The reuse of recovered materials plays an increasingly significant role in improving sustainability in the construction industry. For safe reuse, a detailed understanding of material behaviour is essential. The Old Botlek Bridge in Rotterdam was dismantled in 2017, and this study investigated the fatigue properties of the steel recovered from its main girders. The objective was to determine the fatigue crack growth parameters of the recovered steel using linear elastic fracture mechanics and the Paris–Erdogan equation.

Results from eleven SENB specimens tested under high-cycle fatigue loading at a nominal stress ratio of R = 0.1 were evaluated. Crack propagation was monitored using a crack gauge, a crack opening displacement gauge, and a strain gauge. A semi-automatic algorithm was developed to extract the crack-gauge wire-breakage points from the measured voltage signal. Crack lengths were inferred from the crack-opening displacement and strain measurements using calibrations obtained from two-dimensional finite element models. Crack growth rates were calculated using seven-point local polynomial differentiation for the crack-gauge data and neural-network regression with automatic differentiation for the COD- and strain-gauge-based data. The finite element models used linear-elastic material behaviour and mainly a geometrically linear formulation, while a geometrically nonlinear formulation was also applied as a sensitivity study. Stress intensity factors were extracted using contour integrals. The experimental crack growth rates were then combined with the specimen-specific stress intensity factor ranges to determine the Paris-law parameters.

All three measurement methods captured the same general crack growth trend. The calculated crack growth rates ranged from approximately 10⁻⁵ to 10⁻³ mm/cycle, but the three methods gave different results for the same specimen during the same experiment. Based on the specimen-balanced pooled results from the geometrically linear formulation, the crack-gauge-based evaluation gave m = 4.062 and C = 3.50 × 10⁻¹⁰, the COD-based evaluation gave m = 4.109 and C = 1.71 × 10⁻¹⁰, and the strain-gauge-based evaluation gave m = 3.618 and C = 8.93 × 10⁻¹⁰. These values apply within the method-specific fitting ranges, with da/dN expressed in mm/cycle and ΔK in MPa√m. The geometrically linear stress intensity factor results showed close agreement with the Baratta and Fett reference solution, with a mean absolute difference of 0.21% and a maximum absolute difference of 0.44%. The difference between the geometrically linear and nonlinear results increased towards larger crack lengths, showing that geometrical nonlinearity can affect the stress intensity factor in the final stage of crack propagation.

These differences show why the complete measurement and post-processing procedure must be reported alongside fatigue crack growth results and considered when comparing them. Over the common range of ΔK = 20–40 MPa√m, all three Old Botlek Bridge Paris-law curves were positioned above the selected literature curves for S235 J2, S355 J2, and old crane steel. The evaluated specimens therefore showed higher crack growth rates at the same ΔK and lower apparent fatigue crack-growth resistance. However, this difference cannot be attributed exclusively to the previous service history of the bridge steel because unused reference specimens from the same steel were not available. ... Read more

Public organisations increasingly depend on external parties for the knowledge that underpins their core tasks. Under New Public Management (NPM), execution was outsourced to private parties. With this, the knowledge embedded in execution followed, described as hollowing out. One way organisations can attempt to manage this dependency is through a knowledge sourcing strategy, a deliberate approach to deciding where and how knowledge is obtained. This thesis researches: what barriers influence the execution of a knowledge sourcing strategy in a public organisation?

Knowledge sourcing is not a mature concept in KM literature; this thesis borrows the concept of sourcing strategy from the procurement literature. Together, these fields ground knowledge sourcing as the strategic process through which an organisation determines how to secure the knowledge it needs in order to sustain access to mission-critical knowledge over time.

To answer the research question, a single case study was conducted at Rijkswaterstaat, using interviews and a focus group with employees directly involved in the knowledge sourcing process. The data was analysed through inductive thematic analysis, allowing barriers to emerge from the data itself rather than from a predefined framework.

Rijkswaterstaat was selected for two reasons. First, its evolution from a vertically integrated executor to a governing and commissioning organisation has created a structural dependency on external parties for both execution and the knowledge that underpins it. Second, Rijkswaterstaat's knowledge management is mature, managing its knowledge fields and identifying its knowledge gaps. There is also a dedicated department of knowledge procurement. This maturity means the prerequisites for executing a sourcing strategy are largely in place, allowing barriers to be studied without simply attributing them to an absence of structure.

The analysis identified fifteen barriers, grouped into three categories: intra-organisational influences, external influences, and knowledge sourcing process barriers. These barriers were then interpreted using an enfolding literature approach, comparing the findings against Onofre and Teixeira's knowledge management implementation framework and the conceptual literature developed in Chapter 2.

