Êtes-vous un étudiant de l'EPFL à la recherche d'un projet de semestre?
Travaillez avec nous sur des projets en science des données et en visualisation, et déployez votre projet sous forme d'application sur GraphSearch.
Sufficient daylight exposure contributes to occupants' productivity and health in buildings. However, excessive daylight ingress induces discomfort glare and increases the cooling load in warm seasons. The performance of manual shading systems is limited by usersâ low interaction frequency. Since sky conditions are dynamic, frequent adjustments are necessary, but these are impractical for users to manage. Various shading automation systems have been proposed to better exploit daylight in buildings, the performance of which is limited by a number of factors, including insufficient glare protection, disturbing movement of slats, privacy issues, and commissioning difficulty. In this thesis, an integrated daylighting control system is proposed and demonstrated to regulate daylight in buildings based on real-time lighting computing with the monitored luminance distribution of the sky and landscape, as a decentralized system.
First, an embedded photometric device (EPD) was designed and validated for improved accuracy in daylighting simulation compared to standard sky models. The EPD is composed of an imaging system and a microprocessor. After calibration, the spectral response of the imaging system is close to the photopic luminosity function, achieving a 8.9% spectral correction error. The luminance detection range spans 1.2x10^2 ~ 3.78x10^9 cd/m^2, covering extremes of both the shadowing landscape and the sun orb luminance. The EPD monitors the luminance distribution of the sky vault and the ground. Based on the generated luminance map, the EPD is able to perform on-board lighting computing. The performance of work-plane illuminance (WPI) simulation was cross validated with a lux-meter array in a daylighting test module under different sky conditions achieving a mismatch of below 10%.
Secondly, since the bidirectional distribution function (BTDF) commonly used in lighting simulation involves bulky data, a compression scheme based on planar wavelet transform was investigated and generic error and the influence on daylighting simulation were also studied at various compression ratios. Results showed that both WPI and daylight glare probability (DGP) are relatively immune to a BTDF compression ratio below 100.
Thirdly, an automated Venetian blind was designed which integrates the EPD as both a sensing unit and a controller based on real-time lighting simulation. The EPD determines an optimal shading position according to the simulation results, to offer sufficient WPI, mitigate excessive solar heat gain (SHG), temper discomfort glare, and maximize outwards view. 'In-situ' experiments demonstrated that the automated Venetian blinds were able to regulate WPI efficiently. The expected SHG saving was estimated to reach 47% in warm seasons. It was also demonstrated to mitigate discomfort glare timely, including veil glare from surroundings. A subjective study conducted with 34 subjects showed satisfaction with regulated daylight provision, glare mitigation, and quietness of slat movement.
Finally, the EPD was applied to control tint states of a split-pane electrochromic (EC) window to secure occupants' visual satisfaction. Experimental results in a full-scale testbed showed that the WPI was within the confined range with regulated daylighting under clear skies 83% of time and the DGP 95% of the time; under partly cloudy skies, the WPI was within the range 62%~68.2% of the time and for the DGP 85%~94% of the time to achieve visual comfort