1. Mistake-Proofing Training
John R. Grout
Campbell School of Business
Berry College
Welcome to Design error-proofing. Thank you for attending.
(Self-introduction….)

2. Where does mistake-proofing fit in your quality tool box?
Non-Conformances come from many sources including:
Variation
Culture
Complexity
Mistakes
Culture
Complexity
Variation
Mistakes
READ SLIDE
we’ve already discussed errors and variance (SPC) Now let’s talk about complexity...
Each must be managed
to improve
quality and reliability.

3. Today’s presentation:
Individually errors are rare. However, as a group they are a major cause of failures
Parts and processes can be controlled in ways that dramatically reduce the occurrence of failures due to mistakes.
Today I am going to argue that errors, human beings inadvertently doing things wrong, are a major problem, and that design engineers can eliminate the vast majority of these errors. I am not proposing that every possible error can be eliminated, only that most can be eliminated.

4. Adapted from M. Hinckley, Quality by Design, 1996
Definition
Mistake
The execution of a prohibited action,
the failure to correctly perform a required action or the misinterpretation of information essential to the correct execution of an action
Mistake proofing
the use of process or design features to prevent manufacture of non-conforming product.
Adapted from M. Hinckley, Quality by Design, 1996
(READ SLIDE TEXT)
A caveat about this definition:
We do not use errors and mistakes as synonyms. However, in this presentation we draw on the work of many people, and some authors will use the word mistake as a synonym with error. Where one of these is quoted, we have not changed their words. We do indicate their less precise use of the word mistake by italicizing it on the slide.

5. Everyday Examples
3.5 inch diskettes cannot be inserted unless diskette
is oriented correctly. This is as far as a disk can be inserted upside-down. The beveled corner of the diskette along with the fact that the diskette is not square, prohibit incorrect orientation.
Fueling area of car has three error-proofing devices:
1. insert keeps leaded-fuel nozzle from being inserted
2. tether does not allow loss of gas cap
3. gas cap has ratchet to signal proper tightness and
prevent overtightening.
PARAPHRASE SLIDE
ask for more car examples.
New lawn mowers are required to have a safety bar on the handle that must be pulled back in order to start the engine. If you let go
of the safety bar, the mower blade stops in 3 seconds or less.

6. Why use mistake-proofing?
You can make more money
It works where other techniques don’t
evidence of
effectiveness
the difficulties with human error
The final user deserves it.

7. Evidence of the Effectiveness Source: Productivity Inc
Evidence of the Effectiveness Source: Productivity Inc. and Shingo prize profiles
AT&T Power Systems is first US manufacturer to win the Deming prize. Average outgoing defects reduced by 70%
A washing machine drain pipe assembly line produced 180,000 units without a single defect (6 months).
TRW reduced customer PPM’s from 288 to 2.

8. Evidence of the Effectiveness Source: Productivity Inc
Evidence of the Effectiveness Source: Productivity Inc. and Shingo prize profiles
Cooper Automotive:
95% less defects than nearest competitor
75% less injuries
99.6% less customer defects (13 ppm)
88% in-plant defect reduction
70% less warranty cost
89% scrap reduction (0.7%)
60% productivity increase

9. Devices Tend to be Inexpensive...
Evidence of the Effectiveness
Devices Tend to be Inexpensive...
Cost of Poka-Yoke Devices 1
0.9
0.8
0.7
0.6
Frequency of Occurrence
0.5
Probability
Cumulative Probability
0.4
0.3
0.2
0.1
$25 or less
$25 to $100
$100 to$250
$250 to $1000
$1000 or more
Cost

10. …and Very Effective
The “10:1, 100:1, even 1000:1” rates of return referred to by Bhote above are not unreasonable in practice.
Dana corporation has reported a $500,000 savings resulting from a $6 device. (83,000:1)
AT&T Power Systems (Lucent Technologies) reported net saving of $2545 per device (3300 devices) [Marchwinsky, 1997]. (25:1*)
Weber Aircraft reports saving $350,000 during their first year of implementation of approximately 300 devices. (11:1*)
*Assumes and average devise cost $100

11. Common Mistake-proofing Devices
Guide Pins
Blinking lights and alarms
Limit switches
Proximity switches
Counters
Checklists

12. The difficulties with human error Why existing tools are not enough
Motorola findings:
...it became evident early in the project that achieving a Cp greater than 2 would go only part of the way. Mistake-proofing the design would also be required ... Mistake-proofing the design is an essential factor in achieving the [total number of defects per unit] goal.
Smith, B. IEEE Spectrum 30(9) 43-47
SPC alone provided Motorola 10X more defects than expected.
Sandia Labs (Rook, SCTM93-62(14)) did experiments in 1962 that show undetected omissions occur about once in 33,000 operations. A simple eight step operation will result in 240 PPM due to mistakes. This is much higher than the 3.4 PPM goal of Motorola’s six sigma program.