The first category, intra-organisational influences, captures barriers rooted in the organisation's culture, governance, and management behaviour that prevent a knowledge sourcing strategy from gaining traction in existing operational processes. The second category, external influences, captures conditions outside the organisation's control, including regulatory, institutional, and market constraints, that predetermine which sourcing options are available regardless of strategic in
... Read more

With global warming already exceeding 1°C due to anthropogenic emissions, research in renewable energy and their deployment is more crucial than ever. This work investigates the adaption of the circular liquid encapsulated PV architecture researched by TU Delft and Biosphere Solar into a novel photovoltaic thermal (PVT) collector. In this adapted PVT architecture, the solar cells would be immersed in the cooling agent. By increasing the heat transfer area between the cells and the cooling liquid, the cooling performance and thus thermal efficiency of the PVT should be enhanced. Two different prototypes are made and steps are taken towards characterizing the performance of this novel PVT design. An experimental arrangement is designed and built to characterize the thermal performance of these collectors using a reduced temperature approach. This involves a solar simulator, source measure unit, flow meter, pump, heating element, and temperature sensors. However, combined electrical and
thermal efficiencies cannot be measured due to limitations in the experimental arrangement. While
the thermal efficiency is measured, the electrical efficiency is only recorded for short time intervals.
Separate electrical and thermal efficiencies are respectively recorded up to 20.73% ± 0.60 % for the
cell area and 42% ± 7% relative to the entire collector area. A correlation is established between these
efficiencies and the PVT’s reduced temperature, representing its operating conditions, for specific flow
rates. To do this, the electrical efficiency is estimated throughout the experiments via the cell tempera-
ture estimated from the initial recorded electrical efficiency and the thermal efficiency. The established
correlations between the efficiencies and reduced temperature are used to estimate the combined yield
of this novel design in the field by making some assumptions. In Amsterdam, this results in an annual
production of 344 kWh/m2 of heat and 217 kWh/m2 of electricity.
Nonetheless, most of the electricity and heat produced by a PVT is curtailed as it is generated in
summer when there is little demand while it supplies very little in winter when it is needed. Thus, to
increase the PVT’s performance in a residential heating system, it is integrated with a heat pump and
seasonal storage system. Compared to the base case using only a heat pump, a PVT and heat pump
system reduces the annual electrical energy consumption by 29.4%. Moreover, a combined PVT, heat
pump and seasonal storage system reduced the annual electricity consumption by 1.7% relative to the
heat pump only scenario and the peak electrical power load is shifted to summer ... Read more

Lithium iron phosphate (LFP) is widely used as a cathode material in batteries because of its long service life, good thermal stability and relatively low cost. However, an inherent behavior of this material makes the modeling and estimation of its internal state particularly challenging. Due to the flat voltage response of LFP over a wide range of state of charge (SOC), noticeably different internal states can produce only small differences in terminal voltage. The problem becomes more important in battery packs containing cells connected in parallel, where cells with similar voltages may carry different currents and gradually develop an imbalance that may remain unaccounted for by pack-level measurements and control systems. This behaviour of LFP is associated with phase separation during the intercalation and de-intercalation of lithium in the active-material particles. This thesis examines how phase-separation-related behaviour, represented through different open-circuit-potential and hysteresis formulations, affects current sharing, SOC distribution and thermal response in a parallel-connected battery pack. The investigation was first carried out at the cell scale by comparing three Doyle-Fuller-Newman models under the same operating conditions. Although the models produced broadly similar terminal-voltage curves, clear differences were observed in their internal reaction distributions, local positive-electrode concentrations and retention of spatial heterogeneity following changes in current direction. The physically motivated models showed stronger reaction localisation at low rates, while these distinctions became less apparent at higher rates because of the increasing influence of transport polarisation. The thermal behaviour of the models was subsequently compared over a range of C-rates, with particular attention given to the calculation of reversible heat. Entropic coefficients for the physical models were generated using the same simulated potentiometric procedure, allowing a consistent comparison with the empirical formulation. Differences in reaction localisation and entropic response resulted in different irreversible and reversible heat contributions, even when the predicted terminal voltages remained similar. For the pack-scale study, independent DFN and lumped thermal models were assigned to each cell and coupled through a sequential electrical-thermal solver that accounted for busbar resistance, terminal arrangement, self-heating and heat transfer between neighbouring cells. After numerical verification and comparison with previously reported pack behaviour, the framework was used to study the effect of cell formulation, discharge rate, interruption depth, terminal topology and dynamic loading. The results showed that electrical-path asymmetry initiates unequal current sharing, whereas the flat and state-dependent LFP voltage response controls how the resulting SOC imbalance develops and relaxes. During discharge interruptions, cell voltages equalised much faster than their SOCs, leaving a persistent internal imbalance, particularly when the interruption occurred within the voltage plateau. Higher C-rates and self-heating generally increased the imbalance, while an opposite-end terminal arrangement produced considerably Slimmer current and SOC differences than the same-end arrangement. Finally, varying the regular-solution interaction parameter showed that a stronger phase-separation-related thermodynamic response amplified and prolonged an imbalance created by the electrical topology, but was not itself the original source of that imbalance. Overall, the work shows that pack-level voltage can appear normal while important differences continue to exist between parallel cells, and that the effect of LFP phase-separation behaviour must be considered together with electrical topology, operating conditions and thermal coupling when assessing pack performance. ... Read more

