How to Consider control limits/Standard Deviation for Quarterly Analysis?
October 11, 2021 at 10:08 am #255419
pradeep95Participant@pradeep95 Include @pradeep95 in your post and this person will
be notified via email.
My Client Process: My Client manufacture Turbine Blades all round the year.
They send me the CMM inspected data to me end of every quarter of blades cut in that particular quarter. My job is to load the data to Minitab and create a capability Six Pack report. Then Look at the Control chart and determine if the process is In-Control or not.
There is no change in the Manufacturing Process. It is the same which has been continuing for years.
This means in an year I get four such Six Pack reports/graphs for every feature.
My Observations: Every time when I generate the Six Pack report I get different standard deviations, and hence different UCL & LCL. See attachment.
Q-1 all data is within control limits hence marked Process In-Control, Q2 I have got some outliers hence marked the process is not In-control. Q3 again In-Control, Q4 not.
Now on a summary, Is the Process In-Control Or Not?
If we want to track if our process lies with 3stds (i.e. with control limits) Is that good to use different limits every time we check?
Isn’t that we have to follow the process as below?
Initially when blades Manufacturing is started on the Plant, run few batches determine all the assignable causes eliminate them. Then once for all freeze the process. determine the Standard deviation at that stage (which would be as minimum as it can).
From then, every quarter, use that Std Dev as input Std Dev in your control along with new data. So ur control limits would be same every time and then check if you have any outliers
Please correct me on this if I am wrong at any location.
#Controlcharts #Minitab0October 15, 2021 at 8:55 pm #255454
StrayerParticipant@Straydog Include @Straydog in your post and this person will
be notified via email.
A control chart is basically a run chart that identifies variations that should demand attention. It’s a poor tool for summary reporting as you describe. And yes, since standard deviation, UCL and LCL are calculated from the raw data they will change over time unless the process is remarkably consistent. Perhaps one of the statisticians who contribute to these discussions can suggest a better way to create these quarterly reports.1
You must be logged in to reply to this topic.