Integrated Structural Monitoring using Point Clouds Obtained from Terrestrial Laser Scanning

Terrestrial laser scanning (TLS) technology is by now a mature and widespread geodetic measurement technique, and the accuracy of the commercially available scanners is steadily increasing. TLS is particularly suitable for the task of areal deformation analysis and, despite its popular use for the tasks of structural monitoring, no consistent workflow for structural identification has been yet defined. For this reason, this thesis contributes towards the automatic determination of the material properties of structures modelled and monitored via point cloud technologies. It results in an interdisciplinary work that touches both the fields of geodetic and structural engineering, and forms part of the discipline of structural health monitoring (SHM), which plays a key role in the current engineering practice. The main contribution of this thesis is a novel method to perform the numerical identification of structural members by using point cloud data acquired with terrestrial laser scanners in multiple epochs. The backbone of the developed method is the approach of integrated monitoring, which offers a great flexibility to merge measurements of different origins within a single framework relying on the finite element (FE) modelling of the investigated structure. The underlying criterion of this approach is to match the measured and the FE-calculated displacements by using a least squares adjustment (LSA), which includes the linearisation of the mentioned FE model. In this way, a thorough propagation of the stochastic information from the TLS data to the final estimated parameters can also be performed. The LSA is herein validated with closed-loop and Monte Carlo simulations. Additionally, a method to automatically generate the geometry of the FE model from the point cloud of the surface of the monitored object is proposed, and specific insights about the implementation of the numerical differentiation are provided. Moreover, the observed nodal displacements and their uncertainties are derived from the TLS-based deformation analysis of the structure. Among others, this step can be carried out with three different methods: variance propagation, log-Euclidean interpolation, and nearest neighbour search. Despite being all applicable methods, the third one has proven more computationally efficient. The method is herein implemented for the model update of beams and slabs for which the Young’s modulus is estimated, with the possibility to subdivide the analysed structure in multiple partitions. Moreover, the proposed method can be readily used to estimate other material parameters or even forces, for any shape of structure, including free-form ones. The proposed method is herein analysed and successfully employed for three application cases, of which one with real data and two with synthetically generated point clouds: a cross-laminated timber slab, a steel beam, and an alloy plate. The observations for the application cases relying on synthetic data have been generated with a self-developed point cloud generator. This generator has also been embedded in a tool that enables the planning of TLS monitoring campaigns by predicting the influence of the sensor setup (position, resolution and accuracy) on the estimated parameters. Ultimately, the developed method successfully expands the range of capabilities of TLS, by defining an operational procedure for the numerical identification of structures