Paper Conference

Proceedings of Building Simulation 2021: 17th Conference of IBPSA

     

Evaluating and improving outdoor thermal comfort using surrogate modeling and optimization techniques

Elizabeth Young, Caitlin Mueller, Christoph F. Reinhart
Massachusetts Institute of Technology, Cambridge, Massachusetts, USA

DOI: https://doi.org/10.26868/25222708.2021.30872
Abstract: Biometeorological indices, such as Universal Thermal Climate Index (UTCI), have recently been linked to resident presence in public outdoor spaces offering urban planners and landscape designers a new design criterion to improve these spaces. Example interventions are walkway canopies to enhance pedestrian thermal comfort. While simulation tools for UTCI are becoming available, the underlying workflows are highly computationally intensive, making them less attractive for iterative design space explorations of isolated urban features such as the above mentioned, bespoke canopy. To overcome this challenge, this manuscript introduces a surrogate-modelbased workflow that predicts annual UTCI distributions under a parametrically-defined canopy structure. An optimization algorithm is then used to generate a structure that is optimized for annual comfort conditions. The workflow is applied to two cities - Boston and Singapore. The trained surrogate models for each city are able to predict hourly UTCI with a root mean squared error of 1.7°C for Boston and 0.4°C for Singapore. With the proposed workflow, 6 hours of model-preparation enabled us to explore 500 distinct designs in 5 minutes a feat that would have taken at least a week if done manually.
Keywords: Universal Thermal Climate Index (UTCI), Outdoor Thermal Comfort, Optimization, Urban Interventions, Urban Commute
Pages: 941 - 948
Paper:
bs2021_30872