Tolerance of Transmitters and Gauges

Ahem, that's the same problem I faced in 1994 (and still do).

While working for our ISO9000 accreditation, I asked the guys in production for their process tolerances. Two answers were all I got:

"As accurate as you can make it."
"Zero error."

The first answer is as vague as you can get. The second answer is worse...it's impossible to achieve.

Speaking with the corporate automation head, he said that the production tolerances are supposed to be in the factory manager's office, only available on request by certain persons and for eyes only (not to be taken out of the vault).

What we did was to take the accuracy figures stated by the manufacturer and multiply them by four. For example, if your instrument has an accuracy of 0.25%, we set the calibration tolerance as 1%. Worked for some loops but not all. We had difficulty factoring installation and environmental effects (especially true for weighscales)

Several years later, we're still groping in the dark.

If your production people can tell you how deviation in psi/bars/MPa of steam they can tolerate, you can specify a slightly tighter tolerance for your calibration. If they can't tell you, sit down with them and come to an agreement, on paper.

regards,

Vulcan

Why do not we take the basics of the measurement ?

What we need to measure and what is our required tolerance (for the system)

Say we measure flow and it shoulf be not less than 100 lpm and not more than 120 lpm then our value is 110≠ 10,

The gauge should have tolerance not more than 1/10 of this ie ≠ 1 LPM

Now convert it to whatever way you want - FSD or otherwise.

What you need is first to know the process tolerances. You cannot define the uncertainty of a measurement without knowing what it is used for.

From a probability point of view you should define the risk limit for your measurement and after it define with which reference (uncertainty of the reference) you will try to calibrate it.

There is many times a misunderstanding between tolerances and uncertainty. The tolerance is prescribed as the limits within the result of the process should be found with certain level of risk.

The tool used to measure presents an uncertainty which is the consequence of its calibration according to procedures against a reference which also was calibrated and presents as well an uncertainty.

This is the chain of traceability to the basic reference at International Bureau of Standards. This chain is the guaranty that all measurements where ever they are done and with which ever devices are comparable with respect to given results.

Instrument tolerance values should vary with the cricticality of the application in the process, e.g. custody transfer, SIS, monitor & control, indication only. In addition to this, test equipment capability and manufacturer tolerance specs must be taken into consideration.

Are there any standards that can guide in determining these tolerance values?