13. Errors are difficult to manage using statistics.
normal
variation
omitted
operation
Probability
Here’s an example:
If a worker forgets to drill a hole, variance is not an effective way to describe it. Since a worker is not likely to repeat the error consistently until a sample is taken, it probably won’t end up in a sample. Even if it does, it will be averaged in or treated as an “outlier“ (not indicative of future process performance).
Typically, human error is considered a common cause NOT a special cause (Chase and Stewart, 1994).
variance tools tend to be ineffective controlling human error.
Slot width

14. Poka-yoke & SPC

15. Error-proofing & SPC
SPC is good at detecting shifts in the process mean or variance. Changes to the process must be ongoing to be readily detected.
Human errors tend to be rare, intermittent events. They are not readily detected by control charts.
Use error-proofing (not SPC) to reduce defects caused by human error
You need both error-proofing and SPC.
(READ SLIDE)
If any of you are really interested in the specifics of how SPC and error-proofing are related the study guide accompanying this presentation indicates some articles that address this aspect in detail.
Now for the basics of error-proofing …(NEXT)
Motorola got an order of magnitude closer to their goal
using a combination of SPC and error-proofing.

16. Individually mistakes are rare
Task Type Probability
Detection of deviation or inspection 0.07
Alpha input per character
Numeric input per character
Assembly per task element
As a group they are common
Research study #1 (Harris) >0.80
Research study #2 (Rook)
Research study #3 (Voegtlen)
Research study #4 (Headlamps) >0.70
NASA mishaps >0.50
FAA Maintenance problems >0.94
At the start of this presentation we said that individual errors were rare events. This is shown by the data above
Here are some human error rate data. You’ll notice that the error rates differ for each task type. Notice that the more routine tasks are performed more reliably.
Does this data suggest we should spend our time addressing knowledge-based mistakes (since they are more likely at any given opportunity)? No, workers experience far more opportunities to make skill-based errors. The number of opportunities to make an error performing skill-based actions far exceeds the number of opportunities to make knowledge-based errors.
Data from “Human Reliability Data - The State of the Art and the Possibilities”
Jeremy C. Williams, 1989 CEGB

17. To err is human Have you ever done the following:
Driven to work and not remembered it?
Driven from work to home when you meant to stop at a store?

18. It happens to workers too.
Workers finish the shift and don’t remember what they have done.
After building green widgets all morning, the workers put green parts on the red widgets in the afternoon.
The mistakes I would like to focus on are some times called slips. They occur when an action is executed on “autopilot” but has an unintended result.
Mistake-proofing usually(but not always) will involve precluding behaviors that under other circumstances would be correct.
Putting green parts on widgets is a correct behavior when the widgets are green, but not when they’re red.

19. Corrective action
I recently polled the Quality newsgroup on the internet. A majority reported at least 20-30% of corrective actions were “worker reprimanded and retrained.”
The admonition to “be more careful” or “pay attention” are not effective for humans, especially in repetitive environments.
Survey results here! xxx

20. “Be more careful” not effective
“The old way of dealing with human error was to scold people, retrain them, and tell them to be more careful … My view is that you can’t do much to change human nature, and people are going to make mistakes. If you can’t tolerate them ... you should remove the opportunities for error.”
“Training and motivation work best when the physical part of the system is well-designed. If you train people to use poorly designed systems, they’ll be OK for awhile. Eventually, they’ll go back to what they’re used to or what’s easy, instead of what’s safe.”
“You’re not going to become world class through just training, you have to improve the system so that the easy way to do a job is also the safe, right way. The potential for human error can be dramatically reduced.”
Chappell, L The Pokayoke Solution. Automotive News Insights, (August 5): 24i.
LaBar, G Can Ergonomics Cure ‘Human Error’? Occupational Hazards 58(4):

21. A new attitude toward preventing errors:
“Think of an object’s user as attempting to do a task, getting there by imperfect approximations. Don’t think of the user as making errors; think of the actions as approximations of what is desired.”*
Quality professionals are not the only group interested in the outcomes of processes. Psychologists have been interested in human error. Here’s what Donald Norman said in a very interesting book titled The design of every day things
The human brain’s default mode of operation is pattern recognition and autopilot execution. If the pattern is familiar, a behavior that has been successful in the past is “launched.” It’s only when feedback suggests that things are not going as planned that more in-depth though is called up.
*Source: Norman

22. A New Attitude toward Preventing Errors
Make wrong actions more difficult
Make it possible to reverse actions —to “undo” them—or make it harder to do what cannot be reversed.
Make it easier to discover the errors that occur.
Make incorrect actions correct.
Donald Norman has some recommendations for responding effectively to how the human mind works.

