… on Mistake-proofing, Poka-Yoke, and ZQC
By John R. Grout, and Brian T. Downs
Shigeo Shingo was one of the industrial engineers at Toyota who has been credited with creating and formalizing Zero Quality Control (ZQC), an approach to quality management that relies heavily on the use of poka-yoke (pronounced POH-kah YOH-kay) devices. Poka-yoke is Japanese for mistake-proofing. These devices are used either to prevent the special causes that result in defects, or to inexpensively inspect each item that is produced to determine whether it is acceptable or defective.
A poka-yoke device is any mechanism that either prevents a mistake from being made or makes the mistake obvious at a glance. The ability to find mistakes at a glance is essential because, as Shingo writes, “The causes of defects lie in worker errors, and defects are the results of neglecting those errors. It follows that mistakes will not turn into defects if worker errors are discovered and eliminated beforehand”[Shingo 1986, p.50]. He later continues that “Defects arise because errors are made; the two have a cause-and-effect relationship. … Yet errors will not turn into defects if feedback and action take place at the error stage”[Shingo, 1986, p. 82]. We suspect that Shingo and Deming would have a protracted discussion about whether workers or management are responsible for defects. No resolution of that issue is undertaken here.
An example cited by Shingo early in the development of poka-yoke shows how finding mistakes at a glance helps to avoid defects. Suppose a worker must assemble a device that has two push-buttons. A spring must be put under each button. Sometimes a worker will forget to put the spring under the button and a defect occurs. A simple poka-yoke device to eliminate this problem was developed. The worker counts out two springs from a bin and places them in a small dish. After assembly is complete, if a spring remains in the dish, an error has occurred. The operator knows a spring has been omitted and can correct the omission immediately. The cost of this inspection (looking at the dish) is minimal, yet it effectively functions as a form of inspection. The cost of rework at this point is also minimal, although the preferred outcome is still to find the dish empty at the end of assembly and to avoid rework even when its cost is small. This example also demonstrates that poka-yoke performs well when corrective action involves trying to eliminate oversights and omissions. In such cases, poka-yoke devices are often an effective alternative to demands for greater worker diligence and exhortations to “be more careful.”
An example of a poka-yoke device at General Motors (GM) was described by Ricard [ Ricard, L.J., “GM’s just-in-time operating philosophy”, in: Y.K. Shetty and V.M. Buehler, (Eds.)., Quality, Productivity and Innovation. Elsevier Science Publishing, New York, 1987, pp. 315-329.]: “We have an operation which involves welding nuts into a sheet metal panel. These weld nuts will be used to attach parts to the car later in the process. When the panel is loaded by the operator, the weld nuts are fed automatically underneath the panel, the machine cycles, and the weld nuts are welded to the panel. You must remember these nuts are fed automatically and out of sight of the operator, so if the equipment jams or misfeeds and there is no part loaded, the machine will still cycle. Therefore, we have some probability of failure of the process. An error of this nature is sometimes not detected until we actually have the car welded together and are about to attach a part where there is not a nut for the bolt to fit into. This sometimes results in a major repair or rework activity.”
“To correct this problem, we simply drilled a hole through the electrode that holds the nut that is attached to the panel in the welding operation. We put a wire through the hole in the electrode, insulating it away from the electrode so as it passes through it will only make contact with the weld nut. Since the weld nut is metal, it conducts electricity and with the nut present, current will flow through, allowing the machine to complete its cycle. If a nut is not present, there will be no current flow. We try to control the process so that the machine will actually remain idle unless there is a nut in place.”
Shingo identified three different types of inspection: judgment inspection, informative inspection, and source inspection. Judgment inspection involves sorting the defects out of the acceptable product, sometimes referred to as “inspecting in quality.” Shingo agreed with the consensus in modern quality control that “inspecting in quality” is not an effective quality management approach, and cautioned against it.
Informative inspection uses data gained from inspection to control the process and prevent defects. Traditional SPC is a type of informative inspection. Both successive checks and self-checks in ZQC are also a type of informative inspection. Successive checks were Shingo’s response to the insight that improvements are more rapid when quality feedback is more rapid [1986, pp. 67-69]. Work-in-process undergoes many operating steps as it is moved through a manufacturing facility. Often inspections are conducted at intermediate stages in the process. Shingo’s concern was that the inspections may not occur soon enough after production to give the best information necessary to determine the cause of the quality problem so that it can be prevented in the future. By having each operation inspect the work of the prior operation, quality feedback can be given on a much more timely basis. Successive checks are having the nearest downstream operation check the work of the prior operation. Each operation performs both production and quality inspection. Effective poka-yoke devices make such an inspection system possible by reducing the time and cost of inspection to near zero. Because inspections entail minimal cost, every item may be inspected. Provided that work-in-process inventories are low, quality feedback used to improve the process can be provided very rapidly.
