This PhD uses new data capture methods (e.g. laser scanning and structure from motion photogrammetry) alongside numerical simulations to automate detection and quantification of damage in both rock outcrops and masonry buildings to improve our ability to predict and prevent failures.
Maintaining buildings and infrastructure is critical for sustainability of our built environment, the economy and the societal resilience of the region they service. English Heritage assert that every £1 invested in heritage regeneration has generated £1.60 in additional economic activity . Furthermore, many prominent and valuable masonry structures are also commonly located by rock outcrops. Historic Environment Scotland (HES) currently has 221 masonry-built properties under its care, with 17% of them on or near outcrops requiring constant monitoring and management.
Climate change projections for the UK  suggest that rock materials composing both the masonry and the natural environment, such as outcrops, will be placed under significant strain, which raises fundamental challenges for the monitoring and maintenance-ability of those structures.
Several computational strategies and tools have been developed to understand the mechanical behaviour of masonry-built infrastructure, and similarly rock outcrops. The state of the art employs advanced non-linear computational formulations based on the Finite Element Method (FEM) and Discrete Element Method (DEM). These models rely upon effective determination of the geometry and material characteristics associated with the structure analysed. Regrettably, such inputs have been historically approximated, reflecting the limitations of traditional surveying methods. Recently, modern reality capture technologies, e.g. terrestrial laser scanning and photogrammetry, have enabled engineers to develop numerical models that more faithfully reflect the structures, with clear benefits to the quality of the results . However, continuing assumptions of material homogeneity remain misguided.
This PhD will build on work on the automated detailed modelling of masonry constructions  to assess its value to FEM/DEM, and will transfer this approach to rock outcrop analysis .
- Can more accurate FEM/DEM models be effectively created automatically from reality capture methods (photogrammetry, laser scanning)?
- How do photogrammetry and laser scanning compare for the generation of FEM/DEM models?
- Can machine learning be effectively used to detect structural and weathering defects in rock outcrops that are of similar type to (masonry) built structures (e.g. stone decay and structural failure)?
Building on exciting recent advances in the field, this work will develop a machine learning based approach to detect stone defects in both rock outcrops and masonry structures, and assess the impact of such modelling improvements on structural analysis (FEM/DEM) results. Experimental work will consist of data acquisition and analysis for actual sites under the care of HES, including some masonry structures of significant heritage value. Research results will be disseminated to both academic and professional audiences.
Year 1. Research training. Familiarisation with point cloud data processing and FEM/DEM for masonry construction and outcrop analysis. Familiarisation with machine learning techniques.
Year 2. Integrate machine learning-based defect detection techniques into an automated stone structure modelling approach.
Year 3. Validate the technique with examples from both masonry constructions and outcrops.
A comprehensive training programme will be provided comprising both specialist scientific training and generic transferable and professional skills. Point cloud processing training can be done ad-hoc with support from Dr Bosché and his lab team. Additional formal training in FEM/DEM and machine (deep) learning will be arranged as required. There will be opportunities to attend international summer schools, e.g. the Annual European Computing in Construction Council (EC3) Summer School.
For more information click "LINK TO ORIGINAL" below.
This opportunity has expired. It was originally published here: