Estimating construction waste generation in the Greater Bay Area, China using machine learning

Area of Science:

• Environmental Science
• Waste Management
• Data Science

Background:

• Reliable construction waste generation data is essential for effective waste management but is often scarce in developing economies due to inadequate recording systems.
• The Greater Bay Area (GBA) in China experiences significant construction activity, highlighting the need for accurate waste generation estimates.

Purpose of the Study:

• To develop a reliable and accessible method for estimating construction waste generation using limited, publicly available data.
• To estimate construction waste generation specifically within the Greater Bay Area (GBA) in China.
• To compare the effectiveness of various waste quantification models.

Main Methods:

• Collected 43 sets of annual data (2005-2019) on socio-economic factors, construction activities, and construction and demolition (C&D) waste generation from local authorities.
• Analyzed the data using four machine learning models: multiple linear regression, decision tree, grey models, and artificial neural network.

Main Results:

• All calibrated machine learning models demonstrated acceptable performance, with testing R-squared values ranging from 0.756 to 0.977.
• The study estimated that the 11 cities in the GBA generated approximately 364 million m³ of construction waste in 2018.

Conclusions:

• The developed methodology provides a reliable approach for estimating construction waste generation even with limited data.
• The findings support urban metabolism monitoring, carbon emission quantification, circular economy development, and strategic waste management planning in the GBA.