Basic Knowledge

IoT Basics: What is Predictive Maintenance?

| Author/ Editor: Jakob Schreiner / Alexander Stark

Predictive maintenance is one of the most tangible applications within Industry 4.0. It allows status data to be obtained from machines and proactive maintenance to be carried out on systems. This paper uses a definition and practical examples to explain how predictive maintenance works.

Related Companies

Predictive maintenance is an essential pillar of Industry 4.0.
Predictive maintenance is an essential pillar of Industry 4.0.
( Source: Photo by Ramón Salinero / CC UNSPLASH NaN )

By definition, predictive maintenance refers to a maintenance process that is based on the evaluation of process and machine data. It is used primarily in the context of Industry 4.0. The real-time processing of underlying data makes it possible to make forecasts that form the basis for needs-based maintenance and consequently the reduction of downtimes. Besides the interpretation of sensor data, this requires a combination of real-time analysis technology and an in-memory database in order to achieve a higher access speed to the data compared to hard disk drives. If everything works out, a technician can be assigned to solve a problem before it occurs. Because: business objectives can only be achieved if plants, machines and processes function perfectly.

With the help of Predictive Maintenance Technologies, the condition of machines is evaluated in order to predict when maintenance needs to be performed. As a result, cost savings can be achieved over routine or time-based preventive maintenance, as tasks are only performed when they are needed.

The main objective of predictive maintenance is to provide the most precise advance maintenance planning and to avoid unexpected breakdowns. Knowing when a particular machine needs to be serviced makes it easier to plan resources for maintenance work such as spare parts or personnel. In addition, system availability can be increased by converting "unplanned stops" into ever shorter and more frequent "planned stops". Further advantages include potentially longer service life of the plants, increased plant safety, fewer accidents with negative effects on the environment and optimized spare parts handling.

How Does Predictive Maintenance Work?

Predictive maintenance categorizes the condition of plants by checking them either periodically (offline) or continuously (online). In the ideal case, it is possible to make upcoming maintenance not only as cost efficient as possible, but also performance efficient, i.e. even before the machine is losing performance. In order to keep interruptions to regular system operation to a minimum, most predictive inspections can be carried out during plant operation.

For the assessment of the actual state of a machine, an examination by means of infrared, acoustics (partial discharge and ultrasound), corona detection, vibration analysis and sound level measurements can be used. Of course, such test procedures must not impair the function of the machine or even damage it. A more recent approach is to link the information gained from this with process performance data, which is made available by collaborative process automation systems (CPAS). If you want to implement predictive maintenance in your company effectively and on a long-term basis, you should follow the following three steps:

  • capturing, digitalization and transmission of data,
  • storage, analysis and evaluation of the data collected, and
  • calculation of probabilities of occurrence of certain events.

Predictive Analytics and Big Data

A great difficulty in dealing with predictive maintenance is the processing of huge amounts of data. In order to be able to make reliable statements about the condition of machines and plants and thus to be able to detect malfunctions as quickly as possible, it is necessary to collect large amounts of data. These data must be stored, processed and analyzed using intelligent algorithms.

Due to the huge amount of data, technologies and databases of the big data environment, such as Edge Computing, are suitable for predictive maintenance purposes.

This data not only includes the condition of the machines and plants themselves, but also of their environment: For instance, parameters such as temperature or humidity are also captured and evaluated. Altogether, the data can differ very much and is available in different formats. Due to the large number of different data and formats as well as the large amount of data, databases must have huge capacities. Therefore, it is important to be aware of the fact that the size of the database and the intelligence and performance of the analysis algorithm are essential for the quality and reliability of the information obtained: The larger the database and the more intelligent and sophisticated the algorithm, the more reliable the results will be. After the data is captured, the measured values and diagnostic data are transmitted through networks to service centers or to the manufacturer. The basis for this process is the Internet of Things (IoT).

Nevertheless, it is essential to constantly update and process the large amounts of data collected during predictive maintenance in order to identify trends and developments.

