Statistical-based sensitivity analysis and multi-objective optimizations as a design approach to window-integrated light-shelves based on daylight performance

• Optimization of daylighting performance associated with window design integrated with lightshelves are studied. • A multi-stage parametric sensitivity analysis and multi-objective optimization approaches are proposed. • Parametric sensitivity analysis are used to simplify and facilitate design variables of window design and light-shelves. • The multi-stage approach involves three stages of parametric sensitivity analysis to find the best window configurations and light-shelves parameters • Multi-objective optimization was used to find the directions of the effects of highly-sensitive light-shelves design parameters. This paper aims to address the deficiency of facade daylighting design techniques that balanced daylight provision and multi-criteria daylighting requirements by incorporating the window and light-shelf as two independent variables applied to a case study under the tropic’s climate of Malaysia. The study suggests improvements in energy utilization and daylight provision by implementing various window and light-shelf arrangements. An assessment is conducted to determine the effectiveness of optimizing different light-shelf characteristics in conjunction with ideal window setups. Grasshopper and Rhinoceros are used to parametrically modelled the building and design variables of window and light-shelf. The selected design parameters are genetically optimized by the Octopus plugin using a multi-objective daylighting design that indicating if the zone satisfies the daylight need without uncomfortable glare and lower thermal energy performance. The method's originality and novelty lie in its revolutionary approach to optimisation through the integration of optimisation algorithms with statistically-based sensitivity analysis. The final selection of the optimal window value range with different light-shelf configurations in different window designs is based on the cases that meet multi-criteria daylighting requirements with high regression. Overall, the increase in performance by final optimal cases selected from multi-objective optimization compared to the base case achieved a slight decrease in thermal energy performance with an average between 1.67% to 1.89%, and enhanced daylight availability and visual comfort with an average between 2.56% to 3.91% and 100% for glare.