S
I agree with Bev that a runchart to specification limits is absolutely needed to check for potential out-of-control conditions including runs larger than 6 in a row, stratified groups (hi to low to high), and so forth.
Is there a reason for a 10 pc. sample? Nine would afford 3 sub-groups of 3.
Once you prove a state of control, the sampling frequency may be intuited from the display. I think your earlier list of NCR's, Cpk & Cp (or CR) estimate, etc. is good.
Is there a reason for a 10 pc. sample? Nine would afford 3 sub-groups of 3.
Once you prove a state of control, the sampling frequency may be intuited from the display. I think your earlier list of NCR's, Cpk & Cp (or CR) estimate, etc. is good.