Estimation of construction waste generation based on an improved on-site measurement and SVM-based prediction model: A case of commercial buildings in China. (1st May 2021)

• Record Type: Journal Article
• Authors: Hu, Ruibo; Chen, Ke; Chen, Weiya; Wang, Qiankun; Luo, Hanbin
• Is Part Of: Waste management. Volume 126(2021)
• Page Start: 791
• Page End: 799
• Publication Date: 2021-05-01
• Subjects: Construction waste management -- On-site measurement -- Waste generation rate -- Prediction -- Support vector machine; Hazardous wastes -- Periodicals; Refuse and refuse disposal -- Periodicals; 363.728
• DOI: 10.1016/j.wasman.2021.04.012
• Languages: English
• ISSNs: 0956-053X

Abstract

Highlights: Construction waste was classified into five types based on its chemical properties. Using an improved on-site measurement to collect construction waste generation data. Developing a support vector machine-based model to estimate waste generation rate. Case study illustrates the performance and applicability of the proposed approach. Abstract: Estimation of construction waste generation (CWG) at the field scale is a crucial but challenging task for effective construction waste management (CWM). Extant field-scale CWG modeling approaches have faced difficulties in obtaining accurate results due to a lack of detailed CWG data, and most of them fail to consider the complex relationship among predictive variables. This study attempts to tackle this issue by proposing a novel CWG modeling approach that integrates improved on-site measurement (IOM) and a support vector machine (SVM)-based prediction model. To achieve this goal, 206 ongoing commercial construction sites were investigated to obtain the predictor values and waste generation rates (WGRs) of five types of waste (i.e., inorganic nonmetallic waste, organic waste, metal waste, composite waste, and hazardous waste) generated at three construction stages (i.e., the understructure stage, superstructure stage, and finishing stage). The data were introduced to the SVM to develop the relationships between predictive variables and WGRs. An actual commercial building under construction was used to demonstrate the applicability of the proposed approach. The results showed that the superiority of the IOM can be used as a basis to implement robust CWG data collection. In addition, the SVM-based WGR prediction model (SWPM) can obtain more accurate prediction results (R 2 = 86.87%) than the back-propagation neural network (R 2 = 75.14%) and multiple linear regression (R 2 = 61.93%).