The Zero-Variation-Manufacture Contradiction
Kobus Cilliers | On 19, May 2019
Darrell Mann
I spoke at a Robust Design conference last month. One of the other speakers was extolling the virtues of ‘zero-variation’ manufacture (Reference 1). The concept immediately struck me as one of those naïve, ill-conceived motivational posters the manufacture sector occasionally falls into the trap of promoting (‘zero defects’, ‘zero waste’, ‘zero breakdowns’, etc, etc). It also struck me as having nothing at all to do with creating systems that are more robust. Unless, of course, we take the definition of ‘robust’ that means everything operates okay until a catastrophic failure occurs. It felt like there was a need to do some deeper thinking.
On the positive side of Zero-Variation Manufacture (ZVM), I’m inclined to think about the old automotive industry story from the 1980s where one of the US manufacturers had occasion to compare US-made and Japanese-made versions of a particular gearbox design. The Japanese boxes were turning out to be significantly quieter, smoother and more efficient than their American analogues, and so management instructed that a number of gearboxes be stripped down and inspected to try and work out what the differences were. The results were unequivocal. All of the components from all of the gearboxes were within the stated manufacture tolerances. Every gearbox was compliant.
But then, when the team decided to look more deeply into the details, it was also clear that the dimensions of the Japanese-made components were consistently much closer to the nominal drawing dimensions than those of the American components. The only difference between the two manufacturers, in other words, was that the Japanese components had much smaller variation than the American ones. The story subsequently came to highlight a subtle but significant difference between Japanese and American manufacture philosophies: In Japan the aim was always to achieve the precise target, as opposed to the US, where so long as things were ‘within tolerance’ everything was good:
Figure 1: Difference Between Eastern and Western Manufacture Philosophies
Variation-wise, the clear conclusion the story points towards is that ZVM is a good thing.
So far so good. But, unfortunately, the world of reliability does not quite operate as simply as this. Reliability is all about designing the system as a system, and as such it is not possible to separate the world of manufacture from the rest of the business. By giving the Manufacture people a ‘zero-variation’ goal, we’ve implicitly told them to sub-optimise relative to the overall system.
Zero variation in this regard is akin to equally naïve targets like ‘zero tolerance’. When the designers upstream of the manufacture operation hear these kinds of phrase, they use it as license to design components with tighter and tighter tolerances. Because – ‘obviously’ tighter is better. Continue this thinking for more than a couple of years and what you end up with are systems that perform absolutely impeccably and absolutely consistently on day one of service operation. But then, as the aerospace industry rapidly found out when they started flying aircraft in desert conditions, where jet-engines are ingesting kilograms of airborne dust and sand, performance of those beautifully engineered ‘Swiss watch’ engines rapidly degrades. A new engine is not the same as an engine that has flown for 20 hours of desert conditions. Sand and other environmental contaminants, it turns out, have an uncanny knack of damaging high-speed rotating components. And so a one micron tolerance seal rapidly finds it has become a seal with 50 micron leaks.
By designing the ‘perfect’ solution – from the manufacturers zero-tolerance, zero-variation perspective – we’ve created something capable only of operating in perfect conditions. And the real world is anything but perfect. It is full of dust and sand, rain, snow and ice, leaves, pollen, insects and a million other ‘imperfections’.
Designing the ‘perfect’ solution for an imperfect world means doing the precise opposite of what ZVM and zero-tolerance aim at. Far better to let things move and vary because that’s exactly what the environment will cause anyway if we don’t. Far better to let the system move and shift to continually find its own optimum from moment to moment. Far better to allow the system to self-organise. That’s the way to resilient systems. Or, better yet, ones that are antifragile.
Here’s a simple example to illustrate the point. Car wheels. The automotive industry still thinks it is a good idea to tighten up tolerances and reduce manufacture variation in order to give drivers a smoother ride. Ditto the tyre manufacturers. Then, because they’re still not very good at it, every time we change a tyre, the garage uses an expensive – also accurately manufactured – wheel-balancing machine. Then the hammer small weights onto the appropriate part of the wheel rim to give the driver a beautifully balanced wheel. Balanced, that is, until the tyre starts to wear, or picks up bits of debris, or is subjected to a bad skid. Now the wheel is no longer balanced.
The whole balance problem could have been very easily solved by allowing the wheel to dynamically ‘balance-itself’. A feat that is very easily accomplished dropping a handful of tiny beads (or, better yet, ‘sand’) into the tyre when it is fitted to the wheel rim
Figure 2: Self-Balancing Car Wheel Beads Â
Now what happens is the beads ‘magically’ orient themselves to counteract any out-of-balance forces. The wheel remains as balanced on day 2000 as it was the first day the wheel was installed. No effort, no need for tight tolerances. Resilient not robust.
Pulling matters together, as ever, it seems we have a contradiction here. ZFM is both a good and a bad thing. We might map the contradiction to look something like this:
Figure 3: The Variation-Zero-Variation Contradiction
Once formulated, we also know that every contradiction becomes solvable. Looking at the variation-zero-variation physical contradiction, my first instincts are to say it is most readily solvable using a ‘separation on condition’ strategy: if the manufactured system will never encounter the real world, or will only encounter it for short periods of time, ZVM is probably a good idea (especially if we can achieve it for free – another contradiction). If, however, the system we’re designing is expected to operate for prolonged periods of time in the real world, a completely different design and manufacture strategy is called for. One in which we don’t let the Manufacture silo forget that optimizing their bit is very definitely making everyone else’s life, especially the reliability engineers, significantly worse.
Reference
- Boorla, S.M., Eifler, T., McMahon, C.A., Howard, T.J., ‘Product Robustness Philosophy – A Strategy Towards Zero Variation Manufacturing (ZVM)’, Management and Production Engineering Review, 9(2), 3-12, DOI: 10.24425/119520, 2018.
-
Wonderful article! There is still some room for good debate with the viewpoint that efforts are aimed towards higher “CONTROLLABILITY” of the system. Even if we have most innovative solution, we still would like to make it controllable to sustain the performance in space and time. As in example of tyre, if the beads sizes, packing density, material property or assembly process have huge variation then the system carries the failure risk or risk of unknown effect thereby needing control on the process and product specifications.
Variation control and Innovation then could be made complimentary than contradictory by design.-
Thanks for the comment. The beauty of ‘self-organisation’ is it removes the variation problem. In the tyre case, it doesn’t matter about the size distribution of the particles since they ‘automatically’ position themselves to counter any out-of balance forces. Quite literally, a handful of sand picked off the ground will do the job.
-
Comments