A digital twin approach for geometric quality assessment of as-built prefabricated façades. (September 2021)

• Record Type: Journal Article
• Authors: Tran, Ha; Nguyen, Tuan Ngoc; Christopher, Philip; Bui, Dac-Khuong; Khoshelham, Kourosh; Ngo, Tuan Duc
• Is Part Of: Journal of building engineering. Volume 41(2021)
• Publication Date: 2021-09
• Subjects: Automated construction -- Quality assessment -- Prefabrication -- Modular construction -- Digital twin; Building -- Periodicals; 690.05
• DOI: 10.1016/j.jobe.2021.102377
• Languages: English
• ISSNs: 2352-7102

Abstract

Abstract: We develop a new framework for geometric quality assessment of as-built prefabricated façades during construction process using a 3D as-designed model and a 3D as-built semantic model. The framework is based on a digital twin approach which facilitates an automatic quantitative comparison between the 3D as-built digital replica, reconstructed from the 3D as-built lidar point cloud, and the 3D as-designed model. The framework consists of automated correspondence identification and element-based comparison on three new criteria, which ensure comprehensive measurement of the geometric quality of as-built façades regarding accuracy, completeness, and correctness. Experiments on a synthetic façade system and a prefabricated façade of an actual construction project demonstrate the ability of the framework to detect inconsistencies and perform a quantitative evaluation and localization of geometric errors in as-built prefabricated façades in an efficient and timely manner. Ultimately, the proposed framework enables efficient visual quality assessment during construction process for prefabricated construction. Highlights: A comprehensive framework for evaluating geometric quality of prefabricated façades. Measurement of geometric quality in terms of completeness, correctness, and accuracy. One-to-one correspondence mapping of as-built elements to as-designed counterparts. An efficient solution for localization and visualization of construction errors.