3
$\begingroup$

I have created a process control chart that describes the fraction of deliveries sent to the ideal destination over time. The deliveries all eventually end up at the proper destination, however I am trying to track the fraction that get to the ideal destination directly (no stops along the way). Here is the R code to reproduce a similar p-chart

library(qcc)
nonconforming <- c(12673,13311,8271,10554,12531,11623,
               8458,7852,5149,5964,4206,8035,10636)
sizes <- c(16114,27604,16956,18896,24138,14725,
       10483,8870,5654,8176,13476,20094,22467)
qcc(nonconforming, sizes=sizes, type="p")

enter image description here

The sample size is the whole population over a fixed period of time.

My question is, should I expand the width of the control limits so that more of the points plot in control? With this process I would have to expand the nsigma to a very large number to get about 3/4 of the points to fall withing the control limits

qcc(nonconforming, sizes=sizes, type="p", nsigmas = 52)

enter image description here

Or is it not even worth adding control limits to the chart until the process comes into control?

$\endgroup$
1
$\begingroup$

The purpose of the control chart is to let you know when you have a process in statistical control. The charts provide you with the Voice of the Process.

With points outside of the control limits to this extent, the process is telling you that it is unstable. This is when you would need to decide if you are going to work to reduce the variation in your process or create a whole new process. The main point of consideration should be related to the process capability—is your process meeting the Voice of the Customer.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.