1. Process Measurements Division The New NIST Gas Pressure Scale: Reduced Uncertainties Based on Large Diameter Piston Gauges Douglas A. Olson National Institute of Standards and Technology James Schmidt, Greg Driver, Walt Bowers, and Dana Defibaugh NCSL International Workshop and Symposium August 5, 2008

2. Process Measurements Division Outline of Talk Existing gas pressure scale Primary standard piston gauges (PGs) and their characterization Revised pressure scale based on PGs Characterization of PGs traceable to primary Validation of uncertainties via international comparisons Status of Calibration Measurement Capabilities (CMCs)

3. Process Measurements Division Pressure Realization Using Piston Gauges masses, m,  m gravity, g cylinder piston fluid pressure, p a ambient density,  a Artifact-based SI unit derived from fundamental quantities of mass and length Pressure results from force generated in gravitational field Artifact performance characterized by “effective area”, A e If A e determined by dimensional metrology, PG is primary standard

4. Process Measurements Division Present Gas Pressure Scale: Circa 1990 Pressure realization through gas piston gauge (PG) artifacts, 10 kPa to 110 MPa –Overlaps with Hg manometer (UIM) up to 360 kPa Multiple overlapping ranges of decreasing piston diameters (336 mm 2 to 8.4 mm 2 ) Working standards are GEIS (Ruska) 2465 and 2470 piston gauges Traceability to Gas Thermometer Manometer (GTM) at 27 kPa and 95 kPa in 1989 Standard uncertainties: 6.5 ppm to 20 ppm Up to 7 MPa, distortion is neglected, incorporated as higher uncertainty

5. Process Measurements Division Greg Driver performing crossfloat calibration

6. Process Measurements Division Why Change? GTM dismantled early 1990s, original traceability no longer possible Discrepancy exists for PGs traceable to GTM and UIM Direct traceability was limited to 95 kPa Customers want lower uncertainties –other NMIs can do better. Present NIST values conservative –High quality commercial PGs have much better repeatability than NIST uncertainty Large diameter (35 mm & up) PGs can now be characterized as primary standards Easier calibration of secondary standard PGs to primary PG than to manometers

7. Process Measurements Division PG38 and PG39: Acquired by NIST 1989 from Ruska Instrument Co. Used as check standards for UIM J. Schmidt performed extensive modeling of dimensional data and distortion Metrologia 43 (2006) establishes PG38 and PG39 as primary standards Upright Inverted

8. Process Measurements Division Summary of Characterization and Validation Dimensional measurements and force modeling Comparisons to NIST 3 m UIM Crossfloat comparisons of PG38 vs PG39 Characterization equation: Uncertainties: u(A 0 ), u(b)

9. Process Measurements Division Dimensional Measurements and Force Modeling PG38: NIST Dim 1989, PTB dim 2003 PG39: NIST Dim 1989, PTB dim 1999, 2003 PTB dim: diameters, roundness, and straightness of piston and cylinder –12.5 nm and 25 nm standard uncertainty on piston and cylinder diameters translates to 1 ppm std. unc. on effective area. –However, forces on piston flanks must be modeled and included. Change in diameters between 1999 and 2003 was 0.1 to 0.8 ppm A 0 (Schmidt) assembled dimensional data into “bird-cage”. Used various flow models in gap (viscous, viscous/molecular) b from elasticity theory, FEA, and flow models A 0, b chosen from distribution of modeling results

10. Process Measurements Division Dimensional Measurements and Force Modeling u(A 0 )/A 0 = 3 ppm b = 8.97x10 -12 Pa -1, u(b) = 1.12x10 -12 Pa -1 At 1 MPa: u(A e )/A e = 3.2 ppm

11. Process Measurements Division Comparisons to NIST 3 m UIM to 360 kPa Range of b / 10 -12 Pa via UIM PG38PG39 5.7 to 8.03.1 to 8.8 A 0 agrees within 1-2 ppm. Stable over time b via UIM slightly lower, but limited to 360 kPa

12. Process Measurements Division Crossfloat Comparisons of PG38 and PG39 Operating both inverted/upright allows testing 4 distortion ratios Agreement between experiment and theory confirms b, u(b) Area ratio further confirms dimensional data, UIM

13. Process Measurements Division Revised Pressure Scale to 17 MPa Primary Standard UIM 0.3 MPa Dimensions and Model PG38 1 MPa PG39 1 MPa PG37 1.4 MPa PG34 1.4 MPa PG13 7 MPa PG35 7 MPa PG32 17 MPa PG23 17 MPa PG22 0.1 MPa PG36 0.1 MPa PG28 0.3 MPa PG29 0.3 MPa DHI380 7 MPa b Gauge Max P (MPa) 2465 Series Area (mm 2 ) PG22, PG36 0.2TL336 PG28, PG29 0.35TTL336 PG34, PG37 1.4C84 PG13, PG35 7V8.4 PG23, PG32 17D8.4 DHI380: DHI PG7601, 49 mm 2

14. Process Measurements Division Propogation of Scale TL, TTL, and C series: direct crossfloat vs PG38 and PG39 in gauge mode –Elimination of intermediate step, less extrapolation Crossfloat vs “twin” of series as check V series (7 MPa): crossfloat vs C series –elasticity theory on PG7601 –crossfloat to PG13 to DHI 380 to set b D series: crossfloat vs V series, extrapolation of distortion

