The MRA establishes three types of national measurement standard comparisons of NMIs, which differ in the problems they solve and in the requirements for preparing, carrying out, and reporting the results of comparison:
CIPM key comparisons (CIPM KC);
RMO key comparisons (RMO KC): for example, key comparisons of the EURAMET, APMP, COOMET, and other regional organizations;
supplementary comparisons.
CIPM key comparisons are performed by consultative committees. As a rule, the pri- mary national measurement standards participate. The CIPM key comparisons solve two main problems:
(1) determination of a reference value of key comparisons (KCRV) and the degree of equivalence of national measurement standards (Appendix B of the BIPM data- base);
(2) confirmation of CMCs claimed (Appendix C of the BIPM database).
The number of participants in CIPM key comparisons is limited. Therefore, the prob- lem of disseminating the metrological equivalence to a greater number of participants interested in the mutual recognition of measurement results and calibrations is urgent.
This problem is solved within the framework of regional key comparisons carried out
by regional metrology organizations. Within the framework of RMO KC two tasks are solved:
(1) determination of the degree of equivalence of national measurement standards;
(2) confirmation of the CMCs claimed.
It is important to stress that the reference value of key comparisons within the frame- work of a RMO KC is not determined, and the equivalence degree is established rela- tive to a KCRV through results of a linking NMIs which take part in both comparisons.
Supplementary comparisons of national measurement standards solve only the task of confirming CMCs for which no degree of equivalence of national measurement standards have been established.
An analysis of reports on key comparisons has shown a variety of schemes used to perform the key comparisons. These schemes are characterized by properties of a traveling transfer measurement standard, number of participants, time limits, and so on. ’The tasks of planning and developing a schedule, preparing and investigating the traveling measurement standard, developing a protocol of comparisons, processing comparison results, and coordinating a report on the comparisons are fulfilled by a pilot laboratory. Sometimes pilot laboratories receive the assistance of other laboratories which are participants of comparisons.
The following classification features for selecting a method for the evaluation of comparison data are important.
Scheme of performing a comparison.Comparisons can be of a circular or radial type.
When comparisons are circular, a traveling measurement standard leaves a particular pilot laboratory and, according to a schedule of comparison, measurements with this traveling measurement standard are carried out by all comparison participants. After that, the traveling measurement standard returns to the pilot laboratory. At the end of comparisons the pilot laboratory carries out measurements with the purpose of make- ing sure that the traveling measurement standard is in an appropriate operational state.
To reduce the time of comparisons it is possible to use a number of identical travel- ing transfer measurement standards. In this case a problem of linking different loops of one and the same key comparison arises. This problem should not be confused with the problem of linking results of RMO key comparisons with CIPM key com- parisons. The scheme of circular comparisons implies a good stability of a traveling measurement standard and relatively short duration of comparisons. Provided that the comparisons require a long period of time and that a check of the traveling standard stability is needed in the process of comparison, then the radial scheme of comparisons is applied.
In the radial scheme of comparisons the traveling standard returns to the pilot lab- oratory several times in the comparison process, and the pilot laboratory carries out the comparisons while keeping track of a potential systematic drift or instability of a measurand.
Section 1.2 Classification of key comparisons of national measurement standards 7 Kind of measurand.In comparisons a measurand can be
the value of a material measure or values of set of measures. The measures from a set can be of equal nominal values (in case of possible damages of these measures during transportation) or of significantly different nominal values;
the value of a calibration coefficient of a measuring instrument;
functional dependence – for example, the dependence of a calibration coefficient of a measuring instrument on a frequency. This implies the fulfillment of measurements at different frequencies.
A priori information about compared measurement standards.Participants of comparisons present to a pilot laboratory the measurement results, the uncertainty bud- get, and information about their measurement standard and measurement procedure. In particular, if secondary measurement standards participate in comparisons, then while processing it is important to take into account a possible covariance of the results of participants, which is caused by the traceability of these secondary measurement stan- dards to the primary ones which are also participating in the given comparisons.
Form of reporting the results(the form of a set of reference values and degrees of equivalence). In most cases, at least at the initial stage of comparison data processing, the reference value is understood to be the best estimate of a measurand. Therefore, in processing one frequently obtains a set of reference values and, correspondingly, a set of degrees of equivalence. Strictly speaking, only in specific cases are there grounds for calculating a mean of a number of reference values: for example, the case when a set of material measures with similar nominal values is used as a traveling measurement standard. A tendency to get a laconic presentation of the measurement results leads to the application of different procedures for averaging a number of reference values.
In this case the direct connection of the reference value with a particular measurand disappears. When from the very beginning the reference value can be represented by certain dependence, the choice of a certain fixed point, for example, a frequency given for a joint presentation of results obtained by different participants, seems to be a more correct action.
As an example the scheme of key comparisons CCM.M-K11is given in Figure 1.2.
This scheme combines into one the participants of CIPM key comparisons and regional comparisons of the APMP, EUROMET and COOMET.
One practical result of national measurement standard comparisons is the confirma- tion of the calibration and measurement capabilities of a particular national laboratory.
The laboratory carries out measurements in according to a routine procedure, presents to a pilot laboratory its measurement results, budget of uncertainty, information about the measurement standard, and measurement procedure.
1 http://kcdb.bipm.org/AppendixB/KCDB_ApB_info.asp.
NMIJ
NIMT KRISS NIM
KRISS
KIM-LIPI CSIR-NML
NIMT MUSSD
KIM-LIPI
CMS/ITRI MSL
NML-SIRIM SCL
SPRING Singapore BIPM
NIST NRC
NIM CENAM
INRIM
NMi-VSL NPL
PTB
VNIIM SMU LNE-INM
NSC IM BelGIM
LNE JV
SP MIKES
DFM SMD CEM GUM
CMI
SP JV MIKES
DFM
CEM SMD
VMT/VMC
PTB PTB
PTB
NPLI NMIJ
KRISS
VNIIM SMU INRIM SMU
NPL
NPL MKEH
INRIM
NIMT BEV
METASw
METROSERT
NML(IE) LNMC
MIRS EIM
BIM INM (RO)
IPQ
BIPM
COOMET.M.M-K1 EUROMET.M.M-K4 EUROMET.M.M-K1 EUROMET.M.M-K4.1
CCM.M-K1 Package1 CCM.M-K1
Package2
APMP.M.M-K7 APMP.M.M-K7.1
Figure 1.2.Scheme of the key comparisons CCM.M-K1. Here: ı– laboratory participant of the key comparisons;– pilot laboratory of the key comparisons.
Therefore, key comparisons are, as a matter of fact, the tool for a practical check of laboratory competence, the quality of metrological services rendered by a particular laboratory. The key comparisons can take more than a year. Sometimes discussions on the results and a search for a consensus on a method of data processing take even longer than the actual time required for performing the measurements. A high status of key comparisons is caused by the fact that the standards being compared are the reference national measurement standards in a chain of transferring the unit of a quantity.
Section 1.3 Basic approaches to evaluating key comparison data 9