23. Judgment Inspection
Involves sorting the defects out of the acceptable product, sometimes referred to as “inspecting in quality.”
The consensus in modern quality control is that “inspecting in quality” is not an effective quality management approach.
Judgment inspection does not improve process and should be used only in the short term.

24. Successive checks & Self-check (post-production product inspection)

25. Source Inspection (preemptive process inspection)
Self-Correcting Process
Error opportunity elimination (Mistake-Proof Design)

26. Inspection techniques

27. Setting Functions

28. Setting Functions The real question you need to ask:
How are you going to detect an error?
automatic, not dependent on human attention
fail in “detect” mode
simple & low cost if possible

29. Regulatory Function (Cues)
Control Methods

30. Regulatory Function (Cues) The real questions you need to ask:
How are you going to stop the process?
the worker needs to get the message?
By audible or visible warning
By prohibiting further processing
How are you going to eliminate the possibility of error?
The Contrapositive of Murphy’s Law
Simplicity
Symmetry
Statement: If P then Q
Contrapositive: if not Q, then not P
the statement and its contrapositive are logically equivalent
If anything can go wrong, it will
If it can’t go wrong, it won’t

31. Examples

32. Hierarchy of Techniques
Source inspection
Self-checks
successive-checks
Judgment inspection
Control methods
Warning methods
Informative
inspection
Better
Better

33. Where it works & where it does not

34. Put “Knowledge in the World”
Precise outcomes without precise knowledge or action?
provide clues about what to do:
natural mappings
affordances
visibility
feedback
constraints
One way to help the worker avoid mistakes is to encode clues about what to do in the design. Donald Norman calls this putting “knowledge in the world.” Here are some examples of how to accomplish this:… (NEXT)

35. Which dial turns on the burner?
Natural Mappings:
Which dial turns on the burner?
Natural mapping means building logical one-to-one correspondences that allow the operation or task to become more obvious. If proper stove operation requires a label, it could be designed better.
In this case, stove B has a natural mapping that makes the knob that turns on the left rear burner obvious. It is a better design than stove A where no natural mapping exists.
Stove A relies on labels to instruct the user on which knob to use. The use of labels or instructions often it could have been designed better.
Stove A
Stove B

36. How would you operate these doors?
Affordances
How would you operate these doors?
Push or pull? left side or right? How did you know? A B C
Affordances are properties that suggest how an object could be used.
Q: How does “A” operate? and how did you know?
A: The door handle in “A” affords (is for) pulling. To use door “A”, pull on the left side (the handle and hinges both help suggest what to do).
Q: Door “B”?
A: On door “B”, push on the right side (push plate signals what to do).
Q: how about “C”?
A: Door C is a problem. the lack of any affordances suggests pushing it but which side? (unknown). If the latch is magnetic, proper operation could be 1) push a little, 2) then let go, and 3) pull open.

37. visibility and feedback
Visibility means making relevant parts visible, and effectively displaying system status
Feedback means providing an immediate and obvious effect for each action taken.
If you were to use a magnetic catch on a door like “C” on the prior slide, you could reduce the uncertainty by making the door glass. Seeing the hinges and catch would enhance visibility and make the door easier to use.
Some new light switches change slowly from on to off. The reason is to make the transition “less shocking to the eyes.” The delay some times causes concern and a flurry of rapid button pushing for people unfamiliar with its operation. They are used to the rapid feedback of standard light switches.

38. "The first sign of an intelligent tinkerer
Lego exercise
Constraints
The contents of the bag will make a toy. Engineers and designers spent months designing, fine-tuning this product. It is intended for use by 5-10 year olds. Please try to put them together CORRECTLY. You should use all the parts.
While we are considering constraints, I’d like to invite (insert volunteer’s name here) to put this together (do not disclose what the toy is or what it looks like).
READ SLIDE
"The first sign of an intelligent tinkerer
is to save all the parts."
Aldo Leopold

39. Here is what your
bag should contain.
24 pieces total

40. Answer revealed here

41. Constraints? Constraint Description Application
Physical Shape and size of objects Front vs rear control their relationship hub
Semantic Relies on clues from Face oriented meaning of the situation correctly
Cultural Adheres to known Front & rear convention lights
Logical Based on making sense of Assembly by the relationships process of elimination
READ SLIDE
Now let’s look at the toy. Does it look right?
Physical: legs inserted in body: only place they’d fit (unique attachment, also true of hands)
Semantic: face looks out through helmet opening
Cultural: Red on back for brake light, yellow on front for head light (adheres to convention)
Logical: the antenna gets put on the back because you know you have to use it and there is no place left.
Source: The Design of Everyday Things, by D.A. Norman, 1988, Doubleday