TS16949 8.1.2 includes a statement that over adjustment should be understood throughout the organization. What ideas related to over adjustment should this include? :confused:

TS16949 8.1.2 includes a statement that over adjustment should be understood throughout the organization. What ideas related to over adjustment should this include? :confused:

Overadjustment is also called "tampering." "Messing" with a good process, resulting in a process experiencing more and more variation.

You may do a search on "tampering" and see what pops up.

I use use a "room thermostat" example to explain over adjustment.

Look here: "Deming on Tampering (http://curiouscat.com/deming/tampering.cfm)."

Stijloor.

In a Chrysler Stamping plant, many years ago, the Plant manager had a plaque on the wall of his office. On the plaque was the following inscription:

"The factory of the future will be run by a man and a dog - the dog is there to keep the man away from the machines........"

All because of 'over adjustment'..........

In a Chrysler Stamping plant, many years ago, the Plant manager had a plaque on the wall of his office. On the plaque was the following inscription:

"The factory of the future will be run by a man and a dog - the dog is there to keep the man away from the machines........"

All because of 'over adjustment'..........

"And the man is there to feed the dog."

Consider doing a web search fro Deming's Funnel Experiment. It will teach you about tampering and "over adjsutement."

Consider doing a web search fro Deming's Funnel Experiment. It will teach you about tampering and "over adjsutement."

That experiment is actually called: "Nelson Funnel (http://www.google.com/search?hl=en&q=nelson+funnel&btnG=Google+Search)."

Search and find the good stuff....:)

Stijloor.

Sure enough... per W.E. Deming's, Out of the Crisis, page 327, the funnel experiment is credited to Dr. Lloyd S. Nelson with support from Dr. Gipsie B. Ranney and Dr. Benjamin J. Tepping.

I have the Quality Gamebox from PQ Systems, Inc. The call it the "Deming Funnel." These electronic games are wonderful for teaching principles in a fun and visual way.

http://www.pqsystems.com/products/sixsigma/QualityGamebox/QualityGamebox.php

TS16949 8.1.2 includes a statement that over adjustment should be understood throughout the organization. What ideas related to over adjustment should this include? :confused:

It is a great idea...but before it can be understood throughout the company, it has to be understood by at least one person in the company. Unfortunately, it is not as simple as the statement makes it seem. For example, anyone using X-bar/R charts for precision machining does not understand overadjustment (they are encouraging it, actually). Most precision machining companies are overadjusting like crazy because they use X-bar/R.

It's not that I want to get into that whole concept again, (see http://elsmar.com/Forums/showthread.php?t=20935&highlight=precision+machining for details) but it is a prime example of the problem, and how pervasive it is in industry. Even the auditors do not know about this particular problem. But, they are only human...? :cool:

X Bar/R charts do not cause overadjustment or tampering. It is the understanding of variation, and the improper use statitiscial data that may allow the overadjustment.

Run charts are not bad in and of themselves. They do not prove anything. With propeor use, they aloow one to see varioation that may not have been perceived otherwise. With proper use, charts allow one to "see" variation.

X Bar/R charts do not cause overadjustment or tampering. It is the understanding of variation, and the improper use statistical data that may allow the overadjustment.


OK, technically using X-bar/R charts do not cause overadjustment (they are just a piece of paper - hard to blame them specifically:sarcasm:). But, using them to indicate when an adjustment is needed in precision machining is the root cause of overadjustment in those processes, because they are the wrong chart, using the wrong statistics from the wrong distribution. It is easily proven. And, although not intentional, it is the equivalent to tampering.

Sure, you will "see variation", but using an X-bar/R chart will show you variation that has little to do with the fundamental variation of precision machining you are trying control. (See link previously mentioned for details):cool:

I have some exaples of overadjustments that I have used to explain it during an internal training of new employees:whip::


- wrong decision due "aliasing".
the sample time period is larger then half of source period where larger control period means possible bigger troubles (it may be an X-R chart with "every 2 hours control" with a noise/disrtub source with a time period of 3 hours). You can see some very strange behaviour of the data, think about the wheels of a car on films that spin in the opposite direction of moving.

- wrong decision due to delay on system responce. Every production system can be thought as a feedback control system with delay. You change something but it takes time to get the new results and some oscillation may occur if there is no strategy to dump it. I have tried to explain with a shower. You are in a hotel and your wife open the cold and hot water of the shower for You. there is no indication on the taps which is for cold water and which for hot water. As soon as You enter into the shower You wont cooler water so You try with the right tap but immediately the water seems hotter. therefore You turn the right tap to the original position and try with the left tap and quite soon the water becomes too cold. You turn the left tap to the original position and stat everything again thinking the right tap is for hot water and the left one is for cold water. The problem is the right tap control the cold water and left one the hot water. When You turn the right one You increase the flow so it seems the water is hotter but You stopped before the cold water column comes out. The cold water column touches your skin soon after You turn the left tap etc. you started a neverending adjustment:lmao:

I have some examples of overadjustments that I have used to explain it during an internal training of new employees:whip::

These are very good examples of why it is important to use correct frequencies for sampling.

For normal distributions, people tend to listen to production, and make the frequency low enough that they stop complaining that they do not have time to do the measurements. :frust:

For the uniform distribution, it is easy to get the correct frequency - dividing the time (or number of parts) it takes to go from one control limit to the opposite control limit, and divide by 5. Not comfortable with that? Well, divide by 7 (the number of checks that would indicate a run, if it had been a normal distribution). The only time that is risky (as in too infrequent) is if it takes a week to go from one control limit to the other (yes, I have witnessed that). If so, you may want to sample more often so that if a special cause created an issue, you would not have to sort a whole day's worth of work.

As far as overadjustment, that comes from making adjustments to the process before the control chart tells you to do so. Classic example: I asked an operator how often they charted a grinding process. They said every two hours, just like the control plan said. Then I asked how often the adjusted the process. "Every 15 minutes" they said. That is overadjustment. In fact, the process he was charting was his adjustments, not the machine. He had become the process! You really want to avoid having the operators become the process!