Flow meters that rely on measuring the pressure drop across a laminar flow element can be accurately calibrated using a benign surrogate gas (such as helium or nitrogen) and then used to meter process gases (Wright et al., 2012), provided that accurate values of the surrogate-to-process-gas viscosity ratios are available.
Pope et al., 2012 demonstrated that the fluid-dependence of the complex calibration curves of turbine meters can be understood as a function of fluid's "kinematic viscosity", which is the ratio of the shear viscosity to the density. The database SEMIPROP can be used to calculate the densities of process gases from the virial equation of state and the NIST Chemistry WebBook provides densities of heat transfer fluids.
The gas flow through critical flow venturis (nozzles) depends upon the stagnation temperature and pressure upstream of the nozzle and on the real gas "critical flow function." (See Wright and Johnson, 2009). The real gas critical flow function is now available in NIST Standard Reference Database 23 (REFPROP), Version 9.1 or later.
NIST has two on-line databases that include accurate gas viscosity values:
[Note: as of February 2014, the NIST Chemistry WebBook provides data from Version 7 of "NIST Standard Reference Database 23 (REFPROP)"; The latest version of this database is for sale through NIST's Standard Reference Data program.]
The Fluid Metrology Group contributed highly accurate viscosity values to these databases using a novel, highly effective combination of theory, critical assessment of literature data, and direct measurements.
The Fluid Metrology Group applied gas viscosity data and a physical model of the laminar flow meter (LFM) to improve the extrapolation of LFM calibrations from one gas to another. Wright et al., 2012 calibrated three models of commercially-manufactured, laminar flow meters at four pressures (100 kPa, 200 kPa, 300 kPa, and 400 kPa) with five gases (N2, Ar, He, CO2, and SF6) over a 10:1 flow range using NIST's primary flow standards as references. We combined three items: (1) the calibration data acquired with N2, (2) gas-property data from NIST's database REFPROP 9.0, and (3) a physical model for each LFM that accounts for the effects of viscosity, entrance and exit effects, gas expansion, gas non-ideality, and slip. This combination predicted the calibrations for the flow of Ar, He, CO2, and SF6 with a maximum error of 0.8% for Reynolds numbers Re < 500. Under these conditions, the present LFM model allows prediction of calibration results for other gases with approximately 3 times lower uncertainty than conventional approaches that plot the flow coefficient as a function of the viscosity coefficient or Re.