Hello Bill,
I'd support the other Peter's view w/r/to a direct modeling of the greenhouse's cement foundation. Being a theoretically rather uneductated practical scientist
I'd even suggest to model the whole system as 3-D, as the calculational power that is required afterwards is not too excessive and you can directly relate to a scientific/engineering hands-on experience.
This suggestion brings me back to my former idea of a (somewhat) flat insulation of the surrounding instead of a deeper excavation of the greenhouse foundation. Imagining the 3-D situation my idea behind this is as follows: If you regard the heating volume of interest in the ground as roughly being a sphere-like volume of ground below the greenhouse, with a deep dug insulation just following the foundation you would cut a sector in the ground that the house will draw heat from. If you laid out the insulation flat on the ground (or slightly below), you would just tilt the angle of the sectorizing from vertically into the ground to horizontal, thus creating a larger semi-sphere of warm ground even in colder atmospheric conditions.
There are studies on the atmospheric heat penetration into the ground, saying that already in a depth of about 2 m the seasonal amplitude is near to neglegible (sorry, but I do not remember the precise figures). So, if you lay out your insulation flat for 2 m in the vicinity of the greenhouse the cold would have to creep this distance (from the unprotected ground to the outer limit of the greenhouse) to affect the heat balance, and would protect the soil below from cooling out.
I'd consider the additional amount of preserved heat - namely the one in the soil in the angle between vertical and horizontal from the corner of the greenhouse - as mostly neglegible from a pure heat balance point of view. But you would definitely gain (and certainly not lose) a small amount, and would certainly require much less digging efforts for its realization. You would, of course, create a margin of cool soil surface around the greenhouse, but I guess this is not an opposing argument.
With respect to simulation I'd take the soil temperature in a depth of about 4 or 5 meters as constant, would research the heat capacity and heat conductance values of the soil you have at your place, take these as input parameters in a respective full-volume simulation and leave out any artificially set heat gap parameters. Additionally you could model the the surface temperature of the non-insulated soil and might take respective values from the many EnergyPlus climate models that would give you a respective temperature time series. But I guess you might as well forget about this effect.
Unfortunately, Goggle Sketchup is not available in Linux (lest you used wine, which I have not had a great deal of success with).
I wouldn't go for Wine but a full-fledged Windows virtualizer: I've made the best experiences with VirtualBox. In fact I need it in an about three-monthly turn to make my tax statements, as this area of applications is not yet sufficiently covered by Linux applications.
Regards,
Peter
(the one in Germany
)