Final report to the CCT on key comparison CCT-K6 Comparison of local realisations of dew-point temperature scales in the range -50 C to +20 C

Transcription

1 NPL REPORT ENG 57 Final report to the CCT on key comparison CCT-K6 Comparison of local realisations of dew-point temperature scales in the range -50 to +20 S Bell, M Stevens, H Abe, R Benyon, R Bosma, V Fernicola M Heinonen, P Huang, H Kitano, Z Li, J Nielsen, N Ochi, O A Podmurnaya, G Scace, D Smorgon, T Vicente, A F Vinge, L Wang, H Yi April 2015

2 2

3 Final report to the CCT on key comparison CCT-K6 Comparison of local realisations of dew-point temperature scales in the range -50 to +20 S Bell 1, M Stevens 2, H Abe 3, R Benyon 4, R Bosma 5, V Fernicola 6, M Heinonen 7, P Huang 8, H Kitano 3, Z Li 9, J Nielsen 10, N Ochi 3, O A Podmurnaya 11, G Scace 8, D Smorgon 6, T Vicente 4, A F Vinge 11, L Wang 12, H Yi 9 1 Engineering Measurement Division, National Physical Laboratory (NPL), Teddington, UK 2 United Kingdom Accreditation Service (UKAS), Feltham, UK 3 National Metrology Institute of Japan (NMIJ), Tsukuba, Japan 4 Instituto Nacional de Técnica Aeroespacial (INTA), Madrid, Spain 5 Van Swinden Laboratorium (VSL), Delft, The Netherlands 6 Istituto Nazionale di Ricerca Metrologica (INRIM), Turin, Italy 7 Centre for Metrology and Accreditation (MIKES), Espoo, Finland 8 National Institute for Standards and Technology ( NIST), Gaithersburg, USA 9 National Institute of Metrology (NIM), Beijing, China 10 Danish Technological Institute (DTI), Århus, Denmark 11 National Scientific and Russian Research Institute for Physical, Technical and Radiotechnical Measurements East Siberia Branch (VNIIFTRI-ESB), Irkutsk, Russia 12 National Metrology Centre, Agency for Science, Technology and Research (NMC), Singapore 3

4 Queen s Printer and Controller of HMSO, 2015 ISSN National Physical Laboratory Hampton Road, Teddington, Middlesex, TW11 0LW Extracts from this report may be reproduced provided the source is acknowledged and the extract is not taken out of context. Approved on behalf of NPLML by Teresa Goodman, Optical Measurement Group 4

5 CONTENTS 1 INTRODUCTION ORGANISATION OF THE COMPARISON PARTICIPANTS, AND STANDARDS USED IN THE COMPARISON INRIM INTA MIKES NIM NIST NMC NMIJ NPL VNIIFTRI VSL COMPARISON SCHEME COMPARISON SCHEDULE COMPARISON METHOD TRANSFER STANDARDS REPORTING IMPARTIALITY COMPLIANCE WITH OTHER REQUIREMENTS PERFORMANCE OF THE TRANSFER STANDARDS PROBLEMS WITH THE TRANSFER STANDARDS CHECKS OF SAFE TRANSPORTATION CONSISTENCY OF PERFORMANCE OF INSTRUMENTS THROUGHOUT THE COMPARISON STABILITY OF EACH HYGROMETER Pilot measurements INTA measurements of Hyg Drift estimates and discussion PARTICIPANT RESULTS BILATERAL EQUIVALENCE KEY COMPARISON REFERENCE VALUES (KCRV) CALCULATION OF KCRV CHI-SQUARED TEST COMPARISON RESULTS PRESENTED IN TERMS OF KCRV DISCUSSION CONCLUSION ACKNOWLEDGEMENTS REFERENCES APPENDIX 1: TECHNICAL PROTOCOL APPENDIX 2: RESULTS REPORTED BY THE PARTICIPANTS APPENDIX 3: UNCERTAINTY ANALYSES OF PARTICIPANTS

6 1 INTRODUCTION At the 19 th (1996) meeting of the International Bureau of Weights and Measures (BIPM) Consultative Committee for Thermometry (CCT), the Working Group on Humidity Measurements (CCT/WG6) was formed. The Working Group was charged with organising an international comparison in the field of humidity standards. Preparations for a key comparison began with an initial protocol drafted by NIST and reviewed by CCT WG7 (key comparisons). At its 2001 meeting, the CCT appointed NPL as pilot, and NMIJ as assistant pilot. It was seen that the time-consuming nature of comparison measurements in this field would require that the number of participants be limited in order to keep the duration of the comparison within reasonable limits. The CCT agreed a limit of 10 participants. The balance of representation between Regional Metrology Organisations (RMOs) was proposed by the CCT based on the level of humidity standard activity within the RMOs. During 2002, RMOs nominated institutes to participate, and the comparison protocol was revised and agreed with participants. At this time, the role of pilot for EUROMET.T-K6 was transferred from NPL to MIKES (Finland), and the CCT-K6 and EUROMET.T-K6 protocols and reporting templates were developed in close alignment. The CCT-K6 protocol was reviewed by WG6 when they met in September 2002, approved by K6 participants in March 2003, with agreed minor changes arising from comments from CCT WG7 who approved the protocol in July At the start of preparations, measurement comparison methods in the field of humidity were not well established, so in-depth trials of (initially) three travelling standards were carried out. In consultation with participants, two preferred instruments were identified for use. In late 2002 NPL performed the initial pilot set of comparison measurements using the travelling standards, marking the effective start of the comparison. Measurements by participants proceeded until 2009, interspersed by a large number of delays, repairs, additional checks and participant changes, which all contributed to extending the duration of the comparison. The details of the incidents and checks are given in Section 3. Despite the extended timescale and several minor repairs to the instruments, the checks of consistency and drift of the instruments provide satisfactory evidence that the results are reliable enough for the purpose of the comparison, as discussed in Section 4. By the time of conclusion of CCT-K6, the corresponding RMO comparisons APMP.T-K6 [1] and EUROMET.T-K6 [2] had already been completed, and results were in use to support claims of calibration and measurement capability (CMCs) of NMIs involved. These comparisons are therefore available for linkage via common participants, limited only by the continuity of the realisations at those institutes. 6

7 2 ORGANISATION OF THE COMPARISON The purpose of the comparison was to establish the degree of equivalence between realisations of local scales of dew/frost-point temperature of humid gas, in the range -50 to +20, among the participating national measurement institutes. The key comparison is a comparison of the measurand realisation of local scale of dew-point temperature at the participating national institutes. The dew-point scale comprises dew points and (below the freezing point of water) frost points. The comparison was designed to provide Estimates of bilateral equivalence between every pair of participants at each measured dew point A key comparison reference value (KCRV) for each nominal value of dew/frost point in the comparison. Estimates of equivalence of each participant to the KCRV The technical protocol for the comparison is shown in Appendix 1. The comparison was made by circulation of a pair of travelling transfer standards. Each transfer standard was used to independently measure dew/frost-point temperature of a sample of moist gas (air or nitrogen) produced by a participant's standard generator using the same measuring process. Measurements were made at dew point nominal values of -50, -30, -10, +1 and +20. The points were chosen to test the main range of interest, covering frost and dew regions. The nominal value of +1 represents the range near 0 while being far enough above it to avoid ambiguities that can arise around the freezing point of water. The comparison took the form of a closed circulation in two consecutive loops. There was one pair of hygrometers, which were at all times measured nominally simultaneously. Simultaneous measurements using a pair of standards gives information about the withinlaboratory consistency of the measurements, the reproducibility of the instrument performance, and continuous feedback about the successful transport of the instruments without any major shift in performance. The values of dew-point temperature reported for the travelling standards are arbitrary values calculated from the measured resistance. The travelling standards are used simply as comparators. Further below, values of Key Comparison Reference value (KCRV) are calculated, but it is important to note that any KCRV in this comparison has no absolute significance it does not represent a reference value in the SI, only a comparison parameter. 7

8 2.1 PARTICIPANTS, AND STANDARDS USED IN THE COMPARISON A list of the participants is given in Table 1, including concise indication of their standard generator types, and references to publications. Table 2.1 List of participants. (Institute current names and abbreviations are given at the start of the comparison several of them were known by other names). The type of standard is also given, as twopressure (2-P), or single-pressure (1-P). Participant Acronym Standard type Centre for Metrology and Accreditation (Finland) [3] MIKES 1-P Instituto Nacional de Technica Aeroespacial (Spain) [4,5,6,7] INTA 2-P Istituto Nazionale di Ricerca Metrologica (Italy) 1 [8,9] INRIM 1-P National Institute of Metrology (China) 1 [10,11] NIM 2-P National Institute for Standards and Technology (USA) [12,13] NIST 2-P National Metrology Centre, Agency for Science, Technology and NMC 2-P Research (Singapore) 1 [14, 15] National Measurement Institute of Japan (Japan) [16,17] NMIJ 2-P National Physical Laboratory (UK) [18,19]] NPL 1-P National Research Institute for Physicotechnical and Radio VNIIFTRI 2-P Engineering Measurements (Russia) East Siberia Branch 2 [20] ESB Van Swinden Laboratorium (Netherlands) 1 [21, 22] VSL 1-P Details of participant facilities used for the comparison are given in the following subsections. Participants have confirmed that no significant changes have been made to their dew point standards between the time of measurement and the time of reporting. This ensures that the results of any participant can form a valid link to the KCRV through other comparisons up to the date of this publication at least. The only exceptions to this are NIST and VSL. The NIST Low Frost Point Generator used for the -50 and -30 points in this comparison was significantly modified after the comparison measurements (as discussed in Section 7.1). The VSL generator used here for CCT-K6 was used also for EUROMET.T-K6 [2], but since then has been replaced with a newly constructed standard [23, 24] INRIM The INRIM primary standards used in the comparison were INRIM-designed single-pressure generators [8, 9]. For the comparison values -50, -30 and -10, the IMGC02 (now INRIM 02) standard frost-point generator was used, and and at +1 and +20 the IMGC01 (now INRIM 01) standard dew point generator was used. 1 At the start of planning the comparison, several participant institutes were known by previous names: INRIM was IMGC, NIM was NRC-CRM, NMC was SPRING, and VSL was NMi. Throughout this report, current names of institutes are used. 2 Initially, VNIIM was the planned participant for Russia, but later VNIIFTRI was nominated. 8

9 In the INRIM 02 generator, a stream of N 2 gas flows through a heat exchanger and, then, over a surface of ice inside an isothermal saturator held in a temperature bath. The gas is recirculated many times over the isothermal surface, achieving an equilibrium saturation whose temperature is monitored with two PRTs installed at the outlet of the saturator in the ice layer and in the air flow. The system was constructed with vacuum-grade fittings and electropolished tubing. A centrifugal pump is used to re-circulate the carrier gas (N 2 ) at any flow value between 3 L min -1 and 10 L min -1. A hygrometer in calibration is connected in a branch of the main loop with gas flow rate to be set between 0.5 Lmin -1 and 1.5 L min -1. The INRIM 01 generator is a recirculating-type generator to cover the dew/frost point range from -15 to +90. The carrier gas can be N 2 or air. It is designed to accommodate the hygrometer in calibration either into an isothermal chamber (closed-circuit mode) or in a secondary branch with a draw-off flow up to 1.5 L min -1 (open-circuit mode). The saturating unit consists of a box where the air is forced to flow over a surface of water or ice for a length of 1.2 m. The temperature is monitored at three different sections of the saturator by means of three pairs of platinum resistance thermometers (PRTs): one thermometer is located immediately underneath the water surface (the water depth never exceeds 10 mm) and the other is placed in the gas, immediately above the first one. One pair of PRTs is at the outlet where the saturator temperature is defined. The traceability of the realisation is in terms of temperature through the calibration of the saturator platinum resistance thermometers with traceability to the SI (ITS-90) via INRIM temperature standards. Supporting electrical and pressure measurements are traceable through calibrations at INRIM INTA In the range from -50 to -10, the INTA Low-range standard humidity generator is a modified Thunder Scientific model 4500 two-pressure generator with saturation performed with respect to ice. In the range from 1 to 20 the INTA High-range standard humidity generator is a modified Thunder Scientific model 9000 two-pressure generator used with saturation with respect to water. The generators were operated in two-pressure mode at nominal flow rates of 2.7 and 50 L/min, respectively and with a saturator pressure below 300 kpa. The realised frost/dew-point temperatures in both cases were determined from independent temperature and pressure measurements using standard platinum resistance thermometers calibrated at the ITS-90 fixed points, precision resistance bridges (1 ma / 75 Hz) and low AC/DC difference standard resistors and precision digital absolute pressure gauges. Temperature and pressure measurements are traceable to CEM and DC resistance, AC/DC difference and voltage ratio are traceable to CEM, NPL and PTB, respectively. Both generators were run on CO 2 -free air obtained from oil-free compressors fitted with heat regenerated molecular sieve adsorption driers, and using high-purity deionized water of nominal resistivity 18 M -cm [4,5,6,7] MIKES MIKES used the MIKES Dew/Frost-point Generator (MDFG) as the dew-point temperature standard in this comparison [3]. The generator comprises three saturators: LRS2 (-80 to +15 ), LRS1 (-60 to +15 ) and HRS (0 to +90 ). In this comparison, LRS1 was 9

10 used at measurement points -50, -30 and -10. HRS was used for measuring the points +1 and +20. All three saturators operate on the same principle. Air flowing at a rate of typically less than 2 l/min is saturated with water vapour by a single pass through a precision saturator located in a temperature-controlled liquid bath. A pre-saturator ensures that the dew-point temperature of air entering a main saturator is slightly higher than the main saturator temperature. Thus, condensation takes place in the inlet heat exchanger tube of the main saturator. Saturation with respect to plane water or ice layer is completed by forcing air flow on water or ice surface in a saturation chamber connected to the outlet of the heat exchanger tube. Saturated air flows through an internally electropolished tubing to the hygrometer under calibration. The tubing is heated when needed to prevent any water condensation in it. The generated dew-point temperature is determined from the measured saturator temperature, the saturator pressure and the air pressure in the hygrometer under calibration. Being the primary realisation for dew-point temperature, MDFG provides traceability to the SI through traceable temperature and pressure measurements. The traceability of these measurements is established through calibrations at MIKES within its CMCs published at the BIPM website NIM A two-pressure generator constructed by NIM [10,11] was used for this comparison at the comparison values -50, -30, -10, +1 and +20. The two pressure generator involves saturating a continuous stream of carrier gas with water vapour at a known temperature and pressure. The uncertainty of the device is determined by the uncertainty of the temperature and pressure measurements. The two-pressure generator consists of the heat exchanger, the saturator and the test chamber. All pipe lines were constructed by the use of internally polished tubing. The carrier gas flow rate range is 0.1 L/min to 2 L/min. The dewpoint/frost-point range is -75 to +25. The temperature and pressure measurements are traceable to NIM s temperature and pressure standards NIST The NIST primary standards used in the K6 comparison were NIST-designed thermodynamic generators For the comparison values -10, +1, and +20 the NIST Hybrid Humidity Generator (HHG) [12] was used, and at -50 and -30 the NIST Low Frost-point Generator (LFPG) [13] was used. The HHG was operated using the two-pressure principle. When the HHG is operated this way, air flowing at a rate of typically 30 l/min is saturated with water vapour by a single pass through a temperature-controlled saturator. Saturation takes place with respect to the surface of water by vapour diffusion and mixing at a measured controlled pressure that is often elevated. The generated amount fraction (mole fraction) is determined using the temperature and pressure at the location immediately before the outlet of the saturator. The generated dew/frost point is calculated using the generated amount fraction and the pressure at the location of interest. The LFPG operates in a similar manner as the HHG, except that 1) the operating gas is nitrogen, 2) the flow rate through the saturator is 2 l/min, 3) the saturator contains ice, 4) the pressure difference Δp between the saturator and the location of interest is very small, and 5) the frost-point temperature at the location of interest is determined as the measured saturator temperature with a small correction for Δp. Also, the LFPG uses internally electropolished tubing in its construction to minimize adsorption/desorption on the surface of the tubing. The SI traceability of the realisation is in 10

