In the early 1980s, during one of my first benchmarking trips to Japan, I was intrigued by a fixture a young
Fuji-Xerox worker proudly showed me. His job was to assemble two small pieces of metal with two screws, and it seems that some time before he had assembled a small part incorrectly. After the
error was brought to his attention, he had used his lunch hour to make a small fixture in the woodworking shop. The two metal pieces fit in the wooden fixture only one way—the correct way. His supervisor timed him demonstrating the old way and the new way. The old way took 38 seconds, and the new way took only 18 seconds. With this, I was introduced to what they called
"foolproofization." I later learned that the Japanese word for this technique is poka-yoke,
for which the more proper English translation is "mistake-proofing" or "error-proofing." During that trip and many later visits, I encountered many other forms of foolproofization. In Komatsu, I saw a worker tightening bolts on a large bulldozer. If his wrench didn't count 18 quarter-turns, a loud bell started ringing when he tried to move the bulldozer down the line. Only when he had tightened all 18 bolts would the bell stop. I've seen many similar examples in electronics and automotive companies.
I recently read about a New York hospital's novel idea for foolproofing, called "sign for quality." After several mistakes involving operating on the wrong arm, leg or even
patient, the hospital created a new procedure. Before the patient is anesthetized, the surgeon marks the place for the incision directly on the patient's body and both the surgeon and patient
sign their names agreeing that this is the right location. This makes it almost impossible to operate on the wrong area. Although the concept of mistake-proofing has been
around for many years, it's still practiced too infrequently in the United States. Americans seem to be far fonder of implementing complex statistical process control systems or inspection
systems to detect errors after they've been made than trying to invent clever prevention methods. In "Operations," Section 22 of Juran's Quality Handbook, Fifth Edition
(McGraw-Hill, 1999), Frank Gryna gives an excellent discussion of error-proofing the process. He provides five categories of error-proofing with numerous examples of each category. These five
categories are fail-safe devices, magnification of senses, redundancy, countdowns, and special checking and control devices. There are many types of fail-safe devices. Among
the most common are interlocking sequences, where operation B cannot start until A is finished (e.g., operation A creates a hole and operation B starts with a probe in the hole that operation A
created). Other methods include alarms and cutoffs that signal or even stop a line when materials run out, temperature exceeds a certain limit or other problems are detected. The foolproof
fixture at Fuji-Xerox fits in this category. Another category is redundancy. Many companies use multi-identity codings, redundant actions or multiple tests to reduce the chance
of error. For example, a number can be both read from a barcode and entered from a printed label, and only if both numbers match in the computer is the number accepted. In information quality
systems, data are often entered independently by two different operators, and an automated verification system identifies any discrepancies. Countdowns represent another
category. Checklists are actively used to ensure that every step is accomplished. Spacecraft launches are visible examples of the use of countdowns. A fourth category of
foolproofing methods covers special checking and control devices. One familiar example is the automated credit card checking done by many e-commerce sites during ordering. Invalid numbers are not
accepted, and feedback is given instantly. This instant feedback is one of the core principles of foolproofing. We should try to build systems where the work performance
provides information to the worker about whether the task was completed correctly. Takeshi Nakajo and Hitoshi Kume performed a classic study of foolproofing in the early
1980s: They examined and classified about 1,000 examples of fool-proofing collected from assembly lines. Their classifications were elimination, replacement, facilitation, detection and
mitigation. A table summarizing their findings is given in Gryna's "Operations" section (mentioned previously) with the objectives of each of these principles and examples.
Another excellent reference is the small, extremely easy-to-read book Mistake-Proofing for Operators: The ZQC System (Productivity Press, 1996). This book is based on Shigeo Shingo's
outstanding reference book, Zero Quality Control: Source Inspection and the Poka-Yoke System (Productivity Press, 1986). The concept of foolproofization should be part
of all quality systems, and the tools described in the above references should be in all quality and operations managers' toolboxes. These tools may be simple, but they are very powerful means of
preventing both small and critical errors. About the author A. Blanton Godfrey is
chairman and CEO of Juran Institute Inc., a leading international research, consulting and training company focused on quality management. He is also the co-editor in chief of the fifth edition
of Juran's Quality Handbook. Contact him by e-mail at agodfrey@qualitydigest.com . |
