AI System for Predicting Embodied Carbon (ASPEC) in infrastructure projects
Project summary and details on ASPEC.
Construction and infrastructure projects account for 36% of the global energy use and generate 39% of all carbon emission in the world (United Nation Environment Programme, UNEP, 2017). The World Green Building Council (WGBC) reported that the carbon emission released before built assets are used (known as upfront carbon) is expected to be up to 50% of the whole carbon footprint of new constructions by 2050. The built environment sector therefore has vital role to play in responding to climate emergency (WGBC, 2019).
The UK government has set 2050 target for achieving a net zero greenhouse gas emission (Department for Business, Energy & Industrial Strategy, 2019). In the construction and infrastructure sector of the economy, the government has also set a target of a 50% reduction in carbon emissions, through the Construction 2025 strategy to transform the sector (UKGBC, 2017).
Following the BREEAM sustainability requirements, construction organisations and clients have continuously worked to achieve the carbon emission reduction target set by the government through the use of various embodied carbon estimation techniques. The construction organisations over the years learnt a lot about embodied carbon performance on projects. These lessons learnt have continue to influence the way organisations address the issue of embodied carbon on infrastructure projects. The current methods for calculating embodied carbon of infrastructure projects are tedious and requires a lot of man-hour (Embley, 2019). In addition, existing methods do not provide design support for driving down the embodied carbon and carbon footprint of projects.
This project will develop an AI system for Embodied Carbon Analytics and prediction of infrastructure projects based on BIM designs, materials carbon data and lessons learnt on past projects. The embodied carbon data of past projects will be used to train and develop deep learning models (such as deep neural network, convolutional neural network). Advanced big data analytics techniques will be used to develop the embodied carbon analytics platform.
Two years (November 2020 – October 2022).
- Winvic Construction Limited
- Costain PLC
- Edgetrix (UK) Limited.
- AI-based Embodied Carbon Calculator (ECC): This subsystem will use historic embodied carbon data from previous construction projects to develop machine learning models to predict the embodied carbon based on construction project design. Parametric data from the construction project design will be used to train and develop the machine learning models.
- Embodied Carbon Analytics Platform (ECAP): The ECAS will provide a platform where what-if analysis of embodied carbon of projects will be carried out with the aim of identifying alternative design specifications that reduce embodied carbon of the project. The tool will use advanced big data analytics method that include predictive analytics, prescriptive analytics and visualisation.
- BIM-Based Design Support Tool for Embodied Carbon Analytics (B-DST): The B-DST will provide a platform to support the design team at the design stage of infrastructure projects. The B-DST functionality will be provided through the ASPEC 's Application Programming Interface (API) that will be available as a plug-in to the existing BIM development software (e.g. Revit).