Most plants realize the importance of condition monitoring and predictive maintenance for at least their most critical equipment. Many have invested in oil analysis or vibration, temperature or ultrasonic instrumentation in efforts to support predictive programs. But most are not using these tools effectively, either because they don’t have the necessary expertise to take, interpret and act on their readings, or because staff shortages, cutbacks and breakdowns keep shifting their predictive programs to the back burner.
Then there are the plants that haven’t started a program because they simply don’t have the manpower or expertise to begin.
Monitoring the condition of power-transmission equipment is conceptually simple: Analyze the oil and/or measure the vibration, temperature or sound levels continuously or at intervals; detect trends that indicate impending failure, diagnose the fault and repair the machinery at your convenience -- before it breaks.
But the devil is in the details. Practitioners must obtaining consistent, comparable readings, recognize what is abnormal despite external variables such as ambient temperature and load, and understand the equipment and its failure modes well enough to turn abnormal readings into a confident, accurate diagnosis and prediction of time-to-failure. They must be able to do this for many components of any of myriad pieces of equipment deemed essential to plant operations.
“When predictive maintenance was first being implemented 20 years ago, people didn’t really understand the equipment,” says Kevan Slater, director of service for lubrication-management services company Schematic Approach (www.schematicapproach.com). They didn’t know, for example, what type of bearings it had, or which parts were brass. That makes it very difficult to interpret vibration readings or oil analysis results.
“When you know the equipment, you can give weight to the results,” Slater says. “There’s nothing worse than telling a roomful of production people you have to take a piece of equipment down, and they don’t believe you.”
Concerns about condition can go beyond equipment reliability. “There’s some increase in concern about safety-related failures,” says Greg Cober, training manager, Warner Electric (www.warnernet.com). “People come to the realization that it’s better to shut a process off intentionally than as part of an accident.” No matter how quickly the equipment might be brought back online, “It takes a while to bring a process back up from a crushed finger,” he says.
Some of the information required to implement effective condition monitoring is plant- and application-specific; much of it is peculiar to the equipment itself. So why should users have to develop so much of the necessary equipment-specific knowledge? What are power transmission component suppliers and vendors of industrial machinery doing to facilitate predictive maintenance?
A little help from your OEMs
Most equipment OEMS have been slow to provide built-in condition-monitoring capabilities. You might think it’s because they make more money selling service contracts, parts and replacement equipment than they would if their customers could predict failures, but they say it’s a matter of cost and demand.
“Machinery OEMs tend to be very good at mechanical design and metal cutting, but they’re not very familiar with electronics and software,” says Jim McGuinness, Pump SystemAlert program manager, Texas Instruments Sensors and Controls Group (www.ti.com). Some say they want a nonintrusive smart device that can be embedded in a piece of equipment and give information about its condition, but very few are willing to pay for it. They’re waiting for end users to demand it; meanwhile they see it as R&D money with no return.
“End users have to drive this,” McGuinness says. “OEMs are not going to spend any money where they don’t see the return. If users start to insist, OEMs may start picking it up.”
So far, it’s essentially been the local facility’s concern if it wants to do condition-based maintenance, says Ralph DeLisio, business unit manager, integrated condition monitoring, Rockwell Automation (www.rockwell.com). “Most products do not come from OEMs or manufacturers with that ability as standard. But as the cost-per-channel is coming down, some OEMs are starting to offer it as an option.”
Vendors of basic power transmission components like belts, chains, clutches and cylinders do little to facilitate condition monitoring, focusing instead on making their products durable and making sure they’re used properly. “We don’t have customers beating down the door wanting to do predictive,” Cober says. “But they are trying to get longer lives out of our clutch and brake components.”
In the area of drive belts, “We’ve had some discussions about integrating some technologies to improve predictive maintenance, but we concentrate on reducing maintenance requirements and increasing durability,” says Justin Aschenbrenner, vice president, power transmission development, Gates (www.gates.com). “In a lot of cases, belts have been misapplied. People need to get a design that works on the piece of equipment, and we have teams whose job is to get the right design on the application.”
On the other hand, even a simple conveyor belt can warrant a $30,000 condition-monitoring system if it’s big, expensive and critical for plant operations.
“One of our power plant clients has a 12,000-ft. coal feeder belt that’s absolutely essential for keeping the plant running,” says Tom Armold, vice president, product management and marketing, Applied Industrial Technologies (www.applied.com). “If an idler breaks, it can tear the belt for 5,000 ft., which costs as much as a million dollars for the belt plus installation, and takes eight days to repair.”
Belts are now available with a tear sensor that lets them stop the damage within 400 ft., and Applied recommended the switch. “It was not easy to convince them to spend the money, but they did, and within four weeks the new belt tore,” Armold says. “The sensor system stopped the tear within 400 ft., which they could repair for $20,000 in one day.”
