Monday, October 26, 2009

Speaking precisely -- or accurately

A discussion about "group size" among gun nuts leads to this bit of muttering...

Accuracy as a term used by gunnies is really "precision." This is a bread and butter part of my work a day world. It's remarkable how hard it is for folks to grok. Even among degreed engineers only a few are consistently able to synthesize and apply it to their situation.

Put less precisely than the dictionary definitions but more in the vernacular that seems to "stick" with the folks I deal with:

How tight the group = how precise the (weapon) system.
How far the group deviates from the bullseye = how accurate the (weapon) system.

Lessons from quality improvement trenches:

The key to improvement is reduction of variation. Is there a life lesson lurking in there?

Along the way what all QE's should be able to articulate but not 1 in 100 Certified Quality Engineers I've interviewed can actually tell ya:

1. Make sure your measurement system has at least 10:1 better resolution than the less significant digit you're measuring.

2. Make sure you measurement system can adequately discriminate. i.e. Gauge's variation should no worse than 1/10 of your system variation (99 times out of 100 Gage R&R is a waste of time and you're better of doing an Isoplot for this).

3. Always work to reduce range of variation (R) first. Then work to shift the average (X bar) to the target (nominal).

4. Identify and reduce the largest source of variation. Working on reducing lesser sources of variation will just drive you nuts.

If you can think in mathematical terms maybe it'll help to consider that variation accumulates as the sqrt of the sum of the squares of individual variation.

e.g. Given, you have 3 factors with variations of 3,4 and 5.
Then total of each is squared and summed and sqrt is taken for final result(variation = 9 + 16 + 25 = 50sqrt).

You see that? The largest source of variation "5", has the same total effect on variation as the 3 and 4 combined(!).

There. You're smarter than 99 out of 100 QE's I've interview and probably 9 out of 10 Six Sigma blackbelts.

So is there a life lesson in there?

pfftt. Who knows. Probably.

At a minimum that understanding and a persistent curiousity is enough to make a decent living. /heh

No comments: