13. December at 10.15 in the Senat Hall of the University of Tartu (Ülikooli 18) will defend Viktor Vabson his doctoral theses in physics "Measurement uncertainty in Estonian Standard Laboratory for Mass"
Supervisors: dr Mart Noorma and dr Toomas Kübarsepp
Oponents: dr Lars Nielsen from Danish Institute of Fundamental Metrology and prof Tauno Otto Tallinn University of Technology.
"Measurement uncertainty in Estonian Standard Laboratory for Mass"
A high-level well-working national measurement infrastructure is essential for competitiveness of country, for the advancement of science and technology, and for quality of life. Since 2001 in Estonia at the Central Office of Metrology, Metrosert, the national standards for mass, length, temperature and electric quantities are established at the internationally secondary level. National measurement standard usually represents the top level competence of the country in respective scientific-technical field. However, in order to use its capabilities effectively confidence in the measurement uncertainty and international equivalence of offered services is needed. Technical level and international equivalence of the standard are validated by international experts on the basis of inter-comparisons and peer evaluations, whereby measurement uncertainty is a key element for meaningful determination of the degree of equivalence between the standards and measurement results.
Estonian standard laboratory for mass (NSLM) is realizing and representing the mass scale from l mg to 50 kg, being able to calibrate the mass and conventional mass value of the weights with the highest OIML E1 accuracy class. NSLM can test the conformity of magnetic properties of the weights to the requirements of the respective accuracy class. NSLM can calibrate the density of the weights in the range from 1 g to 2 kg. Mass laboratory is equipped with three automatic mass comparators, with more than 100 weights, and with many auxiliary instruments; laboratory is accommodated in air-conditioned measurement rooms with temperature and humidity control, and with filtered air. At the NSLM preferably four measurement models are used; these models are carefully validated for the whole relevant measurement range. For evaluation of the uncertainty in measurement for major part of instruments extensive calibration histories are available. Additionally, some special studies are conducted, in order to get better data base for uncertainty estimation as provided by routine measurements.
Uncertainty of the weights representing the mass scale at the NSLM is estimated following the GUM standard procedures. In general, uncertainty evaluated according to GUM performs satisfactorily for majority of applications. Nevertheless, there are some situations in practice allowing improvement if sufficient measurement information is available. Not easy to handle is the effect of nonzero serial correlation. Another similarly complicated problem is revealing possible systematic effects in comparator readings in order to eliminate them or at least to take them into account in measurement result and uncertainty estimate. If not treated the uncertainty may be substantially underestimated. Methods proposed in this study will at least partly solve both problems and reduce risk of underestimation, and they are applicable in many other measurement fields. The agreement between the inter-comparison results presented by the NSLM and comparison reference values demonstrated to date shows that measurement methods, calibration procedures and respective uncertainty estimates developed and tested at the NSLM can reliably be applied in practice.