Measurement Uncertainty A Measure of Confidence - National ...
Measurement Uncertainty A Measure of Confidence - National ...
Measurement Uncertainty A Measure of Confidence - National ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
<strong><strong>Measure</strong>ment</strong> <strong>Uncertainty</strong><br />
A <strong>Measure</strong> <strong>of</strong> <strong>Confidence</strong><br />
Chua Sze Wey<br />
Head (Electrical Metrology) &<br />
Principal Metrologist<br />
<strong>National</strong> Metrology Centre
Agenda<br />
• A touch on measurement uncertainty<br />
• Evolution <strong>of</strong> the GUM<br />
• Update on the GUM<br />
& where you can get a copy for FREE !
The only certainty in life is uncertainty<br />
http://m.mfrtech.com/articles/25030.html
International vocabulary <strong>of</strong> metrology — Basic and general concepts<br />
and associated terms (VIM) -2008<br />
measurement uncertainty (uncertainty <strong>of</strong> measurement, uncertainty)<br />
non-negative parameter characterizing the dispersion <strong>of</strong> the quantity<br />
values being attributed to a measurand, based on the information<br />
used<br />
• NOTE 1 <strong><strong>Measure</strong>ment</strong> uncertainty includes components arising from systematic effects, such as<br />
components associated with corrections and the assigned quantity values <strong>of</strong> measurement<br />
standards, as well as the definitional uncertainty. Sometimes estimated systematic effects<br />
are not corrected for but, instead, associated measurement uncertainty components are<br />
incorporated.<br />
• NOTE 2 The parameter may be, for example, a standard deviation called standard<br />
measurement uncertainty (or a specified multiple <strong>of</strong> it), or the half-width <strong>of</strong> an interval, having a<br />
stated coverage probability.<br />
• NOTE 3 <strong><strong>Measure</strong>ment</strong> uncertainty comprises, in general, many components. Some <strong>of</strong> these may<br />
be evaluated by Type A evaluation <strong>of</strong> measurement uncertainty from the statistical distribution<br />
<strong>of</strong> the quantity values from series <strong>of</strong> measurements and can be characterized by standard<br />
deviations. The other components, which may be evaluated by Type B evaluation <strong>of</strong><br />
measurement uncertainty, can also be characterized by standard deviations, evaluated from<br />
probability density functions based on experience or other information.<br />
• NOTE 4 In general, for a given set <strong>of</strong> information, it is understood that the measurement<br />
uncertainty is associated with a stated quantity value attributed to the measurand. A modification<br />
<strong>of</strong> this value results in a modification <strong>of</strong> the associated uncertainty.
<strong><strong>Measure</strong>ment</strong> and uncertainty<br />
• Real measurements are never made under perfect<br />
conditions, flaws in the measurement may be visible or<br />
invisible.<br />
• “uncertainty” means doubt, and thus in its broadest<br />
sense “uncertainty <strong>of</strong> measurement” means doubt about<br />
the validity <strong>of</strong> the result <strong>of</strong> a measurement.<br />
• Every measurement is subject to some uncertainty.<br />
• A measurement result is only complete if it is<br />
accompanied by a statement <strong>of</strong> the uncertainty in the<br />
measurement.
<strong><strong>Measure</strong>ment</strong> and uncertainty<br />
• <strong>Uncertainty</strong> <strong>of</strong> measurement is thus an expression <strong>of</strong> the<br />
fact that, for a given measurand and a given result <strong>of</strong><br />
measurement <strong>of</strong> it, there is not one value but an infinite<br />
number <strong>of</strong> values dispersed about the result<br />
• The values are consistent with all <strong>of</strong> the observations<br />
and data and one's knowledge <strong>of</strong> the physical world, and<br />
that with varying degrees <strong>of</strong> credibility can be attributed<br />
to the measurand.
Where do uncertainties come from?<br />
– Measuring instrument<br />
Errors, bias, ageing, wear, drift, readability, noise and many other problems.<br />
– Item being measured<br />
which may not be stable<br />
– <strong><strong>Measure</strong>ment</strong> process<br />
the measurement itself may be difficult to make.<br />
– ‘Imported’ uncertainties<br />
calibration <strong>of</strong> instrument<br />
– Operator skill<br />
some measurements depend on the skill and judgment <strong>of</strong> the operator.<br />
– Sampling issues<br />
the measurements you make must be properly representative <strong>of</strong> the process<br />
you are trying to assess<br />
– Environment<br />
– Etc<br />
temperature, air pressure, humidity and many other conditions
How to estimate uncertainty<br />
• Uncertainties can be estimated using statistical analysis<br />
<strong>of</strong> a set <strong>of</strong> measurements, and using other kinds <strong>of</strong><br />
information about the measurement process.<br />
• There are established rules for how to calculate an<br />
overall estimate <strong>of</strong> uncertainty from these individual<br />
pieces <strong>of</strong> information.<br />
• When the uncertainty in a measurement is evaluated<br />
and stated, the fitness for purpose <strong>of</strong> the measurement<br />
can be properly judged.