15. Process Measurements Division PG28 (TTL472) Type B uncertainty from measurement equation using PG38/PG39 uncertainty Type A from crossfloat and data fit -4.8 ppm shift from 1994 traceability Uncertainty bars: k=1 4.2 ppm (min)

16. Process Measurements Division Area Ratio, PG29/PG28 Direct crossfloat confirms traceability to PG38, PG39 Area ratio consistent with existing traceability

17. Process Measurements Division TL and C Series PG22 (TL460) PG34 (C74) C series requires modest extrapolation to 1.4 MPa u/A: 5.2 ppm (min) u/A: 4.3 ppm (min)

18. Process Measurements Division V Series (PG13: V373) b via calibration to DHI380 (PG7601) –PG7601: cylinder diameters uniform –b DHI380 from elasticity theory –b DHI380 = -2.36x10 -12 Pa -1 b PG13 = 2.66x10 -12 Pa -1 Primary MethodIntermediate Stepsb (Pa -1 ) / 10 -12 Elasticity theory, DHI380PG13 v DHI380 Xfloat2.66 FEA model, PG13none3.30 FEA model, PG23 (2465 D)PG13 v PG23 Xfloat2.64 Capacitive meas., PG13none1.95 Capacitive meas., PG23PG13 v PG23 Xfloat1.16 b of PG13 determined from independent methods 7.9 mm D32 mm D DHI380

19. Process Measurements Division PG13 (V373) A 0 from crossfloat to C series with fixed b 10:1 area ratio in crossfloat Uncertainty u(b) from PG38, PG39 u(A e ) ranges from 5.8 to 9 ppm (was 20 ppm)

20. Process Measurements Division D series: PG23 (D49) Crossfloat to PG13, PG35 (V series) to 7 MPa b fixed at common average for PG23 and twin PG32 –PG23 v PG32 crossfloat shows same b in both Uncertainty due to b becomes dominant above 7 MPa u/A: 7 ppm (min)

21. Process Measurements Division New vs old uncertainties of standards TL, TTL, C Series V and D Series Improvement through direct traceability to PG38/39 Improvement through better distortion estimation

22. Process Measurements Division Summary of Revised Uncertainties of Standards StandardPressure Range Existing Uncertainties (k=2) Revised Exp. Uncertainties (k=2) UIM1 Pa to 360 kPa5.2 ppm PG38/PG3920 kPa to 1 MPaN/A6.0 to 6.4 ppm PG22/PG3610 kPa to 150 kPa23 ppm10.3 to 23.6 ppm PG28/PG2920 kPa to 350 kPa13 ppm8.4 to 11.1 ppm PG34/PG3740 kPa to 1.4 MPa18 ppm8.5 to 11.6 ppm PG13/PG350.35 MPa to 7 MPa40 ppm11.6 to 18 ppm PG23/PG320.7 MPa to 17 MPa33 ppm14 to 40 ppm

23. Process Measurements Division Validation of uncertainties via international comparisons CCM.P.M-K1c Key Comparison 5 participants, NIST possesses transfer standards (C-415, V-762) 2 ranges: 79 kPa to 896 kPa, 621 kPa to 6.79 MPa Recent data can be compared to Key Comparison Reference Values –C-415 calibrated against PG38 and PG39 –V-762 calibrated against PG13

24. Process Measurements Division C-415 agrees to KCRV NIST area agrees to KCRV to within 3.5 ppm or less Agreement to KCRV always better than standard uncertainty Agreement to other NMIs always better than standard uncertainty NIST vs KCRV k=1 unc. bars

25. Process Measurements Division V-762 agrees to KCRV NIST agreement to KCRV always better than standard uncertainty Agreement to other IMGC, LNE, NRLM always better than standard uncertainty. Agreement to PTB better than expanded uncertainty. NIST vs KCRV k=1 unc. bars

26. Process Measurements Division Summary of international comparisons Direct validation 79 kPa to 6.8 MPa Bilateral comparison with NPL India: –Used NPLI artifact (NPLI-4, V-607) –Confirms equivalence 0.4 to 4.0 MPa

27. Process Measurements Division Status of Calibration Measurement Capabilities (CMCs) CMCs are statements of the measurement capabilities of each NMI in a database maintained by the BIPM and publicly available on the Web Established as part of Mutual Recognition Arrangement Changes in CMCs must be approved by regional metrology organization (RMO), submitted to BIPM, then approved by the remaining 4 RMOs The revised NIST gas pressure scale: –Submitted for SIM review: May 2008 –Approved by SIM, submitted to BIPM: June 2008 –Currently undergoing inter-regional review

28. Process Measurements Division Summary NIST gas pressure scale to 17 MPa now based on large diameter piston gauges –Dimensional measurements, modeling, UIM comparisons, and crossfloats confirm results Uncertainties reduced over most of scale –Direct calibration to primary std. improves u for PGs to 1.4 MPa –Better distortion estimation improves u for 7 MPa and 17 MPa PGs –Uncertainties depend on pressure Shifts in effective area of standards –5 ppm lower for standards < 350 kPa –9 ppm or less, most other ranges Distortion coefficients included in most PG ranges Results supported by Key Comparisons New uncertainties expected in NIST cal service late 2008 early 2009