A ML-assisted climate-responsive framework for sustainable building envelope design from preliminary phases

Abstract

The construction sector consumed around 40% of global energy and 25% of resource consumption, plays a key role in global warming.

Growing environmental concerns necessitates constructing buildings with sustainable design.

Achieving a sustainable design, however, is a challenging task as it requires to satisfy conflicting criteria including embodied energy (EE), operating energy (OE), cost, and demolition and waste generation (DWG). This gets more critical in the initial design stage where there is a limited information available.

Although researcher in this field have suggested optimisation strategies for identifying sustainable design solutions, the necessity for significant number of simulations hinders their feasibility and commercial adoption.

To address gaps in prior studies, this study aims at developing a machine learning (ML)-based integrated optimisation system to achieve sustainable designs from the initial design stage.

For this purpose, an integrated framework consisting of climate-responsive ML-based models is proposed to estimate material cost, DWG, and EE, as well as predict OE, complementing a constrained multi-objective optimisation developed via genetic algorithm.

The findings indicate that the proposed framework is a viable avenue for addressing the challenges associated with simulation-based optimisation approaches. The outcomes of this research will guide designers in finding sustainable design alternatives for buildings.