Correlation of Leak Test Equipment
September 30, 2005 at 2:05 pm #40872
I am currently doing a correlation study between Two Leak Testers. One is located in the US the other in the UK at our sister Diecasting facillity. The purpose of the this project is to try to detect porosity prior to shipping castings across the big pond.
The way I have chosen to attack this issue is to separate 32 samples by location of the porosity and the leak rate that is close to the 5scc/m leak rate threshold (USL). I am also sorting these by Standard Deviation**2 since I have tested each casting 5 times, to help eliminate any other noise factors and ensure these castings are indeed leakers from porosity and not an assembly seal leak.Theory is porosity will be consistant leaker when tested.
My question is would I be better off looking at the coefficient of variances or will standard deviation**2(individual variance) give me what I need in terms of eliminating frames that are inconsistant? Or are these the same? Slightly confused.
CT0September 30, 2005 at 4:57 pm #127737
Because you are using an absolute threshold 5scc/m, I would think you should want to use Standard Deviation and not CoV.You will lose perspective of how far apart the measurements for a given casting are if you use CoV. On the other hand, if you are deciding to pass/fail a casting based on the relative spread with respect to the mean, without any consideration for the threshold, you can use CoV.
I would use standard deviation. Hope this helps.0September 30, 2005 at 6:18 pm #127742
Thanks AB, Thats exactly what I thought, but our QC manager whom is also MBB suggested the CoV, After going back and doing some more reading which is what I should have done to begin with. I fully understand why now; guess the guy was just trying to abuse me.
The forum ‘General’ is closed to new topics and replies.