BILATERAL KEY COMPARISON SIM.T-K6.3 ON HUMIDITY STANDARDS IN THE DEW/FROST-POINT TEMPERATURE RANGE FROM 30 C TO 20 C

Transcription

1 SIM.T-K6.3 NIST / INMETRO Fina Report BILATERAL KEY COMPARISON SIM.T-K6.3 ON HUMIDITY STANDARDS IN THE DEW/FROST-POINT TEMPERATURE RANGE FROM 30 C TO 20 C Peter Huang 1, Christopher Meyer 1, and Juio D. Brionizio 2 1 Nationa Institute of Standards and Technoogy (NIST), USA 2 Instituto Naciona de Metroogia, Quaidade e Tecnoogia (INMETRO), Brazi Abstract A Regiona Metroogy Organization (RMO) Key Comparison of dew/frost point temperatures was carried out by the Nationa Institute of Standards and Technoogy (NIST, USA) and the Instituto Naciona de Metroogia, Quaidade e Tecnoogia (INMETRO, Brazi) between October, 2009 and March, The resuts of this comparison are reported here, aong with descriptions of the humidity aboratory standards for NIST and INMETRO and the uncertainty budget for these standards. This report aso describes the protoco for the comparison and presents the data acquired. The resuts are anayzed, determining degree of equivaence between the dew/frost-point standards of NIST and INMETRO. Keywords: Comparison, Humidity, Dew Point, Frost Point, Degree of Equivaence. 1. Introduction Key Comparisons determine differences between measurement standards of different Nationa Metroogy Institutes (NMIs). They pay an important roe in ensuring that the standards of a NMIs are in agreement. At its 20 th meeting in Apri 2000, the Consutative Committee for Thermometry (CCT) caed for a Key Comparison on humidity standards to be conducted by a major Nationa Metroogy Institutes. It asked CCT Working Group 6 (WG6) on Humidity Measurements (WG6) to draw up a technica protoco for an Internationa Committee on Weights and Measures (CIPM) key comparison named CCT-K6. The Nationa Physica Laboratory (UK) and the Nationa Metroogy Institute of Japan were chosen to be the piot aboratory and assistant piot aboratory, respectivey. The Nationa Institute of Standards and Technoogy (NIST, USA) participated in this key comparison. The Instituto Naciona de Metroogia, Quaidade e Tecnoogia (INMETRO, Brazi) did not participate in CCT-K6. Therefore, to reate the humidity standards of INMETRO to those of the CCT-K6 participants, a Regiona Metroogy Organization (RMO) Key Comparison of dew/frost point temperatures T was carried out by NIST and INMETRO between October 2009 and March 2010; this comparison was designated as SIM.T-K6.3. Here, it is assumed that T is the dew-point temperature TDP for T 0 and T is the frost-point temperature TFP for T < 0. As an NMI, INMETRO meets the Mutua Recognition Agreement requirements for participation in a key comparison. NIST was the piot for this biatera

2 comparison. This biatera comparison foowed the same technica procedures as for the CCT-K6, except that ony one transfer standard was used. Aso, a range of 30 C T 20 C was used instead of 50 C T 20 C. 2. Participants NIST Peter Huang (now retired) Christopher Meyer (current contact) Nationa Institute of Standards and Technoogy 100 Bureau Drive Gaithersburg, MD USA INMETRO Juio D. Brionizio Instituto Naciona de Metroogia, Quaidade e Tecnoogia Av. Nossa Senhora das Graças, 50 Xerém, Duque de Caxias, RJ CEP: , Brazi Te.: Fax: e-mai: cmeyer@nist.gov Te.: Fax: emai: jdbrionizio@inmetro.gov.br 3. Comparison Method The comparison between dew/frost-point temperature standards at NIST and INMETRO was performed through use of a transfer standard (a chied-mirror hygrometer). At a given nomina dew/frost point, each participant used his generator to produce moist air having a constant dew/frost-point temperature. The participant then used his aboratory standard to determine the dew/frost point temperature to be T. For NIST, the aboratory standard was the generator itsef. For INMETRO, the aboratory standard was a chied-mirror hygrometer measuring the dew/frost point from the generator. The transfer standard determined the dew/frost-point t temperature of the generated gas to be T. The difference between the two vaues was T T T t The comparison of NIST and INMETRO humidity standards was then performed by comparing the vaues of ΔT determined using the NIST aboratory standard, ΔT(NIST), with those of the INMETRO aboratory standard, ΔT(INMETRO). 4. Laboratory Humidity Standards The NIST aboratory humidity standard used was the NIST Hybrid Humidity Generator (HHG). Its principe of operation depends on the desired vaue of T.

