2019 marks an inflection point in the maturity of Industry 4.0 and the application of real-world predictive maintenance as industrial companies move from pilots to real deployments – with significant ROI.
As a provider of industrial predictive maintenance analytics to Fortune 500 companies, it is very much our area of expertise. We use Machine Learning to monitor the condition of industrial machinery and spot the often small but significant variations in vibration, pressure, temperature, torque, electrical current and other sources that indicate when a machine will fail up to 6 months in the future.
Here are our top five predictions for the global manufacturing sector in 2019:Accurately predicting the future of an entire industry such as manufacturing is much more difficult – the variables are many more than you’d find on a typical piece of industrial equipment! We do, however, come into contact with hundreds of manufacturers around the world every year, so we like to think we have some idea about where the industry is heading.
- The cost of smart sensing solutions to connect legacy machinery and enable Industry 4.0 will continue to decline. Manufacturers want data from older machines but have been forced to bolt together their own systems due to a lack of off-the-shelf products. There will be a land grab from systems vendors that have recognized the opportunities for retrofitting machinery. 2019 won’t be the year that these become commoditized, but that point is close.
- Building upon the availability of inexpensive data, 2019 is going to be all about the value-add that it brings; using it to monitor machinery, predict problems before they impact on production, and optimize the efficiency and throughput of manufacturing environments. Manufacturers will turn increasingly to ‘holistic’ cloud platforms such as Siemens Mindsphere and use the data they contain at a greater scale (and ease) than ever before. This focus on data provides an opportunity for IT to move beyond problem solving and deliver huge amounts of value to the organizations they serve.
- Automated machine-learning driven predictive maintenance will become mainstream. Predictive maintenance has been used in regulated industries such as aerospace for years, relying on humans to collect and analyze the data for signs of problems. Advances in this area and the ubiquity of cloud computing, together with the ability to gather machine data mean we can now automate condition monitoring and prognostics at a scale and cost that gives an ROI of less than three months in most cases. There have been lots of trials for this nascent technology in recent years, and we’ll see lots of these expanded factory-wide globally in 2019.
- While engineers will make greater use of data, using computer software and tools to access and interpret it, we won't see them spending more time at their desks. Engineers will be mobilized to spend more time on the factory floor, armed with rugged mobile devices and a range of industrial apps. These will make jobs such as monitoring machine health incredibly easy, being done by a computer in the cloud and the critical bits of information served directly to engineers on the factory floor. 2019 will see real-world case studies for this start to emerge publicly as companies build the confidence to share this information with the world.
- 2019 will be the year in which machine manufacturers recognize the opportunity presented by servitization. More OEMs will move to selling capacity and uptime rather than simply a production asset, and this change will require more visibility into how machines perform and greater data sharing between the users of those machines and their OEMs.
Of all of those, what are we most excited about? Servitization! It will represent the biggest step-change for the industrial sector since the introduction of Industry 4.0 (though notice how these step changes are arriving increasingly rapidly).
The software industry has demonstrated that a scalable ‘… As a Service’ model can be effectively integrated into all levels of modern business. Providing the business function of a machine is no different though requires far more complex data processing and interpretation.