Winter 2009

A Quarterly Newsletter of the Instrumentation Testing Association

ITA Enews

Spring 2011

Instrumentation Testing Association  (ITA)

 

Toll Free Inside the U.S. and CANADA 1-877-236-1256

Global Phone: 702-568-1445

Web: www.instrument.org

Email ITA

TOTAL MEASUREMENT UNCERTAINTY

Usually, when an uncertainty analysis is performed, the basis of the assessment is the calibration of the meter and its installation in accordance with suppliers’ recommendations. It is therefore assumed (erroneously) that the original laboratory calibration is transferred directly into the users’ installation.   It is clear from very many reported cases that the local pipe work does have a different influence to that used for the original calibration.  It also cannot be assumed that just because the Reynolds number is high that the velocity profile is fully developed and swirl free. 

Figure 4 shows an uncertainty analysis undertaken on a fiscal orifice plate in a South American oil refinery.  This example is chosen because the oil industry requirements are more stringent that for the water sector.  It was assumed that because all the standards were complied with that the measurement error was within +/-1%.   

The analysis carried out shows this was far from the case.  Figure 4 shows that velocity profile and installation errors account for more than 70% of the total uncertainty of +/- 4.89%. It also shows that dimensional effects, fluid temperature, pressure and other factors also have relatively minor influences. This confirms the methodology that most people use gives a result that is within acceptable limits.  It is because the result is low that it is accepted. A final result of +/-4.89% is usually dismissed.  Verifications of old installations in the UK by RPS show that even this figure is low in some extreme cases!   

ITA Enews is a newsletter of the Instrumentation Testing Association
A non-profit technical and educational association dedicated to promoting reliability and performance of environmental instrumentation and automation technologies. Learn More About Becoming A Member

The various influences for the estimation of total uncertainty can be grouped as follows:

1  Calibration facilities

2  Meter factors

3  Installation factors

4  Operational factors

5  Time dependent factors

6  Data collection and transmission factors

Figure 4: Total Uncertainty analyses for fiscal

flowmeter using fluid dynamic principles

Within each of these are numerous influences some large and some small but all have to be estimated for the total metering uncertainty to be calculated.   Usually items under 3, 5 and 6 are assumed to be negligible and neglected.  Recent RPS work suggests this is not correct and indeed these areas contribute the greatest amounts to the total uncertainty.

 

Items 1 and 2 are included when the meter is supplied (usually better than 0.5%),  installation effects are assumed zero (because suppliers’ recommendations are met), item 4 is zero because the meter is sized correctly and operated within the rated range,  item 6 is checked when the installation is first commissioned and item 5 is not even considered. This leads to the commonly mistaken belief that the overall uncertainty is therefore better than 1%.  Further, because meters read within 2% of each other it is assumed the whole system balances.    This is a dangerous assumption as discussed in the following section.