3 For T 15 C, the HHG operates as a conventiona two-pressure generator, saturating air with water at a temperature Ts and pressure Ps to produce moist air with a moar fraction xg given by e(ts ) xg= f (Ts,Ps ) 1) P s Here, e (Ts) is the water vapor pressure at Ts, cacuated using [1-2] and f(ts, Ps) is the watervapor enhancement factor, cacuated using [3]. The saturator temperature is measured by a standard patinum resistance thermometer (SPRT) immersed in the same temperaturecontroed bath as the saturator. The saturator pressure, which can vary from ambient to 500 kpa, is measured by a strain-gauge pressure transducer that is connected by a tube to the saturator at a point near its outet. For T 15 C, the HHG uses the divided fow method, which invoves diuting the saturated gas with dry gas using precisey-metered streams of gas. The moar fraction after diution is x g n x n s s p xp 2) N where n s and n p are the moar fows of the saturated gas and pure (dry) gas, respectivey, and N is the tota moar fow. Aso, xs is the moar fraction of water in the saturated gas and xp is the residua moar fraction of water in the pure gas. For the HHG in divided fow mode, the saturator is operated at a temperature of 1 C and a pressure of 300 kpa, resuting in xs The generated dew/frost-point temperature is obtained from xg by measuring the pressure Pc using a strain-gauge pressure transducer at the inet of the chied-mirror hygrometer. T is then obtained by iterativey soving the equation DP,FP e TDP,FP xg = f T DP,FP,Pc 3) P Here, T DP, FP ew T DP,FP cacuated using [1-2]. Aso, T DP, FP ei T DP,FP vapor pressure for ice, cacuated using [4-5]. The vaue of T, P c e for T 0 ºC, where ew is the saturated vapor pressure for water, e for T < 0 ºC, where ei is the saturated DP, FP A more compete description of the NIST HHG may be found in [6]. f is cacuated using [3]. To ensure the stabiity of the HHG resuts, the HHG pressure gauges are caibrated yeary. The HHG SPRT resistance at the tripe point of water RTPW is aso caibrated yeary. The pressure gauge and SPRT caibrations are performed at NIST. The poicy of the HHG aboratory is that if the change in RTPW from that of the origina caibration ever corresponds to a temperature drift of more than 10 mk, a fu caibration wi be performed. Finay, NIST empoys check c

