I want to perform a Gage R & R on weld strength. We are using a force gage to break two welded pieces apart, and we're recording the push off force. In Minitab one is supposed to use the "Nested" Gage R & R analysis tool. Instead of measuring the same 10 parts twice (impossible in a destructive test), one is supposed to measure two parts from each of 10 different batches or manufacturing lots of parts. There's an assumption that two parts from the same batch should be nearly identical. I don't believe that is a safe assumption for our welding process. Does anyone have any other approaches that have worked?
Sign In to get started!
mehryarsalim 8/27/2004
R&R is not just for operator error .... remember the other part of the error due to the guage itself ..... you can do an R&R study using Minitab without considering multiple operators .... you can use the same person/operator for all samples to reduce variability and diminish the effects of operator error ... but in actual practice you will be missing some error due to loading of the part by the automated mechanism .... a true R&R considers all the factors that can have an influence on the variability of the measurement system ...
mkomarmy 8/24/2004
"Gauge R&R" is normally used to evaluate the variability of a gauge when measuring the same parts multiple times (repeatability), and also with multiple operators to tell you if the operators have an influence (reproducibility). In the case of auto gauges, operators may not have much of an influence. However, you can use the AIAG long form range method (10X3X3) or ANOVA to evaluate other sources of variation. Instead of using 3 operators, use 3 pallets (if the gauge is loaded from pallets). The reproducibility value will tell you if there is an interaction between the pallets and the gauge. If there are no pallets, you can calibrate between trials to see if the calibration process contributes to the variability of the gauge.
As for the data gathering part, most auto gauges have the ability to output data. If a csv file is available, you can use excel and/or minitab to evaluate your results. If csv files are not available, perhaps it has the ability to print results via an rs-232 port. If this is the case, the results can be then entered manually into minitab or excel after the study is run.
I agree that manual loading could be misleading. However, if there is no alternative to manually recording data, then the best option would be to run the gauge in single cycle or manual mode, but by loading and gauging in the same manner as in automatic.
Mitch