Non-collocated methods to infer deformation in steel structures: The magnetomechanical effect in cylindrical structures subjected to impact loads

Increasing demand for energy from renewable sources has resulted in ambitious plans to construct a large number of offshore wind farms in the coming years. In relatively shallow water depths, the preferred support structure for wind turbines is the steel monopile, which is a thin-walled cylindrical structure. To decrease the cost of the generated electricity, larger wind power generators are commissioned, which has led to a significant increase of the size of the foundation piles. Currently, monopiles are most frequently driven into the seabed by means of hydraulic impact hammering. Aided by the compressive stress wave generated by each hammer blow, the pile gradually progresses to the desired penetration depth. The stress generated by each hammer blow can inflict plastic deformations at the pile head, which can jeopardise the delicate alignment required for the bolted connection between the superstructure and the monopile. Furthermore, the repeated elastic deformation of the pile leads to material fatigue, which reduces the service life of the structure. Hence, monitoring the deformation and stress resulting from the hammer blows is vital to assess the structural health. Offshore, however, dedicated sensors are seldom employed, due to time constraints and the harsh marine environment. In addition, contact sensors can easily be damaged by hammer-induced high-amplitude strains. To this end, this thesis develops several alternative methods to monitor the deformation in a monopile during installation. These methods are non-collocated (i.e. a quantity is measured at certain location to infer the structural quantity of interest at another position), and, preferably, non-contact. By considering the propagation of elasto-plastic waves, a non-collocated method to quantify the amount of plastic deformation inflicted by a hammer blow is first proposed. As a part of the energy contained in the stress wave excited by the hammer blow is used to permanently deform the structure, the stress wave becomes distorted. At a certain distance below the pile head, the energy flux is determined that is carried out by the stress wave through a cross-section of the pile. The difference between the measured value and the expected energy flux from a linear-elastic simulation with the same hammer forcing provides an upper bound for the amount plastic deformation inflicted by a hammer blow. The main benefit of this proposed method is that the sensors are employed outside the region where the highest strains occurs, reducing the risk of damaging the sensors. However, data is collected with sensors which are attached to the pile, leaving the aforementioned restrictions to the sensor deployment in place. To enable the widespread monitoring of steel structures subjected to dynamics loads, non-contact methods are needed. For the development of a non-contact method to infer the hammer-induced deformations, the magnetic stray field of the steel structure is analysed, which permeates the space around it. As the structure's magnetisation depends on elastic and plastic strains through the magnetomechanical effect, it is expected that the magnetic stray field, which is generated by the magnetisation, conveys the information about the present strain state of the structure to the sensor. Contrary to experiments on the magnetomechanical response of structural steel reported in literature so far, a steel cylinder has a significant demagnetising field due to its geometry, creating a non-uniform spatial distribution of the magnetisation. Additionally, magnetomechanical data under dynamic loads are scarce. Hence, a unique laboratory-scale experiment was designed, in which a steel cylinder was repeatedly impacted by a free-falling concrete mass, providing the first insights into the magnetomechanical effect in dynamically-loaded structures with a substantial demagnetising field. In between impacts, the magnetic stray field was mapped to analyse the evolution of the remanent stray field, i.e. the stray field when the structure is unloaded. Due to repeated impacts which only generate elastic strains in the structure, the remanent stray field evolves towards a metastable magnetic equilibrium. When a new peak strain is introduced, the stray field converges towards a new equilibrium, displaying a tendency towards a global magnetic equilibrium. However, as soon as plastic deformation forms, the evolution of the remanent field deviates from this trend as a result of the increased dislocation density, which, in turn, reduces the material's ability to remain magnetised. This behaviour serves as a basis for a non-contact method to detect and localise regions of plastic deformation in a steel structure subjected to repeated impact loads. This novel method is the first non-contact technique to infer structural deformation proposed in this dissertation. In the lab-scale experiment, strain gauges and a magnetometer registered the transient magnetomechanical response during each impact. When the magnetisation is at a magnetic equilibrium, a strong correlation is found between the axial strain and the magnetic field variation around the remanent state. The amplitude and direction of the transient magnetic stray field varies with the circumferential position of the magnetometer, indicating that the response is partly determined by the magnetisation in the vicinity of the sensor. To simulate the measured response, an isotropic magnetomechanical model was developed in this thesis that, for the first time, accounts for the demagnetising field of the structure. The capability of this model to reproduce the measurement results are limited though. It is envisaged that the model may be improved by accounting for anisotropy and by including the remanent magnetisation. To date, limited data have been published on the in-situ magnetomechanical response of large-scale steel structures in a weak ambient magnetic field. Consequently, an in-situ measurement campaign was performed to measure the magnetomechanical response of a monopile installed onshore with a hydraulic impact hammer. During the campaign, several magnetometers were employed using different sensor configurations. Similar to the lab-scale experiment, the position of the magnetometer relative to the pile determines the amplitude and direction of the transient magnetic field. Next to a good correspondence between the strain and magnetic signals, a polynomial relation was found between the peak strain and the maximum deviation from the remanent field expressed along the major principle axis. Using the inverse of this relation and a magnetometer which retains its position with respect to the pile, a novel method to infer the elastic strain from the transient stray field is proposed, which shows a promising correspondence between the inferred and measured strain signals. Additionally, the working principles for a new alternative technique to monitor the pile penetration using non-contact sensors are proposed. For each of the four non-collocated methods introduced in this work, directions for improvements and steps to generalise the techniques are discussed. The main benefit of the non-contacts methods in particular is that they eliminate the onerous process of attaching the sensors, enabling swift deployment and providing the opportunity to reuse the sensors. Although the new methods in this dissertation have mainly been applied to the installation of monopiles, the potential application of these non-collocated methods is much wider. Ultimately, they could be used to monitor any large-scale steel structure subjected to dynamic loads.