Revision history [back]

Modeling the multi-dimensional heat transfer into the ground from a building foundation can be very computationally intensive. The actual ground temperature varies significantly in three dimensions and depends on the climate, the foundation shape, and the level of insulation applied to the foundation. There are few whole-building energy simulation tools available that actually simulate multi-dimensional heat transfer into the ground.

Because the actual computation is so intensive, there have been several approaches to come up with simplified or pre-processed results. Of the tools available now, I think that the best approach (considering computation time and overall capability) is described pretty well in this DOE-2 article (the method can be applied to just about any simulation engine).

That said, there are several people working on improvements to estimating ground heat transfer:

• TRNSYS has a ground-coupling library that performs detailed 3D foundation heat loss calculations. This library requires a fair amount of work from the modeler to create the numerical calculation domain (e.g., meshing, defining cell properties, etc.) and takes a considerable amount of time to run.
• The folks at Oklahoma State are working to improve on the pre-processors used for basements and slabs in EnergyPlus. This work will integrate the calculations into EnergyPlus's time-step heat balance. These are also 3D and require a bit of effort from the user to define the numerical calculation domain. I'm not certain about when this will be available, but I would expect it sometime in the next two releases.
• Big Ladder (Disclaimer: I'm talking about my own work here) is working on a separate foundation heat transfer tool called Kiva. Kiva is more of a framework for testing and creating simpler faster models. It can run (slow) 3D calculations, or quick 2D simulations. Kiva is still under development, but it's source code is open-source and hosted on GitHub for those interested in learning more.

With every numerical problem, there is a significant trade off between accuracy and computation time. I am not aware of a tool that strikes the right balance for typical building simulation analyses, but I'd like to think that we (as a community) are getting closer.

Modeling the multi-dimensional heat transfer into the ground from a building foundation can be very computationally intensive. The actual ground temperature varies significantly in three dimensions and depends on the climate, the foundation shape, and the level of insulation applied to the foundation. There are few whole-building energy simulation tools available that actually simulate multi-dimensional heat transfer into the ground.

Because the actual computation is so intensive, there have been several approaches to come up with simplified or pre-processed results. Of the tools available now, I think that the best approach (considering computation time and overall capability) is described pretty well in this DOE-2 article (the method can be applied to just about any simulation engine).

That said, there are several people working on improvements to estimating ground heat transfer:

• TRNSYS has a ground-coupling library that performs detailed 3D foundation heat loss calculations. This library requires a fair amount of work from the modeler to create the numerical calculation domain (e.g., meshing, defining cell properties, etc.) and takes a considerable amount of time to run.
• The folks at Oklahoma State are working to improve on the pre-processors used for basements and slabs in EnergyPlus. This work will integrate the calculations into EnergyPlus's time-step heat balance. These are also 3D and require a bit of effort from the user to define the numerical calculation domain. I'm not certain about when this will be available, but I would expect it sometime in the next two releases.
• Big Ladder (Disclaimer: I'm talking about my own work here) is working on a separate foundation heat transfer tool called Kiva. Kiva is more of a framework for testing and creating simpler faster models. It can run (slow) 3D calculations, or quick 2D simulations. Kiva is still under development, but it's source code is open-source and hosted on GitHub for those interested in learning more.

With every numerical problem, there is a significant trade off between accuracy and computation time. I am not aware of a tool that strikes the right balance for typical building simulation analyses, but I'd like to think that we (as a community) are getting closer.