11 terms of temperature and pressure. The traceability of the temperature measurement is through the calibration of the saturator s standard platinum resistance thermometer using ITS-90 fixed points maintained at NIST. The SPRT resistance measurements make use of standard resistors calibrated at NIST. The calibrations of the two pressure gauges are traceable to NIST pressure standards NMC Two humidity generators were used as references to compare with the traveling standards [14,15]. Both are Thunder Scientific products. The Model 4500 frost generator is based on two-pressure and two-temperature principle and uses nitrogen as working gas. It was used for the comparison at -50, -30 and -10 frost points. Nitrogen flow of 1.5 litres per minute was used for all measurement points. The Model 2500 humidity generator is based on twopressure principle and uses air as working gas. It was used for comparison at -10, +1 and +20 frost/dew points. Air flow of 10 to 20 litres per minute was used instead. The two transfer standards were connected to each generator in parallel with a leaking valve in between the generator and the transfer standards. Each standard had a flow approximately 0.5 litres per minute as given by the flow indicator of the standard and the leaking valve released the extra gas from the generator. The system accuracy affecting parameters of the two generators, such as pressure transducers and thermometers are regularly calibrated against reference standards maintained at NMC/A*STAR, Singapore NMIJ Two generators, frost-point generator (FPG) [16] and two-pressure generator (2PG) [17], both of which were designed by NMIJ, were used for the NMIJ primary humidity standards in the comparison. The FPG and 2PG were used for the comparison ranges of -50 to -10 and +1 to +20, respectively. The principle of the two generators is on the basis of the generation of saturated water vapour using ice or water maintained at constant temperature. Nitrogen and air were used as matrix gases for the FPG and 2PG, respectively. The flow rates of the gases passing through the FPG and 2PG were 3 L/min and 20 L/min, respectively. SUS316L stainless-steel tubes with an outer diameter of 6.35 mm (1/4 ) whose inner surfaces were electropolished were used to connect transfer standards to the FPG. The pressure and temperature of the saturated gases were measured using pressure gauges and platinum resistance thermometers, which are traceable to the International System of Units (SI) using calibrations services at NMIJ or at calibration laboratories accredited by IA Japan NPL The NPL primary standards used in the comparison were NPL-designed single-pressure generators [18,19]. For the comparison values -50, -30 and -10 the NPL Low Frost-point Generator (LFG) was used, and at +1 and 20 the Standard Humidity Generator (SHG2) was used. Both generators operate on the same principle: air flowing at a rate of typically 0.5 l/min to 1 l/min is saturated with water vapour by a single pass through a precision saturator located in a temperature-controlled liquid bath. Saturation takes place with respect to surfaces of water or (below 0 ) ice, by vapour diffusion and mixing, at measured controlled pressure close to atmospheric pressure. The LFG is specialised for low- 11

12 range operation by the use of internally electropolished tubing in its construction. The SHG2 is specialised for high-range operation by the use of an initial pre-saturation stage, and trace heating for condensation protection. The generated dew-point is defined by the measured temperature of the saturated flowing gas at the final stage within the saturator. The traceability of the realisation is in terms of temperature through the calibration of the saturator platinum resistance thermometer with traceability to the SI (ITS-90) via NPL temperature standards. Supporting electrical and pressure measurements are traceable through calibrations at NPL or at UKAS accredited laboratories in the UK VNIIFTRI The Russian national humidity standard comprises two humidity generators with working range from 5 to 90 and from -60 to +15 respectively. The humidity quantities of relative humidity, dew/frost-point temperature, mole fraction are disseminated. Ordinary hygrometers are traceable to the national primary standard in accordance with the state hierarchical chain for measuring means of gas humidity. The common working range (dew points from 5 to 15 ) allows comparison of the generators. The generators use the phase equilibrium method to generate humid gas defined in terms of dew/frost-point temperature from -79 to +90. The expanded uncertainty in dew/frost-point temperature is no more than VSL The VSL humidity generator used for the comparison was a dew-point generator of the circulating single-temperature, single-pressure type [22]. A pump recirculates air or nitrogen gas over two saturators immersed in temperature-controlled liquid baths in which the temperature is measured using platinum resistance thermometers, and this determines the realised value of dew point for the saturated gas. One or other saturator can be selected, depending on the temperature range within an overall envelope of 60 to +70. Circulation is controlled by means of a centrifugal impeller and switching valves to select the flow path. Traceability of dew point is provided by calibration of the thermometers by comparison against standards calibrated to ITS

13 2.2 COMPARISON SCHEME The comparison scheme is illustrated in Figure 2.1. It is shown in two geographical loops (Europe and rest-of-world) separated by pilot measurements. However the loops ware made in series, not concurrently, and therefore for purpose of analysis the comparison was functionally a single loop. Additional intermediate checks were also made (discussed further below in Section 4). VSL NIST MIKES NPL INTA NMIJ Loop 1 Loop 2 NPL NMC NIM INRIM VNIIFTRI Pilot Assistant Pilot Participants Figure 2.1 Scheme of the comparison in two geographical loops (Europe and rest-of world) made in series, separated by pilot measurements. Loops were sequential, not concurrent. The comparison can be linked to EUROMET.T-K6 through INRIM, INTA, MIKES, NPL and VSL. A link to APMP.T-K6 is available through NMIJ, NIM and NMC. NIST provides a link to SIM, and VNIIFTRI provides a link to COOMET. CCT bilateral comparisons will link to CCT-K6 via key comparison participants. 2.3 COMPARISON SCHEDULE The approximate measurement dates for the comparison are shown in Table 2.2. Each laboratory initially proposed an estimated duration of measurements and shipping (typically 8 weeks). Although many participants complied with these estimates, there were many and frequent disruptions to the schedule, mainly due to instrument breakdowns and repairs, and consequent need for extra measurement checks by the pilot and others. The outcome of these repairs is not believed to affect the instrument readings. More detailed discussion of this is in 13

14 Section 4. However the simple fact of extending the comparison due to multiple delays requires a particularly careful consideration of the impact of any possible long-term drift in the instruments. NIST obtained agreement to make two sets of measurements, against two different humidity generators, but afterwards agreed to submit results for just one of these to the comparison. The 2006 measurements by NMIJ were intended to provide an extra mid-comparison check of instrument stability. However these were not reported or used, due to a number of extra checks made by NPL and INTA. Table 2.2 Approximate measurement dates of the comparison. Measurement start Measurement finish NPL Jul-02 Jul-03 NMIJ Sep-03 Oct-03 VSL Oct-03 Dec-03 MIKES Jan-04 Mar-04 INTA Mar-05 Apr-05 INRIM May-05 Jul-05 NPL Sep-05 Feb-06 NMIJ Mar-06 Jul-06 NIST1 Nov-06 Dec-06 NIST2 Jan-07 Feb-07 NMC Mar-07 Apr-07 NIM Oct-07 Dec-07 NPL Feb -08 Mar 08 VNIIFTRI Dec-08 Mar 09 NPL Apr-09 May 09 14

15 3. COMPARISON METHOD 3.1 TRANSFER STANDARDS After some discussion and trials of several hygrometers, the participants agreed on the use of two transfer-standard condensation-principle dew-point hygrometers one Michell Series 4000 serial number , owned by NPL (identified as Hyg1) and one MBW model DP 3DSH III K-1806, serial number / 91527, owned by INTA (identified as Hyg2). Further details of the instruments are given in the protocol in Appendix 1. These individual instruments were selected because they had the resolution required for the comparison, and in both cases had an established history of reliability, minimal drift, and established operating conditions in terms of supplementary cooling, flow rate, short-term stability, etc. The hygrometers contained integral refrigeration units to provide reproducible supplementary cooling. No additional changes in the instruments were made for the comparison. No specific separate calibrations were made of key components (such as the mirror PRT). (Although this can in modern instruments provide extra confidence and scope for checking, it was considered sufficient to treat the instrument as a whole ). Towards the end of the comparison, transit data loggers were transported with the hygrometers to monitor both temperature and mechanical shock during transportation. No significant events were recorded. For the purpose of establishing instrument history, past calibration data for Hyg2 were supplied to the pilot by INTA in confidence, not revealed to other participants, and did not compromise the blindness of comparison measurements by INTA or other participants. 3.2 REPORTING Each laboratory realised and reported: 5 dew/frost points each dew/frost point separately reproduced 4 times each realisation measured using two travelling standards simultaneously, resulting in 40 individual transfer standard results (20 per transfer standard). Participants were instructed to re-form the condensate layer for every separate measurement. Participants realised and measured values within 0.5 of the comparison nominal values. Participants reported: applied dew/frost point from the participant standard generator measured values (both travelling standards simultaneously) difference (applied dew point minus measured dew point) for both travelling standards uncertainties associated with these (including short-term standard deviation of travelling standards) values are the results compared and analysed for the comparison. 15

16 Supporting information was reported, including pressure and flow rates of sample gas supplied to the hygrometers, coolant temperatures, and other relevant background information. All measurements were made at nominally (just above) atmospheric pressure and at flow rates of approximately 0.5 l/min through each instrument. The indicated value of dew point for the instrument was derived from the measured resistance of the mirror PRT after stabilisation of the instrument at each condition measured. This resistance value was converted to a nominal temperature indication by using the nominal PRT resistance-temperature characteristic in IEC ( ) (corresponding to EN 60751:2008), defined as follows: For temperature above 0 : and for temperature below 0 : R t = R 0 (1 + At + Bt 2 ) (1) where R t = R 0 [1 + At + Bt 2 + C(t-100)t 3 ] (2) t = temperature (ITS-90),, R t = resistance at temperature t, R 0 = nominal resistance of 100 Ω at 0 A = , B = , and C = Note that this is a deliberately arbitrary nominal temperature value for the purpose of comparison, with no absolute significance. Reporting templates are shown in the appendices to this report 3.3 IMPARTIALITY The impartiality ( blindness ) of the comparison was ensured by the pilot conserving the confidentiality of the data throughout, and no communication of results between partners was allowed during the comparison. This was true during instrument evaluation, during participant measurements, and during interim decisions about the comparison where needed. Any limited data shared during the comparison for the purpose of discussing concerns about the instruments (See Section 4 below) was strictly in coded terms, not absolute values. The use of a general function (IEC 60751) for conversion of resistance to temperature, together with the tendency of these instruments to have (stable) calibration offsets, both also contributed to the blindness of the measurements. For a period overlapping this comparison, some participants also took part in the corresponding RMO comparison EUROMET.T-K6. This involved INRIM, INTA, MIKES, NPL, and VSL. However there is no reason to suggest this affected the blindness of CCT-K6. 16

17 Since NPL, as the pilot, measured at more than one time during the comparison, the first full set of NPL results is used as the reported comparison data, while other NPL results are used for instrument drift assessment. 3.4 COMPLIANCE WITH OTHER REQUIREMENTS Other requirements of the CIPM Mutual Recognition Arrangement [25] are met in respect of the following points. The comparison participants have formally approved this report. Consistency has been ensured between this CCT key comparison and the several corresponding RMO and bilateral comparisons to date. Protocols or reports of corresponding comparisons to date have been reviewed by CCT/WG7 for consistency with this one, even though several of these comparisons, notably APMP.T-K6 and EUROMET.T-K6, were initiated or completed before CCT-K6 itself. Certain of the comparisons have covered more measurement points than CCT-K6. For these cases, it has been noted that linkages to CCT- K6 are possible at values measured in common between comparisons, but not at other values. 4 PERFORMANCE OF THE TRANSFER STANDARDS Stability of the transfer standards is critical to the uncertainty and validity of the comparison. This was particularly a concern for two reasons: the long duration of the comparison, and the several interventions made to diagnose and repair problems with the instruments. The performance of the transfer standards during the comparison was monitored in three ways: On receipt of the instruments, every participant was required to execute comparison measurements at a dew point of 20 as a first step, and to report the results immediately to the pilot. From this, the pilot assessed the consistency of agreement between the two transfer standards as an indication of safe transit, and confirmed to the participant to proceed with full measurements. Secondly, the between-instrument agreement at all measured values throughout the comparison gives an indication of consistent performance of the instruments. Thirdly, drift of the instruments over the whole period of the comparison was assessed from pilot measurements at the beginning and end of the comparison, and at certain times in between, especially at times of repair or other concern raised about the hygrometers. In addition, extra measurements by INTA of Hyg2 before, during and after the comparison gave further evidence of stability of this instrument. The three types of checks are detailed further below in sub-sections 4.2 to

18 4.1 PROBLEMS WITH THE TRANSFER STANDARDS The transfer standards represented the state of the art when the comparison started, and were selected for their history of stable and reliable operation. However, due to the age of the instruments, they suffered a large number of minor failures. There was also a suspected electrical measurement anomaly that was not an instrument failure. The repairs and other interruptions considerably delayed the comparison at many stages. However, no significant effect on the long-term characteristic of either instrument was observed (as detailed in Subsections 4.2 to 4.4 below). At every intervention, the pilot briefed participants and secured participant agreement to appropriate actions. The incidents are summarised in Table 4.1 to put into context the results of checks of instrument stable performance throughout the comparison. Overall there were more than 12 distinct breakdowns or problems with the instruments. Only a small number of these (queries over NMI and MIKES measurements, and head-heater problem at NPL) have any potential implications for the comparison outcome. The checks throughout the comparison assume extra significance because of the need to be able to demonstrate that none of the incidents affected comparison results. Where relevant, the events are marked as A to Q on the graphs in Section 4. It was not obvious how any of these instrument problems could have been avoided at the time. However the recent new generation of similar hygrometers is more reliable, and subsequent comparisons have benefited from that. 18

19 Table 4.1 List of problems with the transfer standards, and interventions made Date Event Problem Action Impact on comparison results Oct to Dec 2003 A After measurements completed, VSL expressed some concern about reproducibility of No action, but scrutiny of subsequent results at all participant labs Direct concern only for VSL data. Not observed to affect other results Jan to May 2004 April 2004 May 2004 Late 2004 Late 2004 Sep 2005 B C D E F G Hyg1 in low range. MIKES results (completed Jan-Mar and analysed up to May) appeared to show discrepancy between AC and DC resistance measurements of hygrometer PRT, of up to Hyg2 light source (part of optical detection of condensate formation and feedback) failed while at INRIM. Hyg2 refrigerator failed at INRIM Hyg2 failure of a compressor valve within refrigeration system Extensive investigation on identical model at NPL could not replicate the effect. Lesser effects (due to stray capacitance, inductance, selfheating) were measurable or calculable none significant at this level. Hygrometer component (measuring head) sent to INTA for replacement of light source, then returned to INRIM where gain of photodiode circuit was adjusted in situ. This element of the instrument is part of the control electronics, but does not determine the instrument reading. Complete replacement of refrigeration system needed. Repaired by manufacturer (necessarily travelling via INTA, as owner). Instrument then checked at INTA (against past calibration history, but retaining blindness of comparison). Instrument was then re-checked at NPL. Returned to manufacturer for further repair, Hyg1 film thickness control reinstated on front panel of instrument (from internal position) for improved usability. Instrument was then re-checked by pilot, alongside Hyg2. Mid-comparison measurements as planned, by pilot (NPL) MIKES accepted that the anomaly could not be reproduced, and chose which results to report. Participants were involved in discussion, but blindness of results was preserved throughout. Delay, but no impact on instrument characteristic. Delay, but checks did not show any departure from previous characteristic. Delay, but no impact on instrument characteristic. No impact on instrument characteristic No adverse findings 19

