J
jfgunn
I have reviewed a couple of threads on this topic and good responses from MichelleKay and Wesley Richardson. I will explain this issue, my plan, and then ask some questions about how others have handled the same issue....
Currently, we are an ISO 17025 accredited commercial lab for many different things. For most of the items we currently calibrate, uncertainty is not a huge factor. We currently state that measuremetn of uncertainty is not taken into account in Pass/Fail Results. We take the reading we get and compare it to the tolerance. If it is good, it passes, if not it fails.
We recently purchased a P&W LabMaster Universal and made some major lab improvements to properly control temperature and humidity. This gives us the ability to calibrate gage blocks and ring gages (and a few other things).
Since we are now nearing a state of the art limitation where uncertainty can be equal to the tolerance of the gage we are testing, I have begun to look at better defining our policy in regards to Pass/Fail decisions. For guidance, I went to ASME B89.7.3 .1-2001. Based upon this document there are basically three options I have (as well as some variations on these options):
1.) The specification of the UUT is the acceptance zone, ignore uncertainty as long as we meet a 2:1 or 3:1 or 4:1 TUR.
2.) Relaxed Acceptance/Stringent Rejection: The acceptance zone equals tolerance PLUS uncertainty. We only reject something if we know it is bad (and by "know" I mean we are 95% sure as that is the confidence interval of the uncertainty).
3.) Stringent Acceptance/Relaxed Rejection: The acceptance zone equals tolerance MINUS uncertainty. We only accept something if we know it is good. Any customer I have that needs this service should be directed to use the most accurate lab they can find for each individual item.
Number 1 and 3 are OK except for items where the uncertainty is relatively large (ie Gage blocks and Class X, XX, XXX Ring and Plug Gages). I feel like I would be telling my customers that they should replace things when I really do not know if that is the case.
Number 2 is Ok for the gage blocks, rings, and plugs except I may be passing things that might be bad.
Of course, the 4th option would be to report the measurement with its uncertainty and make no statement as to a compliance to a specification. This would be great, but does not provide a customer with value.
My plan is to do the following in all quotes to my customers: Use number 1 above for almost everything except for gage blocks, Class X, XX, XX Plug and Ring Gages where I will use number 2. Any certificate of calibration that indicates a Pass where the Relaxed Acceptance/Stringent Rejection Policy was used will have a remark to that effect.
Does this seems reasonable?
Is is sufficient to bury this decision rule in a quote? Should I obtain a more formal written approval to use the rule for each customer?
Has anyone ran into problems when trying to implement decision rules with customers? I picture a new customer asking me why I am making this so difficult when his current provider has never asked these questions. I am always happy to explain what I am doing (or not doing) and what my competitiion is doing (or not doing).
I have an audit on May 2nd-4th to add all of these items. Already passed Gage Block Proficiency Test and I am well on my way to completing all of the uncertainty budgets. This is one of the last issues I need to address. Any help would be apprecaited.
Joe
Currently, we are an ISO 17025 accredited commercial lab for many different things. For most of the items we currently calibrate, uncertainty is not a huge factor. We currently state that measuremetn of uncertainty is not taken into account in Pass/Fail Results. We take the reading we get and compare it to the tolerance. If it is good, it passes, if not it fails.
We recently purchased a P&W LabMaster Universal and made some major lab improvements to properly control temperature and humidity. This gives us the ability to calibrate gage blocks and ring gages (and a few other things).
Since we are now nearing a state of the art limitation where uncertainty can be equal to the tolerance of the gage we are testing, I have begun to look at better defining our policy in regards to Pass/Fail decisions. For guidance, I went to ASME B89.7.3 .1-2001. Based upon this document there are basically three options I have (as well as some variations on these options):
1.) The specification of the UUT is the acceptance zone, ignore uncertainty as long as we meet a 2:1 or 3:1 or 4:1 TUR.
2.) Relaxed Acceptance/Stringent Rejection: The acceptance zone equals tolerance PLUS uncertainty. We only reject something if we know it is bad (and by "know" I mean we are 95% sure as that is the confidence interval of the uncertainty).
3.) Stringent Acceptance/Relaxed Rejection: The acceptance zone equals tolerance MINUS uncertainty. We only accept something if we know it is good. Any customer I have that needs this service should be directed to use the most accurate lab they can find for each individual item.
Number 1 and 3 are OK except for items where the uncertainty is relatively large (ie Gage blocks and Class X, XX, XXX Ring and Plug Gages). I feel like I would be telling my customers that they should replace things when I really do not know if that is the case.
Number 2 is Ok for the gage blocks, rings, and plugs except I may be passing things that might be bad.
Of course, the 4th option would be to report the measurement with its uncertainty and make no statement as to a compliance to a specification. This would be great, but does not provide a customer with value.
My plan is to do the following in all quotes to my customers: Use number 1 above for almost everything except for gage blocks, Class X, XX, XX Plug and Ring Gages where I will use number 2. Any certificate of calibration that indicates a Pass where the Relaxed Acceptance/Stringent Rejection Policy was used will have a remark to that effect.
Does this seems reasonable?
Is is sufficient to bury this decision rule in a quote? Should I obtain a more formal written approval to use the rule for each customer?
Has anyone ran into problems when trying to implement decision rules with customers? I picture a new customer asking me why I am making this so difficult when his current provider has never asked these questions. I am always happy to explain what I am doing (or not doing) and what my competitiion is doing (or not doing).
I have an audit on May 2nd-4th to add all of these items. Already passed Gage Block Proficiency Test and I am well on my way to completing all of the uncertainty budgets. This is one of the last issues I need to address. Any help would be apprecaited.
Joe
