M
I need some advice on developing an effective sampling plan on a centerless grinding process of shaft journals. The process automatically feeds and grinds several ODs on the shaft. After a set amount of grind cycles, the machine automatically dress the grinding wheel and continues. There is no integrated size control in the machine (in-process gauging). The shafts are sampled on an external gage and the machine is comensated based on the average of the subgroups if required.
The current sampling is to measure the 1st 3 parts after the dress and calculate a subgroup, measure 3 parts in the middle of the cycle and calculate a subgroup, then measure 3 right before the dress cycle. As you could see we tend to have a saw-tooth shaped control charts with the ranges being fairly tight. When attempting to recalculate the exsiting control limits based on the process data, the lines become tighter than the avg of the start & end of cycle subgroups, thus showing out of control.
No being an expert of any sort, my first impression is that the sampling plan is not approprate for this process and sample ranges don't show true variablility of the process spread... and that a better sample would include parts throughout the whole dress cycle and not neccesarily the 'consecutive' samples.
Any advice or information would be helpful...
The current sampling is to measure the 1st 3 parts after the dress and calculate a subgroup, measure 3 parts in the middle of the cycle and calculate a subgroup, then measure 3 right before the dress cycle. As you could see we tend to have a saw-tooth shaped control charts with the ranges being fairly tight. When attempting to recalculate the exsiting control limits based on the process data, the lines become tighter than the avg of the start & end of cycle subgroups, thus showing out of control.
No being an expert of any sort, my first impression is that the sampling plan is not approprate for this process and sample ranges don't show true variablility of the process spread... and that a better sample would include parts throughout the whole dress cycle and not neccesarily the 'consecutive' samples.
Any advice or information would be helpful...