20 Date Event Problem Action Impact on comparison results May 2006 H Hyg2 photodiode failed while at NMIJ. August- Sep 2006 March 2007 March 2007 March- April 2007 May to August 2007 May to August 2007 Oct-Dec 2007 I J K L M N P Hyg2 failed to operate at NMC Hyg2 mirror check indicator flashing Hyg1 power supply cut out several times Query over sensitivity of instrument readings to coolant temperature Hyg1 display panel meter failed at NIM (display not used or reported in comparison) Jan 08 Q While at NPL, Hyg2 head heater control failed, leading to excessive heating of head. Measuring head sent to manufacturer agent in Japan for repair. NMIJ made extra measurements to confirm satisfactory operation after repair. The photodiode is part of the control electronics but does not determine the instrument reading. Additional check at NPL to confirm no change due to photodiode replacement Identified as failure of a power supply smoothing capacitor. On manufacturer s advice, remedied by removal of capacitor. False indication, solved by switching to standby and switching off, as needed Cause unsure. No action. Delay, but no impact on instrument characteristic. Delay, but no adverse findings Delay, but no impact on instrument characteristic. No impact on instrument characteristic. No impact on instrument characteristic. Instruments re-checked at NPL. Delay, but no impact on instrument characteristic. Additional checks made by NPL. NIM completed measurements by data-logging electronically as planned. On return to NPL, the loose connection to one pin of the display was repaired by the manufacturer. Head heater control was repaired in situ by the manufacturer. This was the first and only problem with risk of affecting key comparison results. Full set of repeat NPL measurements carried out after repair, to be sure no effect on results. Delay, but no adverse findings No impact on instrument characteristic. Delay, but no adverse findings. 20

21 4.2 CHECKS OF SAFE TRANSPORTATION Results of initial between-instrument consistency checks at a nominal dew point of 20 are shown in Figure 4.1. These served to inform the pilot as in indication of safe arrival at each participant lab (in some cases after remedying instrument problems). The results at 20 were sent to the pilot immediately for review. In every case results were judged sufficiently consistent by inspection. From Figure 4.1, retrospective analysis shows that: Although scatter at 20 varies from participant to participant, no lab or labs have significantly worse scatter or deviation than the typical. Instrument events (A to Q) are shown relative to the progress of the comparison. By inspection, no clear effects of breakdown or repair are visible at a dew point of 20. Further discussion is included in Section 8. Figure 4.1 Between-instrument initial consistency checks at a nominal dew point of +20, on arrival at each participant laboratory. Standard uncertainty of individual measurements ranged between 0.02 and 0.05 approximately. Data shown are differences (Hyg1 minus Hyg2). Event markers A, B, C etc. refer to Table

22 4.3 CONSISTENCY OF PERFORMANCE OF INSTRUMENTS THROUGHOUT THE COMPARISON The consistency of between-instrument differences at all measured values throughout the comparison gives a measure of consistent performance of the instruments throughout the comparison. Figure 4.2 shows the between-instrument differences summarised for all participant sets of measurements at all five nominal dew-point values measured. The difference data for NMIJ, VSL and NIM showed some untypical features. However, the consistent pattern of results among a majority of participants at beginning, middle and end of the comparison gives one form of assurance of the stability and consistency of the instrument pair. No clear progressive drift or trend is evident. If the difference were changing suddenly or progressively with time, this would imply a change in at least one instrument. This was not observed. Although the pilot reviewed difference values for the 20 point immediately, as a check of safe arrival, difference values at lower ranges showed some greater disparities (not analysed until later). I Figure 4.2 Graph showing consistency of the difference between the transfer standards as reported by the participants, four measurements (usually), at all five nominal values. Each data point is the difference between results of two hygrometers measuring simultaneously. NPL additional data (NPL 2, 3, 4, 5 and 6) were limited checks, rather than full sets of four measurements. Standard uncertainty of individual measurements ranged between 0.02 and 0.08 approximately. 22

23 By inspection, data from NMIJ, VSL and NIM appear to have more scatter than others. NMIJ and NIM data appear to show an unusual pattern between results in different parts of the range. Other isolated cases of scatter in individual measurements can be seen. 4.4 STABILITY OF EACH HYGROMETER Pilot measurements Measurements were made by the pilot NPL to monitor the stability of both travelling standards relative to the NPL standards. These checks were made as full sets of measurements at planned stages at beginning, middle and end of the comparison, plus some more limited checks at other stages. Figure 4.3 shows a graph of data for Hyg1 and figure 4.4 for Hyg2. The data are plotted as difference (pilot applied values minus instrument dew/frostpoint temperatures calculated from the resistance measurement results). Only the mean values are shown. For measurements in 2003, 2005, 2008 and 2009 these were means of full sets of reproduced measurements in most cases sets of four. At other dates, measurements were fewer (single results at each dew point). Event markers A to Q are also shown. Figure 4.3 Graph showing overview of pilot stability checks on Hyg1 hygrometer (lines joining the data to guide the eye). 23

24 Figure 4.4 Graph showing overview of pilot stability checks on Hyg2 transfer standard, (lines joining the data to guide the eye) INTA measurements of Hyg2 As owner of Hyg2, INTA performed measurements before and after the comparison, as well as their own comparison measurements, plus another measurement set after one of the instrument repairs. The results are shown in the graphs in Figure 4.5 and 4.6. Each data point in these figures represents the result of a single calibration; hence the uncertainties as shown by the error bars are somewhat larger than uncertainties elsewhere in this report which are for aggregated multiple results. 24

25 Figure 4.5 INTA data for overall calibration characteristic for Hyg2, years 2000 to Uncertainties shown are at coverage factor k=2. Figure 4.6 Graph showing INTA measurements for Hyg2 relative to INTA standard generator(s), with straight-line weighted fits to the data to indicate trend. Uncertainties shown are at coverage factor k=2. In general, the INTA data show much less scatter than the NPL data, but with larger uncertainties. This is considered in the analysis of drift that follows. 25

26 4.4.3 Drift estimates and discussion The data for each instrument were analysed to establish whether any significant secular drift should be taken into account in the comparison results and uncertainties. There are two possible cases: drift estimated to be not detectably significant, counted as zero drift with some assigned uncertainty or a significant rate of drift detected and estimated with some uncertainty, and a correction for drift applied to results where relevant. The pilot data for travelling standards, at several occasions spread through the whole comparison, were analysed for drift by evaluating a best-fit curve for data at each dew-point temperature. The fitting used an NPL-developed validated function in MSExcel, XLGENLINE V1.1 [26], to provide a weighted best-fit straight line, together with uncertainty in gradient taking into account both residuals and number of data. In each case a check was made to confirm that a higher order fit did not appear better (smaller residuals). The results are shown in Table 4.2, which gives the drift as gradient in degrees Celsius per year, and as a total drift for the whole comparison, and the uncertainty in both of these taking into account the measurement uncertainty together with the residuals from fitting. In addition, the outcomes of INTA drift checks on Hyg2 are also shown. The uncertainties for INTA values are larger than those or NPL, partly because of the larger uncertainty of the INTA references used, and partly due to fewer INTA data. Table 4.2 Estimates of hygrometer drift, per year, and for the total duration of the comparison, Hyg1 data from NPL, Hyg2 data from NPL with additional results from INTA. Standard Instrument Dew-point temperature () Gradient ( /yr) Standard uncertainty in gradient ( /yr) Total estimated drift (gradient x time interval) uncertainty in total drift () Hyg Hyg Hyg Hyg Hyg Hyg2 (NPL) Hyg2 (NPL) Hyg2 (NPL) Hyg2 (NPL) Hyg2 (NPL) Hyg2 (INTA) Hyg2 (INTA) Hyg2 (INTA) Hyg2 (INTA) Hyg2 (INTA) Drift can be approximated as zero where the best-fit gradient is not significantly different from zero when considered along with its k=2 uncertainty. According to this criterion, drift was found to be negligibly different from zero in the range below 0. However, for Hyg1 at 1 and 20, and for Hyg2 at 1, the criterion was not met. In these cases it is necessary 26

27 to decide whether to treat this as a sign of significant drift, and whether to make an allowance for this as a correction or additional uncertainty. An overview of the drift data is shown in the graphs in figures 4.7, 4.8, and 4.9. These show for Hyg1, Hyg2, and for the two together, estimated drift, and uncertainty in this, at all five comparison values. Figure 4.7 Estimated total drift of Hyg1 hygrometer during comparison based on pilot NPL measurements. Error bars show standard uncertainty of drift estimates. Figure 4.8 Estimated total drift of Hyg2 hygrometer during comparison based on pilot NPL measurements (squares) and INTA measurements (triangles). Error bars show standard uncertainty of drift estimates. 27

28 Figure 4.9 Summary graph of estimates of total drift of both hygrometers during comparison based on pilot NPL measurements (Hyg1 diamonds, Hyg2 squares) and INTA measurements (Hyg2 triangles). Error bars show standard uncertainty of drift estimates. Overall, the estimates of drift from the INTA measurements had larger uncertainties then those from the NPL measurements. However, INTA generally obtained particularly reproducible results (as can be seen from the low scatter in figures 4.5 and 4.6). In addition the relatively smooth trend found by INTA across the range (Figure 4.8) is more believable than the variations that would be suggested by the NPL results. The drift estimates and their uncertainties are summarised in Table 4.3. Drift can be considered negligible if the estimated amount of drift, relative to its uncertainty, is not significantly different from zero. This is the case for Hyg1 at -50 and -30, and for Hyg2 at all values except +1. In addition, all but one of these individual drift values are small approximately 0.01 or less (one is less than 0.02 ). The drift estimates needing special consideration are those for Hyg1 at and above -10 and for Hyg2 at +1. Each of these taken in isolation suggests possible drift of more than 0.02 which, if true, could be considered significant relative to the participant uncertainties. Impact on the KCRV, and on the agreement of individual results with this, could be up to half the magnitude of total drift (up to of order 0.01 ). For Hyg2 at +1, the two drift estimates are discrepant INTA results suggesting a minimal drift upwards of order ±0.022 (standard uncertainty) over the whole comparison, while the NPL estimated downwards drift is ±0.012 (standard uncertainty). Their k=2 uncertainties would overlap, however. In spite of having larger uncertainty, the INTA estimate of drift is credible, because of the small scatter, and because across the range any physically feasible drift is unlikely to take the overall form implied by the NPL +1 result. 28

29 For Hyg1, at 1 the NPL estimate of total drift is and at 20 the estimate is These both differ from zero by more than 2 standard uncertainties, and would both be considered significant levels of drift. However drift of this type seems implausible, because it does not fit known, physically feasible explanations it would be surprising to have little drift below 0 tending towards downward drift at 1 and upward drift at 20. In addition, the doubt about the NPL estimate of drift for Hyg2 at 1 may also cast doubt on the estimate for Hyg1 at the same value. In considering drift, the main component capable of drifting is the instrument PRT. For this, there are a small number of physically feasible mechanisms of drift. One would be drift in R 0 (resistance at 0 ) which would result in drift in one direction, for all measured values. Another would be change in PRT sensitivity, which would be reflected in a trend (slope) of data points in Figures 4.7 to 4.9. A third possibility would be loss of thermal contact of the PRT with the measurement head, which would result in a change likely to be more significant at the lowest values measured. The discrepant drift data do not clearly fit any of these patterns. Checks were also made to be sure the anomaly was not an inversion of results ( error versus correction ). One further consideration is whether the pilot reference (NPL generator) could have performed anomalously in the range, from -10 upwards. However the overall comparison results in Section 5 do not suggest this: NPL results generally show good agreement with the KCRV. Overall, what evidence there is for significant drift from -10 upwards is too conflicting to allow a clear conclusion one that would support the application of a correction for drift, and allow a value of such a correction to be confidently proposed. The potentially anomalous drift estimates cannot be rejected, because no cause is identified. Nor can any of the anomalies be treated as a single rogue value: each is the combination of multiple results. Therefore, instead, an additional uncertainty to allow for the doubt about drift has been added to the instrument-related uncertainty. It should be noted that the impact of any drift (and its uncertainty) in either instrument is partly mitigated by the derivation of the reported comparison results from the average of both instrument readings. According to this, the combined effect of the estimated drift is shown in Table 4.3 below. 29

30 Table 4.3 Table comparing outcomes of using estimates of drift, and uncertainty in this, based on either NPL data for Hyg2 or INTA data for that instrument. Estimated drift and its standard uncertainty, shown for Hyg1 and Hyg2 individually, and combined effect. Uncertainty values marked (*) are used in the subsequent analysis. Dew point Hyg1 (NPL estimates and standard uncertainty) Hyg2 (NPL and INTA estimates, and standard uncertainty) Combined estimates (drift of mean of Hyg1 and Hyg2, and combined standard uncertainty of mean) Proposed correction for drift Total proposed standard uncertainty in (zero) correction due to drift, enlarged for discrepancy Estimated Estimated Estimated u u Lab drift Lab drift mean drift u * * -10 NPL NPL * * NPL As above As above INTA * In summary, for the range of the comparison below -10 there is little evidence for significant drift. At and above -10 the evidence of drift is inconclusive. For the values where there is more doubt, a larger estimated uncertainty is assigned to allow for this, conservatively, by directly adding a component to the uncertainty so that when a coverage factor of 2 is applied, the entire bounds of the apparent discrepancies will be included in the interval. In the rest of the analysis that follows, the values from Table 4.3 marked with asterisks (*) are used as standard uncertainties due to instrument drift. Those for -50 and -30 are based on the pilot (NPL) data, which appear reliable; at other values the larger uncertainty estimate of the two (NPL or INTA) is used. Instrument drift does not contribute uncertainty to the individual laboratory results, but needs to be taken into account in the assignment of KCRV and in the calculations of equivalence. 30

31 5 PARTICIPANT RESULTS Participant data are reported for measurements using the instrument pair simultaneously, for four measurements separately reproduced at each of the five dew-point values. The full data sets are reported in the Appendices of this report. NPL and INTA both made extra measurements, due to providing several checks and drift estimates for the hygrometers. However just one set of measurements was considered for the comparison: for NPL the first complete set of measurements was used, and for INTA the main scheduled comparison measurements at March to April 2005 were used. For each participant, at each nominal dew point, the data were aggregated by taking a mean of pooled results for both instruments to provide a single result for comparison and calculation of KCRV. (The calculated mean of simultaneous readings of two instruments is the mid-point between the two instrument mean readings, and the combined uncertainty is the uncertainty in the value of that mid-point.) Following the same notation as used for EUROMET.T-K6 [2] already reported, the result R lab i at each dew point can be given as R labi ( RHyg 1 RHyg2) ( RHyg1 RHyg2), (3) 4 i i 1 where (R Hyg1 and R Hyg2 ) are the results of the two transfer standards, and where R Hyg n tdref tdind, (4) with t dref the laboratory reference value for the applied dew/frost-point temperature and t dind the dew/frost-point temperature indicated (calculated) from the PRT resistance. The uncertainty of the results are those reported by the participants taking into account both Type B estimates as routinely reported, plus type A estimates of components such as shortterm variation (standard deviation) in repeated readings during the comparison measurements. All participants have independent dew-point scale realisations, including independent traceability of supporting temperature measurements to national realisations of the International Temperature Scale (ITS-90). The uncertainties can therefore be expected to be uncorrelated between participants. Where within-laboratory correlations are known, participants have taken these into account in the values reported. Below, results for each participant at all dew points are plotted for Hyg1 in Figure 5.1, and for Hyg2 in Figure 5.2. Mean results R lab for the two instruments are shown for all participants at all measured values in Figure 5.3. Each set of results was initially reviewed by the pilot on receipt, to check for anomalies in case of misreporting or other causes. In only one case, results as received appeared unusual, and that participant was notified of a possible anomaly at certain measured values (but not the magnitude or sign of the apparent discrepancy as specified in Guideline CIPM MRA D- 05) [27]. After making additional checks, the participant did not propose any change to reported results. 31

