Hundreds of sensors register every movement, every noise, and every temperature change. Neural networks extract comparative data from big data deserts within milliseconds and automatically optimise engine speed and other parameters. The life of expensive components is significantly extended and production yields or volumes rise significantly. Maintenance work is carried out exactly when it becomes necessary, but in any case before a component really fails and paralyses production. These are images that may come to mind when you speak of Predictive Maintenance (PM). No one thinks about technical documentation.
The Introduction "Predictive maintenance [...] is one of the key innovations brought forth by Industry 4.0" of the study "Predictive Maintenance – Servicing tomorrow – and where we are really at today" from the VDMA (German Mechanical Engineering Industry Association) will be directly signed-off by many. In this study, VDMA 2017, together with Deutsche Messe AG and Roland Berger GmbH, asked its members about the current status and expectations, with regard to predictive maintenance. Many will understand the VDMA's concern about competition from the many software strongholds of the world: "New players with a background in the 'digital world' will enter the market for service business in the production industry [...]. The stakes are high for the German Engineering sector, which must respond by taking the lead in defining, implementing and disseminating PM solutions." The ambitions of the companies surveyed are correspondingly strong: According to their own statements, 81% are already working intensively on predictive maintenance. 80% of those surveyed see growth impetus for their own service business. Nearly 89% said that they still lack the ability to understand the needs and requirements of both their customers and end-customers.
To make the subject of maintenance more tangible, experts distinguish between five levels of maturity:
- In event-based or reactive maintenance (Level 1), machines and systems are repaired after they have failed. There is no technology for maintenance or prediction.
- A Maintenance Plan and a Calendar are required for time-based maintenance (Level 2). Here, wear parts are replaced at fixed intervals.
- For use-based maintenance (Level 3), on the other hand, a database (or similar) is already required, in which the actual (i.e. current) operating times of the individual components are recorded. Once a component has reached its maximum operating time, it is replaced.
- Now sensors come into play that detect during condition-based maintenance (Level 4) that, for example, a vibration threshold value on a component has been exceeded. An alarm is triggered and the component is replaced.
- In the fifth stage (predictive maintenance), models based on previous sensor data are derived for wear, and possibly for synergies between various components. Software then monitors the sensors and recommends the most appropriate time to replace a specific part.
Maintenance work accounts for only 2% of machine downtime.
Whoever directly invests in sensors and software may be sacrificing the potential of the first three levels of maturity. In an article in the Technology Review, a study by the consulting firm Staufen-Neonex on this topic was recently highlighted. As Jochen Schlick, co-founder of the company reported, that on average, only 2% of machinery downtime is caused by wear or defects. The fact is that the machine is simply not used or being retooled, or new tools are being set up, and the machine operators' break times would account for the majority of downtimes. As a rule, the machines are productive only 52% of the time!
If one were to simply come to grips with the defects and the wear-and-tear, one would at best achieve 54% utilisation of the machines. Based on his experience, Mr Schlick advises first to do the homework in reactive maintenance and counter the 17% time savings that is used for retooling and set-up work: by keeping the appropriate machine and spare part documentation readily available on the machine – preferably digitally and clearly assigned to the relevant component via QR Code or Augmented Reality. In combination with a digital Repair Logbook, this should facilitate considerably shorter downtimes.
It would be presumptuous to claim that the 17% mentioned above would melt away, once the technical documentation had been fully optimised. It is more likely that a critical eye on production planning will lead to less frequent changeover and set-up work, and thus increase the lion's share of the savings potential. But still: Good on-the-spot information for the technician, presented to him digitally, up-to-date, in-context, and made easily understandable, will contribute to his success. Even if we risk looking into the crystal ball, four-to-six percent less time loss, which can be achieved with good technical documentation, seems quite realistic. And that would be two-to-three times the efficiency gain, as compared with the 2% maximum savings achieved via predictive maintenance.