Estimating Embodied Carbon in Sri Lankan buildings: An early design stage prediction model

Abstract

Growing concerns over carbon emissions and the drive to create net-zero carbon buildings has led to a greater focus on reducing embodied carbon (EC) of buildings. Nevertheless, Sri Lanka is lagging behind in this area. EC estimation is the main driver towards EC reduction. In order to make a considerable impact in terms of EC reduction, it is recommended to estimate EC during the early design stage of buildings, where more reduction opportunities lie. Although a few studies have been conducted on the EC impacts of different buildings in Sri Lanka, they have focused on the later stages of buildings, due to the time-consuming and complex nature of the estimation process and limited data availability during the early design stage. Therefore, this study aimed to develop a parametric model to predict EC during the early design stage of buildings with minimal time and effort. The focus of this research was confined to office buildings in Sri Lanka. The prediction technique of multiple linear regression (MLR) analysis was merged with life cycle assessment (LCA) in developing the model. Assessment was limited to the cradle to gate system boundary and the elements of substructure and superstructure that cover the skeleton of a building: substructure, frame, upper floors, external walls, roof, stairs and ramps, and internal walls and partitions. After performing various statistical tests for model development, adequacy checking, and validation, the EC prediction model with statistically significant independent variables (IVs) was determined. The model included log gross internal floor area (GIFA) and log external wall area (EWA) as the statistically significant IVs to predict EC at the early design stage of office buildings in Sri Lanka. The model demonstrated a coefficient determination (R2) of 97.6%, indicating a good model fit, while ANOVA revealed F-test (F=351.699, p<0.001) and T-test (log GIFA t= 9.640, p<0.05); (log EWA t= 2.555, p<0.05) results indicating the model and its independent variables are statistically significant. Based on very basic IVs, this tool helps built environment professionals to estimate EC of office buildings during the early design stage with minimal time and effort, enabling them to take relevant EC reduction decisions.

Date of Award	25 Feb 2022
Original language	English
Awarding Institution	Northumbria University
Supervisor	Zaid Alwan (Supervisor), Barry Gledson (Supervisor), Rosie Parnell (Supervisor) & Nirodha Fernando (Supervisor)