32 In some cases participants observed supercooled water on hygrometer mirrors at the nominal -10 point, and applied conversions to obtain equivalent frost point, using formulae for saturation vapour pressure curve of water. This is not expected to affect the validity of those results. Figure 5.1 Mean reported values of applied condition minus measured value for Hyg1, shown in participant time sequence. Connecting lines between data are shown to guide the eye. Figure 5.2 Mean reported values of applied condition minus measured value for Hyg2, shown in participant time sequence. Connecting lines between data are shown to guide the eye. 32

33 Figure 5.3 Results of entire comparison shown as reported values for mean (mid-point) of Hyg1 and Hyg2 results combined (R lab ), grouped by dew-point value (data points staggered in x-direction for visibility). Error bars show participant reported standard uncertainties (k=1). As shown here, uncertainty allowance for instrument drift is not included. 33

34 6 BILATERAL EQUIVALENCE Bilateral equivalences at each dew point were calculated from differences D ij between participants i and j, where Dij = R R, (5) labi lab j The bilateral degree of equivalence (DoE) is determined as (D ij, U ij ) = (D ij, ku(d ij )), (6) where the coverage factor k=2 provides a coverage probability of 95 % for sufficiently large effective number of degrees of freedom of u(d ij ). [28]. In this case, u(d ij ) is given by u 2 (D ij ) = u 2 (R lab i ) + u 2 (R lab j ) + u 2 drift, (7) where u 2 drift is the uncertainty in the comparison due to drift of both hygrometers at a given dew point value, with drift having been assigned an expectation value of value of zero as in Section 4. For simplicity here, u 2 drift, is assigned a single generalised value at each dew point, irrespective of whether participants measured in immediate succession or separated in time. The DoE was calculated for each pair of participants at each nominal measurement point. The results are summarised in tables 6.1 to 6.5. In a small number of cases, where participants assigned a coverage factor k greater than 2, due to a low effective number of degrees of freedom of an uncertainty estimate, the larger coverage factor is used to obtain the 95 % coverage interval for equivalences. 34

35 Table 6.1 Degree of equivalence between the participants of CCT-K6 at the frost-point temperature -50. DoE is expressed as (D ij, U ij ) in degrees Celsius. Instrument drift uncertainty is included in the uncertainty shown. -50 NPL NMIJ VSL MIKES INTA INRIM NIST NMC NIM VNIIFTRI NPL NMIJ VSL MIKES INTA INRIM NIST NMC NIM VNIIFTRI Table 6.2 Degree of equivalence between the participants of CCT-K6 at the frost-point temperature -30. DoE is expressed as (D ij, U ij ) in degrees Celsius. Instrument drift uncertainty is included in the uncertainty shown. -30 NPL 0 NMIJ VSL MIKES INTA INRIM NIST NMC NIM VNIIFTRI NPL NMIJ VSL MIKES INTA INRIM NIST NMC NIM VNIIFTRI

36 Table 6.3 Degree of equivalence between the participants of CCT-K6 at the frost-point temperature -10. DoE is expressed as (D ij, U ij ) in degrees Celsius. Instrument drift uncertainty is included in the uncertainty shown. -10 NPL NMIJ VSL MIKES INTA INRIM NIST NMC NIM VNIIFTRI NPL NMIJ VSL MIKES INTA INRIM NIST NMC NIM VNIIFTRI Table 6.4 Degree of equivalence between the participants of CCT-K6 at the dew-point temperature +1. DoE is expressed as (D ij, U ij ) in degrees Celsius. Instrument drift uncertainty is included in the uncertainty shown. +1 NPL NMIJ VSL MIKES INTA INRIM NIST NMC NIM VNIIFTRI NPL NMIJ VSL MIKES INTA INRIM NIST NMC NIM VNIIFTRI

37 Table 6.5 Degree of equivalence between the participants of CCT-K6 at the dew-point temperature +20. DoE is expressed as (D ij, U ij ) in degrees Celsius. Instrument drift uncertainty is included in the uncertainty shown. +20 NPL NMIJ VSL MIKES INTA INRIM NIST NMC NIM VNIIFTRI NPL NMIJ VSL MIKES INTA INRIM NIST NMC NIM VNIIFTRI

38 7 KEY COMPARISON REFERENCE VALUES (KCRV) 7.1 CALCULATION OF KCRV In comparisons of dew/frost-point temperature scales, comparison reference values have no absolute significance outside the comparison. However the availability of a comparison reference value is essential to the use of comparison results for review of CMC claims. In this comparison and other corresponding RMO comparisons, a reference value is calculated for each nominal value of dew point, treating them as separate data populations for this purpose. For each nominal dew point value, a key comparison reference value (KCRV) [28] was calculated as the weighted mean, y, of results x i + +x N u 2 (x N ) y = x 1 u 2 (x 1 ) 1 u 2 (x 1 ) + +1 u 2 (x N ), (8) this method of calculation having been agreed by CCT Working Group 6. For comparison, values of arithmetic mean and median were also calculated. The uncertainty in weighted mean due to dispersion was calculated from [28] 1 = u 2 (y) u 2 (x 1 ) u 2 (x N ) (9) After collection of participant results, but before circulation of Draft A, results of NIST at -50 and -30 were identified as outliers. NIST confirmed that they recognised an inconsistency with expected results at these values (relative to additional data for another NIST generator). A multimeter used for comparison measurements was identified as a possible cause of the inconsistency. However it was not possible to make further measurement checks, because of a lengthy breakdown of the NIST generator refrigeration system, and eventual updating of the LFPG facility. Calculations of weighted mean were made both with and without the outlying results. Values of arithmetic mean and median were also calculated. These are summarised in Table 7.1. At +20 and -50 the median is somewhat lower than the general trend, which suggests consideration of whether a skew or outlier is present in the data. As well as the uncertainty in weighted mean due to dispersion, an additional uncertainty in KCRV was included for the drift of the hygrometers (expectation value zero, with standard uncertainty as given in Section 4.4). The uncertainties are summarised in Table

39 Table 7.1 Values of weighted mean, arithmetic mean and median at each nominal dew point, estimated from all results, and also after exclusion of outlying results at -50 and -30 (in italics) Dew-point value Weighted mean Arithmetic mean Median Table 7.2 Standard uncertainties due to variance of weighted mean, combined effect of hygrometer drift, and their quadrature sum, for each dew point, estimated from all results, and also after exclusion of outlying results at -50 and -30 (in italics) Dewpoint value Standard uncertainty of weighted mean Standard uncertainty due to combined hygrometer drift Quadrature sum CHI-SQUARED TEST A chi-squared test [28] was carried out on the results with and without the identified NIST outliers, as a measure of the consistency of the data and uncertainties. Based on the participant reported uncertainties alone, the test fails. Repeating the test with correct inclusion of uncertainty allowance for instrument drift, the test succeeds for results above - 10, but initially fails for the full set of participant results at -10, -30 and -50. Discrepant results can be identified using the criterion [28]: R labi 2 2 R 2 u ( R ) u ( R ) (10) KCRV labi Inspection of the data and the test for discrepancy show that several participant results deviate from the KCRV by between 2 and 3 standard uncertainties, u(r lab i ). Only 1 participant (NIST) had any result with greater deviation relative to uncertainty by more KCRV 45

40 than 4 standard uncertainties at -50. Removal of the NIST results from the KCRV and chi-squared test at -50 and -30 allows the remaining results to pass the test at those points. At -10 the chi-squared test fails, but is passed if the VSL result is removed. The results that cause the chi-squared test to fail at -30 and -10 are not dramatic outliers. The decision whether to exclude marginally-outlying data is also a matter of considering the impact on the KCRV. Inclusion of the NIST outliers at -50 and -30 influences the KCRV significantly at those points reducing it by and respectively. In contrast, exclusion of the VSL -10 result would affect the KCRV at that value by only 0.004, which can be considered not a significant influence. Accordingly, the NIST results -50 and -30 have been excluded from the calculation of KCRV, but the VSL result at -10 has been included. Overall, the chi-squared test is only narrowly passed, suggesting that the uncertainties are probably not generally overestimated. 7.3 COMPARISON RESULTS PRESENTED IN TERMS OF KCRV The results of all laboratories relative to the KCRV are shown in Table 7.3 below, and in figures 7.1 to 7.5, with uncertainties as shown in Table 7.4. The error bars in the graphs show a combination of the participant reported error with the uncertainty allowance due to hygrometer drift, at coverage probability of 95 %, using a coverage factor k=2 in most cases except where participants assigned a higher coverage factor. Table 7.3 Participant result, R lab, minus KCRV (weighted mean), in degrees Celsius NPL NMIJ VSL MIKES INTA INRIM NIST NMC NIM VNIIFTRI

41 Table 7.4 Uncertainty in difference between participant result and KCRV, at 95 % coverage probability, in degrees Celsius NPL NMIJ VSL MIKES INTA INRIM NIST NMC NIM VNIIFTRI Figure 7.1 between participant results and KCRV, at the nominal frost-point temperature -50. Error bars show the expanded uncertainties at coverage probability of 95 %. Estimated uncertainty due to instrument drift is included. NPL Final values are shown but not included in evaluation of KCRV. Figure 7.2 between participant results and KCRV, at the nominal frost-point temperature -30. Error bars show the expanded uncertainties at coverage probability of 95 %. Estimated uncertainty due to instrument drift is included. NPL Final values are shown but not included in evaluation of KCRV. 47

42 Figure 7.3 between participant results and KCRV, at the nominal frost-point temperature -10. Error bars show the expanded uncertainties at coverage probability of 95 %. Estimated uncertainty due to instrument drift is included. NPL Final values are shown but not included in evaluation of KCRV. Figure 7.4 between participant results and KCRV, at the nominal dew-point temperature +1. Error bars show the expanded uncertainties at coverage probability of 95 %. Estimated uncertainty due to instrument drift is included. NPL Final values are shown but not included in evaluation of KCRV. Figure 7.5 between participant results and KCRV, at the nominal dew-point temperature +20. Error bars show the expanded uncertainties at coverage probability of 95 %. Estimated uncertainty due to instrument drift is included. NPL Final values are shown but not included in evaluation of KCRV. 48

43 8 DISCUSSION The comparison results mainly demonstrate consistency with the key comparison reference value, within the estimated uncertainties. A small proportion (8 %) of results are not within two standard uncertainties of the KCRV. Only one participant (NIST) had any result deviating by more than three standard uncertainties. It was understood, from the outset of the comparison, that the reproducibility of the available state-of-the-art travelling standard hygrometers would be barely sufficient to provide a stringent test of equivalence of standards at the level of uncertainty claimed by the participants. However, the travelling standards were at least expected to be reliable, given their history, and so the failures of components were unexpected. Fortunately, there is no evidence that the instrument values were affected by any of the breakdowns or repairs. Future dew-point key comparisons can benefit (or already have done) from a new generation of transfer standard hygrometers that have been improved in several respects. The long duration of more than six years for participant measurements was far from ideal. It raises two concerns: potential drift of the travelling standards; and the question of linkage to regional comparisons that are separated in time by several years. The drift of the travelling standard hygrometers has been considered in detail in Section 4. They appear to have been sufficiently stable, but some of the evidence in the drift assessment is conflicting. Because of that ambiguity, the uncertainty allowed for instrument drift is slightly conservative in this analysis. Less conservative treatment of drift was also considered, but would not significantly change the broad outcome of which participant results were consistent with the KCRV within the uncertainties. In general, the results of participants with the smaller uncertainties were the most affected by the inclusion and the extent of drift-related uncertainty. In evaluating the magnitude and significance of travelling standard drift, and the impact on KCRV and bilateral equivalences, various other measures have been considered, not detailed fully here. These included: the character of the within-laboratory deviations; the pilot initial and final values as a simple measure of drift; the impact of potential drift on agreement of those measuring early and late in the comparison; and others. None of these considerations contradicted the chosen approaches to the analysis of the comparison. Between-instrument consistency (as discussed in Section 4.3) was also reviewed in case it might provide further insight about drift. While the degree of consistency appears worse for some of the participants whose agreement with the KCRV is weaker, there was no obvious further conclusion to be drawn from this. Two participants, MIKES and NIST, reported doubts about the electrical measurements they had made during the comparison, using multimeters. For VSL, some results had unexpectedly high scatter which also could possibly have arisen from electrical measurement problems although the cause was not identified. Linkages between key comparisons are generally of interest and the KCRV is relevant to these. However, in this field of measurement, the value of the KCRV is only a comparison parameter, with no absolute significance in the SI (unlike, for example, a fixed-point temperature in thermometry). To enable valid linkage it is important that RMO and CC 49

44 comparisons have consistent protocols. For dew-point key comparisons, this consistency is generally being ensured through review by the CCT Working Group on Key Comparisons. The consideration of linkages between this key comparison and those of regional metrology organisations is a matter for further discussion, with the primary responsibility being with the coordinators of other comparisons requiring links to this one. 9. CONCLUSION This comparison was lengthy and challenging, because of the numerous difficulties encountered with the travelling standard hygrometers. However, careful study of the results does not reveal any shift in hygrometer performance attributable to instrument failures or repairs. It is therefore believed that the instrument problems did not compromise the results of the comparison. Instrument stability over the long period of the comparison was assessed, and drift was concluded to be low, although with some inconsistency in the evidence at and above -10. The assessment of drift at these values was not conclusive enough to merit correction for drift, but an additional uncertainty allowance was made, because of the associated doubt. With the provisos above, a key comparison reference value was evaluated, together with bilateral equivalences between pairs of participants, and between participants and the KCRV. Mainly good agreement was demonstrated between participants. The comparison was effective despite instrument difficulties. However, comparison uncertainties would potentially have been reduced if more reliable travelling instruments had been available at that time, if comparison measurements had been more quickly completed, and if evidence about drift had been unambiguous. 10. ACKNOWLEDGEMENTS In addition to the listed authors, Leena Uusipaikka is acknowledged for contributions to the measurements at MIKES. REFERENCES [1] Wang Li et al 2007 Final report of APMP.T-K6 (original name APMP-IC-1-97): Comparison of humidity measurements using a dew point meter as a transfer standard Metrologia [2] Heinonen, Martti, et al. "Investigation of the equivalence of national dew-point temperature realizations in the 50 C to +20 C range." International Journal of Thermophysics (2012): [3] Heinonen M, ISHM 2002 Taiwan. Papers from the 4th International Symposium on Humidity and Moisture, ITRI 2002, pp [4] Benyon R and Huang P, A comparison of INTA and NIST humidity standard generators, Papers and Abstracts from the Third International Symposium on Humidity and Moisture, National Physical Laboratory, UK, 1998,

