Image Image Image Image Image Image Image Image Image Image
Scroll to top


Perfect Tolerances

Perfect Tolerances

| On 19, Apr 2018

Darrell Mann

“In the landscape of extinction, precision is next to godliness.”
Samuel Beckett

This one seems to keep coming back to haunt me. Time, I thought, to try and stem the seemingly unstoppable tide.

Manufacturing tolerances, high or low?

In Operational Excellence World, the answer is a no-brainer. Perfection is good; variation is bad. Therefore, ‘continuously improve’ by progressively tightening the tolerances, and, better yet, make everyone aim for the middle of the target. In theory, too, if we do the job right, there should be no impact on cost. ‘Quality is free.’

Well, it is, if organizations understood the concept of solving contradictions. But that’s something that tends not to be a part of the Operational Excellence world, so it tends not to happen. That was certainly the case with a recent experience working on a problem with a client.

Without getting in to too many details, the problem we were tasked with solving involved an electric motor with an output shaft which, for a variety of reasons, had to have its axial position controlled very accurately. That’s ‘accuracy’ as in plus or minus less than a micron. The original manufacturers of the electric motor had done a very competent job of controlling their manufacture processes in order to maintain the required positional tolerance during operation. Unfortunately, the cost of the motor was very high. Because the company decided they couldn’t bear this cost any more, they decided to out-source manufacture to a part of the planet where the manufacture costs were known to be lower. Several companies in the country concerned put in bids to take over the manufacture. The cheapest credible bid was a factor of five cheaper than the original cost of the motor. They got the contract. They started making motors. Very quickly it became apparent that, contrary to their promise, they were not able to manufacture to the required tolerances, and hence the axial position of the shaft was not able to be controlled to anything like the right levels.

The out-source company was told they had better fix the problem. Still in Operational Excellence mode, they decided the only way to get everything within tolerances was to buy some new manufacture equipment. A bit like the factor-five cost problem with the motor, the new equipment, purportedly able to achieve the required tolerances, was almost five times the cost of the equipment the company had paid for their existing equipment. They couldn’t afford the price. They also couldn’t afford to lose the contract. Operational Excellence Stalemate was reached. Everyone was stuck.

Whenever I look at any technical problem like this, my instincts tend to take me straight to the Trends Of Evolution. And when it’s a manufacture tolerance problem, my mind goes direct to two Trends in particular. The ‘Macro To Nano’ Trend (Figure 1) is a trend that, in theory at least, tells me that things get smaller. More specifically, ‘the smallest engineering dimensions get smaller to in order to deliver a functional benefit’. This is the trend that’s supposed to tell me that smaller tolerances is the ‘right direction’.

Figure 1: Macro To Nano Trend

On the other hand we have the Dynamization Trend. It would also seem to have something to say about the tolerance question. Especially, in this specific case, in terms of whether the right thing to do is to stop the motor shaft from moving axially – Figure 2.

Figure 2: Dynamization Trend & ‘Tolerances’ 

Now, admittedly, this connection between Dynamization and ‘tolerances’ is not so obvious, but once we make it, the Trend is very definitely trying to tell us that the tighter we make our tolerances (the more we ‘immobilise’ the system), the further from IFR we are moving. The Dynamization Trend tells us that physical things become more ideal as they become more flexible. Tolerances ‘want to become flexible’ in the same way that structures and other things that engineers and architects tend to want to ‘stiffen’ want to become movable and adaptable.

It would appear that, when it comes to tolerances, the Macro-To-Nano Trend and the Dynamization Trend are in conflict. Which might seem a little odd. If both Trends represent signposts towards the Ideal system, why are the signposts apparently saying different things?

Sounds like a contradiction. We want tolerances to be ‘tight and loose’.

Sounds like something that requires a separation on condition strategy to solve: we want the tolerances to be tight if the component (per the Macro-To-Nano Trend) is becoming smaller; we want the tolerances to be loose (per the Dynamization Trend) if we wish to reduce manufacture cost, increase the adaptability of the system and increase the resilience of the system.

Plus, of course, we know that the Dynamization Trend sits at the top of a hierarchy of Trends. If we couldn’t solve the contradiction, the looser tolerance direction should win because that’s the direction the Dynamization Trend tells us to travel.

Now, I should say, when I tried to have this discussion with my client, all I seemed to get back were furrowed brows and looks of confusion. ‘What has this got to do with my motor problem?’ I could see them thinking, ‘Do we have to buy some expensive new machines or not?’

Answer: Not.

What does the ‘tight and loose’ problem statement mean in terms of the motor shaft?

It means – in an ideal world – a) we are able to manufacture the motor with the existing machinery and with the current ‘too loose’ tolerances, and, despite the shaft potentially being able to drift axially, it somehow maintains the required constant axial position. “The loose-tolerance shaft aligns itself.”

Needless to say, when we re-framed the problem in this way, the whole story became a whole lot easier to not only imagine, but also, solve. A simple look at some of the other Trends and some of the features of the motor and what the motor shaft had to align with highlighted several geometric features and untapped resources already present in the system (e.g. the shaft was running open to the atmosphere and so was fundamentally surrounded by air) that could very easily be configured to make sure the shaft remained exactly where the application needed it to be.

The moral of the story, as I’ve been trying to impress upon every engineer and designer I see that I realize have been taught the opposite, is that tight tolerances are a last resort and should only ever be allowed as a solution if the overall system is making a jump towards a smaller size-scale… and, even then, will only be a temporary fix. The ideal system is made terribly and performs perfectly. The ideal system aligns ‘itself’, balances itself, compensates itself. All ‘tolerance’ questions are ultimately red herring questions. Despite what the Operational Excellence World and, in particular, the Six Sigma community might have to say on the matter…. Now there’s a tide worth fighting.