Efforts to combat climate change demand global action to promote a sustainable development worldwide. Given the construction industry's significant energy consumption, it is crucial to curtail energy consumption and CO2 emissions in this sector. This study aims to predict and trade-off four essential objectives: time, cost, energy consumption, and CO2 emissions. Initially, various factors affecting time and cost was identified using the CoCoSo method. Subsequently, 11 machine learning algorithms were employed to predict all four targets. Results revealed that support vector regression achieved an accuracy of 95.77% in project duration prediction, while the neural network algorithm demonstrated accuracy of 93.64%, 97.63%, and 97.99% in predicting cost, energy consumption, and CO2 emissions, respectively. Additionally, the research findings from six scenarios highlighted the non-linear relationship among project objectives. Deviating from the optimal point resulted in increases on both sides, with more significant impacts observed at values less than optimal duration. This research equips engineers with a valuable tool for accurately predicting project outcomes by adjusting input values, ensuring successful project execution and efficient resource allocation. Additionally, the framework's predictive abilities are beneficial for companies with energy and CO2 policies, influencing various sectors, regions, and policies tied to energy, emissions, costs, climate change, ESG, and UN SDGs. Its application extends to shaping regulations, strategies, and targets.
Read full abstract