EUROMET TC-M Meeting 2007 Thursday, 01 March 2007 EUROMET Project 702 EUROMET.M.D-K4 Comparison of the calibrations of high resolution hydrometers for liquid density determinations Salvatore Lorefice INRIM, Torino Salvatore Lorefice Thursday, 01 March 2007
Temporal sequence Salvatore Lorefice EUROMET project 702 - EUROMET. M.D-K4 Thursday, 01 March 2007 2 January 2003 Answers to questionnaire March 2003 Registration of EUROMET project September 2003 October 2003 November 2003 Agreement on Technical Protocol Start of measurements (Pilot Laboratory) Stop due to the breakage of hydrometers New Hydrometers as Transfer Standards March 2004 May 2005 February & August 2006 February 2007 June/July 2007 Measure of new hydrometers (Pilot Laboratory) End of measurements Preliminary Results Draft A Draft B?????????????????????????????????
The participants Salvatore Lorefice EUROMET project 702 - EUROMET. M.D-K4 Thursday, 01 March 2007 3
The aims Salvatore Lorefice EUROMET project 702 - EUROMET. M.D-K4 Thursday, 01 March 2007 4 to recognize and compare the different experimental setups and calibration methods used by the participants to ascertain the consistency of their calibration results to harmonize the stated different levels of NMI laboratories for hydrometer calibration in the density range between 0,600 g/cm 3 and 1,300 g/cm 3 at 20 C to compare the participants methods for the evaluation of the uncertainty
The transfer standards (hydrometer samples) Salvatore Lorefice EUROMET project 702 - EUROMET. M.D-K4 Thursday, 01 March 2007 5
The organization Salvatore Lorefice EUROMET project 702 - EUROMET. M.D-K4 Thursday, 01 March 2007 6 Hydrometer 21964 Hydrometer 21971 Hydrometer 5941 Hydrometer 21958 0,600 5-0,609 5 0,610 5-0,619 5 0,810 5-0,819 5 0,800 5-0,809 5 Hydrometer 0001 1,000 5-1,009 5 Hydrometer 6905 0,990 5-0,999 5 Hydrometer 58431 1,290 5-1,299 5
Selected conditions Salvatore Lorefice EUROMET project 702 - EUROMET. M.D-K4 Thursday, 01 March 2007 7 The participating laboratories were asked to calibrate the assigned hydrometers at four graduation marks of the scale and for each of them the correction C, was calculated at the 20 C. The test points and the surface tension values of the liquid, in which each hydrometer was intended for use, were stated in advance. The participants were free to perform all measurements using their own procedure. It was however required that the hydrometers be kept at the laboratory only for the time necessary for calibration and not longer than the time planned. The participants submitted to INRiM, as the pilot laboratory, the information concerning the status of the transfer standards, the apparatus used during the comparison and the measurement results by noting on the corresponding formats.