<strong><strong>Measure</strong>ment</strong> <strong>Uncertainty</strong> Statement<br />
In1930’s, C.H. Meyers et al. at NBS and his colleagues<br />
conducted an elaborate experiment to determine the<br />
specific heat <strong>of</strong> ammonia.<br />
After several years <strong>of</strong> hard work, they completed the<br />
experiment and wrote a paper reporting their results.<br />
Toward the end <strong>of</strong> their paper, Meyers declared:
<strong><strong>Measure</strong>ment</strong> <strong>Uncertainty</strong> Statement<br />
“We think our reported value is good to 1 part in 10,000: we<br />
are willing to bet our own money at even odds that is<br />
correct to 2 parts in 10,000.<br />
Furthermore, if by any chance our value is shown to be in<br />
error by more that 1 part in 1000, we are prepared to eat<br />
the apparatus and drink the ammonia.”<br />
C.H. Meyers et al. (NBS 1930s)<br />
"Round Table Discussion <strong>of</strong> Statement <strong>of</strong> Data and Errors,"<br />
Nuclear Instruments Methods 112, 391 (1973).
The need <strong>of</strong> general rules for evaluating and<br />
expressing uncertainty in measurement<br />
• Reporting the result <strong>of</strong> a measurement requires a quantitative indication <strong>of</strong><br />
the quality <strong>of</strong> the result be given so that those who use it can assess its<br />
reliability.<br />
• Without such an indication, measurement results cannot be compared,<br />
either among themselves or with reference values given in a specification or<br />
standard.<br />
• It is therefore necessary that there be a readily implemented, easily<br />
understood, and generally accepted procedure for characterizing the quality<br />
<strong>of</strong> a result <strong>of</strong> a measurement, that is, for evaluating and expressing its<br />
uncertainty.<br />
• A worldwide consensus on the evaluation and expression <strong>of</strong> uncertainty in<br />
measurement would permit vast spectrum <strong>of</strong> measurement results in<br />
science, engineering, commerce, industry, and regulation to be readily<br />
understood and properly interpreted.<br />
• Imperative that the method for evaluating and expressing uncertainty be<br />
uniform throughout the world so that measurements performed in different<br />
countries can be easily compared in this era <strong>of</strong> the global marketplace
Development <strong>of</strong> GUM<br />
• Initiated by the International Committee on Weights and <strong>Measure</strong>s<br />
(CIPM) in 1977.<br />
• Recommendation INC–1 (1980), ‘Expression <strong>of</strong> experimental<br />
uncertainties’ by Working Group on the Statement <strong>of</strong> Uncertainties<br />
convened by the International Bureau <strong>of</strong> Weights and <strong>Measure</strong>s<br />
(BIPM).<br />
• CIPM approved the Recommendation in 1981, and reaffirmed it in<br />
1986.<br />
• International Organization for Standardization (ISO) formed<br />
Technical Advisory Group on Metrology (TAG4) with 6 other<br />
international organizations to develop a detailed guide based on the<br />
Recommendation.<br />
– BIPM,<br />
– International Electrotechnical Commission (IEC),<br />
– International Federation <strong>of</strong> Clinical Chemistry and Laboratory Medicine (IFCC),<br />
– International Union <strong>of</strong> Pure and Applied Chemistry (IUPAC),<br />
– International Union <strong>of</strong> Pure and Applied Physics (IUPAP) and<br />
– International Organization <strong>of</strong> Legal Metrology (OIML).
Development <strong>of</strong> GUM<br />
• Guide to the Expression <strong>of</strong> <strong>Uncertainty</strong> in <strong><strong>Measure</strong>ment</strong>’ (GUM) was<br />
published in 1993 and reprinted with minor corrections in 1995.<br />
• In 1997 a Joint Committee for Guides in Metrology (JCGM), chaired<br />
by the Director <strong>of</strong> the BIPM, was created by the seven international<br />
organizations that had originally prepared the GUM.<br />
• The JCGM Working Group ‘Expression <strong>of</strong> uncertainty in<br />
measurement’, has the task <strong>of</strong> promoting the use <strong>of</strong> the GUM and<br />
preparing supplements for its broad application and assumed<br />
responsibility for GUM from ISO TAG4.<br />
• In 1998 a further organization joined these seven international<br />
organizations, namely, the International Laboratory Accreditation<br />
Cooperation (ILAC).
