@Lovborg, @jac.lu

The question of modeling ventilation in an atrium with EP has come up multiple times on the Unmet Hours forum, so you can search through there to get a detailed idea of the various nuances and trade-offs associated with the problem. I would summarize the general response as: there is no good way to do this in EP without the integration of a CFD, but if you don’t have access to a CFD and don’t mind the loss of accuracy your best option is to model the atrium with a single zone with the AirflowNetwork (AFN) object.

The primary limitation with EP in this context is that the zone air is assumed to be a well-mixed volume of air with no temperature stratification, and EP doesn’t model the more complex physics associated with the combined heat and mass flow through zones. The AFN object partially solves the mass flow problem by explicitly modeling the conservation of mass via the pressure difference across the envelope and through zones (which drives airflow dynamics), but this occurs at the resolution of the network of zones and surfaces, where each zone is still assumed to contain a well-mixed volume of air. There is the RoomAir object which does model sub-zone airflow, but requires some prior inputs (about the temperature stratification I believe), which requires information from a CFD. Additionally, the AFN can be implemented in HB-Energy, but not the RoomAir object, so without access to CFD data, or within HB, your best option is to use the AFN to model the atrium zone.

Regarding the subdivision of thermal zone debate, the airflow experts seem to lean towards not subdividing the zone, given (historical) limitations of EP in modeling long-wave and short-wave radiation through zones, and that there is no reason to assume the temperature stratification in stacked EP zones correspond to a large vertical space. I have seen some modelers suggest splitting the zone and using the ZoneMixing and ZoneCrossMixing objects to model stratification/pressure difference, but I assume they mean with prior data/engineering assumptions to set appropriate airflow rates, which again, leads back to the CFD problem.

Personally, I do wonder if the tradeoffs have shifted on the bias towards subdivision though, since the updated Airwall construction models short and long-wave radiation through zones, and airflow. There’s still the limitation of the multi-zone stratification not neccessarily corresponding to an actual single-zone stratification, but given that the AFN can model two zones with a horizontal opening, a minimal two-zone subdivision, with pressure difference modeled through a large horizontal opening might be more accurate then assuming the whole space is well-mixed. That might be a good question to ask on the Unmet Hours forums.