I need to model for the thermal performance of putting semi-transparent (1/2 opaque PV and 1/2 clear glazing) PV modules on the roof of a glazed atrium with an ventilated air gap of 15cm. A simple PV module is considered to be fully opaque on energyplus and therefore does not take into account the transparent portion of the module. In order to take into account the thermal properties (transmittance, reflectance etc) of the PV panels and allow for the air gap, I modelled it as a roof diffuse shade. I entered the values for the transmittance, reflectance etc of the semi-transparent PV panels as a "shade" located 15cm away from the roof glazing, allowing for air to flow. I then modeled the same thermal characteristics as an integrated roof glazing, without an air gap and compared the cooling energy. The cooling energy required for the "shade model with an air gap" is higher than the "Integrated roof glazing without an air gap". This seems unreasonable because of the air gap that would undoubtedly cool the shade. For example, the SHGC of the integrated roof glazing is 0,13 and the the combined (glazing:0,6+Shade: 0,20) SHGC is 0,12. So either they should give the same cooling energy demand or the ""shade model with an air gap" should give a lesser cooling demand. However, this is not the case. The "Integrated roof glazing without an air gap" gves a lower cooling energy demand. Why is this so? Am I modelling this scenario correctly? Are there other ideas to model a semi-transparent PV mounted on the glazed roof of a building with an air gap?