Quote:
Originally posted by Marc Smith:
"Poke-Yoke is used to mistake-proof processes, typically by the use of sensors (read: Identify Defects) in equipment in a production line. The idea is to identify defects as early in the process as possible where it costs the least to 'repair' or react to. The bottom line to Poke-Yoke is it is in fact 100% mechanical or electronic or electro-mechanical inspection."
|
Although you are correct that many PokeYokes are simple mechanical 100% inspections other Poke Yokes are not. The kind you describe are known as Detection PokeYokes. There are two other types. These 2 types are Preventive. And are much more effective than SPC in their specific applications.
First let me define the 2 types: One type detects an error BEFORE it becomes a defect. typically, the "rework" required for the error is very simple and quick and obvious. (Ex: a part is placed upside down on an assembly. the Poke Yoke detects this before the part is actually assembled or attached. "rework" is that the operator re-orients the part. 2nd example. An assy needs 4 bolts. The operator is distracted and forgets to install the last bolt. A sensor detects this before the assy moves on and the operator installs the bolt) OK at this point you still have some "hidden rework" and cycle time delay, but it is minimal. (and one might debate the semantics of the error vs defect, but that is how Shingo classified it and there is a real world distinction in cost and time.
The other is a true prevention Poke Yoke in every definition of the term. It is typically a mechanical "thing" that prevents the error from being made in the first place. (Ex: that upside down part from the previous example can only be installed in the correct way. Usually the use of a pin or slot or other device that makes misassemble impossible - at least without the use of a high force istallaltion device: read large mallet.)
Now SPC and the last two types of Poke Yokes that I have just described are actually complementary. They do different things. The significant difference is that Poke Yokes are targeted at MISTAKES. Mistakes are NOT variation based. They are true attributes data. They cannot be described by a continuous distribution function. SPC cannot be used effectively on the input side of the MISTAKE (at least not until we can monitor brain waves and such...) Output SPC (the event must happen to be plotted) is an after the fact kind of thing and has limited utility. It CAN detect when mistakes are starting to occur more often and one could look to environmental changes (morale, training levels, distractions, etc) to maintain a minimum level of mistakes. (should always be used for safety/accident data!) However, most manufacturing mistakes are relatively cheap to PREVENT and thus this is a much better approach than SPC from a defect reduction standpoint.
The first type of poke Yoke (100% detection of the defect) is absolutely the best approach for those defects that have a serious affect should they occur. Especially if the occurence rate is fairly rare. SPC will detect shifts in the occurence rate, but will not (nor is it intended to) guarantee that all defects are caught. Hence the first type of Poke Yoke is an absolute for rarish events that can cause serious injury, or very high warranty costs or large cycle time affects.
Particularly if the defect is caused by a mistake, rather than variation in some input factor. And of course, we can't put SPC on every input factor that can cause a defect. (especially since we'll never know what they all are!) So this type of Poke Yoke is a perfect safety net (pun intended!).
SPC and Poke Yokes really live in two different worlds with only a slight overlap...Both are equally valuable, just in different situations.
Hope this clarifies thing a little.
Bev Daniels
beverly.daniels -- Intel.com
(The opinions expressed herein are not necessarily those of my employer. In fact, they probably bear absolutely no resemblance to those of my employer)
SPC (minimizing defects) Vs Poka Yoke (zeroing out defects)
Linear Mode
