Intelligent Components Simplify Maintenance
| By: Kristin Lewotsky, Contributing Editor
Maintenance has an image problem: Despite all that it does to keep the plant churning out products, it’s perceived as a cost center. It consumes resources. It frequently requires that assets be taken off-line, which on the face of it reduces productivity. Most of all, it costs money in terms of staffing, tools, and equipment. All of these factors make the maintenance department vulnerable any time budget cuts hit.
It’s an unfair situation. Considered in the larger context, maintenance is actually a revenue driver. Proper maintenance maximizes throughput, minimizes downtime, optimizes quality, and positions an organization to peak profitability. An hour of unplanned downtime in an automotive plant might cost $50,000 in lost production, while an hour of downtime in a semiconductor fab represents closer to $150,000. Taken in this context, the money “saved” by skipping maintenance is easily exceeded by the cost of downtime resulting from poor practices.
Even when downtime is avoided, neglecting maintenance can have an impact. In a CNC machine tool, for example, wear can begin affecting the dimensions of parts. It can start out small, trending in a negative direction but still within tolerance. It can be tempting to delay maintenance by a week or a month. Sooner or later, though, it will be a problem. “Eventually, you get to the point where you’re out of spec and it just isn’t good enough,” says Tom Baric, engineering manager custom products/applications, Parker Hannifin Corp. “Maybe it hits on a weekend. Now, your machinery is stopped and you’re forced to call in the maintenance guy on a Sunday. It’s just as bad as a catastrophic failure. If you continue to run your machine until you’re producing parts that are out of spec, it shuts you down one way or the other.”
As a discipline, maintenance has evolved over time:
- Reactive maintenance: repairing/replacing failed components
- Preventative maintenance: performing tasks to prevent faults
- Predictive maintenance: using sensor input and data to repair/replace components as they near end of useful life
- Proactive maintenance: addressing machine issues that could cause failure if left unchanged
The most expensive approach, and the one with greatest impact on overall throughput, is reactive maintenance. Running equipment to failure results in significant amounts of unplanned downtime and lost production. Depending on the component, it can lead to a chain reaction of damage in connected parts. If the equipment in question supplies elements for other steps in the process, it can trigger stoppages on other parts of the line or even other plants in the organization.
The reactive approach can also be expensive in terms of replacement components. With a line down, maintenance techs under pressure to get things moving again will frequently swap out components indiscriminately without taking time to troubleshoot. Once they are producing project again, they package up all the swapped out components that are still under warranty and ship them back to the vendor for repair, not knowing whether they are broken or not. As a result, integrators and vendors frequently receive returns of fully functional equipment, with one sales engineer estimating that an average of one component in 25 returned actually has a failure.
Meanwhile, the swapping process itself can introduce additional issues. If the good component is never put back in place, the company has used a spare unnecessarily. In some cases, remembering to swap the healthy component back in can create a bigger problem. “What maintenance departments don’t understand is when you put that perfectly good spare into the machine and run it for a while, the machine might introduce a problem in the spare,” says Ron Morgan, Operations Manager, K+S Services Inc. If a maintenance tech swaps out the spare when the “faulty” part comes back with a clean bill of health, they could actually be putting a bad part back into inventory, which could present a serious problem during the next unplanned downtime.
And when it comes to reactive maintenance, there will always be another unplanned downtime. “Once you get into where you’re really reactive, it’s the most unpleasant situation for a maintenance professional to be in because you can’t win,” says Morgan. “It’s constant, one thing after another. It’s very discouraging.”
At the very least, companies should institute preventative maintenance programs, probably none more important than lubrication. According to an analysis by Applied Industrial Technologies, improper lubrication is responsible for roughly 40 percent to 50 percent of bearing failures, alone. Lubrication separates moving parts, it decreases friction, which minimizes wear and heat (see figure 1). Done properly, it can significantly increase the lifetime of a machine. This seemingly simple task is fraught with pitfalls, however. All too often, even maintenance departments with the best of intentions can create problems for themselves.
The primary pitfalls in lubrication include using the wrong lubricant and applying it incorrectly. The key characteristics of a lubricant are viscosity, type (mineral oil or synthetic), thermal characteristics, protective characteristics, and stability characteristics. Oil is a fluid lubricant while grease is simply oil with additives and thickeners that hold the oil in place. One common mistake users make with greases is to attach too much importance to the thickener type (lithium, polyurea, etc.). The most important characteristics are the viscosity and the additives. These need to be matched to the application and compatible within any one component. Mixing unlike lubricants can lead to dramatic changes in consistency and early breakdown.
In order for lubrication to provide protection advertised, it must be properly applied (see figure 2). Best practices for lubrication include, somewhat counterintuitively, minimizing replenishing activities. More is not necessarily more--any time the reservoir is open, it presents the opportunity for error:
- Using the wrong lubricant
- Introducing moist air, which can cause lubricant breakdown
- Introducing contamination, which acts as an abrasive, causing the lubricant to damage the very surfaces it is supposed to protect. This, in turn, creates more contamination in a spiral toward component failure.
