This thesis aims to explore the influence of borehole alignment accuracy and other soil and pipe parameters on the pullback forces during the pullback phase in Hor- izontal Directional Drilling (HDD) operations. To that end a model inspired by PipeForce V1 was implemented and expanded upon in Python. The model was used to calculate the pullback forces based on the measured product pipe naviga- tional coordinates obtained from actual HDD projects. Calculated pullback force data was compared to actual measurements. It was found that pullback forces tend to decrease with increasing borehole alignment accuracy. The extent of the influ- ence depends on the length of the installation under consideration. Furthermore, through analysing ballasting and specific gravity of slurry, it was found that a bal- ance between those factors resulted in a decrease in the maximal pullback forces, through optimising the effective vertical force on the product pipe. Lastly, it was found that the originally proposed model PipeForce V1 should not be applied to longer installations due to an almost quadratic growth in one of the calculated fac- tors, which was not checked for in the original publication.

Failure of subsea wind turbine cables are the main cause for wind farm downtimes. Furthermore, 80% of insurance payouts to wind farms come from cable repairs and maintenance. Cable tracking is part of the wind farm cables maintenance scheme. Surveys are required for localisation and determining the depth of burial of the cables. Conventional surveying methods have shortcomings when it comes to localising offshore buried power cables, which can be resolved by using electromagnetic methods. From the measured magnetic field created by the cable, its position and burial depth can be inverted. Forward modelling the magnetic response of cables improves the understanding of their electromagnetic behaviour in one dimensional complex submarine environments. Maxwell's equations are the cornerstones of understanding the behaviour of electromagnetic fields, forming the basis of the models. Implementing numerical modelling packages like Empymod accommodate for boundary conditions in one-dimensional layered media to more accurately investigate the magnetic response of electric cables. Land measurements show that Empymod's numerical models can model cables in the accuracy range of 2-15% depending on the distance. New numerical model analyses show the influence of water depth, underlying layers, depth of burial, geometrical complexities and adjacent cables on the magnetic field. Surrounding resistivity contrasts together with changes in water depth change the measured magnitude of the magnetic field by up to 10%. Geometrical complexities have large effects on the magnetic field, sometimes completely changing the expected response. Therefore, twisted and irregularly shaped cables can pose problems when inverting for their position during surveys.

In this thesis, we discuss two pattern recognition techniques, template matching, and neural networks. We discuss how these techniques have been used for the development of two earthquake detection algorithms. The first algorithm is based on template matching and the second is based on deep learning. The algorithms are designed for the detection of small <0.5M events in the subsurface of Zeerijp, Groningen. These two algorithms have been compared to assess their earthquake detectability and practicality. The systems have been compared using field data from Zeerijp. The algorithm based on deep learning (a neural network) produced too many false positives considering the amount of seismic data we would like to use it for. The algorithm based on template matching did not produce any false positives during testing. The template matching system has been fed six months of continuous seismic data fromZeerijp. This resulted in the detection of at least 22 new events.

Unconsolidated granulates exhibit complex, nonlinear behaviour when subjected to dynamic forces. The presence of granular contacts gives this type of material a relatively low stiffness and provides hysteretic energy losses. These features make unconsolidated granulates suitable for railway ballast as it provides dissipation of vibrational energy from passing trains which is important to minimise vibrational disturbance. However, simulating the response of the railway superstructure under dynamic loads becomes difficult due to then onlinearity of the ballast. In order to develop better prediction tools, the elastic behaviour of unconsolidated rocks is first investigated experimentally by quasi-static and dynamic stress-strain experiments yielding the Young's modulus, nonlinear resonance shift and analysis of harmonic generation. In addition, the transmission of structural waves through granulates is investigated by assessing the transfer function for different thicknesses of granulates, different
particle sizes and different materials with varying viscous damping. Three granulates are used, small-scale ballast, a gravel, and two sizes of uniform steel spheres. All three materials exhibit a combination of classical and hysteretic nonlinearity where the strain depends on the stress amplitude and history.

A completely new finite element approach is taken to model the hysteretic nonlinearity, based on an existing phenomenological static model. Multiple spring-slider elements with gaps are used, as opposed to implementing a homogenised material model. It is shown that only 50 elements can reproduce the hysteretic nature of the material, which is a significant advantage to a traditional material model requiring the discretisation of the entire ballast volume. Each spring-slider element is parameterised by two springs constants, a yield force and an initial gap. A distribution of these parameters across the 50 elements is found that reproduces the quasi-static stress cycles acquired experimentally. In addition, a parametric study of the model parameters during dynamic excitation reveals that key indicators of nonlinearity can be simulated. The finite element simulations prove that using a set of spring-slider elements
to model the behaviour of unconsolidated granulates is viable method. With experimental tests performed on true ballast and further work on the finite element model to understand optimal parameter distributions, a more accurate and efficient railway superstructure model can be produced.