45 [5] Mackrodt P, Benyon R and Scholz G, State-of-the-art calibration of high-range chilled-mirror hygrometers and their use in the intercomparison of humidity standard generators, Papers and Abstracts from the Third International Symposium on Humidity and Moisture, National Physical Laboratory, UK, 1998, [6] Actis A, Bell SA, Benyon R, Cretinon B, De Groot M, Heinonen M, Scholz G and Steiner A, The use of a humid air generator as a reference method for measuring humidity. Papers and Abstracts from the Third International Symposium on Humidity and Moisture, National Physical Laboratory, UK, 1998, [7] Benyon R and Vicente T, Consistency of the National Realization of Dew-Point Temperature Using Standard Humidity Generators, International Journal of Thermophysics, September 2012, 33, Issue 8-9, pp [8] Actis A, Banfo M, Fernicola V C, Galleano R and Merlo S, Metrological performances of the IMGC two-temperature primary humidity generator for the temperature range -15 to 90, Papers and Abstracts from the Third International Symposium on Humidity and Moisture, National Physical Laboratory, UK, 1998, 2-9 [9] Actis A, Fernicola V C and Banfo M, Characterization of the IMGC frost point generator in the temperature range -75 to 0, Proceedings of TEMPMEKO 1999, pp [10] Hong Yi, Zhan-yuan Li and Chang-qing Ren, Evaluation of uncertainty on standard device of two pressure humidity generator, Metrology & Meas. Tech. 26 (z1) (2006) [11] Hong Yi et al, A Hybrid Low Frost Point Generator, ACTA Metrologica Sinica, 29(5A) (2008) [12] Scace Gregory E, Meyer Christopher W, Miller William W and Hodges Joseph T, An overview of the NIST Hybrid Humidity Generator, 5th International Symposium on Humidity and Moisture ISHM 2006 Brazil, May 02 05, 2006 Rio de Janeiro, Brazil [13] Scace G E and Hodges J T. "Uncertainty of the NIST low frost-point humidity generator." Proceedings of TEMPMEKO [14] Wang Li and Victor Tan, "Facilities for Humidity Calibration and Their Characterisations"; Proc. Tempmeko'96, 6th International Symposium on Temperature and Thermal Measurement in Industry and Science, Torino, September 10-12, 1996, pp [15] Wang Li and Victor Tan, Characterization of PSB Frost Point Generator by Using a High Precision Dew Point Meter ; Proc. Tempmeko 2001, 8th International Symposium on Temperature and Thermal Measurement in Industry and Science, Berlin, June 19-21, 2001, pp [16] Ochi N, Takahashi C and Kitano H, Uncertainty of a new NMIJ frost-point generator, Papers from the 4th International Symposium on Humidity and Moisture 2 (2002) [17] Takahashi C, Kitano H, Ochi N and Yokota T, Uncertainty in dew-point hygrometer calibration by a two-pressure two-temperature humidity generator, Papers from the 4th International Symposium on Humidity and Moisture (2002) [18] Stevens M and Bell S A, The NPL Standard Humidity Generator - An analysis of uncertainty by validation of individual component performance Measurement Science and Technology 3 (1992) pp [19] Stevens Mark, The new NPL low frost-point Generator in Proceedings of TEMPMEKO '99, 7th International Symposium on Temperature and Thermal Measurements in Industry and Science, (Delft, 1999), pp [20] The Russian national standard of gases humidity and traceability system of humidity measurements. Book of Abstracts. Vol A, Joint International Symposium on Temperature, Humidity, Moisture in Industry and Science 31 May- 4 June 2010.Portoroz. Slovenia, p

46 [21] de Groot M J, Papers and Abstracts from the Third International Symposium on Humidity and Moisture, National Physical Laboratory, UK, 1998, pp [22] Nielsen J and de Groot M J, Revision and uncertainty evaluation of a primary dew point generator, Metrologia 41, n.3 pp (2004) [23] Bosma R and Peruzzi A, Development of a dew-point generator for gases other than air and nitrogen and pressures up to 6 MPa, Int.J.Thermophys. September Issue 8-9, pp [24] Bosma R, Mutter D and Peruzzi A, Validation of a dew-point generator for pressures up to 6 MPa using nitrogen and air, Metrologia 49 (2012), [25] CIPM, MRA. "Mutual recognition of national measurement standards and of calibration and measurement certificates issued by national metrology institutes." Comité International des Poids et Measures (1999). [26] Smith, Ian, Software for determining polynomial calibration functions by generalised least squares: user manual NPL REPORT MS 11, December 2010 (Teddington, UK: National Physical Laboratory) [27] Measurement comparisons in the CIPM MRA, CIPM MRA-D-05, Version 1.4 June 2013, (accessed 4 October 2013) [28] Cox M, The evaluation of key comparison data, Metrologia 39 (2002)

47 APPENDIX 1: TECHNICAL PROTOCOL The following pages show the technical protocol for CCT-K6, together with its Appendix 5 listing conditions to be reported as background information. The names of participating institutes and the measurement sequence shown are those at the time the comparison started. In addition (not shown here) the protocol included appendices giving: participant contact details; reporting template to document safe receipt of instruments and checklist of associated items; IEC (EN 60751) formula for nominal resistance-temperature relationship (see Section 3.2); and MS Excel reporting templates for comparison results and uncertainty. 53

48 International Committee for Weights and Measures (CIPM) Consultative Committee for Thermometry (CCT) CCT-K6 Key Comparison of Humidity Standards Dew/Frost-Point Temperature 50 to +20 Technical protocol 54

49 1. INTRODUCTION 1.1 Under the Mutual Recognition Arrangement (MRA) 3 the metrological equivalence of national measurement standards will be determined by a set of key comparisons chosen and organized by the Consultative Committees of the CIPM working closely with the Regional Metrology Organizations (RMOs). 1.2 At its 20th meeting in April 2000, the Consultative Committee for Thermometry, CCT, considered a Key Comparison on humidity as imperative for the related laboratories. It was decided that the Working Group on Humidity Measurements (WG 6) be called upon to draft a key comparison protocol. 1.3 To date, the Working Group consists of 13 members, the National Institute of Standards and Technology, USA (NIST, Chair), the National Physical Laboratory, UK (NPL), the National Metrology Institute of Japan (NMIJ-AIST), the Korea Research Institute of Standards and Science, Republic of Korea (KRISS), the Standards, Productivity and Innovation Board, Singapore (SPRING Singapore), the National Research Centre for Certified Reference Materials, China (NRCCRM) the Consiglio Nazionale delle Ricerche - Istituto di Metrologia "G. Colonnetti", Italy (IMGC), the Bureau National de Métrologie - Cetiat, France (BNM-Cetiat), the D.I. Mendeleyev Institute of Metrology, Russia (VNIIM), Nederlands Meetinstituut Van Swinden Laboratorium, the Netherlands (NMi-VSL), the Physikalisch- Technische Bundesanstalt, Germany (PTB), the Measurement Standards Laboratory, New Zealand (MSL) and the Ulusal Metroloji Enstitüsü, Turkey (UME). 1.4 This technical protocol has been drawn up by the working group described above, and in consultation with the nominated participants listed in Section The procedures outlined in this document cover the technical procedure to be followed during measurement of the transfer standards. The procedure, which follows the guidelines established by the BIPM 4, is based on current best practice in the use of dew/frost-point hygrometers and takes account of the experience gained from the regional comparisons and that of the working group over the years. 1.6 This comparison is aimed at establishing the degree of equivalence between realisations of local scales of dew/frost-point temperature of humid gas, in the range -50 to +20, among the participating national measurement institutes. 3 MRA, Mutual Recognition Arrangement, BIPM, T.J. Quinn, "Guidelines for key comparisons carried out by Consultative Committees," Appendix F to the MRA, BIPM, Paris. 55

50 2. ORGANIZATION 2.1 Participants A list of participants representing RMOs of SIM, APMP, EUROMET and COOMET has been approved by the CCT. Details of mailing and electronic addresses are given in Appendix 1. The nominated institutes5* are: Centre for Metrology and Accreditation (MIKES) Finland DI Mendeleyev Institute of Metrology (VNIIM) Russia Instituto Nacional de Técnica Aeroespacial (CEM/INTA) Spain Istituto di Metrologia G. Colonnetti (IMGC) Italy National Institute for Standards and Technology (NIST) USA National Metrology Centre (SPRING) Singapore National Metrology Institute of Japan (NMIJ) Japan National Physical Laboratory (NPL) UK National Research Centre For Certified Reference Materials (NRCCRM) China Nederlands Meetinsituut (NMi) Netherlands NPL is the Pilot of the key comparison, taking main responsibility for running the key comparison NMIJ is assigned as Assistant Pilot in verifying the data analysis for the draft A. The assistant will also perform additional measurements as required By their declared intention to participate in this key comparison, the laboratories accept the general instructions and the technical protocol written down in this document and commit themselves to follow strictly the procedures of this protocol as well as the version of the "Guidelines for Key Comparisons" in effect at the time of the initiation of the Key Comparison Once the protocol and list of participants have been approved, no change to the protocol or list of participants may be made without prior agreement of all participants All participants must be able to submit an uncertainty budget of their humidity standard generators. 2.2 Method of comparison The key comparison is a comparison of the realisations of the scale of dew-point temperature at the participating national institutes The comparison will be made by calibration of a pair of travelling transfer standards. Each transfer standard will independently measure dew/frost-point temperature of a * At the time of planning the comparison, several participant institutes were known by previous names: INRIM was IMGC, NIM was NRC-CRM, NMC was SPRING, and VSL was NMi. 56

51 sample of moist gas (air or nitrogen) produced by a participant's standard generator using the same measuring process Simultaneous measurements using a pair of standards gives information about the within-laboratory consistency of the measurements, the reproducibility of the instrument performance, and continuous feedback about the successful transport of the instruments without any major shift in performance The comparison will take the form of a closed circulation in two consecutive loops. There is one pair of hygrometers, which are at all times measured simultaneously. Measurements will start in the pilot laboratory. The assistant will perform the measurements next. The other participants in Loop 1 will then make comparison measurements at the dew/frost-point temperatures required. After loop 1, the travelling standards will return to the Pilot for checks mid-way through the comparison, and optionally to the Assistant Pilot to repeat these checks. The comparison will then proceed through loop 2, and the last participant will then return the transfer standards to the pilot to carry out final measurements to monitor drift. The assistant will also carry out repeat measurements following those of the pilot. The sequence would therefore be*: NPL NMIJ MIKES NMi IMGC INTA NPL ( NMIJ optional) NIST SPRING NRCCRM VNIIM NPL NMIJ. Allowing between 6 and 8 weeks per set of measurements (and additional time for shipping), this set of measurements will take up to 32 months The proposed circulation scheme for travelling standards for CCT dew-point key comparison is illustrated below*. * At the time of planning the comparison, several participant institutes were known by previous names: INRIM was IMGC, NIM was NRC-CRM, NMC was SPRING, and VSL was NMi 57

52 MIKES NIST NMi NMIJ SPRING IMGC Loop 1 Loop 2 NPL NRCCRM INTA VNIIM Pilot Assistant Pilot Participants All results are to be communicated directly to the pilot within six weeks of the completion of the measurements by a laboratory Each laboratory has estimated a time for measurement and transportation. If for some reason, the measurement facility is not ready or customs clearance takes too much time in a country, the participating laboratory must contact the pilot laboratory immediately. Exclusion of a participant's results from the report may occur if the results are not available in time to prepare the draft report In case of serious difficulty with customs, or other delays which might over-run the time period of the ATA Carnet or temporary import licence, the pilot may request the instruments be returned to NPL, or the sequence of participation be changed to the most practical arrangement. 2.3 Handling of artefacts The artefacts should be examined immediately upon receipt at the laboratory. All participants are expected to follow all instructions in the operator's manual provided by the instrument manufacturers for proper unpacking, subsequent packing and shipping to the next participant. During packing and unpacking, all participants should check the contents with the packing list including the operator's manual The transfer standards should only be handled by authorized persons and stored in such a way as to prevent damage. 58

53 2.3.3 During operation of the transfer standards, if there is any unusual occurrence, e.g., loss of heating or cooling control, the pilot laboratory should be notified immediately before proceeding. 2.4 Transport of artefacts The transportation process begins when the artefact leaves the sending laboratory and does not end until it reaches the destination laboratory. All participants should follow the following general guidelines: (1) Plan the shipment well in advance. The recipient should be aware of any customs issues in their country that would delay the testing schedule. The shipping laboratory must be aware of any national regulations covering the travelling standard to be exported; (2) Mark the shipping container "FRAGILE SCIENTIFIC INSTRUMENTS" TO BE OPENED ONLY BY LABORATORY STAFF and with arrows showing "THIS WAY UP"; attach tip and shock indicators if such devices are available; (3) Determine the best way to ship the travelling standard to the next participant; (4) Obtain the recipient's exact shipping address. If possible, have it shipped directly to the laboratory; (5) Coordinate the shipping schedule with the recipient. The sending laboratory should provide the recipient with the carrier, the exact travel mode, and the estimated time of arrival; (6) Instruct the recipient to confirm receipt and condition upon arrival to the sender and the pilot. A form for reporting on the receipt of the travelling standards is shown in Appendix Each transfer standard is supplied with its shipping container, which is sufficiently robust to ensure safe transportation The artefacts will be accompanied by a suitable customs ATA Carnet or temporary import bond (TIB) (as deemed most appropriate by the pilot laboratory) and documentation uniquely identifying the item. Care should be taken with the timing of the ATA Carnet, which only lasts for one year Shipping Costs Each laboratory is responsible for the cost of shipping to the next participant including any customs charges and insurance. The insurance should be sufficient to cover the costs of the travelling standards and any damages that could occur. 59

54 2.6. Timetable Activity Start Month Provisional date Submission of a revised technical protocol to Participants for unanimous approval Re-submitted Feb 2003 Submission of revised protocol to CCT/WG7 for Mar 2003 approval Travelling standards characterized by the pilot 2002-early 2003 Pilot s fist set of key comparison measurements Month 1-2 May-Jun 2003 according to the protocol Travelling standards sent to assistant pilot and Month 3-12 July 2003 successive participants for measurements Travelling standards re-measured by pilot(s) at Month May-Jun 2004 mid-point Completion of measurements Month 28 approx Mid 2005 Draft A ready Month 32 approx Late 2005 Deadline for comments on draft A Month 35 approx Early 2006 Draft B ready and submitted to CCT Month 40 approx Mid DESCRIPTION OF THE TRANSFER STANDARDS 3.1. Artefacts Two travelling standards selected for the key comparison are state-of-the-art, commercially available chilled-mirror type of dew-point hygrometers. They have proven to be robust with known performance characteristics such as repeatability and transportability Details of travelling standards: Travelling Standard #1 Travelling Standard #2 (Figure 1) (Figure 2) Model: Michell S4000 MBW DP 3DSH III K-1806 Serial Number: / Size (in Packing case): 60 cm x 65 cm x 105 cm 60 cm x 55 cm x 60 cm Weight (in Packing case): 55 kg 55 kg Manufacturer: Michell Instruments, UK MBW Elektronik AG, Switzerland Owner: NPL, UK CEM-INTA, Spain Electrical supply: 240 V 50 Hz 240 V 50 Hz Electrical connection:2 UK 3-pin plugs 1 European 2-pin plug Power: 1500 W total 500 W (100 W plus 1400W approx) Approximate value for insurance and customs declaration 60

55 Figure 1. Travelling standard 1, Michell S4000 Figure 2. Travelling standard 2, MBW 61

56 4. MEASUREMENT INSTRUCTIONS 4.1 Measurement process All participants should refer to the operating manuals for instructions and precautions for using the travelling standards. Participants may perform any initial checks of the operation of the hygrometers that would be performed for a normal calibration. In the case of an unexpected instrument failure at a participant institute, the pilot institute should be informed in order to revise the time schedule, if necessary, as early as possible Sample gas generated by a participant's standard generator, is introduced into the inlet of a travelling standard hygrometer through a stainless steel tube terminating with a 6 mm Swagelok fitting for the Michell and ¼ inch Swagelok fitting for the MBW. The instruments should be connected in parallel. For dew points near ambient temperature (e.g. +20 ) normal precautions (heating of pipework) should be used to protect against condensation in sample lines A total of five dew-point temperatures humidity levels are used for the comparison at nominal values of +20 and +1 and frost-point temperatures at nominal values of 10, -30 and 50. The value of +1 nominally represents 0, while avoiding any complication due to phase change between water and ice At 10, the applied condition should be generated with respect to ice in the saturator of a single-pressure generator. Where a two-pressure generator is used, the phase in the saturator at elevated pressure will be according to local procedure, to result in a water vapour pressure corresponding to saturation over ice at 10 at the pressure of the travelling standards. Participants should report the applied condition in terms of frost-point temperature. The phase of condensate apparent on the mirrors of the travelling standards should also be reported. At -30 and 50, all data will be assumed to be with respect to ice unless otherwise reported Measurements should be made in rising order of dew/frost point The condensate should be cleared and re-formed for each value or repetition of dew/frost point The values of dew/frost point applied to the travelling standards should be within ±0.5 of the five agreed nominal values for the comparison, and ideally closer than this. Deviations greater than this may increase the uncertainty in the comparison, for a particular result The conditions for operation of the travelling standard Michell S4000: (1) Set the Michell S4000 to Standby and Manual. (2) Clean the mirror surface using cotton tips with distilled or de-ionised water. This may be preceded by initial cleaning with alcohol if necessary (3) Set the coolant temperature to 30 above the generated frost point, for 62

