Parametric embodied carbon prediction model for early stage estimating.

The focus of carbon management is shifting from operational carbon to embodied carbon as a result of the improved operational energy efficiency of buildings. Measuring and managing embodied carbon right from the early stages of projects will unlock a range of opportunities to achieve maximum reduction of emissions which could not be achieved otherwise during the latter stages. However, measuring embodied carbon during the early stages of design is challenging and highly uncertain due to the availability of limited design information. Therefore, the research presented in this paper addresses this problem in a structured and an objective way. A parametric embodied carbon prediction model was developed using regression analysis to estimate embodied carbon when only minimal design information is available and with less uncertainty. The model was developed by collecting historical data of office buildings in the UK from four different data sources and estimating embodied carbon by combining several estimating techniques. Wall to floor ratio and the number of basements were identified as the model predictors with a model fit of 48.1% (R2). A five-fold cross-validation ensured that the model predicts within the acceptable accuracy range for new data. The developed model had and accuracy of {plusmn}89.35% which is within the acceptable for an early stage prediction model. In addition, the need for standardising embodied carbon measurements and to develop embodied carbon benchmarks to facilitate embodied carbon estimating throughout the project lifecycle was identified