Comparing Two Measurement Systems in terms of Accuracy and Capability

[tahirawan11]

Hi all,

I am interested in compairing two measurement systems and decide which is better in terms of accuracy and capability. The characteristics in question is the 'Mix ratio' of a compound. The mix ratio is expressed in terms of 'percentage value'. The compound is made up of two 'Ingredients'. The first 'measurement system A' cannot measure the mix ratio directly; instead four different 'mechanical properties' are calculated which are then compared with a standard table which is used to predict the 'mix ratio'.

The second 'measurement system B' is capable of making direct measurement and gives direct 'mix ratio' figures. In order to evaluate which measurement system is better, i have performed a Gauge R&R study. And on the basis of the results; 'measurement system B' has less Gague R&R value (0.9 %) compare to 'measurement system A' (10%).

My question is that' is it enough to prove that 'measurement system B' is better than 'measurement system A'. Do i need to perform a 'Producer risk' and 'Consumer risks' analysis as well and if so how can i do that. Are there any other methods / techniques beside Gauge R&R which can be used to compare two different measurement systems.

thanks

 

[bobdoering]

I think you have a situation that hold several different complications. Before doing a gage R&R, you need to determine if the gaging system is "calibrated", that is how well does each system correlate to known accurate measures. In order to do that, you would need to have a methodology that can verify the mixture more accurately than the systems you are evaluating. You may not have to go any further than that evaluation (as to a Gage R&R study) if the gaging "system" can not accurately determine the mix content. It may be the "deal breaker". Even this may be complicated as it sounds as though the tests are destructive - in that you can not very the same sample several time. You may need to clarify that point.

If you get past this, then you can look to your gage R&R studies. Better gage R&R usually means better gage for the job BUT you still have to evaluate gage costs, stability (Does the technique have any time-function variation? Gage R&R does NOT have a time function!) Better yet, what gage R&R shows is that "every gage is bad, but the one you are evaluating is 'good enough'."

If it meets the conceptual criteria of calibration and gage R&R, if is probably overkill to go to consumer and producer risk.

 

[Bev D]

Bob is correct that simply knowing that one system is mroe repeatable than the other is typically not sufficient. You may be more precisely wrong with system B.

Method Comparison is one way to determine if there is any bias between the two methods and how large that bias may be. This approach only tells you how well the two systems agree. You amy also need to know how accurately they reflect truth as well.

I have posted a word document ( “MEASUREMENT SYSTEMS VERIFICATION AND VALIDATION”) describing the statistical approaches for your situation.


http://elsmar.com/Forums/attachment.php?attachmentid=12094&d=1266255721


If you search for "MEASUREMENT SYSTEMS VERIFICATION AND VALIDATION" you will aslo find many similar posts that might help you

 

[tahirawan11]

Thanks for the informative replies, yes calibration of both the measurement systems have been performed before doing the Gauge R&R study and the results showed that both the systems are performing within the limits. But I am still interested in compariring the two measuerment systems by Gauge R&R study. The reason is that the 'Mix ratio' is a very critical parameter of the process and it must be kept within specs, so a high measurement accuracy needs to be proved.

Since its a destructive test and I dont have any other superior test which can determine the 'true mix ratio'. Therefore the samples for the gauge R&R study have been 'hand mixed' on the basis of 'weight measurement' and comprisise of full range expected within production instead of taking samples from production. The 'bias' will be calculated on the basis of difference from the observed value and 'hand mixed' ratios.

I have one question what do i need to do to say that my measurement system is capable of Six Sigma performance. Do i need to prove that my gauge R&R is less that certain percentage of my specifications?

 

[Bev D]

I have one question what do i need to do to say that my measurement system is capable of Six Sigma performance. Do i need to prove that my gauge R&R is less that certain percentage of my specifications?

Why do you want to say your system is capabel of six sigma performance? what is the value in that? what does six sigma perfomance mean to you?

 

[tahirawan11]

when i say my measurement system is capable of six sigma performance, i mean to say that the variation from my measurement system is small enough that i can use my current measurement system for process improvement / monitoring purposes such as use it DOE / SPC.

Six sigma performance means that my end product has a consistant quality and it always meets its customer requirements.