Leak Test Specialist

Thank you. My problem is we are testing for the attribute of LEAK TIGHTNESS as defined as no hole in the casting larger than the equivalent hydraulic diameter of an 8 micron sharp edge orifice. This is determined by measuring air flow through the leak and rejecting everything above the flow through the master orifice. Quality says that makes it a variable gauge doubling the cost.

If the air flow rate is fixed and constant, it's an attribute (leak, no leak).

If the air flow rate is variable with upper and lower limits of flow, and those flows are recorded, charted and analyzed, then it becomes a variable statistic.

If the "measurement" is Go / No Go or Pass/ Fail, its attribute data.

If its quantifiable by real numbers it is variable data. here's a definition, SImilar to the ones in most statistics courses and AIAG.

http://www.sixsigmaspc.com/dictionary/discreteattributedata.html

Short answer, if you record a number like 10.8 cc of air flow per minute (or whatever, then it is variables.

However if you have a cutoff value, and you are recording as strictly that result (Above 10.8 = FAIL) or (Below 10.8 =Pass), then it is attribute data. If you record a discrete number for each test, 10.9, 10.2, 9.8, 12, 11.4, etc. its variables data.

Attribute data is qualitative, (IS/IS NOT; PASS/ FAIL; Present/ Not Present) not quantitative (1.2555; 10.82; 78.9 etc.)

Thats about as clear as I can make it.

milo

Establish the limits of acceptability. Since its an attribute that you are establishing (reject if leaking vs accept if not leaking) your limits are the variables for the attribute. If you determine that a casting can be accepted if leaking between 1 to 2 lpm then that is a GO! If the leak is above that which was determined; then that is No-Go!