Analysing the data from three different cavities of one mould

[ingov]

Dear Cove members,

I have same gears from three different cavities of one mould and two pins in gearbox on which these gears are mounted and creates a gear train. We have the problem that in some combinations of the gears the gear train can't rotate smoothly. Is it possible a creation of a disigned experiment to find the right combinations(with that there is a smootly rotation)?

for instans I've tried the following design that covers all combinations of the gears, but i can't analyse it with Minitab because pin 1 and pin 2 are not factors (they are places where the gears are mounted)
gear 1 - pin 1 - gear 1 - pin 2
gear 2 - pin 1 - gear 1 - pin 2
....
and so on.

Any help will be highly appriciated!
Thank you in advance!

 

[Steve Prevette]

Re: analysing of the data

I'd first suggest some good measurement of the gears. If there is too much variation in the measurements, on a random basis some gears won't work with others, and testing pairs will just react to that random variation. However, if you find variation is small within each cavity, and there are signifcant differences between cavities, then you may be on to something.

 

[Statistical Steven]

Re: analysing of the data

Dear Cove members,

I have same gears from three different cavities of one mould and two pins in gearbox on which these gears are mounted and creates a gear train. We have the problem that in some combinations of the gears the gear train can't rotate smoothly. Is it possible a creation of a disigned experiment to find the right combinations(with that there is a smootly rotation)?

for instans I've tried the following design that covers all combinations of the gears, but i can't analyse it with Minitab because pin 1 and pin 2 are not factors (they are places where the gears are mounted)
gear 1 - pin 1 - gear 1 - pin 2
gear 2 - pin 1 - gear 1 - pin 2
....
and so on.

Any help will be highly appriciated!
Thank you in advance!

You can have pin be a factor in the study if you like...regardless you have the following combinations gears and pins

Pin 1 Pin 2
1 1
1 2
1 3
2 1
2 2
2 3
3 1
3 2
3 3

If you have repeats on the 9 combinations you can use a one-way ANOVA with 9 levels OR you can use pin as a factor and two-way ANOVA.

 

[ingov]

Re: analysing of the data

Thank you very much! I've just tried the two-way ANOVA, it gives good picture of which gear where can be mounted.

 

[Bev D]

Re: analysing of the data

Have you done a quick MSA on the repeatability of the measurement of 'smooth'?

I would treat pin as a factor (because it is) and the two different pins as levels. Just like treating the 3 different molds as levels in the factor gear.

I would run the experiment with muliptle replicates of hte gears and pins. If you can collect gears and pins that came from assemblies htat are known to fail and known to be good it will reduce your replicate set for the DoE.

Once you have pairs that are very smooth and that are very 'not smooth', Steve P's idea of measuring the dimensions on the parts would work in telling you which factor(s) affect smoothness. measuring after the DoE will provide the large contrast in smoothness to be able to detect the characteristics with confidence...