Indoor daylight has significant influences on building energy use and occupant health, and studies suggest a targeted illuminance range to achieve an overall performance, the duration of which has been adopted to evaluate the urban design. In previous studies on urban form, solar availability, and daylight, there are two streams of research, namely, the parametric design, and the design optimization. As urban design often involves diverse urban form patterns, the former stream focuses on whether a uniform pattern or random pattern has better solar availability or daylight while the latter directly searches for the best performative design through optimization. However, in the parametric stream, the definitions and sampling for uniform and random design patterns were largely limited; in the optimization stream, the form parameters to optimize were often limited, and searching for the best design only provided one-sided information as design heuristics for generalization. Moreover, most of the studies in both streams focused on general solar availability and daylight, and it is largely unclear how urban form influences the comprehensive daylight duration metric. This study investigates the indoor daylight performance of different urban forms with two specific sets of questions: First, is random urban form better than regular urban form for daylight as stated in the first research stream? And second, what are the best and worst urban forms for daylight and how much performance difference they make, extended from the discussions in the second research stream? To answer these two sets of questions, this study used a grid-based hypothetical design setting as the test case. The study has two parts. In the first part, 500 parametric models were developed to compare daylight availability between five groups of urban form with different levels of uniformity/randomness. In the second part, designs with nearly best and worst daylight are generated using a genetic algorithm and compared. It was found that on average, randomness improves daylight duration, but some random forms perform poorly compared to uniform urban forms, which is echoed by a large performance variation of random urban forms from the optimization results. The findings provide a better understanding of design performance for daylight, which can be used as a general heuristic to inform urban design practice.
Keywords solar availability, urban design, daylight, optimization, parametric design