Individual or Moving Range Chart? Parts run in lots of 7 with 4 key wall thicknesses

[MikeG]

We are currently using individual moving range charts, we run these parts in lots of 7 and have 4 key wall thicknesses that we are required to do SPC on.
What I want to do is change to an average moving range chart subgroups of 7 and chart the range for each run. Lots are ran back to back on a continual basis. My goal is to show that this is a capable process so I can have objective evidence to support a reduction in 100% CMM inspection of these parts.

Thanks in advance.
Mike

 

[Miner]

We are currently using individual moving range charts, we run these parts in lots of 7 and have 4 key wall thicknesses that we are required to do SPC on.
What I want to do is change to an average moving range chart subgroups of 7 and chart the range for each run. Lots are ran back to back on a continual basis. My goal is to show that this is a capable process so I can have objective evidence to support a reduction in 100% CMM inspection of these parts.

Thanks in advance.
Mike

What is the basis for the lot size of 7? Is it arbitrary, a standard pack quantity, the number a cavities in a mold, etc.? What type of process is it?

The type of control chart and the sampling scheme must be rationally based on the process and the types of expected variation from the process.

 

[MikeG]

What is the basis for the lot size of 7? Is it arbitrary, a standard pack quantity, the number a cavities in a mold, etc.? What type of process is it?

The type of control chart and the sampling scheme must be rationally based on the process and the types of expected variation from the process.

Its a machining center that produces leading edge ribs for commercial aircraft. The lot size is controlled by the build size which requires 7 pieces. Our customers weekly order is grouped in lots of 7.

 

[Miner]

We are currently using individual moving range charts, we run these parts in lots of 7 and have 4 key wall thicknesses that we are required to do SPC on.
What I want to do is change to an average moving range chart subgroups of 7 and chart the range for each run. Lots are ran back to back on a continual basis. My goal is to show that this is a capable process so I can have objective evidence to support a reduction in 100% CMM inspection of these parts.

Thanks in advance.
Mike

Okay, based on your reply, my understanding of the process is that you machine one part at a time. You are grouping them in lots of seven based on customer demand, but this does not necessarily mean that you shut down after seven. You may immediately start into another group of seven, then another.

If this is correct, an XBar/R chart is more appropriate than an IMR chart. However, the sample size should be determined by something other than the somewhat arbitrary grouping of seven based on customer order patterns. The grouping should be determined by the process itself.

Ideally, the subgroup should be just large enough to include random/common sources of variation within the subgroup. You can usually achieve this with less than seven samples. Variation that occurs over time such as tool wear should show up between subgroups.

Next, you need to establish a time interval between subgroups. You do not want to measure 100% of the parts. This should be based on significant events in the process such as changes in lots of material, but may be time based to detect tool wear.

In your situation, you want to collect enough data to demonstrate a state of control, then use the data to calculate a Pp/Ppk that shows capability.

 

[Caster]

We are currently using individual moving range charts, we run these parts in lots of 7 and have 4 key wall thicknesses that we are required to do SPC on.
What I want to do is change to an average moving range chart subgroups of 7 and chart the range for each run. Lots are ran back to back on a continual basis. My goal is to show that this is a capable process so I can have objective evidence to support a reduction in 100% CMM inspection of these parts.

Thanks in advance.
Mike

Mike

Your goal is to show capability to reduce inspection.

How much data do you have from you current IMR charts? If you have 25 or so groups, you should be able to do a cap study with this data?

Capability studies can be somewhat independent of chart type. A bad capability result will be bad no matter if you have IMR or XbarR charts. There will be a slight difference in the number, but the information will be the same.

If you have some SPC software, you can use your data set to try every type of chart you can imagine to see which one gives you the best information.

To me the idea behind rational subgrouping is to understand how the part comes to you and what you might want to do as a result of the measurements.

A screw machine making thousands of parts an hour lends itself to taking a rational subgroup of a handful of parts, XbarR.

You seem to have a large complex machining job. You likely want to make any needed changes before starting the next, so IMR makes sense.

 

[MikeG]

Thanks for the input

Sounds like we are on the right track, thanks for the input