Capability on Parallel Processes such as Individual Mould Cavities

Bev D

Heretical Statistician
Leader
Super Moderator
i would love to look at your model but I don't have EXCEL 2007 and the Microsoft converter trashes your file...
 
N

ncwalker

Hah.

You are about to press one of my hot buttons. We keep getting force fed new versions of Windows and what are they REALLY adding? Other than tying up all my CPU resources and forcing me to buy ANOTHER computer. And we as consumers TAKE IT!!!!!

In what other place do they FORCE features on us? I STILL don't like "new Excel." All they have done is added window trappings and pretty formats. How about an Anderson Darling Normality test function? NO. Still kurtosis and skewness which require LOTS of measurements to be valid. So I wrote my own.

But I can sure as heck color the mess out of my cells. Funny thing that. Looks great on my monitor, but when I go to a big screen for a meeting, suddenly you can't read it. Pleasing but no contrast. I have to use it as a TOOL. Rant over.

I am a Linux fan. So here is the same thing in ods format (GASP! It actually WORKS! And it was FREE!) And I went ahead and saved it in your Excel format, no guarantees though. "Backwards Compatible" is I think maybe the name of Microsofts garage band, don't think it has any meaning over there.

Here is what you will see, by sheet:

Individual: This is where you model the cavities. Fill in the yellow cells, the rest are calculated. In top section, you put in some spec limits. You also put in an overall mean and sigma. This will be the mean an sigma of your random normal variables. Number of measurements is how many for each process, and number of processes is how many parallel processes you want to model. It gets sticky a line or two below. Mean Nudge and Sigma Nudge are where you nudge the individual processes. Leave them blank and each process will generate random normal numbers based on the overall mean and sigma. Put in a value and they will be nudged. If my overall mean is 12.000 and I put in a nudge of .005, the mean of that process will be 12.005. A positive sigma nudge will add variability to that process, a negative on will reduce variability.

Plot - Individual: This basically graphs each process stream. It shows the limits. The ACTUAL average of the resulting streams will be the plusses and the +/- 3 sigma limits will be the minuses. Sort of visual feedback you have set your model up the way you want.

Collective: This is where you model your receiving inspection sampling. You change the value in D13= Pct of Total Values to Use. If in my Individual sheet I set up 20 processes in parallel with 30 measurements, that sheet will have a table of random numbers 30 x 20 = n = 600. If on the Collective sheet I set my Pct of Total Values to 10%, it is going to randomly grab 10% of the 600 (give or take) resulting in 60 random pulls from the individual streams.

Plot-Collective: This is then the histogram of those random pulls. At the bottom, the capability indices.

I'm pretty proud of the model, actually. It's pretty flexible. And if you don't know, every time you hit F9 it will regenerate ALL the random numbers to the parameters specified.
 

Attachments

  • MultiCav Capability Model.xls
    456 KB · Views: 244
N

ncwalker

And of course Elsmar won't accept a Linux file upload.

So I cheated. I changed the extension of the file from .ods to .txt

Change it back and it should open up with OpenOffice or Excel.
 

Attachments

  • MultiCav Capability Model.txt
    103.9 KB · Views: 122
Top Bottom