馃攳
Six Sigma: Root Cause Analysis Examples - YouTube
Channel: QuikSigma
[12]
Here are some thoughts for you about
[15]
root cause analysis. First, some basic
[19]
principles. Number one, for the most part
[22]
things happen for a reason, and we call
[26]
those reasons causes. The second idea is
[31]
that not all causes are equally
[34]
important.
[35]
Some are much more important than others.
[37]
And finally, if we can understand root
[42]
causes and control them then we can get
[46]
the results that we like out of our
[48]
process. One of the commonly used root
[52]
cause analysis tools, is what I call the
[55]
two-year-old approach. You know, if you've
[58]
ever had the pleasure of having a
[60]
two-year-old in your life, you know they
[62]
ask, "why?" a lot. So the point of the
[66]
exercises to ask why at least five times.
[69]
And here's an example that sometimes is
[72]
used. Hey, there's oil on the floor.
[75]
Well, the immediate reaction is, well,
[78]
let's clean it up so we have a clean
[79]
workplace. But, it's better to ask, why is
[84]
there oil on the floor?
[86]
Well, there's oil on the floor because
[88]
the machine overhead is leaking. Oh, why
[93]
is the machine overhead leaky? Well, the
[98]
machine overhead is leaking because it's
[100]
got a bad gasket. Why does it have a bad
[103]
gasket? Because we have a policy and we
[106]
got such a good deal on a large batch of
[109]
these low quality gaskets. There's
[113]
another tool, that similar in spirit, to
[117]
the 5 why approach and that sometimes is
[120]
called variation breakdown or it's
[122]
sometimes called the thought map. And all
[125]
it does is, like five whys, keeps asking
[129]
what influences this or why does this
[131]
happen? But it does not make the
[134]
assumption that it's always just one
[136]
cause. So one question is, what might
[139]
affect the amount of time that it takes
[142]
me to get to work in the morning?
[144]
Well, off hand I can think of some
[147]
different things. I might choose
[149]
different routes. For example, I might
[151]
have the scenic route, or the freeway
[153]
route, or I might prefer the route past
[156]
the coffee shop.
[158]
Certainly the level of traffic on the
[161]
road is going to make a difference. How
[163]
early I leave is going to make a
[165]
difference. And I might choose different
[167]
modes of transportation. That'll affect
[169]
how long it takes me to to get to work.
[172]
Then, for some of these, one level is
[178]
all you need to go to. For example, I may
[182]
only have three routes to choose from
[184]
and so this tree ends right here but for
[189]
the amount of traffic on the road, well,
[192]
that might be caused by the number of
[195]
accidents, by the road conditions, by the
[198]
amount of road work that's going on, or
[200]
by the time of day. If I can
[203]
avoid rush hour, so much the better.
[206]
Well then you ask, what affects road
[209]
conditions? And the road conditions are
[212]
mostly affected I think by the weather,
[215]
and it can be snowy, or it can be rainier,
[217]
it can be clear. So basically you go
[220]
through and make this little chart that
[223]
lets the causes flow down, and the lowest
[228]
level items on the chart are then what you
[232]
would consider your root causes. Now, in
[235]
some cases you're going to come up
[237]
against one of these where you say I
[239]
don't know, and that's a good place to
[241]
start an investigation. Another tool
[244]
that's very useful is the Pareto Chart.
[247]
This nested Pareto Chart is our patented
[251]
invention, and it's very handy for
[255]
finding where, and when, and by whom,
[258]
defects were made.
[259]
That's a really good simple root cause
[263]
analysis tool. For a very thorough root
[267]
cause analysis, we use a trio of tools in
[270]
concert. And that starts with the process
[273]
map. And of course in the process map
[277]
one of the important things that you're
[279]
doing is looking at your knob variables.
[282]
The input variables that make things
[285]
what they are and cataloging all of
[289]
those. Then we next move to the
[292]
cause-and-effect matrix because remember
[294]
we said not all variables were important.
[297]
Not all causes are equally important, and
[300]
in the cause-and-effect matrix, we
[302]
prioritize and bring to the surface
[304]
those variables that are likely to be
[307]
most influential. And finally we go to
[311]
the FMEA and drill down to take a good
[314]
look at the variables that the
[317]
cause-and-effect matrix indicated were
[319]
likely to be most important, and that's a
[322]
very nice root cause analysis. For a root
[326]
cause analysis tool with a little bit
[329]
more mathematical rigor, we can turn to a
[332]
process behavior chart or as some people
[335]
call them control charts. And all that
[339]
does is put your data in the order that
[342]
they happened.
[344]
It provides a center line, which is the
[347]
average or sometimes the median, and it
[350]
provides some limits here that define
[354]
what constitutes an unusual event.
[357]
Well, if I get something like this, where
[360]
I've got a point out of limit or if I've
[363]
got shifts in my process, the rules of
[369]
the process behavior chart will detect
[372]
those. Then I have a very good basis for
[375]
going and asking, what was going on right
[379]
here?
[380]
What happened that caused this to happen?
[383]
This other stuff isn't very
[387]
interesting because statistically it's
[390]
all the same.
[391]
We really can't tell one of these points
[394]
from another. But this one is different.
[395]
That makes it interesting. Sort of the
[400]
granddaddy of the root cause analysis
[402]
tools is what we call data mining, or
[405]
sometimes exploratory data analysis,
[410]
and that's just a way of finding the
[412]
hidden relationships in your data. Now,
[415]
this may look a little forbidding, but
[417]
believe me, it's not. It's extremely
[420]
simple to use. What we have down here is
[423]
some data that we have taken. We've kept
[428]
track of how many pounds of potatoes we
[431]
got and as input variables or causes, the
[436]
temperature, and the water, and the type
[438]
of soil that we planted in, and whether
[441]
or not we used to pesticide. What the
[446]
computer then does is it builds us a
[449]
model that we can play with and we can
[451]
slide this back and forth. And here's our
[454]
average harvest weight right here,
[456]
238.6, and if I get warmer weather, what
[462]
happens? Well, it goes up to 252.8. And if I
[467]
plant in sandy soil, I get only 240 pounds.
[474]
So, this gives me a wealth of information
[477]
about the relationship between the root
[482]
causes and the outcomes that I get. So
[487]
that should give you some good ideas
[489]
about root cause analysis tools and how
[493]
to use them.
[494]
Thanks for watching!
Most Recent Videos:
You can go back to the homepage right here: Homepage