Informational SPC in Batch Processing

[Paweln86]

Dear All,
I'm facing problems with establishing SPC in my production processes.
The production process has few sub-processes. Firstly one machine is feed with steel and make around 300 kg of sub-product. Then it goes to another machine that make some grinding, then it goes to furnace for heat treatment. Then few other machines that do different types of grinding. Batch created on the beginning of the process will go through it and finish as one batch also. Parts are small therefore in 300 kg I can have up to 400 000 parts.
Let's consider heat treatment. I take 5 samples of every batch and measure hardness with Rockwell hardness tester.
I tried Average and Range Control Chart but I have a lot of points out of CLs. This is because I have very small variation withing subgroups (short-term variation, part-to-part within the batch). Small Range -> tight Control Limits.
In my case batch-to-batch variation is significantly greater.
Now I know that this in not apropriate type of control chart. What do you suggest to try next?

 

[Bev D]

What you are looking for is sometimes called a 3D Chart. 3D refers to the 3 characteristics that are plotted when you have batch to batch variation as the largest component of variation. Wheeler calls is it a 3 way chart: Wheeler, Donald, “The Three-Way Chart”, Quality Digest, March 2017. There are other references for this chart (Innovative Control Charting,Wise and Fair is a great book if you get a copy).
Essentially you plot the subgroup averages on an I, MR chart and then add a range or SD chart for the within subgroup variation. I use it a lot and it is very effective.

 

[Miner]

I had forgotten about these. This would also work for multi-cavity molds/dies.

 

[Welshwizard]

Hi,
Yes indeed, a lovely little tool, more commonly called over in the UK between/within charts. Here is also a good reference and introduction too from Dr. Bill McNeese: www.spcforexcel.com/knowledge/variable-control-charts/xbar-mr-r-betweenwithin-control-chart.

Hope this helps.