Andrews
12th March 2002, 07:00 AM
We are a company manufacturing precision turned parts using multi-spindle automatic machines like Wickman , Davenport etc.. While Xbar-R charting on these machines , we found that more than 96% of the points lie close to the average line (Xbar).Now we know that if more than 2/3 rd of the points lie within 1/3 rd of the control limits, then it means the process is unstable.On referring the AIAG Statistical Process Control manual we found that this might be because each subgroup contains measurements from more than one stream i.e in our case more than one spindle and it advises us to chart each spindle separately. My doubt is will the method of charting each spindle separately meet the requirements of customer because I feel they may be more concerned in knowing the process control and process capability of the process as a whole ( all spindles together) rather than each spindle separately.One more point to be noted is that if the customer during receiving inspection takes 100-150 nos. and tries to check the process capability of the lot as a whole then it will be always lesser than the process capability of each spindle separately ( that we calculate). Can anyone suggest a way to check the stability of the process in such cases without being misleading and also wihout increasing the load on the operator (by adding more charts). Any publication or article that discusses this point in more detail will help
