Part three of a series.
Industry 4.0 can demonstrably only work if the following conditions are met:
1. Completely reliable machines that can be changed over very quickly
Each product is made individually to customer requirements and moves itself – based on its digital twin information- through the plant from one machine to another. The machine may require a different setup between each product, so machines have to be ready instantaneously.
Measuring thousands of machines, in my experience, the vast majority of machines have a mean time between failure (MTBF) of much less than eight hours. The time-honored and proven total productive maintenance techniques, including autonomous and preventive maintenance, only become more important as equipment becomes more complex and vulnerable. Most equipment I know of can hardly be set up within single cycles and with a vertical ramp-down/ramp-up. So single-minute exchange of dies (SMED) will simply have to be a built-in feature of the process and must evolve into SCED: single cycle exchange of dies.
2. Completely reliable production processes.
In the 4.0 concept, there is absolutely no time to check products, let alone correct them. It is perfectly possible to produce zero defect, but then all processes must be perfectly under control.
Henry Ford already in his day saw the need to standardize to a large extent, and W. Edwards Deming taught by using statistical process control (SPC) to find and distinguish normal versus abnormal deviations. Without SPC, it is hard to tell. Where many manufacturers today are satisfied with quality figures of 95+%, the question will be whether we will maintain sufficient flow in a dream 4.0 environment with 99.99% quality.
3. Completely reliable internal processes
In addition to the production processes, all other business processes must also be reliable. In overall equipment effectiveness (OEE) measurements, time and again, the causes of machine disturbances turn out to lie somewhere in an office. My favorite method for getting administrative and supporting processes on track is Makigami process improvement.
However it is done: The shopfloor can no longer be the place where flaws from other processes need to be fixed.
4. Completely reliable supply chain
What applies to the companies’ processes also applies to the processes in the supply chain. Neither the products themselves nor the moment of delivery may disrupt the value streams. That is easier said than done. From real 3.0 manufacturers we know in some cases, they know the processes up to Tier 4 just as well as their own and are confident that they are just as well-controlled. As long as buyers believe quality can be achieved through (input) controls, “reasonable doubt” should be in place.
Whenever a condition in the above four prerequisites is not being met, it will become visible in the OEE measurement (when implemented correctly). Therefore OEE is an indispensable indicator to move towards controlled processes.
5. Everything can communicate with everything
a. A not unimportant detail to get 4.0 up and running: All transport and conversion devices must be able to communicate with each other as well as with a central controller. Most of the machines I encounter in common factories are decades old and never designed to communicate with their environment (nor, for that matter, to be able to change-over within one cycle and have a very high reliability and uptime.)
b. This communication must be fed from the digital twin. Therefore, for every product it must be recorded in detail without error what the product looks like, what the specifications and tolerances are, how and where it is produced etc. In my experience, the product specifications and certainly the tolerances are poorly recorded for many products. This is especially true for process parameters.
I would venture to say that these requirements would anyhow be very helpful in any production environment. In a well-organized 3.0 environment, that will certainly be the case to a large extent. When things go wrong in a 3.0 factory, the self-regulating capacity of the well-developed people who are present in the process, day in day out, keep things going and use such events to further improve the process. I see no way how that could still be the case in a 4.0 environment.
Has anyone tried it before?
Fiat, Volkswagen and most recently Tesla have tried to a certain extent. All three have given up. All three have learned that you don't ‘solve’ complex processes with even more complex solutions. But above all, they have learned how important the self-regulating and problem-solving abilities of people are. All three applied a high degree of automation, but the human factor was brought back into the process and they learned to experience the beauty of 3.0.
There seems to be a tendency to solve problems by ‘automating’ problems and people away, sometimes costing billions. It makes much more sense to do it the other way around:
- First solve the problems, get the process in order, then automate it
- Instead of eliminating the human factor, embed it and use its unique features
Automation is a great solution for quantitative tasks. But when there are qualitative problems, it is much more effective to simplify things first; to know the real value streams and bring them under control instead of trying to “organize” the underlying problem with complex technology. Doing so just makes the problem bigger.
What is ALWAYS needed?
Whatever manufacturing concept has preference, some things are always beneficial or even indispensable:
1. People who know perfectly what they are doing, why and how.
Only when people know what they are doing and why can we ensure that humans can master their technology and respond appropriately to unexpected situations, of which there always will be.
2. Deviations that are immediately visible and responded to.
Only when we know and can see that something is deviant can we react to it and possibly stop or correct it before it is too late.
3. An improvement system that eliminates root-causes.
The larger and more complex the manufacturing-process becomes, the more things can go wrong. Only when we are able to turn each mistake into an improvement can we grow.
4. Extensive standardization.
Only when we know what is the best way to do something, and ensure that it is always done that way, can processes be in control. By putting together complex products from standardized components, the complexity is much more manageable.
5. Machinery and processes are reliable.
The preceding points are the basis for processes and machines that are reliable. A reliable process is the cheapest, fastest, safest, most schedulable, etc.Arno Koch has over 25 years of experience in process improvement and process control. His improvement goals are defined in terms of “halving” and “doubling.” He teaches process improvement at the CETPM at a German university, is partner at OEE Coach BV and owner of Makigami BV, and has written three books on OEE and two on Monozukuri (’the art of making things").