4 standards during every customer caibration for the purpose of detecting any possibe errors or ong-term drifts. The INMETRO aboratory humidity standard used was a chied-mirror hygrometer (Miche S4000, indicator S/N and sensor S/N ) [7] with a caibration traceabe to the SI through the Nationa Physica Laboratory (NPL, UK). The caibration certificate number was E and it was issued by NPL on 26 August The caibration provided both dispay vaues and mirror-prt resistance vaues for the Miche S4000, but ony the dispay vaue caibrations have been used by INMETRO for biatera comparisons and customer caibrations. The hygrometer operates in the dew/frost point temperature range from 75 ºC to +20 ºC. In this instrument, ight shines onto a poished mirror surface, the temperature of which is controed by a thermoeectric heat-pump. A sensitive photo-detector measures the intensity of the direct refection. When the mirror is cean and dry, the intensity of the refected ight is at its maximum. Conversey, a cod mirror with water vapor condensed on its surface scatters the ight, resuting in ess ight directy refected and in reduced signa intensity. Using this received ight signa as feedback in a cosed oop contro system, the mirror may be cooed to the temperature at which the thickness of the condensed ayer, detected through the intensity of the received ight, remains constant. A condensate ayer of constant thickness, with no further net increase or decrease in condensation, is in dynamic equiibrium with the gas surrounding the mirror. In this equiibrium condition, the dew or frost point temperature of the gas is determined by measuring the temperature of the mirror. If the condensate is known to be in iquid form, even for temperatures beow freezing, then the measured mirror temperature is taken as the dew point. If the condensate is known to be in a soid form as ice or frost, then the measured mirror temperature is taken as the frost point. Specifications for the hygrometer may be found in [8]. To ensure the stabiity of the INMETRO aboratory Standard, it is caibrated at NPL at threeyear intervas. The caibration used for the comparison reported here was performed in The INMETRO aboratory humidity standard measured the dew/frost point of moist air produced by a commerciay-made humidity generator. This generator (Miche divided-fow generator, mode DG-4) [7] operates over the dew/frost point range from 75 ºC to +20 ºC. In this generator, dried gas is divided into two streams of which one passes through a water saturator and is mixed with the other stream to produce a certain gas sampe. Dew/frost point temperatures can be seected via a front pane keypad, through factory pre-set vaues, or by manuay mixing the wet and dry gases by means of metering vaves mounted on its front pane. In the comparison reported here, the Miche DG-4 was used for T < 0 ºC. The other equipment used for the generation of the air sampes was a Weiss Technik cimatic chamber, mode WK3-340/40, which operates in the dew/frost point temperature range from -20 ºC to +94 ºC. In the comparison reported here, the Weiss Technik WK3-340/40 was used for T 0 ºC.

5 5. Transfer standard Mode: MBW 373 Seria Number: Size (in Packing case): 63 cm x 53 cm x 40 cm Weight (in Packing case): 20.1 kg Manufacturer: MBW Eektronik AG., Switzerand Owner: INMETRO, Brazi Eectrica suppy: 220 V / 50 Hz Approximate vaue for insurance and customs decaration: US$ 18, Measurement process Sampe air with T reaized by a participant's generator was introduced into the inet of the transfer-standard hygrometer through a stainess stee tube. The tube was attached to the transfer standard using a 6.35 mm Swageok fitting. The dew point temperature shown on the t hygrometer dispay was then recorded as T. For INMETRO, the sampe air was aso introduced into the inet of its standard chied-mirror hygrometer through a stainess stee tube in parae with the tube eading to the transfer standard; the dew/frost point temperature shown on the hygrometer dispay was then recorded as T. For NIST, the dew/frost point temperature cacuated from measurements of the generator s reevant parameters was recorded as T. A tota of four dew/frost-point temperatures were used for the comparison: 20 C, 0 C, 10 C and 30 C. Each participant made four independent measurements for each dew/frost-point temperature, reforming the condensate on the hygrometer s mirror each time. At each measured dew/frost point, hygrometer dispay readings were monitored unti they drifted ess than C over a period of 20 minutes (dew points) or 40 minutes (frost points); at that point they were assumed to be in a steady state. Afterwards, mutipe readings of the dew point temperature were recorded, and the mean and standard deviation of these readings were recorded. 7. Measurement data Tabe 1 shows the resuts of the generator/hygrometer comparisons for both INMETRO and NIST.