- Overfilling, which can change pressure, affecting the way the lubricant is distributed in the system. In the case of greasing a motor bearing, the lubricant can push grease past the seal into the windings, causing overheating and premature failure.
For optimum results, the maintenance department needs to develop standards and processes. How clean should the oil be? How much moisture is acceptable? Who should be allowed to handle lubrication activities? After all, even the best lubrication program can be undone by a well-intentioned weekend operator dumping in oil from the wrong can. Operator manuals and labeling can help prevent these mistakes.
Lubrication in a facility with a large number of lines and axes can be time-consuming, especially for operations running 24/7. In such cases, quarterly or monthly scheduled downtime may be so packed with action items that there is neither manpower nor time available to deal with lubrication. That doesn’t mean that it’s smart or effective to ignore the topic altogether, though. A better technique is to rationalize lubrication activities, putting more effort into activities that bring the greatest benefit.
Start by identifying the components that need lubrication, then evaluate components on a case-by-case basis. Where can lubrication maintenance intervals be extended? Synthetic oils in general tend to last longer. Another approach is known as bleed and feed, in which you refresh the oil by draining out a portion and adding more while the line is still running. It’s not as good as a complete change out, but it refreshes additives. When using this technique, it’s essential to ensure that oil is not allowed to fail--adding new oil to oxidized oil will simply oxidize the new oil.
Another technique is to identify components for which lubrication activities can be skipped altogether. In these cases, the benefits may not justify the risk for the time required. Depending on the duty cycle, “lubed for life” bearings may be available to last a lifetime of the machine. For longer lifetimes, running components to failure may still be a more efficient approach.
Inspecting the oil regularly can allow organizations to stretch maintenance intervals. Of course that loads one more task on the shoulders of already overloaded maintenance techs. Or does it? “Anyone who’s familiar with machinery can be trained very quickly to inspect,” says Jarrod Potteiger, Education Services Manager and Senior Technical Consultant at Des-Case Corp. Tasks like monitoring oil levels; looking for negative characteristics like darkened, cloudy, or bad smelling oil; and checking for leaks are easy enough to do with a little training (see figure 3). “The operators are already working in close proximity to the machinery, so they can perform routine daily inspections. That frees maintenance staff to focus on activities that require a greater depth of knowledge or skill.”
Other preventative maintenance includes cabling, which is the single most common point of failure in motion systems. In this case, it’s difficult to look for advance signs. Cable tends to flex and fail, so it should be replaced on a manufacturer-recommended schedule. Preventive maintenance should also include other useful, if mundane, activities like cleaning, wiping down surfaces, and replacing seals.
Preventative maintenance certainly beats no maintenance at all. Many companies plan around regularly scheduled downtime and replacing components at established intervals. The problem is that depending on the performance and demands on the machine, that schedule may be too frequent. “You’re not getting the full life out of your components,” says Derrick Stacey, Solutions Engineer at B&R Industrial Automation. “You’re replacing them before they reach the end of their useful life and that is just throwing away money.”
At the other extreme, equipment may not be replaced frequently enough. After all, the time between failures is only a statistical likelihood--in practice, components may not last nearly as long. Premature wear not only introduces unscheduled downtime, it can potentially impact components elsewhere in the machine, decreasing their lifetime, too. Having access to detailed, real-time diagnostic information allows users to monitor their equipment and take action.
Today’s devices contain diagnostic features that can monitor component health and wear. A drive can monitor the number of times it has been powered up, for example, tracking the number of inrush charges on the DC bus. The capacitors on the bus have limited lifetimes. Obviously, a system powered on every eight hours undergoes more cycles than one that runs 24/7. Still, the machine builder probably takes that factor into account during design. The real problem pops up when the system is restarted more frequently than expected. “If somebody is hitting the e-stop three or four times a shift, it will undergo a much larger number of inrushes,” says Bryan Knight, automation solutions team leader at Mitsubishi Electric. “Knowing how many times you’re actually getting an inrush versus thinking you’re only getting one per shift is critical to building an effective maintenance schedule.”
Monitoring motor temperature can also provide early warning of problems. Perhaps there are packaging issues caused by too many heat-producing components in a single spot. Maybe a fan or cooling system has a blockage that is compromising its ability to manage heat. Electronic design software only estimates system lifetime and performance. Intelligent components with onboard diagnostics allow the user to compare performance and look for issues.
Machines today feature accelerometers that can be used to monitor vibration. Every machine has characteristic resonances. When a bearing, for example, begins to wear, the baseline signature changes as characteristic damage frequencies emerge. Rather than merely catching failure before it happens, vibration analysis can flag a problem early in the process. This gives the end user the freedom to choose their response. Instead of replacing the bearing because a certain number of hours have passed, they might calculate that it will remain functional for days, weeks, or even months. They can leave it in place up to a safety margin--say 80% of lifetime--which is a good balance between maximizing use of the resource and avoiding unplanned downtime.