ISO/IEC Guide 98-3:2008 <strong>Uncertainty</strong> <strong>of</strong> measurement<br />
Part 3: Guide to the expression <strong>of</strong> uncertainty in<br />
measurement
JCGM 100 : 2008<br />
Evaluation <strong>of</strong> measurement data- Guide to the<br />
expression <strong>of</strong> uncertainty in measurement
OIML G 1-100<br />
Evaluation <strong>of</strong> measurement data- Guide to the<br />
expression <strong>of</strong> uncertainty in measurement
Steps to evaluate the overall uncertainty<br />
1. Decide what you need to find out from your measurements.<br />
2. Identify what actual measurements and calculations are needed to produce<br />
the final result.<br />
3. Carry out the measurements needed.<br />
4. Estimate the contributing uncertainty <strong>of</strong> each input quantity that feeds into<br />
the final result.<br />
5. Evaluate each contribution by<br />
– Type A evaluations - uncertainty estimates using statistics<br />
– Type B evaluations - uncertainty estimates from any other information.<br />
6. Express all uncertainties in similar terms<br />
7. Decide whether the input quantities are independent <strong>of</strong> each other. If not,<br />
then some extra calculations are needed.<br />
8. Calculate the result <strong>of</strong> your measurement, including any known corrections<br />
for things such as calibration.<br />
9. Find the combined standard uncertainty from all the individual aspects.<br />
10.Expand the uncertainty in terms <strong>of</strong> a coverage factor together with a<br />
uncertainty interval, and state a level <strong>of</strong> confidence.<br />
11.Write down the measurement result and the uncertainty
New Development<br />
• Wide international dissemination and acceptance <strong>of</strong> the<br />
GUM<br />
• Its release in the mid-1990s has had a dramatic effect in<br />
improving the evaluation and reporting <strong>of</strong> measurement<br />
uncertainty<br />
• Timely to revise the GUM<br />
• The GUM revision will take the form <strong>of</strong> supplemental<br />
guides.
Evaluation <strong>of</strong> measurement data –<br />
Guide to the expression <strong>of</strong> uncertainty in measurement<br />
• The fundamental reference document<br />
JCGM 100:2008 Evaluation <strong>of</strong> measurement data – Guide to the<br />
expression <strong>of</strong> uncertainty in measurement<br />
• JCGM 104:2009 Evaluation <strong>of</strong> measurement data – An introduction<br />
to the GUM and related documents<br />
• JCGM 101:2008 Evaluation <strong>of</strong> measurement data – Supplement 1 to<br />
the GUM – Propagation <strong>of</strong> distributions using a Monte Carlo method<br />
• JCGM 102:2011 Evaluation <strong>of</strong> measurement data – Supplement 2 to<br />
the GUM– Extension to any number <strong>of</strong> output quantities<br />
• Evaluation <strong>of</strong> measurement data – The role <strong>of</strong> measurement<br />
uncertainty in conformity assessment<br />
• Evaluation <strong>of</strong> measurement data – Concepts and basic principles<br />
• Evaluation <strong>of</strong> measurement data – Supplement 3 to GUM –<br />
Modelling<br />
• Evaluation <strong>of</strong> measurement data – Applications <strong>of</strong> the least-squares<br />
method
‘Guide to the Expression <strong>of</strong> <strong>Uncertainty</strong> in<br />
<strong><strong>Measure</strong>ment</strong>’<br />
• Provides general rules for evaluating and expressing<br />
uncertainty in measurement<br />
• Applicable to a wide range <strong>of</strong> measurements and for use<br />
within standardization, calibration, laboratory<br />
accreditation and measurement services.<br />
• Reference document for evaluating and expressing<br />
uncertainty in a broad spectrum <strong>of</strong> measurements rather<br />
than detailed, technology-specific instructions.<br />
• May be necessary to develop particular<br />
guidelines/examples based on the Guide that deal with<br />
the problems peculiar to specific fields <strong>of</strong> measurement<br />
or with the various uses <strong>of</strong> quantitative expressions <strong>of</strong><br />
uncertainty”
http://www.bipm.org/en/publications/guides/gum.html
End <strong>of</strong> Presentation<br />
Any Question ?<br />
Email: chua_sze_wey@nmc.a-star.edu.sg