control limits
Six Sigma – iSixSigma › Forums › Old Forums › General › control limits
 This topic has 10 replies, 11 voices, and was last updated 14 years, 5 months ago by jberilla.

AuthorPosts

December 10, 2007 at 9:53 am #48884
I have encountered this Problem: A process with very narrow control limits and many points out of limits but when measuring the characteristics of product tolerances seem to be Ok. There is no issue with mistakes at mesuring procedure neither at control limits calculation. How can this be explained;
0December 10, 2007 at 10:20 am #165934
Six Sigma guyMember@SixSigmaguy Include @SixSigmaguy in your post and this person will
be notified via email.process out of control but with a very widen spec limits thats why though ur process is meeting customer expectations but not under control… check if there is any process change? check if the baselines needs to be revisted…probably u could also be clubbing two different processes in same control chart.
0December 10, 2007 at 11:32 am #165935
Harish GoyalParticipant@HarishGoyal Include @HarishGoyal in your post and this person will
be notified via email.i dont think its concerned with specification limits. Yhere may be some calc mistake there in process.
why dont you put the data here so that someone can see and give suggestion.
Regards
Harish0December 10, 2007 at 12:01 pm #165936data from two process can be represneted by bimodla distribution having 2 or more peaks in histogram .
FYI
Regards
Amit0December 10, 2007 at 12:18 pm #165937
Dhananjay HegdeParticipant@DhananjayHegde Include @DhananjayHegde in your post and this person will
be notified via email.Hi.
The process is able to meet the tolerance limits indicates that it is a capable process.But capability and stability are 2 different things.Stability talks about indication of special cause variation and in your case the problem is about stability.It would be great if you can share the raw data.
Dhananjay0December 11, 2007 at 9:55 pm #166016dimi,
Overly narrow control limits, with a large # of OOC data points, is a potential indication of poor subgroup selection and/or autocorrelation in the data. Another potential contributing factor could be the number of data points being used to create the points that are ultimately plotted this would magnify the impact of poorly selected rational subgroups.
Are you using XbarR or XbarS? Have you assessed the process data to understand where the independent variance component lies? Tools that can help provide this insight include: Multivari studies, COV/REML, and GLM analysis.
Regards,Erik0December 11, 2007 at 10:00 pm #166017Control limits are statistically inferred parameters commonly referred to as the “Voice of the Process” the situation you described is quite normal for a process where the drawing specifications are much looser than the capability of the process.
Remember that control have no relationship to specification limits.
Also be sure you are usingthe correct type of control chart.0December 11, 2007 at 11:09 pm #166021
Dr. ScottParticipant@Dr.Scott Include @Dr.Scott in your post and this person will
be notified via email.dimi,
I suspect you are using binomial data with very large sample sizes.
I suggest you follow the advice of Dhananjay at post your data here so we might help.
Happy Holidays,
Dr. Scott0December 17, 2007 at 8:43 am #166188
k.bhadrayyaParticipant@k.bhadrayya Include @k.bhadrayya in your post and this person will
be notified via email.Dear Dimi
Since many of the points on the control chart are out of specs, it shows that some assignable/ external causes are influencing the process . By reading your out put data and corelating with the changes of inputs taken from the log sheet of production must show some trends. Keenly observe the changes in the input data and relate to out put data. Unless the factors affecting this distrubance are detected and teken remidial action , control charts will not help.
k.bhadrayya
0December 17, 2007 at 3:14 pm #166205Without specific data, I say you are suffering from one or two classical situations, overcontrol and / or measuring nearly identical parts.
Overcontrol: one reading, one adjustment (lots of points are out of control, 50% above the control limit and 50% below). Operators are trying too hard to control the process based on one reading. Try this, never make a process adjustment based on inidividual readings. Only, adjust if the subgroup average is out of control. If you are in an overcontrol situation, statistically, you do not know what your process looks like. You need to instruct your folks to not make any adjustments for 2 days. Let the process define itself.
Measuring nearly identical parts. Once set up, in the short run, the process will produce nearly identical parts, R is nearly zero, hence the control limits will be very narrow. If this happens, you will throw your process into an overcontrol condition. If your process is producing nearly identical parts, then you do not need to monitor your process. You need to monitor your set up, nominal is best.
0December 17, 2007 at 4:32 pm #166212
jberillaParticipant@jberilla Include @jberilla in your post and this person will
be notified via email.You can also consider gage resolution. When you have a limited number of observational values (usually considered <6) displayed in your R or S chart, your subgroup averages will will uncharacteristically small, resulting in tight CL on your mean chart.
Look to the R/S chart and determine if you have 6 or more different values represented. If not, consider:Changing the gage (if not too expensive) if possible, if not
Using larger sample sizes, if possible, if not
Gather the R or S chart subgroup averages over a longer time frame0 
AuthorPosts
The forum ‘General’ is closed to new topics and replies.