Confidence in Condition Monitoring

May 1, 2004
Today’s information technology makes monitoring the condition of machinery viable in more applications.

Lean manufacturing can pay handsome dividends, sometimes even in unexpected places. Such was the case at the Ideal Division, in St. Augustine, Fla., an operating unit of Stant Companies that makes hose clamps for automobiles in both the original equipment and replacement markets. The plant is reporting a significant increase in uptime since it switched on the condition monitoring capability that came with a new human-machine interface (HMI).

The engineering staff had installed the interface on the plant’s eight automatic clamp assembly machines while upgrading the machines’ control system to accommodate new in-line stamping units. Because the monitoring capability was there already, the engineering team decided to activate and use it—and thus joined the growing ranks of manufacturers using such systems to boost the reliability and productivity of their equipment.

“When we finally got all eight machines (now 10) on the network, we saw the potential immediately,” recalls Steve Gilmour, electrical engineering supervisor at Ideal. “We have been developing that capability ever since.”

The assembly machines produce each clamp from a screw, housing and band, completing one every second. Each machine notches and cuts the band to length, crimps a housing to the band hydraulically, rolls the band, and then drives a screw into the housing, winding the clamp to the specified diameter. For a smooth integration into this process, the new stamping units needed a control system that could coordinate four servo axes, provide a recipe-driven, standard operator interface and connect the machines to the plant’s computer network to give technicians access to component and spare part data.

Because Ideal practices lean manufacturing, its engineering department had standardized on several subsystems, each from a different manufacturer. For example, the programmable logic controllers (PLCs) came from Omron Electronics LLC, in Schaumburg, Ill., and the PC830 servos were from Pacific Scientific, in Rockford, Ill. “We wanted to take advantage of PacSci’s comm800 DLL to wrap all the motion systems into one custom operator interface,” recalls Gilmour. “This meant an HMI with solid scripting capability in Visual Basic.”

For this reason, he and his team specified the heart of the control system to be Cimplicity HMI Plant Edition software from GE Fanuc Automation Inc., in Charlottesville, Va., an affiliate of GE Infrastructure. “Its powerful Visual Basic scripting capabilities allowed us to tie in some custom DLLs for motion systems,” says Gilmour. “And combined with the appropriate PLC driver, we were able to wrap these systems up into one neat bundle with little difficulty.”

Moreover, because the software uses Web and supervisory control and data acquisition (SCADA) technology, information is available over the computer network. From their own screens, operators and team leaders can monitor the status of both the machines and the jobs flowing through them.

Management and engineering also can view reports that a subroutine running on the engineering server generates from relevant shift data compiled on the fly. Among the reports are analyses of fault data and work stoppages. Because the software collects, processes, and flags problems as they begin to develop, “a poorly running machine is now identified immediately, instead of in a daily or weekly production report,” says Gilmour.

In one case, the software helped his staff to find the cause of a higher-than-usual number of unwound bands. After the software drew engineering’s attention to falling uptime, a quick query into the database allowed an engineer to identify the common denominator, a screw that several machines had been using. As part of the company’s Six Sigma program, the department is using this information to recommend ways to tweak machines and even alter components to prevent the problem from recurring. All of this boils down to “more stable processes and more efficient manufacturing as we lower costs without compromising quality,” says Gilmour.

Benefit from the shakes

A number of strategies and tactics exist for collecting condition data and interpreting it for taking corrective action. Some involve simply analyzing feedback signals inside a control loop and any fault signals that might be generated by the controllers. Others take the more proactive approach of adding sensors to measure and track changes in a descriptive parameter. For the later approach, many experts recommend tracking vibration, especially on rotating equipment.

Because vibration is a periodic phenomenon that is the sum of various specific, repetitive events, vibration waveforms are actually a string of signatures that repeat to form a kind of event log. They, therefore, are a valuable database of process conditions to anyone who knows how to decode them. Using condition monitoring software or some other type of spectrum analyzer, knowledgeable people can break vibration waveforms into their components. The signatures of these components not only indicate whether a machine is exceeding a control limit, but also point to the reason.

“You can’t do that with traditional process parameters like temperature, pressure and flow because the data is more static,” says John Pucillo, manager, service programs, Rockwell Automation, Integrated Condition Monitoring (ICM), in Milford, Ohio. Rather than being a string of repetitive events, these data are averages by nature and, so, contain no historical details. Consequently, they are useful only for making go-no go decisions. For example, an increasing temperature on a bearing does not point to the cause; it only suggests that corrective action is necessary.

