nist-aharvey
commented
1 month ago If you look in the GUI under Substance, Fluid Information, to see the method used to compute viscosity, you will see that it is just using a corresponding-states model because there was not enough experimental data to make a real correlation (it looks to me like there are no experimental liquid-phase viscosities at all, and only a few points for the vapor). It gives the deviations of the few experimental data sources, including 7.2% from Dunlop. That amount of difference would not be unexpected from a corresponding-states estimation technique. Since you do have some vapor-phase data, it might not be unreasonable to shift the REFPROP results up by 7% or something if you were using the vapor viscosities (especially if the data from other sources like Casparian and Kochubey deviate similarly).
Edited to add: I see that the paper of Dunlop reported measurements for a number of fluids, including well-known things such as methane. So the accuracy of his methane and ethane values would be a good indication of whether you should trust his result for fluoromethane enough to adjust REFPROP vapor values to match it.
ianhbell
AndrewPrice775
I'm wondering if there is a mistake in the REFPROP viscosity output for CH3F (R41), Fluoromethane.
REFPROP OUTPUT = mu = 10.955 uPa*s P= 0.1 MPa T = 298.15 K
From [1,2]: mu = 11.79 uPa*s
A ~7.5% difference between the values.
Versions
REFPROP Version: Version 10.0. Reference Database # 23 Operating System and Version: Windows 10 Access Method: Coolprop Python with REFPROP backend and REFPROP in Excel.
References:
[1] Dunlop, Peter J. (1994). "Viscosities of a series of gaseous fluorocarbons at 25 °C". The Journal of Chemical Physics. 100 (4): 3149–3151. doi:10.1063/1.466405 [2] https://en.wikipedia.org/wiki/List_of_viscosities#Organohalides:~:text=Fluoromethane,%5B12%5D