Report data Salvatore Lorefice EUROMET project 702 - EUROMET. M.D-K4 Thursday, 01 March 2007 8 Details of the instrumentation used by each participant in the project, Information about the measurements and the parameters used for the comparison, Calculated values of the four corrections for each transfer standard at the specified reading marks, Uncertainty budget of the four calculated corrections
The results Salvatore Lorefice EUROMET project 702 - EUROMET. M.D-K4 Thursday, 01 March 2007 9
Experimental facilities at the different NMI s Salvatore Lorefice EUROMET project 702 - EUROMET. M.D-K4 Thursday, 01 March 2007 10 Institute Balance Max capacity [g] / readability [g] Weighing in Hydrostatic air weighing Buoyant liquid INRiM 405 / 0,00001 n-nonane Thermostat Type, capacity double-walled glass vessel, 30 litre OMH 1000 / 0,0001 n-nonane Tamson, 70 litre PTB* 410 / 0,0001 n-tridecane Tamson, 70 litre IPQ 405 g / 0,0001 n-nonane Tamson, 70 litre MIKES 303 / 0,000 1 Ethanol double-walled glass vessel, 5 litre LNE* 3 010 / 0,000 1 405 / 0,000 1 n-tetradecane Tamson, 70 litre BEV 300 / 0,000 1 1109 / 0, 000 1 n-nonane double-walled glass vessel, 80 litre SMU 205 / 0,000 1 n-nonane Tamson, 70 litre GUM* 1 109 / 0,000 1 n-nonane Tamson, 70 litre UME 405 / 0,000 1 n-tridecane Tamson, 70 litre VNIM 205 / 0,000 01 Ethanol Tamson, 15 litre *Weighing: Substitution method Thermometer for the liquid temperature 100 Ohm PRT, a.c bridge Alignment CCD camera automatic Surface tension method Plate Quartz Thermometer Magnifier hand-operated Reference data Mercury glass thermometer CCD camera hand-operated Plate 100 Ohm PRT, Digital Thermometer 100 Ohm PRT, Digital Thermometer 100 Ohm PRT, Digital Thermometer 25 Ohm PRT, Digital Thermometer Mercury glass thermometer 25 Ohm PRT, a.c bridge 100 Ohm PRT, Digital Thermometer Mercury glass thermometer Magnifier hand-operated CCD camera hand-operated Magnifier hand-operated Magnifier hand-operated Cathetometer hand-operated Magnifier hand-operated Magnifier hand-operated Magnifier hand-operated Ring Ring Du Noüy & Wilhelmy Reference data Reference data Ring Plate Reference data
The degrees of equivalence Salvatore Lorefice EUROMET project 702 - EUROMET. M.D-K4 Thursday, 01 March 2007 11 Degree of equivalence with respect to the KCRV Degree of equivalence between pairs of NMIs Mutual recognition of national measurements standards and of calibration and measurement certificates issued by national metrology institutes (MRA), Technical Report, International Committee for Weights and Measures, 1999. (http://www.bipm.org/pdf/mra.pdf).
Key comparison reference value Salvatore Lorefice EUROMET project 702 - EUROMET. M.D-K4 Thursday, 01 March 2007 12 The conditions: Transfer standard having good short-term stability and stability during transport Each institute s measurement was realized independently of the other institutes measurements in the key comparison (no mutual dependence of the institute s measurements) For each institute a gaussian distribution can be assigned to the measurand of which the institute s measurement is an estimate Maurice Cox et al. Metrologia 2002, 39, p 589-595
The steps Procedure A: The weighted mean of the institutes measurements, where the weights are equal to the reciprocals of the squares of the associated standard uncertainties, should generally be taken as the KCRV. It would, however, be inappropriate if some of the institutes measurements were inconsistent with the remainder. if the 2 consistency χ check made within Procedure A doesn t fail One then proceeds to compute degrees of equivalence and the associated uncertainties Salvatore Lorefice EUROMET project 702 - EUROMET. M.D-K4 Thursday, 01 March 2007 13
The deviation from the KCRV Salvatore Lorefice EUROMET project 702 - EUROMET. M.D-K4 Thursday, 01 March 2007 14 D i = x i KCRV u D i = 2 ( x ) + u ( KCRV ) 1 ( 2 u 2cov( x, KCRV ) 2 i i ( D i ) = u D i U 2 Discrepant measurements at the 5 % level of significance are if: Di > 2u Di
The steps Salvatore Lorefice EUROMET project 702 - EUROMET. M.D-K4 Thursday, 01 March 2007 15 Procedure B: is based on Monte Carlo simulations of the median generated from distributions representing the results from each laboratory. A very large number of simulated data sets are generated, the median of each is computed, and then the KCRV is computed by taking the mean of the medians of all the simulated data sets. Use of the median reduces the sensitivity to outliers, and the average over a large number of simulated results should or could reduce the sensitivity to just one result that occurs for the median method for small numbers of participants.
