Generative design of walkable urban cool spots using a novel heuristic GAN×GAN approach

Abstract

Enhancing pedestrian-level outdoor thermal comfort (OTC) and walkability in urban environments is vital for mitigating thermal risks induced by global warming. This study introduces walkable urban cool spots (WUCS), a novel concept which incorporates pedestrian walkability into the design of thermally comfortable urban street blocks. The methodology combines OTC simulation, pedestrian walkability analyses, and surrogate models based on integrated generative adversarial networks (GAN×GAN)-enhanced genetic generative design (GGD) approach. And maximise the design metrics of WUCS through the multi objective optimization algorithm in GGD. The method is tested in To Kwa Wan, an urban renewal area in Hong Kong. Results demonstrate that it can efficiently identify high-performing WUCS solutions, with proportion of WUCS areas increasing from 53.3% to 74%, and the number of walkable urban cool spots increasing by 51%. The average universal thermal climate index (UTCI) in high walkability areas decreasing from 31.7°C to 30.5°C, and total average UTCI during extreme heat days dropping from 31.73°C to 31.22°C. This research thus contributes an efficient design tool for enhancing both pedestrian-level OTC and walkability in urban environments.