GR&R of a gauge to check a new feature of a new product

C

Carmine

I'd like to show this issue related to a GR%R of a gauge for a new product.
Let's start like this. MSA AIAG says that I have to take sample of parts that represent the actual or expected range of process variation not the range that represents tolerance of the part.
If my gauge is going to measure a particular feature of a part. I have some different case:

1)This feature is already created by a similar manufacturing line in my plant ( a line that produces a similar product with the same feature), in this case I have historical data where I can take the process variation and, then , I can conduct a GR&R analysis on my gage taking the sample that represents the range of the process variation.

2) I have a very new product with a new feature (I have no Historic data). I get a new manufacturing/assembly line in my plant. So I have a new manufacturing/assembly line that produces a new feature, but not historical data are available because everything is new. The only informations I have are the tolerance range that this feature has on the drawings.
In this case I don't have the process variation, I have the tolerance of the feature.

In addition to this, I can create this features off-line with a special tool. I can create the parts (and the features) needed to conduct a gauge study to verify how capable my gauge is to check the new feature made on the line.
In this case, without historical data and an actual/expected range of process variation, I can take only a sample whose features are within the tolerance range.

Can I have a good and real GR&R by taking a sample of parts that represents the tolerance range?

And If Taking this sample (that represents the tolerance range) at first time is right. Can I repeat the same GR&R Analysis later when I have more data and an actual process variation?

Thanks in advance.

Carmine
 

Marc

Fully vaccinated are you?
Leader
Another quick "Bump". My Thanks in advance to anyone who can help with this one.
 

Jim Wynne

Leader
Admin
I'd like to show this issue related to a GR%R of a gauge for a new product.
Let's start like this. MSA AIAG says that I have to take sample of parts that represent the actual or expected range of process variation not the range that represents tolerance of the part.
If my gauge is going to measure a particular feature of a part. I have some different case:

1)This feature is already created by a similar manufacturing line in my plant ( a line that produces a similar product with the same feature), in this case I have historical data where I can take the process variation and, then , I can conduct a GR&R analysis on my gage taking the sample that represents the range of the process variation.

2) I have a very new product with a new feature (I have no Historic data). I get a new manufacturing/assembly line in my plant. So I have a new manufacturing/assembly line that produces a new feature, but not historical data are available because everything is new. The only informations I have are the tolerance range that this feature has on the drawings.
In this case I don't have the process variation, I have the tolerance of the feature.

In addition to this, I can create this features off-line with a special tool. I can create the parts (and the features) needed to conduct a gauge study to verify how capable my gauge is to check the new feature made on the line.
In this case, without historical data and an actual/expected range of process variation, I can take only a sample whose features are within the tolerance range.

Can I have a good and real GR&R by taking a sample of parts that represents the tolerance range?

And If Taking this sample (that represents the tolerance range) at first time is right. Can I repeat the same GR&R Analysis later when I have more data and an actual process variation?

Thanks in advance.

Carmine

GR&R using percent of tolerance is OK if the gage is used for inspection purposes. For process control you should be using percent of variation. If you don't have a handle on process variation yet, using percent of tolerance is fine for starters. As you suggest, you might want to do another study when the actual process variation is established and controlled.

You might not need to do a study at all if the characteristic in question and the measurement system are reasonably similar to other characteristics and systems that have already been done.
 
Top Bottom