Simple power transmission components like gears and bearings are getting some attention from gearbox and equipment makers in the form of pre-engineered sampling points and specifications for condition parameters. For example, “Gearbox manufacturers design the oil flow characteristics so larger particles go into pockets by the drains, so you can’t get a representative sample there,” Slater says. “Now they’re putting in sampling tubes.” Other vendors are doing things like defining infrared targets and supplying equipment with sensor mounts epoxied or bolted in the right places.
“End users have become more reliant on the expertise of gear drive manufacturers,” say Rick Lynch and Bruce Neumiller, product managers at Falk Corp. (www.falkcorp.com). “Advances in technologies such as remote monitoring are helping address their maintenance needs. These shifting priorities mean that a gear drive manufacturer's responsibility no longer ends with supplying a high-quality product.”
Meanwhile, emerging technologies in machine diagnostics are rapidly expanding suppliers’ options for improving support. “An OEM of, say, a generator can create a model containing the components and how they fail -- the failure modes -- and how to monitor speeds, temperatures, etc.,” says Kevin Cavanaugh, chief operating officer, Qualtech Systems (www.teamqsi.com). The model includes the failure modes and symptoms, and how the system will perceive each failure. “The equipment engineers can think of those things one at a time and our system puts them together, develops signatures for different failures and identifies any ambiguities, which can be resolved with measurements or more diagnostics.”
The system is small enough to be embedded on a machine, where it can monitor parameters and diagnose in real time. A technician can connect a laptop or tablet that can upload the diagnosis and use the model to define the most cost-effective troubleshooting procedure. Cavanaugh says, “It’s not a static fault tree, it uses live reasoning with dynamic data.”
The approach has been applied to aerospace and turbo machinery equipment, and was deployed by Orbotech on optical scanning machines for printed circuit board fabrication.
“Orbotech modeled the machine, then opened one up and put in a faulty part,” Cavanaugh says. Without using the model, their best technician solved the problem in 15 minutes, but their average technicians took four to six hours or just gave up. “With the model, the average technicians could fix it in five to 15 minutes, and so could the company president.”
More pre-engineered predictive
Users of traditional predictive technologies are also benefiting from more powerful software, computing and networking. The greatest weakness in conventional temperature, vibration and ultrasound route-running (other than actually getting around to it) is obtaining consistent, comparable measurements to accurately track equipment deterioration. The second major stumbling block is actually charting the data, identifying out-of-specification readings and recognizing and reporting trends that call for action.
Predictive technology suppliers are responding with better specifications for how to take readings and what to expect, and with software and networking options that help get consistent readings into actionable reports.
For example, “If you’re going to use infrared as a predictive tool, you need to have a way to make consistent inspections, organize and document them, and make reports,” says Jason Wilbur, thermography business unit specialist, Raytek (www.raytek.com). Where vibration, temperature, oil analysis and ultrasound instrumentation rely on using the same sampling point for comparable measurements, thermal imaging requires the technician to stand in the same place and use the imager the same way, month after month (Figure 1). Wilbur’s company helps them with stored images. “You can compare images month-to-month to get consistent results,” he says.
A trained thermographer runs the route the first time, and downloads images along with descriptions of the locations, alarm limits and emissivity values. “Next time, it doesn’t have to be an expert in thermography,” Wilbur says. The routing information and images are there for each location, so the technician knows exactly where to go and where to stand based on what the last image looked like. He says, “The person taking the images should understand the facility and the operating conditions, but he doesn’t have to learn thermography.”
Of course, no description of predictive technology vendors’ willingness to help would be complete without a mention of services. Those companies that can cost-justify expert condition monitoring but not on-staff expertise can hire it out, either to a team that comes in to take samples or to an online monitoring service.
“Many predictive programs have failed because, due to cost cutting or loss of personnel, the companies have lost their ability to keep it going,” says Rockwell’s DeLisio. His company takes on control system, temperature and vibration monitoring as an outsource, and is adding new Web-based services that include lube oil and infrared monitoring.
“Where the cost of downtime is high, ROI makes it an easy sell,” DeLisio says. Obvious applications include large pieces of equipment such as power plant turbines and generators where the costs of both production losses and repairs tend to run in seven figures.
Relatively inexpensive wireless instrumentation systems are starting to make the approach cost-effective for less expensive equipment that’s still considered critical or essential, and DeLisio believes it will eventually percolate on down through the plant. “So far, it’s not going into low-horsepower transmission areas due to the cost,” he says. “Getting the cost-per-point down is key, and that’s where we’re working hard.”
Like equipment OEMs, predictive technology vendors are following the end-user money, slowly getting prepared to offer more pre-engineered systems while looking for some demand that will guarantee sales. Industry groups, market researchers and end users like you can help move them along by telling them what you need.