It’s an approach that can be applied not just to a bearing on a shaft but to more complex systems like CNC linear actuators. “Now, the user doesn't have to use models or worry whether they got the right specs from the supplier,” says Stacey. “They can have direct access to an area of the machine by accelerometer and understand exactly what’s going on.”
Depending on whom you ask, proactive maintenance may simply be a program designed to use preventive and predictive maintenance techniques to minimize downtime. More recently, a more sophisticated version of proactive maintenance has emerged. This involves using predictive maintenance techniques to detect problems early, but then using that information to identify and address the root causes and improve machine performance. This not only extends the lifetime of the component in question but, potentially, lifetime of the entire machine.
Servo motors are tuned during commissioning to optimize performance. A mechanical problem like a slack belt changes the stresses and performance requirements for the axis, altering its response. “We recently shipped a demo machine that got rattled around pretty badly in shipping,” says Knight. “One of the servo motors was starting to make a good bit of noise because the belt tension had loosened. It was just a small amount but now the resonance point had changed.” The adaptive auto-tuning had compensated for the shift and allowed the machine to continue to operate, but the performance had suffered.
The short-term approach to address this problem would be retuning the servo axis. There was a time this was done manually but more recently, vendors have released devices with auto tuning capabilities sophisticated enough to adjust tuning at the touch of a button. This is traditionally performed when the machine is built but with proactive maintenance, it can also take place on the plant floor. During the time the mechanical issue persists, auto tuning can adjust parameters to deliver best possible performance. Once the problem is fixed, auto tuning can be initiated again to return performance to baseline.
Although this type of auto tuning is significantly easier than the manual approach, it still takes time and effort. State-of-the-art components now offer continuous auto tuning. The components analyze the data in real time and adjust to compensate for any changes without intervention by operators or maintenance (see figure 4). This can be useful not only in the case of system problems but also to address changes to tooling or load, which speeds and simplifies changeovers.
Auto tuning should be applied properly to ensure that the technique does not hide mechanical issues. Let’s return to our belt example. The change in resonant frequencies not only highlighted the problem, it also provided information about the solution. If the new frequency was higher than baseline, the belt would have been tensioned too much. In this case, the resonant frequency was lower than baseline, indicating that the belt was not tensioned enough. Performing auto tuning without capturing that information would have covered up the problem. The machine would have continued to operate, but never as fast or effectively as planned. “By going back and having the system self characterize these things rather than just masking them with auto tuning, you can validate some of the maintenance functions after they’ve been performed,” says Knight.
Ideally, a system with real-time auto tuning would also include a supervisory function to monitor the process. If it detects a resonance offset from the frequency of a preconfigured filter, it would take action. “It would flag maintenance that a resonance point has changed in the machine and say, ‘I am going to correct it, but you should probably know that this has happened,’” says Knight. This type of system would not be able to diagnose the problem, but providing a warning would give maintenance an opportunity to investigate before harm has taken place. As before, once mechanical problem has been addressed, the system can return to baseline, which can be confirmed by diagnostics.
Real-time auto tuning can be a very good solution for the bearing example discussed above. In this case, the bearing exhibits signs of wear but to a tolerable degree. While it remains in place, continuous auto tuning optimizes machine performance to maximize throughput even while the user gets maximum return on their capital investment.
Putting it all together
Establishing an effective maintenance program takes commitment, especially for companies starting at the reactive stage. “Really, maintenance should be proactive in everything that it is doing,” says Morgan. In his previous position at a manufacturer with a 1.6 million sq. ft. facility, Morgan led the transformation from operations characterized by unplanned downtime to a phase of smooth functioning and high productivity. First and foremost, he says, organizations need to commit to the process. “The hardest part is to change the culture. There’s a perception that maintenance is strictly in a facility to fix things when they break. It took us seven years to go from a reactive maintenance regime and into planning some predictive technology. It’s a long process.”
No matter where your organization operates on the maintenance continuum, you can always strive to advance. Today’s smart components make it straightforward for any organization to improve productivity. It’s merely a matter of justifying the cost. In these cases, the maintenance department’s most powerful tool is not a laptop or a screwdriver but a record of faults and downtime. Documentation not only facilitates predictive maintenance, it makes it easier to justify the cost of maintenance to management.
“You have to have some means of capturing your data,” says Morgan. “If you do nothing else, build a simple Excel database because then you can start going back and looking at your history of failures. Every opportunity you get, show the cost and then talk about what it could have been if you had performed it during a planned outage. That’s probably one of big things that helped us get support in the plant. Once we got started, we routinely got capital funding to do upgrades because we could show hard numbers. As long as you stay on that course, at some point I think that you will get the support that you need.”
Get the Most Out of Your Bearings
Troubleshooting Tips – Lubrication
Troubleshooting Tips – Avoid Bearing Failures
Understanding Lubricants -- Part I