Implementation and Calibration of a Model to Treat Naturally Ventilated Complex Fenestration Systems in TRNSYS

Adaptive façade systems offer the opportunity to improve building performance and user experience with their ability to adapt the façade configuration to the dynamic variability of the external environment. Nevertheless, the correct deployment of adaptive systems in real buildings is highly dependent on the ability to predict their performance. This is especially relevant in the case of Complex Fenestration Systems (CFS), which are characterized by a complex behavior both from a thermal and daylighting perspective. Often, such CFS are combined with airflow movement when the façade cavity is either mechanically or naturally ventilated, making the performance even more difficult to be characterized. The possibility to use building performance simulation tools to simulate the behavior of these systems integrated in a whole building is central for the proper use and the penetration on the market of these systems. In this framework, a naturally ventilated window with integrated venetian blinds was modeled in TRNSYS and compared with a FEM-based 2D detailed model, developed in COMSOL Multiphysics. Type56_CFS solves thermal calculation and uses the Bidirectional Scattering Distribution Function for describing optical properties of the façade. This model requires the inlet mass flow rate, which was assessed thanks to an ad-hoc implementation of ISO 15099. The numerical modeling of the coupled heat transfer and fluid flow with COMSOL allowed the TRNSYS model to be calibrated. The calibration was carried out by increasing the model complexity, focusing on the inlet ventilation flow rate parameter: (a) firstly, it was provided as an input to the Type56_CFS from the FEM-based simulation and then (b) it was calculated by the ISO 15099 internal model and provided to the Type56_CFS. Using this methodology, it was possible to compare the ability of TRNSYS to simulate the thermal behavior of the naturally ventilated cavity against a FEM-based benchmark. Results show a difference of 5 % after the fine tuning of all TRNSYS-related parameters (40 % as un-calibrated starting value).