APPLICATIONS OF INNOVATIVE BUILDING MATERIAL AND COMPUTER VISION METHODS IN GEOTECHNICAL ENGINEERING

Abstract

The Hawaiian Islands have been a world-famous traveling spot for their unique tropical island view. But the particular geological formation of the islands and their unique locations have always proposed challenges for geotechnical engineers and geologists. For example, coral sand is widely encountered in coastal areas of tropical or subtropical regions. It can be found on most beaches in Hawaiian islands. Compared with silica sand, it usually exhibits weaker mechanical performance from the perspective of engineering geology. Thus, necessary soil improvements shall be applied to the coral. Considering the fragile and unique ecosystem, sustainable material with less carbon footprint and less environmental impact would be developed and selected priorly. Moreover, the infrastructures along the island have been facing coastal erosion issues from both physical erosion (waves) and chemical erosion (sea wind and seawater). The road embankment, the house embankment, the harbor, etc often require maintenance to sustain their service time. Due to the topography and climate, the windward side of the coastal area on Oahu is suffering from marine microplastics (MP) pollution issues. Furthermore, as an island state, hurricanes and tsunamis could also threaten the safety of islanders and infrastructures. Therefore, as a geotechnical major Ph.D. student, this dissertation would devote some potential solutions to the challenges. Firstly, a novel alkali-activation-based sustainable binder was developed for coral sand stabilization. The alkali-activated slag (AAS) binder material was composed of ground granulated blast furnace slag (GGBS) and hydrated lime with the amendment of biochar, an agricultural waste-derived material. The biochar-amended AAS stabilized coral sand was subjected to a series of laboratory tests to determine its mechanical, physicochemical, durability, and microstructural characteristics as well as durability. Results show that the addition of a moderate amount of biochar in AAS could improve soil strength, elastic modulus, and water holding capacity by up to 20%, 70%, and 30%, respectively. Moreover, the addition of biochar in AAS had a marginal effect on the sulfate resistance of the stabilized sand, especially at high biochar content. However, the resistance of the AAS-stabilized sand to wet-dry cycles slightly deteriorated with the addition of biochar. Based on these observations, a conceptual model showing biochar-AAS-sand interactions was proposed, in which biochar served as an internal curing agent, micro-reinforcer, and mechanically weak point. Secondly, a state-of-the-art deep-learning algorithm, Mask R-CNN, was utilized for the clayey soil crack detection, locating and segmentation. A comprehensive dataset including 1200 annotated crack images of 256×256 resolution was prepared for the algorithm training and validation. The proposed Mask R-CNN algorithm achieved precision, recall and F1 score of 73.29%, 82.76% and 77.74%, respectively. Besides, the algorithm gained a mean locating accuracy (APbb) of 64.14% and a mean segmentation accuracy (APm) of 47.59%. The detection performance of the Mask R-CNN was also compared with the U-Net in three different scenarios. The test results have demonstrated the superiority of the Mask R-CNN over the U-Net algorithm in crack detection, locating and segmentation and the algorithm could automatically process the crack characterization. Then, this dissertation proposed a state-of-the-art deep learning-based approach, Mask R-CNN, to locate, classify, and segment large marine microplastic particles (fiber, fragment, pellet, and rod). The fully trained Mask R-CNN algorithm was compared with U-Net in characterizing microplastics against various backgrounds. The results showed that the algorithm could achieve Precision=93.30%, Recall=95.40%, F1 score=94.34%, APbb=92.7%, and APm = 82.6% in a 250 images dataset with white background. The algorithm could also achieve a processing speed of 12.5 FPS. The results obtained in this study implied that the Mask R-CNN algorithm is a promising microplastics characterization method that can be potentially used in the future for large-scale surveys. Finally, a video instance segmentation algorithm was trained to locate, identify, and segment soil cracks in a real-time video stream. The algorithm could record the cracks' locations and numbers simultaneously. Besides, the crack ratio of clay could be calculated by crack pixels divided by total clay pixels among the entire soil cracking process. Furthermore, Structure from Motion (SfM) has been applied to reconstruct the 3D soil desiccation models. The soil crack can be detected in a 3D point cloud graph and highlighted. A series of 3D parameters like depth, volume, and cross-section profile can be obtained for future analysis. The proposed video instance segmentation method has demonstrated the potential application for real-time crack alerts and monitoring of geotechnical infrastructures via surveillance cameras.