nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

Control chart limits set up by phase (not normal distribution)

Arkawa

Hi all,
I would like to know if it is possible to set up control limits by phase once data are not normally distributed, thus having calculated them by quantile estimation and entered them manually. I know that the CL are automatically set up by phase using the normal distribution by default. But I did not succeed either to fix the distribution for control chart in column properties (it only works for capability) or entered it manually per phase (currenlty I use the reference lines but it crosses out the graph and I have to delete the exceeding by paint

). If not, should I transform the data in column by the distribution observed and set up control limits normally ?

Thank you very much for your support !

Best regards
Arkawa.

Accepted answers

tonya_mauldin

Here is a blog post that details how to set historical control limits for each phase.

Creating a phase control chart with historical limits for each phase - JMP User Community

Arkawa

Hi all,

Tonya, your post is amazing ! Very helpful ! However, the approach was simplified with JMP 15.

Generate the I control chart with the phases containing the 3sigma limits within data normally distributed. Go to the hot spot of the control chart and save limits into data table. In this data table we already have the column formula that you shared in your post. You just have to change each LCL and UCL as you have determined before (ex: quantile estimation). You save the file (ex Force limits). When you return on your control chart you go to the hot spot, then get limits and finally select the data table Force limits. Control limits by phase are therefore automatically set-up from the file and western electric process rules are applicable

.

Thank you !

Arkawa

All comments

statman

Not sure I understand completely, but there is no assumption of normally distributed data for control charts to be effective (See Wheeler and Shewhart). https://www.spcpress.com/pdf/DJW088.pdf

Having said that control charts in JMP subgroups can be set by constant size or by Sample Label.

tonya_mauldin

Here is a blog post that details how to set historical control limits for each phase.

Creating a phase control chart with historical limits for each phase - JMP User Community

Arkawa

.

Thank you !

Arkawa

Arkawa

Hi, thats why I proposed the quantile estimation to be applied on the data distribution observed (after data cleaning).
Thanks !
Arkawa

tonya_mauldin

Yes, this blog post has the following comment "JMP 13.1 and above will create this table for you if you click on the red triangle next to Control Chart Builder and choose Save Limits -> in New Table." The steps that came after that statement were for JMP 13.0 and below.