This was previously posted on LinkedIn.
The factory of the future is a promising concept, yet it is still having a hard time being accepted in today’s plants. So far, plants that are fully equipped with sensors and where computers have replaced humans are very scarce. A great many initiatives conducted by large and mid-sized businesses are failing to fulfil their promises. Not because industry has nothing to gain from digital technology, but because some of those promises have not been realistic enough. So what are the 'myths' about Industry 4.0 that manufacturers need to understand?
Debunking the 'Myths' about Industry 4.0
The first myth is the idea of “artificial intelligence (AI) in a box” which would simply require one to supply data without knowing what it means. In practice, those technologies are useful for tasks that are impossible to model, especially those giving the five senses to computers (to analyze images, noise, vibrations, etc.). But they should be combined with appropriate industrial expertise and a physical model of the machines or processes involved. This is quite different from the commercial internet, where consumers are impossible to model and the margin for error is greater. Achieving 90% of suitable purchasing recommendations is a good performance for a bookstore, but experiencing one air crash for every ten flights that take off would be catastrophic for the airline sector. Simply put--AI still takes a great deal of human intelligence to be effective.
The same goes for the 'magic' of big data, where simply investing in collecting data would be enough to bring about valuable data.
The problem is, some data is not valuable enough to be collected.
Furthermore, other data should be collected but not stored for it can be processed on the 'edge,' meaning within the machine’s embedded system or controller. To obtain a guaranteed return on investment, one should instead start with the technology value levers, meaning the tangible improvements that are expected (faster prototyping using 3D printing, computing power from within the cloud, automated data analysis, agile methodology, etc.). Then one should find a profitable way to confirm and then extract this value on a wide scale (minimal product in a site, a strategy to extend it to other instances and other sites etc.).
Another myth is the belief that traditional skills in materials science, chemistry, or processes could be depreciated by their digital equivalent. In fact, those traditional skills will probably continue to account for 90% of the added value. Admittedly, those businesses among the remaining 10% that are not performing at their top level will be wiped out by the competition. But the same applies to those that might abandon the top 90%. Industry has always thrived through innovation--the future will be no different.
The fourth mistake is to underestimate the human factor and the appropriation of technology.
In the past, a great many air crashes happened with correct software and data, but under conditions (weather, fatigue, or stress) where the pilot was having too much difficulty absorbing all of the information coming at him to perform the right action. The risk exists in any place where there are large volumes of information (control rooms, operating blocks, etc.) or tough working conditions (a dirty or noisy environment), or when the software is ill-suited to the qualifications of its users. Industrialists in critical sectors (aeronautics, health, transport, energy, etc.) are quite familiar with those issues, and there are ways to resolve them. Unfortunately, those ways are often ignored by those who are focusing too much on the factory of the future, and not enough on how to make it function in this day and age!
Seeing Success with Proven Technologies
Conversely, those who have espoused proven technologies have been compellingly successful. For instance:
- 3D printing to produce industrial parts (and no longer just prototypes) for the aeronautics and automotive sectors. It will make it possible to replace the growing amount of parts that have complex structures or procurement constraints.
- Preventive maintenance to cut down on downtime, based on the expertise of workers in charge of the equipment and using the analysis of “early signs” so as to be able to repair a machine before it breaks down and to do so at a time when downtime is the cheapest.
- Optimized production to reduce the amount of scrap by half or more in a tube cutting facility (even though it is known to be run efficiently), using an application that was designed in a few weeks and is intended to be easy to use without any training.
- Going from Lean Management (which improves production, giving operators the possibility of identifying and resolving problems) to Digital Lean Management (which brings in flexible tools to boost the method, primarily for visualizing and analyzing). To specialists in the field, it is a revolution… because Lean Management purists often prefer using pen and paper in order to liberate workers from the rigidity of previous-generation production systems.
- Making use of “digital twins”, meaning digital duplicates of machines, created on the basis of the original blueprints and operational data, which help to better assess the precise condition of a machine and to monitor its progress through time (the wear and tear or replacement of certain parts, temperature, pressure, etc.). In that way one is in a better position to anticipate problems or understand what has caused them without having to shut down the equipment for inspection.
The sheer diversity of those successes suggests a major feature of the digital factory… namely that the idea is to see to it that the entire production system takes full advantage of the latest technologies (low costs of data collection, storage and processing, faster development and deployment, 3D printing, etc.). To that end, a business needs to have foundations (infrastructure, tools, talent, partnerships, etc.) so it can enable every site and every employee to make the products they are selling to their customers cheaper, better, more reliable, and to combine them with improved services. And to do so in multiple ways, most of which will remain unseen by any kind of single approving committee.
This undoubtedly represents the greatest challenge facing top management… indeed, it is less a question of arbitrating projects than of laying out architecture principles, providing tools or inspiring their organization to ensure that the transformation should largely happen without them.
This article was originally posted by Vincent Champain and has been translated from French.