While successive checks provide rapid feedback, having the person who performs the production operation check their own work provides even faster feedback. Self-checks use poka-yoke devices to allow workers to assess the quality of their own work. Because they check every unit produced, operators may be able to recognize what conditions changed that caused the last unit to be defective. This insight is used to prevent further defects. Self-checks are preferred to successive checks whenever possible.
Since the main difference between successive checks and self-checks is which work station performs the inspection, in this research we do not distinguish between the two types of informative inspection. Both successive and self-checks provide information “after the fact.”
Source inspection determines “before the fact” whether the conditions necessary for high quality production exist. (Note that Shingo’s use of the term source inspection is not the practice of having the buyer’s representative inspect the quality of work-in-progress at the supplier’s facility, which is also called source inspection.) Shingo writes, “It had dawned on me that the occurrence of a defect was the result of some condition or action, and that it would be possible to eliminate defects entirely by pursuing the cause” [Shingo, 1986, p.50]. He further writes that “I realized that the idea of checking operating conditions before the operations rather than after them was precisely the same as my concept of source inspection” [Shingo,1986, p.51].
With source inspection, poka-yoke devices ensure that proper operating conditions exist prior to actual production. Often these devices are also designed to prevent production from occurring until the necessary conditions are satisfied. Norman  refers to this type of device as a “forcing function.” The example from GM that “forces” the nut to be present before welding can occur is an example of source inspection.
Source inspection, self-checks, and successive checks are inspection techniques used to understand and manage the production process more effectively. Each involves inspecting 100 percent of the process output. In this sense, zero quality control is a misnomer. These inspection techniques are intended to increase the speed with which quality feedback is received. And although every item is inspected, Shingo was emphatic that the purpose of the inspection is to improve the process and prevent defects, and therefore is not intended to sort out defects (although in some cases that may also be an outcome) [Shingo,1986, p. 57]. Shingo believed that source inspection is the ideal method of quality control since quality feedback about conditions for quality production is obtained before the process step is performed. Source inspection is intended to keep defects from occurring. Self-checks and successive checks provide feedback about the outcomes of the process. Self-checks and successive checks should be used when source inspection cannot be done or when the process is not yet well enough understood to develop source inspection techniques. Additional information about ZQC and failsafing is provided in the poka-yoke reading list.
In Shingo’s seminal book on ZQC , he criticized SPC and suggested that ZQC should supplant SPC as the preeminent tool for defect elimination in quality control. His main argument against SPC was that it is by nature an intermittent form of inspection, and therefore allows for some number of defects to occur. He further argued that SPC is designed to maintain the current level of defects , rather than to aggressively seek to eliminate them. In addition, Shingo claimed that “…a look at SQC methods as they are actually applied shows that feedback and corrective action – the crucial aspects of informative inspections – are too slow to be fully effective.” [Shingo, 1986, p.68]
Given the fact that applications of SPC generally have substantial intervals between the taking of samples, it seems reasonable to argue that feedback will be faster with source inspection and informative inspection in ZQC. However, it is not clear that ZQC should be systematically faster than SPC at insuring corrective actions. Indeed, according to Shingo [Shingo, 1986, p.71], “Defects will never be reduced if the workers involved do not modify operating methods when defects occur.” The willingness to take corrective action is a function of the attitude and commitment of both managers and workers, not an intrinsic attribute of a particular approach to quality management. Shingo’s complaint about the actual implementation of SPC may also apply to ZQC.
A detailed, academic treatment of the relationship between SPC and ZQC is presented in working papers by Grout and Downs (1995). The essence of their conclusions is when used for informative inspection,
- ZQC is not as effective as SPC for defects that result from variance in measurement data
- ZQC is a special case of SPC for defects that result from variance in attribute data.
- ZQC’s source inspection can be used effectively to eliminate mistakes and in conjunction with SPC to eliminate the recurrence of special causes.