57 measuring frost-point temperatures of 50 and 30. The cooling must be switched off for all other points. (4) Set the indicated flow rate of sample gas at approximately 0.5 litres per minute. (5) Monitor the cooling as detailed in below (6) After the cooling has stabilised for 20 minutes press the Initiate button to initiate an optical balance cycle. Set the Optical Balance Control to the centre position. When the balance cycle is complete switch from Standby to Operate. (7) A cable is connected between the 100 cut-out socket of the measurement head and the temperature measurement socket of the monitor. This is to allow monitoring of the operating temperature of the back of the Peltier element. When connected in this way, the hygrometer display and the analogue both indicate the Peltier temperature (not the dew/frost-point). The analogue is nominally 10 mv per degree Celsius, with 0 volts equal to 0. (8) The dew/frost-point indication of the hygrometer is measured directly from the hygrometer PRT resistance, using the supplied cable (See below, 4.2 Data collection) The conditions for operation of the travelling standard MBW K1806: IMPORTANT: Due to the nature of the mechanical configuration of the head and endoscope, the head should only be opened once the following instructions have been read and fully understood. Failure to observe these may result in severe damage to the transfer standard. The unit is provided with a blank outer cover for transport. (1) Ensure the service jack is inserted and set the MBW to Standby, with the automatic mirror check set to Off. (2) Set the mode switch to Cooler Temp and set the temperature to 30. Wait for the mirror temperature indication to reach at least 25. (3) In order to gain access to the mirror for cleaning, carefully follow the instructions below: -Starting from the closed position at 6 o clock, turn the bayonet socket release of the endoscope 90 anti-clockwise until the notch on the endoscope guide tube and the rotatable arm are aligned. - Gently withdraw the endoscope. (Special care must be observed to ensure no bending moment is applied to the endoscope at any time). - Turn the outer blue alloy head cover anti-clockwise using the knurled surface (not the endoscope guide tube). - Remove the grey head cover along the guide pin. - The mirror is now ready for cleaning. (4) Clean the mirror surface using cotton buds with distilled or de-ionised water. This may be preceded by initial cleaning with alcohol if necessary. (5) In order to replace the measurement head cover, carefully follow the instructions below: - Replace the grey head cover, aligning the hole with the guide pin. - Replace the outer blue alloy head cover by turning it clockwise until the endoscope guide tube is at the 12 o clock position. This should only be finger-tight. Correct alignment can be checked by observing the light emitted from the head, ensuring that the full circular cross-section of the tube (reduced by the internal o-ring) is visible. - Slowly insert the endoscope until a slight resistance is felt as the tip touches the o- ring. 63

58 If the endoscope cannot be inserted smoothly, remove and slightly adjust the position of the outer head cover. Once the endoscope passes the o-ring, align the notch on the endoscope with that of the guide tube, at the 6 o clock position. Once the endoscope has been fully inserted, starting from the open position at 3 o clock with the two notches aligned, turn the bayonet socket release of the endoscope 90 clockwise. IMPORTANT: Never turn the head with the endoscope inserted. Special care must be observed to ensure no bending moment is applied to the endoscope at any time. (6) Do not adjust the light intensity potentiometer. If the mirror check fails consult the pilot before proceeding. (7) Control the flow rate of sample gas at approximately 30 litres per hour. (8) Ensure the cooler mode switch is set to Cooler Temp and set the temperature to 30 higher than a nominal frost-point temperature to be measured. This is a setting (nominal value) and the actual value achieved will not be exactly the set value. (Note: For the measurement points below 0, ensure that the hygrometer has been adequately purged by the sample gas to a dew-point temperature below the nominal cooler temperature setting, before setting the cooler temperature.) (9) Monitor the cooling as detailed in below. (10) After the cooling has stabilised for 20 minutes, ensure the service jack is inserted and activate the Mirror Check. When this is complete, take the MBW off Standby. (11) During the mirror check, when the mirror is heated, the pointer should lie just to the left of the boundary between the red and green backgrounds. (12) Once the mirror check has finalised (the illumination of the meter is extinguished), the service jack is replaced by the measurement cable provided and the potentiometer attached to the cable adjusted until the indication of the hygrometer display is nominally +60. This cable provides direct electrical access to the PRT in the mirror. IMPORTANT: Once the measurement cable is connected to the hygrometer, the Manual Mirror Check function must not be activated and the automatic function must remain in the Off position. (13) Hygrometer head heater is to remain off during all measurements Each measurement should be conducted with the instruments measuring in parallel and nominally simultaneously. Each dew/frost-point temperature should be separately repeated (reproduced) four times, to reduce the effect of any irreproducibility of the travelling standards Participants should avoid lengthy additional measurements, except those necessary to give confidence in the results of this comparison The transfer standards used in this comparison must not be modified, adjusted, or used for any purpose other than described in this document, nor given to any party other than the participants in the comparison The Pilot will make an assessment of any drift in the travelling standards during the comparison, based on measurements at the Pilot laboratory at the beginning, middle and end of the comparison period, and in case of doubt using optional extra 64

59 measurements at the Assistant pilot laboratory. If significant drift is found, then this will be taken into account in the final overall analysis of the comparison If unacceptable performance or failure of a travelling standard is detected, the Pilot will propose a course of action, subject to agreement of the participants Data collection In the travelling standards, a 100-ohm platinum resistance thermometer (PRT) is embedded beneath the surface of the chilled-mirror to measure the dew/frost-point temperature. The current input to the PRT should be nominally 1 ma. The resistance of the PRT should be measured using a calibrated multi-meter or a resistance bridge, and then converted to a corresponding nominal dew/frost-point temperature using the reference function of IEC as shown in Appendix 3. This reference function should be used to convert resistance to (arbitrary nominal) temperature At each measured value, the mean and standard deviation of multiple readings of the resistance of the PRT should be monitored. Participants may apply their own criteria of stability for acceptance of measurements. When hygrometer is in equilibrium with the gas sample, the standard deviation of a set of 10 resistance readings, taken over a period of 10 to 20 minutes, is likely to be no more than ohms or approximately As a supporting measurement, the coolant/peltier temperature in the travelling standards should be monitored. The mean and standard deviation a set of 10 readings, taken over the same period as the frost point measurements should be reported. For the Michell, an analogue voltage signal from the USER I/O is monitored while the 100 cut-out socket of the measurement head is connected to the temperature measurement socket of the monitor. The is nominally 10 mv per, with 0 V corresponding to 0. This temperature will not be the same as the set temperature but is an indication of the temperature of the heat exchanger behind the Peltier cooler. For the MBW the analogue voltage from the Cooler Temp plug is monitored (cables supplied). In this case also, the is 10 mv per, with 0 V= Values reported for dew/frost-point temperatures produced by a participant's standard generator should be the value applied to the instruments, after any allowances for pressure and temperature differences between the point of realisation (laboratory standard generator) and the point of use (travelling standards) The data reported for the pair of instruments should be for simultaneous or nearsimultaneous measurement of the same applied condition. 5. REPORTING OF MEASUREMENT RESULTS 5.1 Participants must report their measurement results of four repeated experiments, within six weeks of completing their measurements. 65

60 5.2 The pilot should accumulate data continually and should analyse the results for possible anomalies in the travelling standard. If problems arise, the pilot should consult with the participant that submitted the data as soon as possible, and certainly before the distribution of Draft A of the Report of the comparison. 5.3 The parameter to be compared between laboratories in CCT-K6 is the mean difference found between the laboratory standard generator and the travelling standards. Note that the values of dew-point temperature reported for the travelling standards are arbitrary values calculated from the measured resistance. The travelling standards are used simply as comparators. 5.4 Participants should report results to the pilot in terms of dew/frost-point temperature. The main measurement results comprise: values of dew/frost-point applied to the travelling standards, and associated standard uncertainty values measured using both travelling standards simultaneously (and their associated standard uncertainties derived from standard deviation of the set of readings) values of difference between applied dew/frost point and measured dew/frost point. A provisional template for reporting results is shown in Appendix 4, and can be made available to participants in electronic form as an Excel spreadsheet. Use of this format, including calculations of means and differences, allows participants to see clearly the values and uncertainties of the parameters they are submitting for comparison. 5.5 From the data measured by each participant, results will be analysed in terms of differences between applied and measured dew points. In each case, the difference will be taken between the applied (realised) value and the mean (mid-point) between the two hygrometer values. 5.6 In addition, the difference between the two hygrometer readings on all occasions will be analysed and will serve as a check of consistency. 5.7 The participants should report the conditions of realisation and measurement, as background information to support the main results. These conditions should include, where relevant, pressure and temperature in saturator, pressure difference between saturator and travelling standards, measurement traceability, frequency of AC (or DC) resistance measurement, travelling standard coolant measurements, and other items. A provisional checklist for reporting conditions of measurement is shown in Appendix Participants should provide a general description of the operation of their dew/frostpoint apparatus. 5.9 Participants should also provide an example plot of equilibrium condition (resistance versus time) at a nominal frost-point temperature of -30, over a suggested period of at least one hour. 66

61 5.10 Any information obtained relating to the use of any results obtained by a participant during the course of the comparison shall be sent only to the pilot laboratory and as quickly as possible. The pilot laboratory will be responsible for coordinating how the information should be disseminated to other participants. No communication whatsoever regarding any details of the comparison other than the general conditions described in this protocol shall occur between any of the participants or any party external to the comparison without the written consent of the pilot laboratory. The pilot laboratory will in turn seek permission of all the participants. This is to ensure that no bias from whatever accidental means can occur. These constraints on communication apply until the circulation of Draft A of the report of the comparisons If a participant significantly delays reporting of results to the Pilot, then a deadline will be agreed among the participants. If that deadline is not met, then inclusion of those results in the comparison report will not be guaranteed. 6. UNCERTAINTY OF MEASUREMENT 6.1 The uncertainty of the key comparison results will be derived from some or all of: o the quoted uncertainty of the dew/frost-point realisation (applied dew/frost point) including any uncertainties due to pressure drop or other influences acting between the point of realisation and the point of use (travelling standards). o the estimated uncertainty relating to the short-term stability of the travelling standards at the time of measurement o the estimated uncertainty due to any drift of a travelling standard over the period of the comparison (estimated by the pilot) o the estimated uncertainty in mean values due to dispersion of repeated results (reflecting the combined reproducibility of generator and travelling standards) o the estimated uncertainty due to the resolution of the travelling standards (if found to be significant) o the estimated uncertainty due to non-linearity of the travelling standard in any case where measurements are significantly away from the agreed nominal value o the estimated covariance between applied (generator) and measured (travelling standard) values of dew/frost-point (if found to be significant) and o any other components of uncertainty that are thought to be significant 6.2 Participants are required to submit detailed analyses of uncertainty for their dewpoint standards. Uncertainty analyses should be according to the approach given in the ISO Guide to the Expression of Uncertainty of Measurement. A list of the all significant components of the uncertainty budget should be evaluated, and should support the quoted uncertainties. Evaluations should be given at a level of one standard uncertainty. Type B estimates of uncertainty may be regarded as having infinite degrees of freedom, or an alternative estimate of the number of degrees of freedom may be made following the methods in the ISO Guide. A provisional template for documentation of uncertainties is shown in Appendix 6, and can be made available to participants in electronic form as an Excel spreadsheet. Individual institutes may add to the template any additional uncertainties they consider relevant. 67

62 6.3 The pilot laboratory will collect draft uncertainty budgets as background information to the uncertainties quoted by participants for the comparison measurements. The pilot will review the uncertainty budgets for consistency among participants. 6.4 The uncertainty budget stated by the participating laboratory should be referenced to an internal report and/or a published article. 7. DETERMINATION OF THE KEY COMPARISON REFERENCE VALUE 7.1 The s of the key comparison are expected to be: Results of individual participants for comparison of the hygrometers against their dew point reference in terms of mean values for each hygrometer at each measured value, estimated standard uncertainty of each mean result and of comparison process if necessary. Estimates of bilateral equivalence between every pair of participants at each measured dew point A key comparison reference value (KCRV) for each nominal value of dew/frost point in the comparison. The KCRV might be calculated as the arithmetic mean of all valid results, or a weighted mean. Estimates of equivalence of each participant to the KCRV. This might be expressed in terms of the Degree of Equivalence (DOE) given as a difference and its uncertainty ( ±U), in. 7.2 Values of the above will be reached by an appropriate method proposed by the Pilot, subject to confirmation by the Assistant Pilot and agreement of all participants and confirmation by CCT Working Groups 6 (Humidity) and 7 (Key Comparisons). 7.3 In the field of dew-point standards, the KCRV does not have any absolute significance with respect to an SI unit. It is calculated only for purposes such as the presentation and inter-relation of key comparison data for the MRA. 7.4 The Pilot will make an assessment of any drift in the travelling standards during the comparison. The assessment will be based on initial measurements by the Pilot and Assistant Pilot, together with measurements when the instruments return to the pilot mid-way through the comparison (repeated by the Assistant Pilot if necessary), and final measurement by both Pilot and Assistant Pilot. If significant drift of one or both travelling standards is observed, then this will be taken into account in the final overall analysis of the comparison. This may be by assigning a time-dependent value to the KCRV, or by other suitable method so that estimates of equivalence can be meaningfully calculated between results taken at different times. 7.5 If a travelling standard fails or performs poorly during the comparison, the Pilot and Assistant Pilot will propose a course of action, subject to agreement of the participants. If the results of one of the travelling standards (from some or all participants) are deemed un-usable, and if measurements cannot be re-attempted, the 68

63 KCRV and estimates of equivalence may be based on the results of satisfactory measurements using only one travelling standard. 69

64 APPENDIX 5. PROVISIONAL CHECKLIST FOR REPORTING OF CONDITIONS OF MEASUREMENT The following is guidance for reporting of the background information to the key comparison measurements. This information is likely to be of secondary importance, but will become relevant if there should be any need to resolve anomalies which might appear in the results. Reporting of the main results is outlined in Appendix 4. The report should include the following information: A full description of the humidity generator used in the comparison and the traceability of the realisation to the SI, including o The gas used (air or nitrogen) o The connection between the hygrometer and the standard - tubing material and dimensions o Description of cleaning the mirror o Value of flow rate set for each hygrometer o Frequency of AC (or DC) resistance measurement of hygrometer PRTs, and current used. o Description of any problems with the hygrometers, or with the participant s generator system. For each separate repetition of each measurement point: o Applied reference value(s) (generated dew-point temperature determined by the generator, after any correction for pressure drop to the point of use) o Standard deviation of the applied value(s) o Standard uncertainty of the applied value(s) o Values indicated by the travelling standard hygrometers o Standard deviation of the hygrometer indicated values o between the applied (generator) value and the measured (hygrometer) values o Combined standard uncertainty of the difference o Date when the measurements were carried out o Hygrometer coolant temperature settings o Measured temperatures of MBW coolant and Michell Peltier o Temperature and pressure in saturator of generator o Pressure difference between the hygrometer and the generator, and value of correction(s) applied to compensate for this, if any. o Environmental conditions (temperature, humidity, pressure) o Number of recorded values o Stabilisation time o Time interval taken to record the values o Raw data in units of resistance for the PRT measurements, and in units of voltage for the analogue s 70