6 Nomina T Tabe 1. Resuts of aboratory/transfer standard comparisons Meas. # Hygrometer MBW 373, S/N Laboratory Transfer Standard Standard T T t ΔT INMETRO NIST INMETRO NIST INMETRO NIST INMETRO NIST

7 For the second nomina dew point, INMETRO used a vaue of T = 1 C whie NIST used a vaue of T = 0 C. This introduces two additiona uncertainties for this point. The first uncertainty is due to the assumption that at T 0 C the condensate on the hygrometer mirror is aways water (never ice). Based on the difference between the dew point and frost point temperatures at the dew/frost-point vaues reaized at T 0 C, we estimate the standard uncertainty for this assumption to be C. The second uncertainty comes from choosing the nomina vaue for the second point to be 1 C (for consistency with the CCT K6 comparison). Because of this, an uncertainty must be estimated for the difference between the NIST vaue of ΔT at T = 0 C and that which it woud be at T = 1 C. Based upon the difference between the average vaue of ΔT at T = 0 C and that at T = 10 C, we estimate this standard uncertainty to be C. Tabe 2 shows the difference between the aboratory standard and transfer standard dew/frostpoint temperatures ΔT for four measurements. For a given nomina vaue of ΔT, the resuts of INMETRO and NIST are shown on separate rows. The resuts for each of the four measurements are shown in separate coumns. The mean and standard deviation of these differences ( T and σ(δt), respectivey) are shown in the ast two coumns. The data shown in Tabe 2 is potted in Fig. 1. Tabe 2. Difference between aboratory-standard and transfer-standard dew/frost-point temperatures ΔT for NIST and INMETRO. The mean and standard deviation of these differences ( respectivey) are shown in the ast two coumns. Nomina T NMI Meas. 1 ΔT Meas. 2 ΔT Meas. 3 ΔT Meas. 4 ΔT T and σ(δt ), T σ(δt) 20 INMETRO NIST INMETRO NIST INMETRO NIST INMETRO NIST

8 Figure 1. Difference between aboratory standard and transfer standard dew/frost-point temperatures, ΔT, for NIST and INMETRO. Note: the data at -30 C, -10 C, and 20 C from the two NMIs are sighty offset horizontay to faciitate viewing. 8. Comparison Uncertainty For a set of determinations of ΔT made at a nomina vaue of T, the standard uncertainty of the generator/hygrometer comparison uc(δt) is given by u c 2 2 T u T u T 1/ 2 4) A Descriptions of ua(δt) and u are given beow. T First, ua(δt) is the type A uncertainty for the determination of ΔT. For NIST, this uncertainty incudes the reproducibiity of the generator and the transfer-standard chied-mirror hygrometer. For INMETRO, it incudes the reproducibiity of the aboratory-standard and transfer-standard chied-mirror hygrometers. For both NIST and INMETRO, ua(δt) aso incudes resoution errors arising from rounding off the vaues of ΔT to two digits after the decima point. For simpicity and a more accurate determination, it was assumed that ua(δt) is independent of T. For each NMI, ua(δt) was determined as the average vaue of σ(δt) for the four nomina T vaues. For INMETRO and NIST these average vaues were C and C, respectivey. The individua vaues of σ(δt) are given in Tabe 2. For NIST, u T is the type B uncertainty of the generated vaue of T. The source of the vaues u T for NIST is Ref. 6, which contains a compete uncertainty budget for the NIST Hybrid Humidity Generator. Tabe 3 shows the uncertainty eements and their standard uncertainty vaues for the NIST generator, for the four nomina vaues of T. Tabe 4 shows the contribution of these uncertainty eements to u. T

