Please use this identifier to cite or link to this item: http://theses.ncl.ac.uk/jspui/handle/10443/5336
Title: Efficient and robust numerical modelling of masonry
Authors: Kassotakis, Nicko
Issue Date: 2020
Publisher: Newcastle University
Abstract: Masonry structures constitute a sizeable portion of the built environment, yet the prediction of their structural behaviour remains an extremely complicated task. Despite the advances in structural engineering, safety assessment methods of masonry lag far behind those of modern materials such as steel and reinforced concrete. High-level numerical methods such as the Discrete Element Method (DEM) are the most advanced and effective tools available for modelling the complex structural behaviour of masonry, yet their robustness depends on the accuracy of the geometric and mechanical properties employed. Although abundant work exists for reliably obtaining and representing material properties, methodical strategies lack for the case of geometric properties, which renders such high-level numerical modelling either inefficient or inaccurate. This research develops a methodological framework for the geometrically-accurate and efficient high-level numerical modelling of masonry structures through the employment of non-contact sensing techniques and automation. The framework is holistic, encompassing three stages of structural surveying, geometric model development and structural analysis. The first stage entails structural surveying of the masonry structure with non-contact sensing techniques such as terrestrial laser scanning (TLS) and Structure-from-Motion (SfM) photogrammetry. The second stage encompasses the utilisation of geometrical data (discrete points, orthoimages and point clouds) and computer vision geometric model development. The final stage of the framework consists of numerical model development and structural analysis with the DEM. With the specific numerical method, each block and joint are represented as a distinct entity, achieving a more faithful representation of the discontinuous nature of masonry than other state-of-the-art numerical methods, and thus permitting both the accurate and efficient prediction of the in-service and collapse behaviour the analysed structure. Three main approaches stem from the framework which are implemented on both regular and rubble masonry structures. Firstly, the manual image-based approach is implemented on 25 arch specimens. This approach entails structural analysis of geometric models developed from an orthomosaic of SfM photogrammetry with manual CAD-based block segmentation. By comparing the manual CAD-based geometric models with those of traditional geospatial techniques (i.e. tape measurements), significant differences in: a) collapse load (-1 to 10%); b) stiffness (-2 to 46%); and c) normal forces (-15 to 22%) were found, demonstrating the importance of employing accurate geometric models. Thereafter, the semi-automated image-based approach is implemented on the same 25 arch specimens. Conversely to the previous approach, structural analysis is now semi-automated, incorporating both SfM photogrammetry and image processing techniques (IPTs). By comparing IPT- and manual CAD-based geometric models, a relatively good agreement was found of collapse load, with differences of up to 7%. Stiffnesses, however, showed partial agreement, with differences of up to 7% for 10 specimens and 24% for 15 specimens). These findings, demonstrate both the potential efficiency and robustness of the framework. Finally, the cloud-based approach is implemented on Caerphilly Castle. This entails semi-automated structural analysis, highly irregular of rubble with structures through TLS and voxelization. For a course voxel size of 50 cm, an unprecedented DEM structural analysis of a full-scale masonry tower was achieved in an affordable time of 71 minutes. Therefore, this thesis ultimately paves the way for improving the efficiency and robustness of the structural analysis of masonry structures.
Description: Ph. D. Thesis.
URI: http://hdl.handle.net/10443/5336
Appears in Collections:School of Engineering

Files in This Item:
File Description SizeFormat 
Kassotakis N Final Submission.pdfThesis11.57 MBAdobe PDFView/Open
dspacelicence.pdfLicence43.82 kBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.