Predictive maintenance: definition, benefits, examples ... all with one click!

Predictive maintenance is generally understood to mean the early anticipation and avoidance of an error in machinery or components thanks to data on their condition. This is possible with data-based methods that analyse the condition of your machines and help predict malfunctions, faults and also the time when maintenance work is required.

Industry 4.0 promises greater efficiency in production through networked machines, insights due to data analyses and better machine availability. In this context, predictive maintenance, i.e. a forward-looking maintenance process based on the evaluation of process and machine data, is a crucial component.

Unplanned downtime of production equipment costs money and cuts your profits significantly in the long run. What is certain is that machine and system failures pose a serious threat to the industrial sector. Predictive maintenance is intended to lead to cost savings compared to routine, interval- or time-based preventive maintenance as work is only carried out when necessary.

First of all, various sensors are needed to record functionally relevant data on the machines, such as speed, temperature, noise, bearing vibrations or power consumption. Next, a combination of real-time analysis technology and a database is required to interpret and meaningfully analyse the sensor data. If all this is successful, it will be possible to rectify the machine problem before it actually arises.

No. Predictive maintenance is not only interesting for the manufacturing industry or associated MRO areas (Maintenance, Repair & Operations), but also for all mobility services - whether in the air, in vehicles or on rails. 
It is usually worthwhile for companies that frequently use the same type of machine or do not want to use predictive maintenance for their own production, but for the machines they sell.

• Higher system availability thanks to less downtime
• Maximisation of the service life of systems and machines
• Optimised deployment of maintenance technicians
• Plannable maintenance times
• Higher machine performance and productivity

Of course, there are the one-time investment costs for sensor hardware and software for data analysis and evaluation. In addition, every new method initially requires change and training efforts. However, these can be reduced to a minimum with the ready-to-install smart plastics solutions from igus i.Cee.

Between 50 and 70% less unplanned downtime should be possible, as well as savings of 20 to 40% in maintenance costs, depending on which study by the two renowned consulting firms McKinsey or Accenture you prefer to trust. In addition, Roland Berger claims to have discovered that with predictive maintenance, only 15 per cent of the time is spent on maintenance. With traditional reactive maintenance, this figure is 40 per cent of the time.

According to the authors of a survey conducted by another technology consultancy, Bearing Point, among experts in maintenance, production, logistics and IT, over 80% of users expect predictive maintenance to increase system availability, i.e. OEE (Overall Equipment Effectiveness).

Find out more about the connection between predictive maintenance and OEE here.

The collection and analysis of machine data is essential to this type of intelligent maintenance work. With the help of this data, it is possible to calculate the optimum time for maintenance in order to avoid malfunctions. Machine data is usually collected via sensors attached to the machine or system component, often aptly called predictive maintenance sensors. In igus energy chains, this job is performed by i.Sense condition monitoring sensors from smart plastics. This intelligent sensor technology continuously records the various operating states of the energy supply.

The prerequisite for the most accurate possible indication of fault-free running time is that information on usage is as accurate as possible. This includes information on the place of use with weather conditions and temperatures, distances covered, number of cycles, etc. Thanks to the knowledge acquired in the igus test laboratory and the current condition data, igus is able to predict the service life data precisely. A necessary replacement of a component is then indicated in good time and enables planned and resource-conserving maintenance.

A prerequisite for predictive maintenance is that the machines are networked, i.e. all machines provide their data for storage and analysis in a database. This is ensured by an OPC UA protocol.
However, with igus i.Cee solutions for predictive maintenance, no additional systems or hardware are required. All necessary data is provided by the associated i.Sense sensors, and the special AI algorithms are implemented in the i.Cee module. The associated i.Cee dashboard can be accessed from any standard PC.

In its DIN EN 13306 Terms and Definitions of Maintenance, the German Institute for Standardisation (DIN) defines a maintenance strategy as a procedure for achieving maintenance objectives, such as maintaining a specified condition or extending the life expectancy of a machine. A distinction is usually made between the following three different approaches:

• Reactive maintenance strategy / Damage-dependent maintenance strategy
• Preventive maintenance strategy
• Predictive maintenance strategy

With reactive maintenance, damage to machines is only repaired and production parts replaced when an acute malfunction occurs. This means that companies must react quickly in the event of damage in order to continue production as quickly as possible and avoid high downtimes and costs. This strategy is mostly used for machines that do not need to be repaired often or that can be repaired or replaced inexpensively.

Preventive maintenance is a maintenance strategy in which measures are carried out at specific intervals. These can be time- or quantity-dependent. This means that you do not wait until systems or machine parts break down, but carry out systematic preventive inspections and maintenance measures. The actual degree of wear and tear on the machines and systems is irrelevant. In predictive maintenance, on the other hand, process and machine data is collected during operation via sensors and interfaces, stored and then analysed.