atdotde
Contributor
I am sure the GERG-2004 model is much better researched than the Subsurface calculation (which is not really a model) and has much broader applicability. It is a model because it has a theoretical background (and justification for all the possible terms).
I should say, however, that research into these equations of state is mainly motivated by applications to the oil and gas drilling industry. Thus it is likely that their models are optimized for pressure/temperature/gas mixture values that are relevant there and which probably differ from those relevant to filling scuba cylinders at room temperature with breathing gases.
For us the only relevant variable is the compressibility for trimix mixes between 1 and 300 bar at room temperature. But at least for the three component gases He, O2 and N2 this is simply measured and tabulated and all the Subsurface calculation does is to interpolate these tabulated values. It makes a simple assumption about the compressibility of a mixture like trimix (namely that you can take the weighted average of the compressibilities of the components) but besides that, there is no theoretical justification, simply a pragmatic fit to the actual data.
I am pretty sure that this should be good enough for diving and blending applications.
I should say, however, that research into these equations of state is mainly motivated by applications to the oil and gas drilling industry. Thus it is likely that their models are optimized for pressure/temperature/gas mixture values that are relevant there and which probably differ from those relevant to filling scuba cylinders at room temperature with breathing gases.
For us the only relevant variable is the compressibility for trimix mixes between 1 and 300 bar at room temperature. But at least for the three component gases He, O2 and N2 this is simply measured and tabulated and all the Subsurface calculation does is to interpolate these tabulated values. It makes a simple assumption about the compressibility of a mixture like trimix (namely that you can take the weighted average of the compressibilities of the components) but besides that, there is no theoretical justification, simply a pragmatic fit to the actual data.
I am pretty sure that this should be good enough for diving and blending applications.