervals between required shutdowns, he added.
"For next-generation plants, they are looking at much longer periods between required shutdowns," he said.
One reason is fuel, which could keep the system sealed for perhaps 10 years, he said, versus a period of 18 to 24 months that is typical in present nuclear power plants. "If you were to do that, you would need to have much better monitoring than you have now."
SDMS is in the second year of a three-year program. "The tags work. We've gotten data streams back, and the basic architecture is all working," said Bond. The next step is to go to trials, the first of which are fouling and cleaning as well as pump diagnostics. The research tean1 will use information from the trials to develop prognostic capability, which includes neural networks. "We have created the flow rate, we have created the hardware, and we have performed the initial trials," said Bond.
He said that during the coming year he expects a full system demonstration, including flow, temperature, pressure, and more advanced instrumentation.
The SDMS as well as DSOM systems include protocols in which the sensors can self-check, to make sure that data sets are logical. Potential data corruption from electromagnetic interference in the plant can be dealt with by data encryption as well as by other means, such as using l110re than one sensor or running current or voltage checks to ensure that sensors are functioning correctly.
Once the system is proved out, plans are to scale it up to a full-scale noncritical system on a nuclear power plant. "You want to get operational experience on systems that are not safety critical," Bond said. "That's the first place people are going to let this get into a plant."
Bond expects to complete the first series of trials by the end of tills year and expand their scope during the next fiscal year. The research team is trying to place SDMS on other types of processes besides nuclear power plants that will also let the technology be developed and validated.