Calibration Determination decision

[Jen Kirley]

The OEM doesn’t have anymore idea on how frequently a device is used - they often recommend 12 months! You have to experiment and get data. A screw plug gauge used to check threads in aluminium will wear less than one used on stainless steel. Yes, it’s work, and you may not want to do what is necessary to generate the data on which to base a decision. However, this talk of time based recalls is only OK when you want to throw money at calibration.

Not only does the OEM not know how often the instrument is used, the OEM does not know the environment in which the instrument is stored and used. If we are to question their recommended period (for example, one year) we should be ready to say why we deviate from it.

To be clear, I am completely aligned with the suggestion that usage frequency should be a factor in determining calibration frequency. However, a gage that sits on a shelf with zero usage can become unreliable as is the gage that is used every day. So the sum of considerations should be applied to decide the calibration frequency.

 

[Ron Rompen]

Funboi, i agree with your idea of calibration based on usage - but (for most of us) there are just too many gauges out there to track usage on. Easier (in some ways) to just schedule calibration based on time intervals. This is (or can be) based on associated costs/down time for instruments that have to be sent out for a 3rd party to calibrate.

 

[dwperron]

In the real world you speak of, at least in my real world, we calibrate or verify it in the first place to ensure we are starting off with a good tool and to comply with customer/standard demands.

There are a few items that are cheaper to replace than calibrate, so we consider the risk: how likely is it that one of the pins goes out of tolerance and if it does what is the likely impact? If the risk is low, we replace vs recalibrate.

So again my question is "Why even bother calibrating in the first place if it doesn't matter if the tool goes out of tolerance?"

 

[John Predmore]

i agree with your idea of calibration based on usage - but (for most of us) there are just too many gauges out there to track usage on

I worked in one machine shop that kept less-frequently-used gages in a central gage crib. When I checked out one ring gage to measure one part, the attendant had me sign my name, department and phone #. I explained that I was going to use the gage right there, I didn't need to take the gage back to my workstation, I would return it as soon as I was finished. He still wanted my entry on the log. He said the usage record was entered into the computer, and cumulative days used was how they scheduled re-calibration, instead of a fixed number of months.

I was fascinated by this alternative, and I asked questions. It didn't matter whether I used the gage for 5 minutes or 5 hours, it was recorded as one day of use in the tracking log, they only tracked # days, not fractions of a day, for scheduling purposes. It didn't matter if I gaged 1 part in 1 day or 500, they had decided to use # days as their basis for this gage. Not every gage was treated on this basis, only certain less-frequently used gages. He showed me the computer record for that particular ring gage. It had been used 4 times in 6 years, so I could see why it was an economic decision not to recalibrate every 12 months. I don't remember any longer how many cumulative days use triggered a re-cal. He said they would re-calibrate it after 10 years regardless, just for good measure (I think that was a pun!). While this calibration protocol was different from anything I had previously encountered, the protocol was documented, the rationale was based on gage history and sound science, and presumably the protocol worked for them. The point of the story is there are valid alternatives to fixed timeframes as a basis.

 

[Jen Kirley]

Then there is always "calibrate before use" for seldom-used gages.

 

[Funboi]

It shouldn’t be too difficult to determine uses based on knowledge of the number of parts being made and things like the process control methods used. If low quantities, operators commonly gauge 100% or with higher volumes, to a control plan (spc sample or something less sophisticated) and that should inform the number of times its used. Not much of a task.

 

[John Predmore]

I made this graphic today, and decided to share by posting it here. I found myself re-reading and parsing section 7.1.5 to make sense of what the quality standard says about calibration, trying to explain to others. My company operates in alignment with AS9100, so the grey text shows what is unique to AS9100. Fun with Venn Diagrams is tongue-in-check homage to Sheldon Cooper and his Fun with Flags.

Enjoy!