Although vibration analysis is the workhorse of monitoring the condition of rotating mechanisms, other techniques can either supplement it or sometimes even replace it entirely. Lubricant oil analysis is one, particularly for machinery that has oil reservoirs for gearboxes, hydraulic bearings and the like. Although the practicality of the technique depends on the application, its implementation is similar in each case. A technician withdraws an oil sample from the reservoir and sends it to a laboratory, which measures physical properties of the oil, such as viscosity, and looks for contaminants.

Such analyses uncover clues that speak volumes about the condition of the equipment. The oil’s viscosity, for example, will decrease as a lubricant degrades, and a suddenly abnormal viscosity can indicate that someone has added the wrong makeup oil. The presence of water or dirt can alert maintenance technicians to a failing seal, and the presence of metal fines is an important clue for detecting and finding wear. “If you see an increase of say, lead, in the oil and if the babbitt material of the bearing contains lead, you can tell that the bearing is wearing,” notes Pucillo.

Some condition-monitoring software vendors, such as Rockwell ICM, have laboratories that offer such analyses as services, sending their customers either a hard-copy report or data in electronic format ready for use by their software. Such services allow small facilities to derive some benefit from various types of analyses without having to invest in the otherwise necessary equipment and staff.

The evolution of information technology has done much for automating condition monitoring and making it practical for a wider range of manufacturers. An important thread in this evolution has been the explosive growth of computing power, which has allowed the development of more capable, user-friendly analysis software. “Artificial intelligence isn’t to the point where software packages actually are making decisions based on the data,” observes Pucillo, “but there are techniques that can help you weed through the data much more effectively.”

Advances in communications are another thread in the evolution of information technology that has made automating condition monitoring more practical. For example, the object linking and embedding (OLE) programming interface and ActiveX and other plug-ins allow software developers and users to collect data from various sources and display it all in one place. Condition-based maintenance information can flow rather easily from a machine to a computer screen anywhere through an OPC connection. Consequently, merging oil-analysis and vibration data into one plot on a control interface is relatively easy today.

Despite the importance of vibration analysis to machinery with rotating components, other “static” forms of measurement are valid, and even critical, methods for monitoring the condition of equipment without rotating components. Infrared thermography, for example, is the primary technique for identifying failing components and other problems in electronics, such as motor control centers, electrical distribution systems and transformers.

Problems in such equipment appear as hot spots on temperature maps. Consider a motor control center overseeing a three-phase motor. “If you have a loose connection on one of the phases, you’re going to get a hot spot,” notes Pucillo. The technique also can identify hot spots in rotating equipment and in chemical systems such boilers, heat exchangers and reactors. Build-up of fly ash in a fossil fuel-fired boiler, for example, can create a hot spot that could weaken a boiler tube and lead to a leak.

Header pressure, rather than temperature, was the crucial parameter for Citizens Thermal Energy, the second-largest provider of steam and chilled water in the United States. The boilers operated by this division of Citizens Gas & Coke Utility provide steam and hot water to more than 230 businesses in downtown Indianapolis for a variety of purposes, such as sterilizing surgical equipment and heating facilities in the winter. Because the plant has no back-up system, the reliability of its boilers and control system is of paramount concern. So, the utility upgraded its control system to improve its ability to monitor their condition.

Built in 1893, the plant has two groups of boilers, each with its own control room and network of distributed electromechanical controls. Part of the upgrade was to link the control rooms using OperateIT Process Portal from ABB Inc., in Norwalk, Conn. Besides making the feedback loop 20 times faster and keeping the process in better statistical control, the software allows the operators in the control rooms to keep better tabs on their boilers.

Moreover, it gives them greater freedom to take care of developing problems. Through the computer network, the operator in each control room now has access to the other’s network and can monitor the other’s boilers. “If one has to check a piece of equipment in the plant, he can notify the other operator that he will be away from his screen,” says Orlando Chamberlain, technical manager at Citizens Thermal Energy. “The other operator would then pull up his screens and have full control over them.”

Another benefit is that the plant manager receives efficiency reports continuously in real time and can deploy the plant’s resources accordingly. For example, the plant typically has a supply of five fuels on hand: coke oven gas, natural gas, fuel oil, coal and trash. “Based on this report, he can decide on the best way to use our fuel,” says Chamberlain.

Pucillo, at Rockwell Automation ICM, urges first-time users of condition monitoring technology to start small and build on their successes. “Pick an area of your plant that is critical and has been giving you problems,” he says. “Start there, keeping the program small. Then move forward.”

The most important part of moving forward is having a plan for acting upon the collected data. “Condition monitoring programs aren’t successful if they are just collecting data,” explains Pucillo. “We’ve seen time and time again customers who collect gigabytes of data but still fail because they didn’t treat the program as collecting, analyzing, and correcting. They didn’t close the loop.”