Page 18 - base

“Uncertainty-noise” Le Mans17
Acoustique
&
Techniques n° 40
Treatment of Measurement Uncertainties in International and European Standards on Acoustics
measurement, all relevant uncertainty components which are
of importance in a given situation shall be taken into account
using appropriate methods of analysis”.
It is still apparent, however, that testing laboratories often
are not familiar with the uncertainty concept and may thus be
unable to meet these requirements without further guidance.
With some justification, they expect to receive information on
the evaluation of measurement uncertainties from international
or national standards. However, with some exceptions,
standards generally do not yet provide this kind of guidance.
Therefore, laboratory and accreditation organizations jointly
approached the main international standardization bodies ISO
and IEC some years ago expressing their concern about this
situation and asking for more attention to be paid to this issue
during the development of international standards specifying
measurement methods.
In consequence of the above, respective discussions were
initiated within relevant international Technical Committees
on acoustics, such as IEC/TC 29 “Electroacoustics”, ISO/
TC 43 “Acoustics” and, especially, ISO/TC 43/SC 1 “Noise”.
The respective conclusions of these discussions taken by
ISO/TC 43 and TC 43/SC 1 and recent experience with their
implementation are presented in the following sections. A
special section is devoted to the situation in the corresponding
European Technical Committee which is CEN/TC 211
“Acoustics”.
Policy of ISO/TC 43 “Acoustics” and
TC 43/SC 1 “Noise” with respect to
measurement uncertainty
First steps
The need to put more emphasis on the issue of measurement
uncertainty when developing new or revised measurement
standards was recognized by ISO/TC 43/SC 1 already in
1999. The following resolution was adopted:
“ISO/TC 43/SC 1 requests each of its Working Groups
to consider the “Guide to the expression of uncertainty
in measurement” in the preparation of documents and,
if appropriate, include a statement of measurement
uncertainty”.
It appeared, however, that even among experts appointed
to ISO Working Groups, sufficient knowledge about the
proper treatment of measurement uncertainty in international
standards was hardly available. With a few exceptions, the
results remained unsatisfactory. It became obvious that some
more information and direct guidance had to be provided on
the principles of the GUM and its application to measurements
in acoustics. For this reason, a brief technical seminar was held
for all delegates to Working Group and Plenary meetings of TC
43 and its Subcommittes in Berlin in September 2003.
This seminar was attended by around 100 experts and was
quite well accepted. Though it could not be the intention to
assist Working Groups in their specific problems, the seminar
certainly reduced existing aversion to approach the subject
and submitted the right impression that the application of
the GUM is less complex than it might have been suspected
before.
A short policy paper was introduced and accepted with
slight modifications at the following plenary meetings. It is
considered as a pragmatic approach to the subject and since
then forms the basis for the treatment of uncertainty in ISO
Standards on measurement and prediction of sound. The
paper is reproduced in full in the following section followed
by additional interpretation of some selected aspects.
Treatment of measurement uncertainty in standards
on acoustics (adopted policy paper, document
ISO/TC 43 N 1023, June 2004)
1- If relevant, each standard shall contain a specific clause
on measurement uncertainty in its main text. It shall contain
the following statement:
«The uncertainty of results obtained from measurements
according to this International Standard shall be evaluated,
preferably in compliance with the GUM. If reported, the
expanded uncertainty together with the corresponding
coverage factor for a stated coverage probability of ...%
(preferably 95 %) as defined in the GUM shall be given.
Guidance on the determination of the expanded uncertainty
is given in Annex yx.»
2- The GUM shall be added to the list of normative
references.
3- In a clause dealing with test reports, it shall be stated
that the reporting of measurement uncertainty is considered
mandatory unless, in a particular case, a working group has
good reasons for leaving it optional.
4- The annex on uncertainty should be structured as
follows:
- general information;
- a functional relationship (model) that links all relevant input
quantities to the quantity to be determined (measurand);
- an uncertainty budget in tabular form, containing the input
quantities dominating the total uncertainty, their best estimates,
the standard uncertainties associated with these values, the
assumed probability distributions, the sensitivity coefficients
derived from the model and the resulting contributions to the
combined standard uncertainty;
- the calculation of the expanded uncertainty for the stated
coverage probability.
Annex J of ISO 3745:2003-12-01 provides a useful example
of this principle.
5- The evaluation of measurement uncertainty is the
responsibility of each laboratory performing the measurement.
Even if the requirements of the standard are fully met, the
uncertainty of results from different laboratories may differ
depending on the specific measurement conditions.
It is up to each working group to decide whether quantitive
information on the various uncertainty contributions can be
given or not and whether this information is considered to
represent a typical situation or rather a worst case.
It is emphasized that a detailed uncertainty budget in
accordance with the GUMwill provide useful information on the