Great point about you decide how long it may take to sample enough to get all variables! I imagine some machines need maintenance more frequently for very precise parts, etc.
Thanks for the simple and effective explanations. What about the sample qty and subgroup sizes? Let's assume that my hourly production is 1000 pcs, and 8 hours is enough to see all deviations that might come from the machine, process parameters, material lot change, etc. What should be the # of subgroup (4-5?) and sampling frequency (30-25), total sampling quantity? What should I consider; 8 hours, cycle time or total sample qty that I want to reach?
4 or 5 doesn't matter really if the 8 hours shows all the variation then I would divide it by 50 even slots and take a sub-group every 1/50th of the shift...in theory the samples should be collected randomly, but I don't think that would matter...
Hello Paul, Thanks for you video, always very interesting! I have a question if a made a Cmk calculation on 30 consécutives parts and find a good Cmk. After that I will calculate my cpk on a longer period of time to see if my variables have influence on my process, imagine that I find a bad Cpk, what is the next things I can do to find the variables that have the most impact on the variability of my process? Thanks!
Great point about you decide how long it may take to sample enough to get all variables! I imagine some machines need maintenance more frequently for very precise parts, etc.
Usually yes - but it's really about investigation, identifying for your machines what's needed...
Thanks for the simple and effective explanations. What about the sample qty and subgroup sizes?
Let's assume that my hourly production is 1000 pcs, and 8 hours is enough to see all deviations that might come from the machine, process parameters, material lot change, etc.
What should be the # of subgroup (4-5?) and sampling frequency (30-25), total sampling quantity?
What should I consider; 8 hours, cycle time or total sample qty that I want to reach?
4 or 5 doesn't matter really if the 8 hours shows all the variation then I would divide it by 50 even slots and take a sub-group every 1/50th of the shift...in theory the samples should be collected randomly, but I don't think that would matter...
Hello Paul,
Thanks for you video, always very interesting!
I have a question if a made a Cmk calculation on 30 consécutives parts and find a good Cmk. After that I will calculate my cpk on a longer period of time to see if my variables have influence on my process, imagine that I find a bad Cpk, what is the next things I can do to find the variables that have the most impact on the variability of my process?
Thanks!
Laurent good to hear from you. you need to conduct an investigation of common cause variability using cause and effect diagram..
ua-cam.com/video/T4qWy6aFWSo/v-deo.html watch this video....
where is the link?
oops! Here it is - 7 QUALITY TOOLS ONLINE COURSE - Click the link
paul-allen.mykajabi.com/7-quality-tools-1-day-offer-1