It’s one of the oldest jokes in manufacturing. “Why is this repair bill so high?” demands the accountant. “Twenty quid for the part; £5,980 for knowing which part,” retorts the engineer. It illustrates the simple principle governing the whole science of preventive maintenance: it’s usually a lot cheaper to replace something before it breaks than to fix it afterwards.
As manufacturing has steadily reduced manning levels and increased automation, the truth of this axiom has become even more apparent. The cost of line stoppages and lost production, especially in continuous processes, far outweighs the price of effective planning and unplanned preventive maintenance (PM). What is more, the recession has brought its own imperatives: with many companies hanging on by their fingernails, there is a real pressure to defer the replacement of plant and machinery, nursing old installations to keep them going for a while longer. It’s the kind of climate that has brought condition monitoring, one of the key ingredients of PM, into its own.
At its simplest,
condition monitoring (CM) demands nothing more sophisticated than a good pair of eyes. Leaks, ear and corrosion can often be spotted by a skilled operator long before they have time to cause real trouble and, as more manufacturers develop first line maintenance to production staff, visual inspection has a vital part to play in any workplace. When, however you can’t see inside the machinery, it’s too complex to monitor visually or it is in a hazardous environment, human eyes and ears aren’t much use. For that reason, technologies like
vibration monitoring,
oil analysis,
ultrasonic surveys and
thermography have become increasingly important, either through scheduled application or, for critical equipment, through constant, real-time CM.
All rely on the collection of data about current operational performance over establishment parameters which, by comparing their compliance or deviation from an expected norm, can give early warning of problems. When failure could have serious safety implications, systems are often linked to auto shutdown functions, triggered at the first indication of trouble. The level of sophistication in monitoring systems varies hugely. Some merely record information for interpretations by a skilled engineer; others use computer analysis or even artificial intelligence to predict failure.
The capabilities of CM technology are constantly being improved. Recent product developments include a motor condition monitor system from Artesis that included wireless technology for use in remote and inaccessible locations and fully automated condition assessment reports. It has also developed a predictive maintenance system, the MCM, which uses intelligence mathematical modelling to establish a norm for each item of equipment, after which it continuously monitors the machinery for potential faults. If a fault is detected the system provides a diagnosis of mechanical and electrical problems and indicates the severity of the fault in time to highlight potential failures and to provide advance diagnostics to avoid secondary damage.
There’s no doubt that condition monitoring can be a complex business. It’s not just the actual physical monitoring, although that can demand a lot of highly specialised equipment. The collection and interpretation of the data is also taxing, requiring the kind of expertise that is rarely needed full-time in normal plant running. Many operations therefore choose to contract CM out to service companies, buying in the knowledge without the heavy investment and extra headcount. There’s nothing wrong with this provided potential problems don’t fall down the gap between external reports and internal maintenance procedures.
Proviso Systems and glass manufacturer Saint-Gobain Glass UK (SGGUK) are an excellent example of how to make this kind of partnership work for all concerned.
SGGUK, winner of the Best Factory Award for process plants, is one of the leanest and most highly automated plants in the UK. The main float line runs all day every day. It feeds into a coating line that operates 24 hours a day for a seven-week campaign, with often only a three-day shutdown between for a routine maintenance. The vibration checks alone are over 3,000 points on each. “We only do some monitoring on site through the trade guys – the routine stuff that needs doing every two to three days,” explains engineer manager Mick Dickinson. “But where we want a really detailed analysis of any equipment through thermography analysis and vibration or oil analysis, we use Proviso.” Two of Proviso’s specialists are on site for six days each month for the oil analysis and vibration monitoring, and another spends four days every three months for Thermographic analysis. They submit reports after every visit. “I get a copy as well as the responsible engineer for each area,” says Dickinson. “The report highlights anything that needs looking at as a matter of urgency but we have been working with this company for so long that they don’t leave site without telling us about it face to face. Sometimes the fault has already been rectified by the time I get the report two days later.” The reports themselves flag up areas that need action at a glance. In the case of thermography reports, it also provides clear, graphical details of how the fix should be handled: “That has saved us so much time and prevented so many sudden failures.”
The really admirable part, however, is the way SGGUK dovetails its contractor’s CM recommendations into its own comprehensive preventive maintenance programme. Lean to the core, most of SGGUK’s trades people have additional production skills that absorb about 10% of their working hours. Their priority, however, is planned preventive and reactive maintenance. Dickinson says they perform PM examinations on every piece of equipment every day. Out of a 12-hour period, they might spend six or seven working through their PM task sheets – anything from a visual inspection to a full strip down and replace. “We have around 400 PMs that come out every week and we have a 97% completion rate.
When CM analysis flags up remedial action, the engineer responsible for that area raises a notification on SGGUK’s SAP system, which accumulates new jobs for allocation at the planning meetings, held every week for each line.” The meeting has representatives from all the production and engineering areas so you have the right people there to give the go ahead to stop a process or assess the priorities,” says Dickinson. The jobs are incorporated in SAP’s live labour planning sheets, which give graphical representation of PM loading, the people available for each shift and the jobs allocated to each for two or three weeks ahead. When a shift engineer arrives, he opens up the plan to see what’s been planned for him. The engineer who originally picked up the CM reports will also have left all the equipment for that job in a holding area, together with any parts he has ordered, identified by job number. “The engineer may come in completely cold from a holiday yet he will still now exactly what to do and be able to do it,” explains Dickinson. When the job is complete, the plan is updated. If it can’t be finished, the reasons are also entered. So the manager knows exactly what is happening at any time. No notification ever leaves the system until it is confirmed as completed by engineers.
Call outs in the middle of the night do still happen but they are always followed by a root cause analysis, stored in a database to help solve any similar future problems. “Hidden knowledge is not power,” Dickinson maintains. “If you can tell somebody something that stops the 3am phone calls, then tell them.”
Do this plant’s techniques for melding external monitoring with its own preventive maintenance work? Dickinson openly acknowledges that there will always be a t least two reactive breakdowns a week – not everything is predictable. But the figures speak for themselves. He is measured by plant interruptions due to maintenance breakdown expressed as yield lost. His annual target is a tough 0.4%. It is currently running at a remarkable 0.3%. Case proven.
