# Process capabilities study mechanical properties

Discussion in 'Capability - Process, Machine, Gage …' started by laffranchinis, Jan 7, 2021.

Tags:
1. ### laffranchinisMember

Joined:
Jan 30, 2020
Messages:
12
2
Trophy Points:
2
Hallo everybody,

Does it make sense to calculate Cpk on properties like compressive strengths and moduli?
I mean, the calculation of such indirect properties is based on different single measurements:
- dimensions (length/ width/ thickness)
- mechanical tests

Samuele

2. ### MinerModeratorStaff Member

Joined:
Jul 30, 2015
Messages:
591
512
Trophy Points:
92
Location:
Greater Milwaukee USA
In my opinion, no. First, there are a lot of arguments against the use of Cpk on anything. Second, Cpk is a calculation based on the mean, standard deviation and specifications. The first two of these (mean and standard deviation) have an associated uncertainty about their values (i.e., confidence intervals). Combining two uncertainties into one equation increases your uncertainty in the result. Now, take modulus with four variables, each of which has two uncertain parameters. You now have eight uncertain values. The confidence limits around the resulting Cpk would be very wide. I recommend trying a Monte Carlo simulation to demonstrate that this is true.

Bev D and Andy Nichols like this.
3. ### Bev DModeratorStaff Member

Joined:
Jul 30, 2015
Messages:
613
672
Trophy Points:
92
Location:
Maine
To what end? What do you think it will tell you?

Andy Nichols likes this.
4. ### laffranchinisMember

Joined:
Jan 30, 2020
Messages:
12
2
Trophy Points:
2
Hi Miner,

Thanks for the explanation.
@ Bev: from time to time our wind customers want that we calculate Cpk based on mech. properties. I already tried to explain them that it make no sense. BTW Miners explanation seems to be more professional, so next time I will try to explain all the story with confidence intervals.

Stay healthy
Samuele

5. ### MinerModeratorStaff Member

Joined:
Jul 30, 2015
Messages:
591
512
Trophy Points:
92
Location:
Greater Milwaukee USA
If you can provide the equation, means and variances, I can run a Monte Carlo simulation to show the resulting confidence intervals.

6. ### laffranchinisMember

Joined:
Jan 30, 2020
Messages:
12
2
Trophy Points:
2
Hi Miner,

You wrote that, for example, the moduls has 4 variables. I guess that you are referring, for example to the following formula, right?
G (shear modulus, following ISO 1922) = h (=specimen thickness) * Teta (= slope of linear portion of the forcer-displacement diagram) / l (=specimen length) * b (=specimen width)

thanks
Samuele

File size:
35.6 KB
Views:
4
File size:
35.6 KB
Views:
3
7. ### MinerModeratorStaff Member

Joined:
Jul 30, 2015
Messages:
591
512
Trophy Points:
92
Location:
Greater Milwaukee USA
It's been a while since you last posted. You are correct.

8. ### laffranchinisMember

Joined:
Jan 30, 2020
Messages:
12
2
Trophy Points:
2
yes. its been a while, since form time to time is easier to convince customers not to run Cpk studies for shear strength and moduli, but with new qualifications we have to start discussing again with new people.....
in any case, thank for the confirmation!

Samuele