Gage R&R before all new production - Frequent Production Changes

Hello, have a problem !

If I well understood; a R&R analysis must be realized for a measurement device to know, if this device is able to control products of a specific process (machine). The analysis must be realized before the beginning of the production.

But in my company, production change often (new products dimensions) and we have more than 7000 different devices.

So, how can I do ?

I can't do a R&R every time we begin a new production.

Re: R&R before all new production - Frequent Production Changes

Mr.Cigi said:

Hello, have a problem !

If I well understood; a R&R analysis must be realized for a measurement device to know, if this device is able to control products of a specific process (machine). The analysis must be realized before the beginning of the production.

But in my company, production change often (new products dimensions) and we have more than 7000 different devices.

So, how can I do ?

I can't do a R&R every time we begin a new production.

Click to expand...

Welcome to the Cove.

You make a very good point. First, keep it simple. Unless there is a specific requirement, there is no need to do Gage R&R on all your devices.

You will need to do an R&R on the measurement system to determine the suitability of the gage but then consider the "family" of devices being used. For example if in the 7000 devices you have 1000 calipers, there would be no need to do an R&R on each gage. Be sure they are in calibration and group all calipers in a "family" of gages. Your MSA can be applied to all calipers making like measurements on multiple products. Once gage suitability is determined, a simple check on the calibration status of the gage before each production run should ensure the intended measurement for control.

Dave

Re: R&R before all new production - Frequent Production Changes

D.Scott's advice is correct. I would like to emphasize the "like measurements" aspect.

"Like measurements" means measuring similar features such as diameters. It does not mean that a gage qualified on diameter can be extended to measuing step height without an MSA.

In addition, the similarity of like features must extend to the denominator for the metric. This means that if you are using the %GRR (% Study Variation), the process variation must be the same. If you are using P/T Ratio (% Tolerance), the tolerance must be the same.

There are workarounds to redoing the complete MSA. If the features are the same, but the tolerance is different (or the process variation), you can still use the actual measurement variation from the original MSA and calculate the new %GRR or P/T Ratio using the new denominators.

Develop the "like measurement"

For example, in my company:

I have 5 production lines; on each line, a gage measuring the product diameter is attributed.
But production lines make portions of different types of pieces (there are never the same types on the 5 lines). All types have different sizes. (different tolerances, different variations of process)

Can I group gages in "family" ?

Re: R&R before all new production - Frequent Production Changes

This will require a more complicated approach. If I understand you correctly, you are measuring diameters, but the diameters have different tolerances and process variation.

This does not directly allow you to group these into families, but an alternative approach comes to mind.

I suggest performing MSAs on intervals that cover the entire range of tolerances and process variation. The results could then be modeled using linear or nonlinear regression. If you obtain an adequate model, you should have two sets of lines/curves that show the gage performance for each tolerance and process variation. That would allow you to group the gages into a family and still answer questions pertaining to the MSA for any tolerance or variation.

Re: R&R before all new production - Frequent Production Changes

Hi,

My colleague want to do a R&R in the following way:

For example, we have 10 production lines (10 differents products with differents size). On the 10 lines, there are the same type of control gage.

So, he want do one R&R with 10 parts representative of the 10 production lines (without any tolerance); 3 operators who measure each part three times.

And than, look at the GR&R/TV result to conclude on the capability of the gage.

Is it a good technique ?

In my opinion, I think NO.

Re: Develop the "like measurement"

Mr.Cigi said:

For example, in my company:

I have 5 production lines; on each line, a gage measuring the product diameter is attributed.
But production lines make portions of different types of pieces (there are never the same types on the 5 lines). All types have different sizes. (different tolerances, different variations of process)

Can I group gages in "family" ?

Click to expand...

If I apply Miner's technique. For my example:

Imagine that process variation of the 5 production lines are same.
I should do R&R studies with representative tolerances measured by the gage.

But, How many R&R should I do ?
The first R&R with the smallest tolerance, The second with the highest tolerance and anothers with intermediate tolerances.
Then I can modeled %GRR/TV or/and %GRR/Tol using linear regression.
The two sets of lines corresponding to the tolerance and process variation ?

To conclude, I have a graph showing the gage performance between the minimal and maximal tolerances

After that, Can I group the same gages into a family ?

If you have evidence that the process variation is the same for each line, Run 3 studies at the largest, smallest and mid-range tolerance ranges. This will allow you to draw the line/curve for this specific process variation.

If you find that the process variation does vary, repeat this for the largest and smallest process variation. This will give you 3 lines/curves that should allow you to interpolate for any tolerance/variation combination.