ARC Advisory Group (www.arcweb.com) is doing its part by recognizing the importance of smart condition-monitoring technologies. David Clayton, senior analyst for the influential market research company, has said, “Manufacturers increasingly need plant asset-management and condition-monitoring solutions that go beyond providing reams of raw data, which are capable of pinpointing problem areas and offering appropriate solutions.”
Smart systems proliferating
It can be argued that equipment OEMs and conventional predictive technology vendors are slowly evolving in response to what they see as weak end user demands, but a number of companies that specialize in sensors, controls and software are training their R&D guns on the rising need for full-time, smart condition-monitoring and diagnostic systems.
“Users want to be able to run equipment full-bore, but with feedback so they can maximize output without risking breakdown,” says TI’s McGuinness. His company has developed a system that monitors electrical, mechanical and process parameters -- current, voltage, vibration, flow, pressure and temperature -- to determine the condition of the equipment (Figure 2).
This in-line, continuous data stream of both mechanical and electrical parameters provides real-time information for both operations and maintenance. “Operators are given a tool to allow them to run their equipment on the performance curve to maximize plant output and, at the same time, maintenance is provided continuous data to lengthen essential equipment life,” McGuinness says.
The first manifestation of the TI system on an OEM’s pump was announced at this year’s International Pump Users Symposium. Called Pump SystemAlert, it uses sensors installed throughout a piece of rotating equipment to continuously collect and record maintenance and process data in real time. The data is compiled and stored in the control unit, allowing maintenance personnel to see equipment performance in real–time or during a given period of time.
Multiple control units can be integrated to provide critical process information for rotating equipment throughout a facility. “Continuous, 24/7 data collection and analysis exposes potential problem conditions before they have a chance to develop into a costly breakdown,” McGuinness says. “Repairs can be scheduled when needed, so the problem does not persist and potentially cause system upsets on other equipment in the process. This real–time response can dramatically improve the life and mean time between repair cycles of pumps, motors and other rotating equipment.”
Texas Instruments is not alone in seeing an expanded need for equipment monitors. GE Energy recently started shipping a Bently Nevada general-purpose unit that accepts up to four vibration transducers (accelerometers, velocity sensors, or proximity probes) and can be configured for vibration or thrust position measurements. It simultaneously accepts up to four thermocouple inputs for temperature measurements.
“Users have generally had to choose between high-end offerings designed to monitor critical turbomachinery, or simple devices such as 4-20 mA transmitters that have very limited capabilities,” says Ricardo Artigas, president of energy services, GE Energy (www.gepower.com). “Most small monitoring systems simply don't provide the balance of functionality and affordability that customers require for their general-purpose equipment.”
Applied Industrial Technologies recommended SKF’s Copperhead fault-detection system for rock crushers in a client’s concrete aggregate plane. Crushers that go down unexpectedly have to be emptied by digging them out with shovels, which can take four to six hours.
“Sensors can be added by the crusher manufacturer or by retrofit to keep track of bearings, gear mesh, oil viscosity and temperature so outages can be planned and performed during the seasonal shutdown,” says Bob Boyle, applied product manager, power transmission. “SKF has a program called Copperhead where they come in and determine the needed sensors and setpoints.”
Copperhead fault detection uses vibration and temperature techniques to monitor the whole machine for abnormal conditions. It can detect faults like loose parts, unbalance, gear damage, bearing damage, lack of lubrication, impacts and high temperatures. Faults can be reported by periodic monitoring with handheld data loggers or by monitoring the machinery continuously through connection to an automation system.
Going a step further by combining condition monitoring with an automatic cure, Lord Corp. just introduced a new version of its automatic balancing system that reacts more rapidly and precisely to report and correct vibration in rotating equipment.
The RealTime system consists of five parts: the balancer, actuator coil, sensors, controller and software. Sensors continuously monitor the vibration levels, and when an increase in vibration is detected above a predetermined level, the controller initiates an advisory message as to the change in state of the unbalance. It then instructs the balancer to make an automatic correction. Rather than covering up or masking a balance problem, the system continuously reports the state of the system's unbalance, providing diagnostic information to predict system health and optimize service outages.
Get critical
Equipment, predictive technology and condition-monitoring vendors are offering more support for plants that want to move up to predictive mode. The methods are not yet so simple and expensive that you can use them everywhere, but they are cost-effective enough to offer attractive return on investment for critical and essential equipment.
Your first step is to identify the places where predictive should pay off. Look at each piece of equipment and its role in operations. Categorize it based on its value to the organization: What does it cost? What’s the impact if it goes down? Can production run without it? Is it important for safety or environmental compliance?
Go over its history and determine the best strategy for maintaining it. “For a $5,000 gearbox that’s not critical, that might be breakdown maintenance,” Slater says. But consider getting predictive for those power transmission components that can shut the plant down. “When you really know the value of the equipment, it not only lets you optimize maintenance,” he says. “It also gives impetus to the program.”