9 Uncertainty for NIST generator: Tabe 3. Uncertainty eements and their standard uncertainty vaues for the NIST generator, for the four nomina vaues of T Saturator Temperature Measurement TDP = 20 C TDP = 0 C TFP= -10 C TFP= -30 C Caibration uncertainty C C C C Long-term stabiity C C C C Saturator Pressure Measurement Caibration uncertainty 18 Pa 47 Pa 39 Pa 42 Pa Long-term stabiity 7 Pa 7 Pa 7 Pa 7 Pa Hygrometer Pressure Measurement Caibration uncertainty 18 Pa 18 Pa 18 Pa 18 Pa Long-term stabiity 7 Pa 7 Pa 7 Pa 7 Pa Fow measurement (divided fow method): Caibration uncertainty % Long-term stabiity % Cacuation: Saturation vapor pressure formua(e) 0.15 Pa 0.10 Pa 0.06 Pa 0.04 Pa Water vapor enhancement formua(e) T Tabe 4. Contribution of the uncertainty eements in Tabe 3 to u for NIST, in C, for the four nomina vaues of T. The combined standard uncertainty is shown in the ast row Uncertainty for NIST generator: Saturator Temperature Measurement TDP = 20 C TDP = 0 C TFP= -10 C TFP= -30 C Caibration uncertainty Long-term stabiity Saturator Pressure Measurement Caibration uncertainty Long-term stabiity Hygrometer Pressure Measurement Caibration uncertainty Long-term stabiity Fow measurement (divided fow method): Caibration uncertainty Long-term stabiity Cacuation: Saturation vapor pressure formua(e) Water vapor enhancement formua(e) Combined standard uncertainty:

10 For INMETRO, u T is the type B uncertainty of the vaue of T measured by its aboratory-standard chied-mirror hygrometer. The source of the vaues u T for INMETRO is [9], which provides an uncertainty anaysis for the INMETRO standard. Tabes 5 shows the vaues of these standard uncertainties for the INMETRO standard hygrometer and the combined type B uncertainty, u. T T Tabe 5. Contribution of the uncertainty eements in Tabe 5 to u for INMETRO, in C, for the four nomina vaues of T. The combined standard uncertainty is shown in the ast row. TDP = TDP = TFP= TFP= Uncertainty for INMETRO standard: 20 C 1 C -10 C -30 C Caibration uncertainty of the hygrometer Resoution of the hygrometer Drift of the hygrometer Fitting of the hygrometer correction curve Combined standard uncertainty: Tabe 6 shows the cacuated vaue of uc(δt) and its components for each vaue of T and each participating NMI. Note that in this tabe we have adjusted the NIST vaue of u( T ) at 1 C from C to C to account for the uncertainty due to the NIST measurements being performed at 0 C (see discussion in section 7). Tabe 6. Standard uncertainty of the determinations of ΔT for NIST and INMETRO. The coumn headings are described in the text. Nomina T Participating ua T u( T ) u Institute 20 INMETRO NIST INMETRO NIST INMETRO NIST INMETRO NIST c T 9. Drift of the Transfer Standard The first comparison between aboratory humidity standard and transfer standard was made at INMETRO from October to November in Afterwards, the transfer standard was sent to

11 T, ºC NIST so that it coud perform its comparison measurements. The transfer standard arrived back at INMETRO in June 2010, and in Juy 2010 measurements were repeated at two of the four dew/frost-point temperatures. Drift of the transfer standard during the course of the INMETRO-NIST comparison may be estimated by examining the measurements at the dew/frost point temperatures performed at INMETRO in October/November 2009 and Juy The difference between the average of the October/November 2009 measurements and the Juy 2010 measurements is approximatey 0.06 C, as shown in Figure 2. It is quite possibe that this difference is due to reproducibiity uncertainty rather than to drift. Nevertheess, in our uncertainty budget we have added a type B uncertainty component due to the possibiity of transfer standard drift. Based on the resuts of Fig. 2, we have estimated it to contribute a standard uncertainty of C to the INMETRO- NIST comparison Meas. 1 Meas. 2 Meas. 3 Meas. 4 Juy T,ºC Figure 2. Difference between the INMETRO measurements performed in Oct./Nov (Meas. 1,2,3, and 4) and Juy Degree of Equivaence The vaues ΔT given in Tabe 1 may be used to determine the degree of equivaence DINMETRO(T) between the vaues of T reaized by INMETRO and reference vaues of T : T, Ref D INMETRO T T INMETRO T Ref 5)