Advantages at a Glance

If predictive maintenance is used correctly and efficiently, it can deliver a variety of benefits - both for the manufacturer and the user. These are the most important advantages at a glance:

  • Improvement of economic efficiency: On the one hand, predictive maintenance can reduce machine and plant downtime and the cost of unplanned downtime. On the other hand, regular maintenance of machines and systems can also increase their service life.
  • Ideal maintenance time: With predictive maintenance, the best possible time for an upcoming maintenance can be determined by permanently evaluating the captured data. In addition, maintenance can be optimally integrated into the production process.
  • Improvement of machine performance: The permanent analysis of the collected data makes it possible to improve the performance of the machine and achieve higher productivity in the long run.

Predictive Maintenance Examples

Predictive maintenance is already used in many areas. It is not only an attractive tool for the manufacturing industry in general, but also for all mobility services - whether in aviation, automotive or trains - or wind power plants. With the help of predictive maintenance, the downtimes of wind turbines can be almost completely eliminated.

Predictive Maintenance of Motor Vehicles

Predictive maintenance is also gaining in importance in the area of mobility - for example in the maintenance of motor vehicles. Extensive data collection acquired by sensors in the engine or body of vehicles, helps to avoid expensive repairs or breakdowns at an early stage because it allows preventive activities to be initiated. This includes, for example, the replacement of a damaged vehicle part during the next workshop visit before the part fails completely. Predictive maintenance goes one step further if vehicles are networked: You can transmit data online and automatically to service workshops or the manufacturer.

Predictive Maintenance in the Aerospace Industry

Planes only make money when they're in the air. If an aircraft breaks down due to mechanical damage, this leads to high costs for the airline. Therefore, they are keen on detecting possible damage in advance and thus prevent a breakdown or even an accident. Areas of application for predictive maintenance in aviation include turbines or hydraulic pumps. Airbus launched its "Skywise" data platform in 2017, a project that enables aircraft engineers to intervene at an early stage to detect errors and replace a component before it fails.

Predictive Maintenance in Rail Transport

The application of predictive maintenance in rail transport is similar. There, too, predictive maintenance is used to better plan repairs and gain insights into the life cycle of certain components. Predictive maintenance in rail transport can help to prevent the unexpected failure of a train. It does so by allowing to plan in advance when the damaged part of the train needs to be repaired in order to prevent disruptions to operations. By analyzing and utilizing the captured data, repair instructions can be selected, suitable spare parts can be provided, and downtimes can be kept as short as possible.

Preventive vs. Predictive Maintenance

Predictive maintenance differs from preventive maintenance in that it is based on the actual condition of a machine and not on its average or expected life to predict when maintenance is required.

Preventive maintenance is also aimed at avoiding downtimes. The difference to predictive maintenance, however, is the fact that no actual data collected from the machine is used for maintenance, but that maintenance measures and monitoring are carried out on the basis of (regular) intervals already defined in advance. Preventive maintenance is based on the theoretical failure rate and therefore disregards the actual machine performance. Downtime is planned on the basis of calendar dates or usage. Care should be taken to ensure that the maintenance time is neither too early nor too late. For example, most forklift truck manufacturers recommend preventive maintenance every 150 to 200 hours of operation.

Since companies, their plants and their daily operations can differ greatly, the use of Preventive Maintenance can lead to unnecessary maintenance work. In addition, when using preventive maintenance, a company quickly runs the risk of replacing parts at an early stage that still function perfectly and would still have done so.

This article was first published by Industry of Things.

Original: Julia Moßner / Translation: Alexander Stark

This article is protected by copyright. You want to use it for your own purpose? Infos can be found under www.mycontentfactory.de (ID: 45867053)

Spotlightmetal; Pixabay; Thinkstock; ACE; Asco; ; Photo by Ramón Salinero; Siemens; Hochschule Landshut; Daimler; Hornet; Altair Engineering GmbH; ANDREAS STIHL AG & Co. KG; gemeinfrei; Dr. Ing. h.c. F. Porsche AG