65 APPENDIX 2: RESULTS REPORTED BY THE PARTICIPANTS The participant reported results are shown on the following pages in the form of extracts pasted from the MS Excel reporting template for the comparisons. In general, each result is a standard uncertainty reported at sufficiently high number of effective degrees of freedom that a coverage factor k=2 can be used to give a coverage probability of 95 %. In cases where a lower number of effective degrees of freedom necessitated a larger coverage factor, to give a 95 % coverage probability, participants were asked to calculate and report that, and where relevant details are given in Appendix 3. 71

66 NPL REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: -50 Lab name NPL Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 etc. Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) I column between 2 means (each the mean of 4 results) (consistency indicator - all labs should hope to get the same value for this) Average (mid-point) of 2 means (each the mean of 4 results) ** # (aggregated result - parameter to be compared between institutes) Uncertainty in average ** # (uncertainty in the parameter being compared) REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: -30 Lab name NPL Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 etc. Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) I column between 2 means (each the mean of 4 results) (consistency indicator - all labs should hope to get the same value for this) Average (mid-point) of 2 means (each the mean of 4 results) ** # (aggregated result - parameter to be compared between institutes) Uncertainty in average ** # (uncertainty in the parameter being compared) 72

67 REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: -10 Lab name NPL Results Hygrometer 1(Michell) Applied dew point () Hygrometer 2 (MBW) (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 etc. Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) I column between 2 means (each the mean of 4 results) (consistency indicator - all labs should hope to get the same value for this) Average (mid-point) of 2 means (each the mean of 4 results) ** # (aggregated result - parameter to be compared between institutes) Uncertainty in average ** # (uncertainty in the parameter being compared) REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: 1 Lab name NPL Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 etc. Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) I column between 2 means (each the mean of 4 results) (consistency indicator - all labs should hope to get the same value for this) Average (mid-point) of 2 means (each the mean of 4 results) ** # (aggregated result - parameter to be compared between institutes) Uncertainty in average ** # (uncertainty in the parameter being compared) 73

68 REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: 20 Lab name NPL Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 etc. Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) I column between 2 means (each the mean of 4 results) (consistency indicator - all labs should hope to get the same value for this) Average (mid-point) of 2 means (each the mean of 4 results) ** # (aggregated result - parameter to be compared between institutes) Uncertainty in average ** # (uncertainty in the parameter being compared) 74

69 NMIJ REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: -50 Lab name NMIJ Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Number of data Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Hyg 1 Hyg 2 Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 etc. Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) (average weighted proportional to number of data) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) Effective degree of freedom of uncertainty of mean dew point difference I column between 2 means (each the mean of 4 results) (consistency indicator - all labs should hope to get the same value for this) Average (mid-point) of 2 means (each the mean of 4 results) ** (aggregated result - parameter to be compared between institutes) Uncertainty in average ** (uncertainty in the parameter being compared) Effective degree of freedom of uncertainty in average ** Uncertainty in difference between 2 means Effective degree of freedom of uncertainty in difference between 2 means REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: -30 Lab name NMIJ Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Number of data Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Hyg 1 Hyg 2 Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 etc. Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) (average weighted proportional to number of data) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) Effective degree of freedom of uncertainty of mean dew point difference I column between 2 means (each the mean of 4 results) (consistency indicator - all labs should hope to get the same value for this) Average (mid-point) of 2 means (each the mean of 4 results) ** (aggregated result - parameter to be compared between institutes) Uncertainty in average ** (uncertainty in the parameter being compared) 75

70 REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: -10 Lab name NMIJ Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Number of data Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Hyg 1 Hyg 2 Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 etc. Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) (average weighted proportional to number of data) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) Effective degree of freedom of uncertainty of mean dew point difference I column between 2 means (each the mean of 4 results) (consistency indicator - all labs should hope to get the same value for this) Average (mid-point) of 2 means (each the mean of 4 results) ** (aggregated result - parameter to be compared between institutes) Uncertainty in average ** (uncertainty in the parameter being compared) REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: 1 Lab name NMIJ Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 etc. Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Uncertainty of resistance measurement Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) I column between 2 means (each the mean of 4 results) (consistency indicator - all labs should hope to get the same value for this) Average (mid-point) of 2 means (each the mean of 4 results) ** (aggregated result - parameter to be compared between institutes) Uncertainty in average ** # (uncertainty in the parameter being compared) 76

71 REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: 20 Lab name NMIJ Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 etc. Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Uncertainty of resistance measurement Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) I column between 2 means (each the mean of 4 results) (consistency indicator - all labs should hope to get the same value for this) Average (mid-point) of 2 means (each the mean of 4 results) ** # (aggregated result - parameter to be compared between institutes) Uncertainty in average ** # (uncertainty in the parameter being compared) 77

72 VSL REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: -50 Lab name NMi-VSL Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1(Michell) Hygrometer 2 (MBW) Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 Meas 2 Meas 3 Meas 4 Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) 95% I column between 2 means (each the mean of 4 results) (consistency indicator - all labs should hope to get the same value for this) Average (mid-point) of 2 means (each the mean of 4 results) ** (aggregated result - parameter to be compared between institutes) Uncertainty in average ** 95% (uncertainty in the parameter being compared) REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: -30 Lab name NMi-VSL Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1(Michell) Hygrometer 2 (MBW) Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 Meas 2 Meas 3 Meas 4 Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) 95% I column between 2 means (each the mean of 4 results) (consistency indicator - all labs should hope to get the same value for this) Average (mid-point) of 2 means (each the mean of 4 results) ** (aggregated result - parameter to be compared between institutes) Uncertainty in average ** 95% (uncertainty in the parameter being compared) 78

73 REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: -10 Lab name NMi-VSL Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1(Michell) Hygrometer 2 (MBW) Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 Meas 2 Meas 3 Meas 4 Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) 95% I column between 2 means (each the mean of 4 results) (consistency indicator - all labs should hope to get the same value for this) Average (mid-point) of 2 means (each the mean of 4 results) ** (aggregated result - parameter to be compared between institutes) Uncertainty in average ** 95% (uncertainty in the parameter being compared) REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: 1 Lab name NMi-VSL Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1(Michell) Hygrometer 2 (MBW) Meas 1 Meas 2 Meas 3 Meas 4 Meas 5 Meas 1 Meas 2 Meas 3 Meas 4 Meas 5 Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) 95% I column between 2 means (each the mean of 4 results) (consistency indicator - all labs should hope to get the same value for this) Average (mid-point) of 2 means (each the mean of 4 results) ** (aggregated result - parameter to be compared between institutes) Uncertainty in average ** 95% (uncertainty in the parameter being compared) 79

74 REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: 20 Lab name NMi-VSL Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1(Michell) Hygrometer 2 (MBW) Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 Meas 2 Meas 3 Meas 4 Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) 95% I column between 2 means (each the mean of 4 results) (consistency indicator - all labs should hope to get the same value for this) Average (mid-point) of 2 means (each the mean of 4 results) ** (aggregated result - parameter to be compared between institutes) Uncertainty in average ** 95% (uncertainty in the parameter being compared) 80

75 MIKES REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: -50 Lab name MIKES Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 etc. Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] (Std uncert due to resolution of travelling standard [if needed]) Uncertainty of resistance measurement Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] 9.70E E E E E E E E-07 Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) I column between 2 means (each the mean of 4 results) (consistency indicator - all labs should hope to get the same value for this) Average (mid-point) of 2 means (each the mean of 4 results) ** 0.09 (aggregated result - parameter to be compared between institutes) Standard uncertainty in average ** (uncertainty in the parameter being compared) REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: -30 Lab name MIKES Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 etc. Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] (Std uncert due to resolution of travelling standard [if needed]) Uncertainty of resistance measurement Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] -1.86E E E E E E E E-07 Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) I column between 2 means (each the mean of 4 results) (consistency indicator - all labs should hope to get the same value for this) Average (mid-point) of 2 means (each the mean of 4 results) ** 0.11 (aggregated result - parameter to be compared between institutes) Standard uncertainty in average ** (uncertainty in the parameter being compared) 81

76 REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: -10 Lab name MIKES Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 etc. Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] (Std uncert due to resolution of travelling standard [if needed]) Uncertainty of resistance measurement Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] 1.87E E E E E E E E-06 Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) I column between 2 means (each the mean of 4 results) (consistency indicator - all labs should hope to get the same value for this) Average (mid-point) of 2 means (each the mean of 4 results) ** 0.13 (aggregated result - parameter to be compared between institutes) Standard uncertainty in average ** (uncertainty in the parameter being compared) REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: 1 Lab name MIKES Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 etc. Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] (Std uncert due to resolution of travelling standard [if needed]) Uncertainty of resistance measurement Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] 1.92E E E E E E E E-07 Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) I column between 2 means (each the mean of 4 results) (consistency indicator - all labs should hope to get the same value for this) Average (mid-point) of 2 means (each the mean of 4 results) ** 0.13 (aggregated result - parameter to be compared between institutes) Standard uncertainty in average ** (uncertainty in the parameter being compared) 82

77 REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: 20 Lab name MIKES Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 etc. Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] (Std uncert due to resolution of travelling standard [if needed]) Uncertainty of resistance measurement Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] -1.30E E E E E E E E-07 Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) I column between 2 means (each the mean of 4 results) (consistency indicator - all labs should hope to get the same value for this) Average (mid-point) of 2 means (each the mean of 4 results) ** 0.14 (aggregated result - parameter to be compared between institutes) Standard uncertainty in average ** (uncertainty in the parameter being compared) 83

78 INTA REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON CCT/K6 Nominal value: -50 Lab name INTA Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 Meas 2 Meas 3 Meas 4 Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point temperature [if needed] Combination of these standard uncertainties in quadrature (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) between 2 means (each the mean of 4 results) Average (mid-point) of 2 means (each the mean of 4 results) ** Uncertainty in average ** I column (consistency indicator - all labs should hope to get the same value for this) (aggregated result - parameter to be compared between institutes) (uncertainty in the parameter being compared) REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON CCT/K6 Nominal value: -30 Lab name INTA Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 Meas 2 Meas 3 Meas 4 Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point temperature [if needed] Combination of these standard uncertainties in quadrature (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) between 2 means (each the mean of 4 results) Average (mid-point) of 2 means (each the mean of 4 results) ** Uncertainty in average ** I column (consistency indicator - all labs should hope to get the same value for this) (aggregated result - parameter to be compared between institutes) (uncertainty in the parameter being compared) 84

79 REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON CCT/K6 Nominal value: -10 Lab name INTA Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 Meas 2 Meas 3 Meas 4 Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point temperature [if needed] Combination of these standard uncertainties in quadrature (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) between 2 means (each the mean of 4 results) Average (mid-point) of 2 means (each the mean of 4 results) ** Uncertainty in average ** I column (consistency indicator - all labs should hope to get the same value for this) (aggregated result - parameter to be compared between institutes) (uncertainty in the parameter being compared) REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON CCT/K6 Nominal value: 1 Lab name INTA Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 Meas 2 Meas 3 Meas 4 Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point temperature [if needed] Combination of these standard uncertainties in quadrature (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) between 2 means (each the mean of 4 results) Average (mid-point) of 2 means (each the mean of 4 results) ** Uncertainty in average ** I column (consistency indicator - all labs should hope to get the same value for this) (aggregated result - parameter to be compared between institutes) (uncertainty in the parameter being compared) 85

80 REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON CCT/K6 Nominal value: 20 Lab name INTA Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 Meas 2 Meas 3 Meas 4 Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point temperature [if needed] Combination of these standard uncertainties in quadrature (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) between 2 means (each the mean of 4 results) Average (mid-point) of 2 means (each the mean of 4 results) ** Uncertainty in average ** I column (consistency indicator - all labs should hope to get the same value for this) (aggregated result - parameter to be compared between institutes) (uncertainty in the parameter being compared) 86

81 INRIM REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: -50 Lab name INRiM Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 Meas 2 Meas 3 Meas 4 Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) between 2 means (each the mean of 4 results) Average (mid-point) of 2 means (each the mean of 4 results) ** Uncertainty in average ** I column (consistency indicator - all labs should hope to get the same value for this) 0.06 (aggregated result - parameter to be compared between institutes) (uncertainty in the parameter being compared) REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: -30 Lab name INRiM Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 etc. Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) between 2 means (each the mean of 4 results) Average (mid-point) of 2 means (each the mean of 4 results) ** Uncertainty in average ** I column (consistency indicator - all labs should hope to get the same value for this) 0.12 (aggregated result - parameter to be compared between institutes) (uncertainty in the parameter being compared) 87

82 REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: -10 Lab name INRiM Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 etc. Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) between 2 means (each the mean of 4 results) Average (mid-point) of 2 means (each the mean of 4 results) ** Uncertainty in average ** I column (consistency indicator - all labs should hope to get the same value for this) 0.14 (aggregated result - parameter to be compared between institutes) (uncertainty in the parameter being compared) REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: 1 Lab name INRiM Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 etc. Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) between 2 means (each the mean of 4 results) Average (mid-point) of 2 means (each the mean of 4 results) ** Uncertainty in average ** I column (consistency indicator - all labs should hope to get the same value for this) 0.12 (aggregated result - parameter to be compared between institutes) (uncertainty in the parameter being compared) 88

83 REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: 20 Lab name INRiM Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 etc. Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) between 2 means (each the mean of 4 results) Average (mid-point) of 2 means (each the mean of 4 results) ** Uncertainty in average ** I column (consistency indicator - all labs should hope to get the same value for this) 0.16 (aggregated result - parameter to be compared between institutes) (uncertainty in the parameter being compared) 89

84 NIST REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: Lab name NIST Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 etc. Standard uncertainty of applied condition Type B uncertainty documented in paper Standard deviation of the calculated dew / frost-point temperature (Deg. C) Number of observations Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) The value reported here is the standard uncertainty obtained by combining the type A value in the boxes immediately above with the uncertainty of the humidity generator. between 2 means (each the mean of 4 results) Average (mid-point) of 2 means (each the mean of 4 results) ** Uncertainty in average ** I column (consistency indicator - all labs should hope to get the same value for this) (aggregated result - parameter to be compared between institutes) (uncertainty in the parameter being compared) REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: Lab name NIST Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 etc. Standard uncertainty of applied condition Type B uncertainty documented in paper Standard deviation of the calculated dew / frost-point temperature (Deg. C) Number of observations Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) The value reported here is the standard uncertainty obtained by combining the type A value in the boxes immediately above with the uncertainty of the humidity generator. between 2 means (each the mean of 4 results) Average (mid-point) of 2 means (each the mean of 4 results) ** Uncertainty in average ** I column (consistency indicator - all labs should hope to get the same value for this) (aggregated result - parameter to be compared between institutes) (uncertainty in the parameter being compared) 90

85 REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: Lab name NIST Results Notes: The condensate for all of the -10C points is dew. The applied dew point of Deg. C corresponds to a frost point temperature of nominally Deg. C. A correction of +.005C is applied to the temperatures reported as Deg. C. Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 etc. Standard uncertainty of applied condition Standard deviation of the calculated dew / frost-point temperature (Deg. C) Number of observations Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) The value reported here is the standard uncertainty obtained by combining the type A value in the boxes immediately above with the uncertainty of the humidity generator. between 2 means (each the mean of 4 results) Average (mid-point) of 2 means (each the mean of 4 results) ** Uncertainty in average ** I column (consistency indicator - all labs should hope to get the same value for this) (aggregated result - parameter to be compared between institutes) (uncertainty in the parameter being compared) REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: Lab name NIST Results Note: A correction of +.005C is applied to the temperatures reported as Deg. C. Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 etc. Standard uncertainty of applied condition Standard deviation of the calculated dew / frost-point temperature (Deg. C) Number of observations Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) The value reported here is the standard uncertainty obtained by combining the type A value in the boxes immediately above with the uncertainty of the humidity generator. between 2 means (each the mean of 4 results) Average (mid-point) of 2 means (each the mean of 4 results) ** Uncertainty in average ** I column (consistency indicator - all labs should hope to get the same value for this) (aggregated result - parameter to be compared between institutes) (uncertainty in the parameter being compared) 91

