Quantification of uncertainty in product stage embodied carbon calculations for buildings

• Published in: Energy and buildings 2021-11, Vol.251, p.111340, Article 111340
• Main Authors: Marsh, Ellen, Orr, John, Ibell, Tim
• Format: Article
• Language: English
• Subjects: Buildings; Carbon; Carbon mitigation; Coefficient of variation; Concrete; Construction; Data points; Embodied carbon; Embodied carbon coefficients; Energy consumption; Environmental impact of buildings; Error analysis; Extreme values; Greenhouse gases; LCA (life cycle assessment); Life cycle analysis; Life cycle assessment; Life cycles; Mathematical analysis; Monte Carlo simulation; Product variability; Recycled materials; Superstructures; Uncertainty analysis
• ISSN: 0378-7788; 1872-6178
• DOI: 10.1016/j.enbuild.2021.111340

Decarbonisation of the energy industry and enforcement of strict targets for operational energy consumption means that non-operational greenhouse gases (GHG) emissions, also known as embodied carbon (EC), will soon represent the majority of whole life carbon associated with buildings. EC assessments are often presented as deterministic, single-point values but contain a high degree of variability which is typically unacknowledged. Common sources of uncertainty are variability, data gaps, measurement error and epistemic uncertainty such as absence of detailed material specification (e.g. manufacturer, concrete mix, recycled content etc). Particularly during early design stages when such information is unconfirmed, average material data is used by necessity. While some material databases and LCA software can provide ranges of embodied carbon coefficients (ECC) between some materials and/or the uncertainty within individual manufacturers’ carbon data, the practice of reporting this is uncommon and has limited practicality for whole building assessments. This paper presents a simple procedure that selects the highest impact materials of the EC of an asset and implements a Monte-Carlo simulation to estimate the uncertainty behind the product stage EC assessment. Material coefficients of variation (CoV) are obtained from database values where available, and interpolated values are used in the absence of such data. A product stage EC assessment of a UK educational building, initially undertaken using single data points for each material, gave an EC prediction of 525 kgCO2e/m2 GIFA. Two scenarios were then assessed using our proposed procedure: 1) the full building scope and 2) substructure and superstructure only. It was demonstrated that, for scenario one, the EC can range from 50 to 140% of the original result when considering the extreme results from the Monte-Carlo simulation. Scenario one (considering the full building scope) resulted in an average EC value (mean ± CoV) of 526 kgCO2e/m2 GIFA ± 10.0%. The second scenario (sub- and super-structure only) resulted in an average EC value of 312 kgCO2e/m2 GIFA ± 11.9% with a full range of 45–155% of the original result. This paper shows that a straightforward uncertainty analysis procedure can support designers in understanding the possible range of asset product-stage EC and, therefore, inform construction product selections at an early stage where detailed information is not known. The variation also gives a deg