Zero Quality Control
Source Inspection and the Poka-Yoke System
Shigeo Shingo (translated by Andrew P Dillon)
Shingo writes like an engineer, clear, concise with technical drawings and photographs. This could be because, unlike many other Japanese quality experts, Shingo’s background was not academic. After leaving technical college, he went straight to work in the Taipei Railway Factory and devoted his life to adapting Taylor’s scientific management philosophy. Although he is, perhaps, best known for his work in Mitsubishi’s shipyards in the 1950s, which led to a 50% reduction in supertanker assembly, this is only a small part of his achievements.
This book concentrates on practical ways to reduce defects and non-conformances in manufacturing environments, specifically it shows Shingo’s train of thought in how to achieve zero defects.
One major advantage Shingo had was that he never lost touch with attitudes on the shop floor. He found that shop foremen did not find statistical quality control user friendly. It would get to the point that “merely hearing the words ‘quality control’ gave them headaches”; he also realised that control charts were, effectively, mirrors which only reflected prevailing conditions and were therefore useless unless they were used for corrective action.
Shingo realised that there was a need for 100% inspection, as human-dominated tasks would, inevitably, produce defects, because of factors like forgetfulness, distraction, boredom or pressure of work. “Defects will always crop up” people told him at regular intervals “in any task performed by humans”. His solution was to devise the idea of mechanisms, which would both carry out 100% inspections and, if a defect was detected, then carry out immediate feedback and action. He called this system “poka-yoke” – mistake proofing – and a substantial proportion of this book is devoted to showing actual examples of these.
The advantages of poka-yokes are that they tend to be simple, ingenious and inexpensive. To take just one example at random, a problem was found in the process for casting engine valves of a Japanese company. It was found that backward strikes would occasionally occur. This was remedied by ensuring that a worker would orient the work piece visually; the worker would then place the work piece in the hopper. However, one end of the component was magnetic and the other end was not. The non-magnetic end was the end, which was involved in being cast. A magnetic sensor was installed at the conveyer belt the components travelled along, so that if a component went in the wrong direction, magnetic attraction would occur and the conveyor belt would be stopped. The effect was the backward strikes disappeared; the cost of installing the magnetic sensor was about £30. Savings are not specified, just left as an exercise for the reader’s imagination.
One point Shingo makes very forcibly is that defects should not be seen as the starting point of investigation – as they are when control charts are used – because defects are results, rather than causes. Defects are produced by errors and the whole aim of zero quality control is to detect and treat the causal error before it has the chance to turn into a defect. Shingo doesn’t specify that mistake proofing should be built into the design process – poka-yokes are add-ons to solve problems discovered – and, again, this probably derives from his intensely practical experience of shop floor manufacturing.
The usual objection made to poka-yokes is that they only apply to manufacturing. It’s certainly true that switches, sensors, buzzers, meters and gauges have a wide variety of uses as detection devices. But poka-yokes can be very simple. One of the earliest examples derives from the manufacture of a simple push button device, which is composed of an off and an on button with a small spring under each button. The problem was that occasionally a worker would forget to put a spring under one of them, an inspector would then have to come from another town to check the entire output and the parent company would then complain about defects made in such a simple process. The remedy was to incorporate a check list into the process, whereby the operator put two springs at a time into a small dish at the beginning of the operation. If there was a spring left in the dish afterwards, then it was obvious that it had been omitted and the defective assembly could be easily identified. Shingo acknowledges that workers do not have divine infallibility (something which supervisors tend to take for granted) and may forget things.
One thing I wondered about was Shingo’s assumption that his target readership for this book was the general public. I’m not entirely sure that it would be the first choice for any of my relatives and the cover, while functional, does not appear designed to attract the casual reader. On the other hand, it is a very good book for quality engineers and even engineers in general. Shingo’s message is simple and clearly expressed and, for that reason, the book will probably appeal to those in any sort of business improvement also. Another attraction is that it is a helpful book for people who may have been put off quality control by the perceived mathematics involved. Shingo proposes that the combination of source inspection and poka-yoke may lead to the happy situation where “inductive statistics has been rendered unnecessary in the area of control”.
Still, it must be noted that although poka-yokes are powerful and effective ways of reducing the number of defects, they are techniques, rather than objectives. The success rate of a process can be vastly improved if self checks, successive checks and poka-yokes all work together and that will need – or engender – an atmosphere of continuous improvement which will, ultimately, lead to reliable products, satisfied customers and, hopefully, empowered workers.
As for the people who still – even after reading all this – think that poka-yokes can only work in a manufacturing context, all they need to do is to consider something as mundane as an electric kettle, which automatically switches off when the water boils. Just think for a moment back to the dark ages when the kitchen was filled with steam and the intensity of the domestic disaster if the kettle was left to boil dry. Poke-yoke can be applied in any situation where a process is involved which converts an input into an output. All that is needed is some “out of the box” thinking and a desire to improve the process.
© Fell Services Ltd., 2004