Dynamic Building Envelope with Phase Change Material (PCM) and Multi-Layered Fenestration Using Model Predictive Controls

Abstract

As buildings are designed and operated towards reducing energy demand while maintaining a both thermally and visually comfortable environment for occupants, building envelope systems, including windows, external walls, doors, roof, etc., have the potential to reduce building energy consumption, improving the energy flexibility and mitigate greenhouse gas emissions. In this thesis, we have conducted a simulation-based study of a dynamic building envelope with thermally activated phase change material (PCM) panel and adjustable fenestration system. We first explored an integrated simulation approach to coordinate the simulation of daylighting, thermal and energy performance evaluation of the dynamic building envelope. Specifically, we have developed a Radiance model for daylighting performance estimation, and then this model was validated by carrying out a series of experiments under different conditions. The validation results showed that the Radiance modeling approach can generate reliable indoor illuminance estimations with Coefficient of the Variation of the Root Mean Square Error (CV(RMSE)) values of less than 25% and Normalized Mean Bias Error (NMBE) values of less than 15% for all control points.

Second, we investigated a PCM modeling approach to enhance the current PCM model in EnergyPlus to better account for the hysteresis effects of industrial-grade PCM products. Compared to experimental measurements, the enhanced PCM module can have a good agreement with Mean Bias Error (MBE) values of 0.5oC and 0.59 oC for PCM layer average temperature in both complete and incomplete phase change processes, respectively. Using this improved PCM model, we created an EnergyPlus simulation model for Then, an integration approach was developed to co-simulation of the Radiance model and EnergyPlus model in run-time for the studied dynamic building envelope. Moreover, this integrated simulation model can also enable the evaluation of advanced control strategies such as model predictive control (MPC).

Third, the integrated physics-based simulation models were used as a virtual testbed to develop and test the MPC controller. We developed a set of control-oriented models to predict the indoor thermal and visual conditions and the electricity consumption of both electrical lighting system and HVAC systems. These control-oriented models consist of a zone dynamics sub-model, an indoor illuminance prediction sub-model, and an electricity consumption prediction sub-model. These control-oriented models were then used as surrogate models to the physics-based simulation models, and the evaluation results showed that these control-oriented models can generate accurate and efficient predictions for indoor conditions and electricity consumption with CV(RMSE) values of less than 25% and NMBE values of less than 15% for testing dataset.

Finally, an MPC controller was formulated and implemented to adjust the states of the fenestration system and regulate both the electrical lighting system and air-conditioning system in order to minimize the total electricity cost when a time of use utility rate is applied and without violating both the thermal and visual comfort requirements. Compared to a baseline case, the proposed MPC controller can improve the indoor environment comfort and reduce the total electricity cost up to 25% for a clear day with a typical load profile and up to 45.7% on a cloudy day with a moderate load profile