The coverage interval Salvatore Lorefice EUROMET project 702 - EUROMET. M.D-K4 Thursday, 01 March 2007 16 Among the 95% coverage intervals for the KCRV the shortest interval is chosen Lower Upper Correction at 0,616 5 Mean -0,000071 Std Dev 0,0000065 Left x (Lower end point) -0,000083 Left P 1,5 % Right x (Upper end point) -0,000058 Left P 96,5 %
Hydrometer S/N 21964 Salvatore Lorefice EUROMET project 702 - EUROMET. M.D-K4 Thursday, 01 March 2007 17
Hydrometer S/N 21971 Salvatore Lorefice EUROMET project 702 - EUROMET. M.D-K4 Thursday, 01 March 2007 18
Hydrometer S/N 5941 Salvatore Lorefice EUROMET project 702 - EUROMET. M.D-K4 Thursday, 01 March 2007 19
Hydrometer S/N 21958 Salvatore Lorefice EUROMET project 702 - EUROMET. M.D-K4 Thursday, 01 March 2007 20
Hydrometer S/N 0001 Salvatore Lorefice EUROMET project 702 - EUROMET. M.D-K4 Thursday, 01 March 2007 21
Hydrometer S/N 6905 Salvatore Lorefice EUROMET project 702 - EUROMET. M.D-K4 Thursday, 01 March 2007 22
Hydrometer S/N 58431 Salvatore Lorefice EUROMET project 702 - EUROMET. M.D-K4 Thursday, 01 March 2007 23
Degree of equivalence between pairs of NMIs Salvatore Lorefice EUROMET project 702 - EUROMET. M.D-K4 Thursday, 01 March 2007 24 x i1 x i2 KCRV d ij = Both laboratories are in the same loop ( x KCRV ) ( x KCRV ) = ( x x ) i1 i2 i1 i2 Laboratories are in different loops X i,j2 x i1 x j1 RV 1 RV 2 artefact 1 artefact 2 d ij = ( x RV ) ( x RV ) = ( x x ) ( RV ) i1 1 j1 2 i1 j1 2 RV1 2 2 2 u d = u cov ( 1 1),( 2 2 ) ij k xi RV x j RV ux + u i1 k = xi1, x j 2, RV1, RV2 White D. R., Metrologia 2004, 41, p 121-131 2 ( ) ( ) 2 2 x j 2 1 1
Possible situations in this project Salvatore Lorefice EUROMET project 702 - EUROMET. M.D-K4 Thursday, 01 March 2007 25 Both laboratories are in the same loop Laboratories are in different loops If one of the two laboratories, i.e. the j-th one, measures both the artefacts, that is the measurements it provides are ~ x j 1, 2 Both laboratories i and j participated to both the petals, hence contributing to determine the reference values RV 1 and RV 2
in the range 600 kg/m 3 Salvatore Lorefice EUROMET project 702 - EUROMET. M.D-K4 Thursday, 01 March 2007 26
in the range 800 kg/m 3 Salvatore Lorefice EUROMET project 702 - EUROMET. M.D-K4 Thursday, 01 March 2007 27
in the range 1 000 kg/m 3 Salvatore Lorefice EUROMET project 702 - EUROMET. M.D-K4 Thursday, 01 March 2007 28
in the range 1 300 kg/m 3 Salvatore Lorefice EUROMET project 702 - EUROMET. M.D-K4 Thursday, 01 March 2007 29
Conclusion Salvatore Lorefice EUROMET project 702 - EUROMET. M.D-K4 Thursday, 01 March 2007 30 Some systematic differences and either underestimated or overestimated uncertainties can be identified between the submitted results, although a particular good agreement was found among the results provided by most of the participants. Last but not least the results allowed also to determine the degree of equivalence between pairs of the participating NMIs. In conclusion these results could support: the reduction of some uncertainty contributions of some participants, the degrees of equivalence will serve as a basis for the new Calibration Measurement Capabilities (CMC) entries on hydrometer calibration, the experience acquired in this key comparison shall be taken into account in conducting the planed CIPM key comparisons and further intercomparisons in the future.
Thank you for your attention Salvatore Lorefice EUROMET project 702 - EUROMET. M.D-K4 Thursday, 01 March 2007 31