The Test / Rework / Scrap Cycle

If one's only method of quality assurance is testing the finished product to see whether or not it conforms to the customer's specifications, there is a good chance that some considerable proportion will be non-conforming. As Harold F. Dodge famously said:

Consider the operation of producing a shaft with diameter 100mm and a tolerance of ±0.5mm. Any part which is below 99.5mm will have to scrapped at some cost to the manufacturer. Thus a common practice is for the operator to aim high. This reduces the risk of producing shafts which fall below the lower specification limit, but greatly increases the chance that shafts will exceed the upper limit, and therefore have to be processed further. This has a negative impact for both the customer and the manufacturer.

The customer wants shafts that are 100mm in diameter, but is willing to accept a variation of 0.5mm. The process as it is currently being operated will produce shafts with an average diameter that is closer to the upper specification limit than to the target. In his book Out of the Crisis, W. Edwards Deming makes the following point:

A vice-president in charge of manufacturing told me that half his problems arise from materials that meet the specifications ¹

The manufacturer also suffers by having to absorb the cost of rework. This effectively increases the average production cost for each good part shipped. This does not even take into account the potential extra costs of late shipment penalties, or even loss of business because of customer dissatisfaction.