86 REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: Lab name NIST Results Note: A correction of +.005C is applied to the temperatures reported as Deg. C. Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 etc. Standard uncertainty of applied condition Standard deviation of the calculated dew / frost-point temperature (Deg. C) Number of observations Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) The value reported here is the standard uncertainty obtained by combining the type A value in the boxes immediately above with the uncertainty of the humidity generator. between 2 means (each the mean of 4 results) Average (mid-point) of 2 means (each the mean of 4 results) ** Uncertainty in average ** I column (consistency indicator - all labs should hope to get the same value for this) (aggregated result - parameter to be compared between institutes) (uncertainty in the parameter being compared) 92

87 NMC REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: -50 Lab namenmc/spring Results Hygrometer 1(Michell) Applied frost point () Output in (applied fp - meas fp) in Hygrometer 2 (MBW) Applied frost point () (applied fp - meas fp) in Meas Meas Meas Meas Note: Meas 1 was not a full simultaneous dataset Uncertainties (in ) Meas 2 Meas 3 Meas 4 Meas 5 Meas 2 Meas 3 Meas 4 Meas 5 Standard uncertainty of applied condition except type A Standard uncertainty of applied condition type A (when there are more than one records, max value is taken) Std uncert due to short-term stability (from standard deviation) of measurements of traveling standard (type A) (when there are more than one records, the average of the standard deviations of records is taken) Hygrometer1 Hygrometer Std uncert due to the resistance measurement uncertainty of traveling standard (the uncertainty of the two bridges is 1 mk 95% k=2) Std uncert due to resolution of traveling standard (resolution is 1 mk) Std uncert due to long-term drift of traveling standard [Not observed] Std uncert due to non-linearity of traveling standard [not involved] Covariance between applied and measured values of dew/frost-point [not applicable] between ABC/MMC (insignificant) Drift due to longer waiting time (insignificant) Combined standard uncertainty (8 values) Average of combined standard uncertainty Auc Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences for Michell/MBW Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) (Auc combined with the reproducibility between 2 means (each the mean of 4 results) Hyg2-hyg1 Average (mid-point) of 2 means (each the mean of 4 results) ** Uncertainty in average ** (consistency indicator - all labs should hope to get the same value for th (aggregated result - parameter to be compared between institutes) (uncertainty in the parameter being compared) 93

88 REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: -30 Lab name NMC/SPRING Results Hygrometer 1(Michell) Applied frost point () Output in (applied fp - meas fp) in Hygrometer 2 (MBW) Applied frost point () Output in (applied fp - meas fp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 Meas 2 Meas 3 Meas 4 Standard uncertainty of applied condition except type A Standard uncertainty of applied condition type A (when there are more than one records, max value is taken) Std uncert due to short-term stability (from standard deviation) of measurements of traveling standard (type A) (when there are more than one records, the average of the standard deviations of records is taken) Std uncert due to the resistance measurement uncertainty of traveling standard (the uncertainty of the two bridges is 1 mk 95% k=2) Std uncert due to resolution of traveling standard (resolution is 1 mk) Std uncert due to long-term drift of traveling standard [Not observed] Std uncert due to non-linearity of traveling standard [not involved] Covariance between applied and measured values of dew/frost-point [not applicable] between ABC/MMC (insignificant) Drift due to longer waiting time (insignificant) Combined standard uncertainty (8 values) Average of combined standard uncertainty Auc Aggregation of results Hyg 1 Hyg 2 Mean of 5 dew-point differences Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) (Auc combined with the reproducibility between 2 means (each the mean of 4 results) Hyg2-hyg1 Average (mid-point) of 2 means (each the mean of 4 results) ** Uncertainty in average ** (consistency indicator - all labs should hope to get the same value for t (aggregated result - parameter to be compared between institutes) (uncertainty in the parameter being compared) 94

89 REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: -10 Lab name NMC/SPRING Results Generator Model 4500 Hygrometer 1(Michell) Applied frost point () Output in Hygrometer 2 (MBW) (applied fp Applied - meas fp) frost point in () Output in (applied fp - meas fp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 Meas 2 Meas 3 Meas 4 Standard uncertainty of applied condition except type A Standard uncertainty of applied condition type A (when there are more than one records, max value is taken) Std uncert due to short-term stability (from standard deviation) of measurements of traveling standard (type A) (when there are more than one records, the average of the standard deviations of records is taken) Std uncert due to the resistance measurement uncertainty of traveling standard (the uncertainty of the two bridges is 1 mk 95% k=2) Std uncert due to resolution of traveling standard (resolution is 1 mk) Std uncert due to long-term drift of traveling standard [Not observed] Std uncert due to non-linearity of traveling standard [not involved] Covariance between applied and measured values of dew/frost-point [not applicable] between ABC/MMC (insignificant) Drift due to longer waiting time (insignificant) Combined standard uncertainty (8 values) Average of combined standard uncertainty Auc Aggregation of results Hyg 1 Hyg 2 Mean of 5 dew-point differences Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) (Auc combined with the reproducibility Results Generator Model 2500 Hygrometer 1(Michell) Applied dew point () Output in (applied dp - meas dp) in Hygrometer 2 (MBW) Applied dew point () Output in (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 Meas 2 Meas 3 Meas 4 Standard uncertainty of applied condition except type A Standard uncertainty of applied condition type A (when there are more than one records, max value is taken) Std uncert due to short-term stability (from standard deviation) of measurements of traveling standard (type A) (when there are more than one records, the average of the standard deviations of records is taken) Std uncert due to the resistance measurement uncertainty of travelling standard (the uncertainty of the two bridges is 1 mk 95% k=2) Std uncert due to resolution of travelling standard (resolution is 1 mk) Std uncert due to long-term drift of travelling standard [Not observed] Std uncert due to non-linearity of travelling standard [not involved] Covariance between applied and measured values of dew/frost-point [not applicable] between ABC/MMC (insignificant) Drift due to longer waiting time (insignificant) Combined standard uncertainty (8 values) Average of combined standard uncertainty Auc Aggregation of results Hyg 1 Hyg 2 Mean of 5 dew-point differences Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) Aggregation of results Hyg 1 Hyg 2 Average of 2 generators Uncertainty of mean dew point difference for each instrument (2 values) (average of the uncertainties of the two generators) between 2 means (each the mean of 4 results) Hyg2-hyg1 Average (mid-point) of 2 means (each the mean of 4 results) ** Uncertainty in average ** (consistency indicator - all labs should hope to get the same value for this) (aggregated result - parameter to be compared between institutes) (uncertainty in the parameter being compared) 95

90 REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: 1 Lab namenmc/spring Results Hygrometer 1(Michell) Applied dew point () Output in (applied dp - meas dp) in Hygrometer 2 (MBW) Applied dew point () Output in (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 Meas 2 Meas 3 Meas 4 Standard uncertainty of applied condition except type A Standard uncertainty of applied condition type A (when there are more than one records, max value is taken) Std uncert due to short-term stability (from standard deviation) of measurements of traveling standard (type A) (when there are more than one records, the average of the standard deviations of records is taken) Std uncert due to the resistance measurement uncertainty of traveling standard (the uncertainty of the two bridges is 1 mk 95% k=2) Std uncert due to resolution of traveling standard (resolution is 1 mk) Std uncert due to long-term drift of traveling standard [Not observed] Std uncert due to non-linearity of traveling standard [not involved] Covariance between applied and measured values of dew/frost-point [not applicable] between ABC/MMC (insignificant) Drift due to longer waiting time (insignificant) Combined standard uncertainty (8 values) Average of combined standard uncertainty Auc Aggregation of results Hyg 1 Hyg 2 Mean of 5 dew-point differences Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) (Auc combined with the reproducibility between 2 means (each the mean of 4 results) Hyg2-hyg1 Average (mid-point) of 2 means (each the mean of 4 results) ** Uncertainty in average ** (consistency indicator - all labs should hope to get the same value fo (aggregated result - parameter to be compared between institutes) (uncertainty in the parameter being compared) 96

91 REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: 20 Lab namenmc/spring Results Hygrometer 1(Michell) Applied dew point () Output in (applied dp - meas dp) in Hygrometer 2 (MBW) Applied dew point () Output in (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 Meas 2 Meas 3 Meas 4 Standard uncertainty of applied condition except type A Standard uncertainty of applied condition type A (when there are more than one records, max value is taken) Std uncert due to short-term stability (from standard deviation) of measurements of traveling standard (type A) (when there are more than one records, the average of the standard deviations of records is taken) Std uncert due to the resistance measurement uncertainty of traveling standard (the uncertainty of the two bridges is 1 mk 95% k=2) Std uncert due to resolution of traveling standard (resolution is 1 mk) Std uncert due to long-term drift of traveling standard [Not observed] Std uncert due to non-linearity of traveling standard [not involved] Covariance between applied and measured values of dew/frost-point [not applicable] between ABC/MMC (insignificant) Drift due to longer waiting time (insignificant) Combined standard uncertainty (8 values) Average of combined standard uncertainty Auc Aggregation of results Hyg 1 Hyg 2 Mean of 5 dew-point differences Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) (Auc combined with the reproducibility between 2 means (each the mean of 4 results) Hyg2-hyg1 Average (mid-point) of 2 means (each the mean of 4 results) ** Uncertainty in average ** (consistency indicator - all labs should hope to get the same value f (aggregated result - parameter to be compared between institutes) (uncertainty in the parameter being compared) 97

92 NIM REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: -50 Lab name NIM China Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - meas dp) in Applied dew point () Output (mv) (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 Meas 2 Meas 3 Meas 4 Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) I column between 2 means (each the mean of 4 results) (consistency indicator - all labs should hope to get the same value for this) Average (mid-point) of 2 means (each the mean of 4 results) ** (aggregated result - parameter to be compared between institutes) Uncertainty in average ** (uncertainty in the parameter being compared) REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: -30 Lab name NIM China Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () Output (mv) (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 Meas 2 Meas 3 Meas 4 Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) I column between 2 means (each the mean of 4 results) (consistency indicator - all labs should hope to get the same value for this) Average (mid-point) of 2 means (each the mean of 4 results) ** (aggregated result - parameter to be compared between institutes) Uncertainty in average ** (uncertainty in the parameter being compared) 98

93 REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: -10 Lab name NIM China Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () Output (mv) (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 Meas 2 Meas 3 Meas 4 Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) I column between 2 means (each the mean of 4 results) (consistency indicator - all labs should hope to get the same value for this) Average (mid-point) of 2 means (each the mean of 4 results) ** (aggregated result - parameter to be compared between institutes) Uncertainty in average ** (uncertainty in the parameter being compared) REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: 1 Lab name NIM China Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () Output (mv) (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 Meas 2 Meas 3 Meas 4 Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) between 2 means (each the mean of 4 results) Average (mid-point) of 2 means (each the mean of 4 results) ** Uncertainty in average ** I column (consistency indicator - all labs should hope to get the same value for this) (aggregated result - parameter to be compared between institutes) (uncertainty in the parameter being compared) 99

94 REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: 20 Lab name NIM China Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () Output (mv) (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 Meas 2 Meas 3 Meas 4 Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) between 2 means (each the mean of 4 results) Average (mid-point) of 2 means (each the mean of 4 results) ** Uncertainty in average ** I column (consistency indicator - all labs should hope to get the same value for this) (aggregated result - parameter to be compared between institutes) (uncertainty in the parameter being compared) 100

95 VNIIFTRI ESB REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: -50 Lab name VNIIFTRI ES Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 Meas 2 Meas 3 Meas 4 Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) I column between 2 means (each the mean of 4 results) (consistency indicator - all labs should hope to get the same value for this) Average (mid-point) of 2 means (each the mean of 4 results) ** (aggregated result - parameter to be compared between institutes) Uncertainty in average ** (uncertainty in the parameter being compared) REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: -30 Lab name VNIIFTRI ES Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () Resistanc e (applied dp - meas dp) in Applied dew point () Resistanc e (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 Meas 2 Meas 3 Meas 4 Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) I column between 2 means (each the mean of 4 results) (consistency indicator - all labs should hope to get the same value for this) Average (mid-point) of 2 means (each the mean of 4 results) ** (aggregated result - parameter to be compared between institutes) Uncertainty in average ** (uncertainty in the parameter being compared) 101

96 REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: -10 Lab name VNIIFTRI ES Results Hygrometer 1(Michell) Applied dew point () Hygrometer 2 (MBW) (applied dp - meas dp) Applied dew in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 Meas 2 Meas 3 Meas 4 Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) I column between 2 means (each the mean of 4 results) (consistency indicator - all labs should hope to get the same value for this) Average (mid-point) of 2 means (each the mean of 4 results) ** (aggregated result - parameter to be compared between institutes) Uncertainty in average ** (uncertainty in the parameter being compared) REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: 1 Lab name VNIIFTRI ES Results Hygrometer 1(Michell) Hygrometer 2 (MBW) Applied dew point () (applied dp - Applied dew meas dp) in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 Meas 2 Meas 3 Meas 4 Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type A) Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) I column between 2 means (each the mean of 4 results) (consistency indicator - all labs should hope to get the same value for this) Average (mid-point) of 2 means (each the mean of 4 results) ** (aggregated result - parameter to be compared between institutes) Uncertainty in average ** (uncertainty in the parameter being compared) 102

97 REPORTING TEMPLATE FOR DEW-POINT KEY COMPARISON Nominal value: 20 Lab name VNIIFTRI ES Results Hygrometer 1(Michell) Applied dew point () Hygrometer 2 (MBW) (applied dp - meas dp) Applied dew in point () (applied dp - meas dp) in Meas Meas Meas Meas Uncertainties (in ) Hygrometer 1 Hygrometer 2 Meas 1 Meas 2 Meas 3 Meas 4 Meas 1 Meas 2 Meas 3 Meas 4 Standard uncertainty of applied condition Std uncert due to short-term stability (from standard deviation) of measurements of travelling standard (type Std uncert due to long-term drift of travelling standard [if needed] Std uncert due to resolution of travelling standard [if needed] Std uncert due to non-linearity of travelling standard [if needed] Covariance between applied and measured values of dew/frost-point [if needed] Combined standard uncertainty (8 values) Aggregation of results Hyg 1 Hyg 2 Mean of 4 dew-point differences (for 2 instruments) Type A standard uncertainty due to reproducibility of difference results two values (each derived from standard deviation of 4 values on same instrument) Uncertainty of mean dew point difference for each instrument (2 values) I column between 2 means (each the mean of 4 results) (consistency indicator - all labs should hope to get the same value for this) Average (mid-point) of 2 means (each the mean of 4 results) ** (aggregated result - parameter to be compared between institutes) Uncertainty in average ** (uncertainty in the parameter being compared) 103

98 APPENDIX 3: UNCERTAINTY ANALYSES OF PARTICIPANTS Uncertainty budgets of each participant are shown on the following pages. These were reported either in the MS Excel template provided for comparison reporting or as tabulated information giving a suitable level of detail to cover the main sources of uncertainty. Where the template was used, included here are selected pages detailing uncertainty calculations at +20 and -50, the extremes of the comparison. The protocol required participants to report the effective number of degrees of freedom of the uncertainty estimates. In cases where this was sufficiently large, a coverage factor of 2 is used to obtain an interval for confidence probability of 95 %. In the few cases where a larger coverage factor was needed, the detail is included in this Appendix. 104

99 NPL The uncertainty routinely reported for calibrations is based on the combined standard uncertainty of the standard applied condition, together with an allowance for the resolution and short-term stability of the instrument being calibrated. The result is reported with an expanded uncertainty at a coverage factor k=2 giving a coverage probability of approximately 95 %. 105