Shikang Wen / 温世康

Shikang Wen / 温世康

Research in building energy, computational design, and intelligent built environments.

Daylight and shading optimization

Published research Nov 2022

Overview

This project developed a performance-based framework for selecting and designing shading strategies in a university library. The work balanced glare reduction with satisfactory daylight conditions rather than optimizing either objective in isolation.

Floor plan of the simulated library reading area, an axonometric view of the building, and photographs of glare conditions at the study desks.
The case-study library: simulated region, sensor layout, and observed glare conditions along the west facade.

Method

Annual daylight simulations were combined with an artificial neural network and a multi-objective genetic algorithm. Four strategies - perforated aluminum sheets, vertical slats, and south- or north-facing serrated windows - were evaluated through spatial glare autonomy and a daylight-satisfaction indicator derived from field-survey data.

Flow chart linking daylight simulation, neural-network prediction, and the multi-objective genetic algorithm.
Three-stage workflow: annual daylight simulation, neural-network surrogate modelling, and multi-objective optimization.
Parametric diagrams of a perforated aluminum sheet, vertical slats, and serrated windows with their design variables labelled.
Design variables of the three shading device families used to generate the search space.

Key findings

The four strategies produced distinct Pareto performance curves. North-facing serrated windows were the most effective at reducing glare, while vertical slats provided the most satisfactory illuminance levels. The near-optimal perforated-sheet, vertical-slat, north-facing serrated-window, and south-facing serrated-window designs reduced the complementary glare indicator by 9%, 22%, 8%, and 8%, respectively, while maintaining daylight satisfaction close to the original space.

Pareto fronts of the four shading strategies plotted as spatial daylight visual autonomy against spatial glare autonomy.
Pareto fronts of the four strategies, compared against the performance of the original space.

Contribution

Conceptualization, data curation, methodology, software, visualization, and original-draft writing.