The large-scale development of offshore wind is a complex socio-technical challenge in which technological performance, economic viability, government policy, energy markets, and the decisions of multiple actors are strongly interconnected. Governments use policy and tender procedures to shape offshore wind development, while developers and investors must make long-term design, bidding, and investment decisions under considerable uncertainty. These decisions depend on expectations of future energy production and project revenues. Technical modelling uncertainty can therefore propagate beyond engineering calculations into financial assessments, tender strategies, and ultimately the viability and deployment of offshore wind projects. Improving the information available to decision-makers is consequently important for both individual projects and the wider transition towards a low-carbon electricity system.

Wake prediction represents one important source of uncertainty within this decision-making system. Wind turbines reduce downstream wind speeds, affecting the production of neighbouring turbines and therefore the expected energy yield of a wind farm. Engineering wake models such as Jensen are computationally efficient but rely on simplifying assumptions, while high-fidelity Large Eddy Simulations (LES) provide substantially more detailed representations of wake behaviour at much greater computational cost. This creates a trade-off between the accuracy of information available for decision-making and the computational effort required to generate it, particularly when many layouts or scenarios must be evaluated.

This thesis investigates whether recent advances in artificial intelligence can reduce this trade-off by using a Graph Neural Network (GNN) trained directly on high-fidelity LES data. The central research question is: To what extent can a Graph Neural Network, trained on LES data, improve the accuracy of offshore wind-farm wake predictions while maintaining computational efficiency compared to traditional engineering wake models?

A turbine-interaction GNN was developed in which turbines are represented as nodes and potential wake interactions as directed edges. Through message passing, the model learns how upstream turbines influence downstream wind-speed deficits. The architecture was first verified using synthetic Jensen data and subsequently evaluated using the WINS50 LES dataset. The experiments investigated heterogeneous inflow information, turbulence intensity and atmospheric-stability indicators, graph construction, and alternative message-passing architectures. Finally, a multi-farm GNN was trained on 25 wind farms and evaluated on the unseen future_052 layout, with Jensen/NOJ as the principal engineering-model baseline.

The experiments demonstrate that the GNN can successfully learn turbine–turbine wake relationships. The LES experiments further showed that additional model information and complexity do not automatically improve prediction accuracy. Local wind speed provided useful information about heterogeneous conditions, while local wind direction offered limited additional benefit. Turbulence intensity generally improved performance, and exploratory experiments indicated that atmospheric-stability information may provide further predictive value. Graph construction also revealed a trade-off between accuracy and computational cost: wider interaction corridors improved predictions, whereas full connectivity provided only a small additional improvement at substantially greater computational cost. Similarly, an attention-based architecture did not outperform the edge-conditioned NNConv model. These findings indicate that a relatively simple, physically informed graph representation can efficiently capture relevant wake interactions.

The strongest result was obtained in the final cross-farm experiment. When trained on 25 wind farms and evaluated on the excluded future_052 layout, the GNN achieved a mean absolute error of 0.7888 m/s compared with 1.0859 m/s for Jensen, representing an approximately 27% reduction in prediction error. The GNN also exhibited virtually no mean prediction bias. At the same time, including graph construction, evaluation of the complete test set was approximately 4.5 times faster than the Jensen implementation. The results therefore demonstrate that wake relationships learned from computationally expensive LES simulations can be transferred to an unseen wind-farm layout while retaining computationally efficient evaluation.

Importantly, these improvements can affect the information on which offshore wind decisions are based. An illustrative valuation case translated predicted wind speeds into annual energy production (AEP) and discounted lifetime revenues. For the considered wind farm, the LES reference resulted in an AEP of approximately 5318 GWh. The GNN overestimated AEP by only approximately 0.6%, corresponding to a discounted lifetime revenue error of approximately EUR 22.8 million under the case-study assumptions. The Jensen model produced a substantially larger energy-production and revenue error. Although this does not constitute a complete financial valuation, it demonstrates how technical modelling errors can propagate into economically significant differences in project assessment.

The contribution of this thesis therefore extends beyond improving a technical wake-prediction metric. More accurate and computationally efficient wake predictions can improve one important source of information within a wider system of interconnected technical, economic, and policy decisions. For developers and investors, this can support more reliable comparisons of layouts, expected revenues, investment alternatives, and tender strategies, while efficient evaluation enables a larger number of scenarios to be considered under uncertainty. For governments, better-inf
... Read more