12 Once Draft B for the report of CCT K6 Key Comparison is approved, T Ref wi be T KCRV, the CCT K6 Key Comparison Reference Vaue (KCRV) for T. In the meantime we sha define Ref T NIST T, 6) since NIST is a participant in CCT K6 and wi ater be abe to provide inkage to T KCRV. Therefore, for the purposes of this report, D INMETRO T T INMETRO T NIST T INMETRO T NIST 7) The uncertainty of the degree of equivaence u(dinmetro(t) is the combination of uc T for INMETRO, uc T for NIST, and the uncertainty udrift due to possibe drift of the transfer standard: u 2 D T u T 2 2 u T 1/ 2. 8) INMETRO c INMETRO c u NIST drift The expanded (k=2, 95% confidence eve) uncertainty for the degree of equivaence is U(DINMETRO) = 2u(DINMETRO), 9) The resuts are presented in Tabe 7 and potted in Fig. 3. As can be seen in Tabe 7 and Fig. 3, a vaues of DINMETRO are within their expanded uncertainties. Tabe 7. Degree of equivaence between INMETRO and NIST and its expanded uncertainty (k = 2) in a comparison of four humidity eves Nomina T DINMETRO U(DINMETRO)

13 Figure 3. The degree of equivaence between the four dew/frost-point temperatures reaized by the INMETRO aboratory humidity standard, T INMETRO and the corresponding reference vaues T Ref (represented by T ), as defined in Eq. 7. NIST 11. Summary NIST and INMETRO have competed a biatera comparison of their humidity standards. The quantity compared was the dew/frost-point temperature. The NIST aboratory standard was a humidity generator and the INMETRO aboratory standard was a chied-mirror hygrometer. The transfer standard used was a chied-mirror hygrometer. At NIST, the dew/frost point was produced by the NIST aboratory standard and measured by the transfer standard. At INMETRO, the dew/frost point temperatures were produced by a stabe generator and measured by both the INMETRO aboratory standard and the transfer standard. The nomina dew/frostpoint temperatures used for the comparison were 20 C, 0 C, 10 C and 30 C. The comparisons have determined the degree of equivaence between T INMETRO and a reference vaue for T, presenty defined as T NIST. For a dew/frost-point temperatures over the range studied, the degree of equivaence was within 0.09 C and we within its expanded uncertainty of 0.2 C.

14 12. References [1] A. Sau and W. Wagner, Internationa Equations for the Saturation Properties of Ordinary Water Substance, J. Phys. Chem. Ref. Data 16, (1987). [2] W. Wagner and A. Pruss, Internationa Equations for the Saturation Properties of Ordinary Water Substance--Revised According to the Internationa Temperature Scae of 1990, J. Phys. Chem. Ref. Data 22, (1993). [3] R.W. Hyand and A. Wexer, Formuations for the Thermodynamic Properties of Dry Air from K to K, and of Saturated Moist Air from K to K, at Pressures to 5 MPa., ASHRAE Trans. 89-IIa, (1983). [4] Internationa Association for the Properties of Water and Steam, Revised Reease on the Pressure aong the Meting and Subimation Curves of Ordinary Water Substance (2011), avaiabe at [5] Wagner, W., Riethmann, T., Feiste, R., and Harvey, A. H., New Equations for the Subimation Pressure and Meting Pressure of H2O Ice Ih, J. Phys. Chem. Ref. Data 40, (2011). [6] C.W. Meyer et a, Caibration of Hygrometers with the Hybrid Humidity Generator, NIST Specia Pubication (2008). [7] Certain commercia equipment, instruments, or materias are identified in this paper to foster understanding. Such identification does not impy recommendation or endorsement by the Nationa Institute of Standards and Technoogy, nor does it impy that the materias or equipment identified are necessariy the best avaiabe for the purpose. [8] [9] Brionizio, J. D. Technica Report of the Hygrometry Laboratory (2008).