This thesis introduces methods for accurate 3D reconstruction of FIB tomography data using machine learning. It addresses challenges in obtaining large datasets and ground truth values, proposing a method for virtual FIB tomography data generation and an isotropy-based validation method. ML-based segmentation methods tackle issues in BSE images, like the shine-through effect and image intensity ambiguities. Integrating the slice repositioning method with image inpainting resolves inconsistencies in slice thicknesses in FIB tomography data. Additionally, a multimodal machine learning approach using multivoltage images further enhances 3D reconstruction accuracy.

Magnesium is the lightest metallic construction material and bears high potential for automotive components, medical implants and energy storage systems. However, the practical use of untreated magnesium alloys is restricted as they are prone to corrosion. An essential prerequisite for the inhibition or control of the degradation process is a deeper understanding of the underlying corrosion mechanisms. For instance, identifying the reaction pathways most relevant for the release of gaseous hydrogen—one of the main products during magnesium corrosion—builds the foundation for adapting the hydrogen evolution rate in general. A possible approach to modulate the hydrogen evolution and magnesium dissolution rate is the introduction of small organic molecules that either stabilize the corroding surface or capture corrosive species. Such molecules constitute benign and useful materials to modify the service environment of magnesium for applications that require tailored degradation properties. This thesis illuminates different aspects of magnesium degradation, ranging from atomistic insights into the corrosion process itself to further developing strategies for effective corrosion control. By combining atomistic simulations and machine learning techniques, a holistic framework is developed to improve understanding and enable more efficient magnesium corrosion engineering in the future.

Due to the various types of waste disposal, treatment, utilization and technologies, decision support model for waste management is needed to assist planners and decision makers in finding most suitable way to manage municipal solid waste efficiently. Many planners and decision makers in the area of municipal solid waste have a lack of thorough understanding of the complex chains of waste management system. Therefore the impact for the environment quality and the public health can only be judged at the rudimentary level. However, most existing models are primarily focusing on cost or environmental analysis. Only few consider other crucial factors such as the demographic condition, the characteristics of urban form and urban infrastructure, land transformation aspects due to urban development. Consequently, such models often meet difficulties to cope with cultural requirement. Based on those reasons, a decision support model to set up alternatives of most appropriate technology for sustainable waste management towards a low carbon eco-city on a regional basis is developed in this PhD study. The Low Carbon- and Eco-Region, in particular the contribution of waste management sector, is a vision of living in low rate of carbon generation, using fewer natural resources, and encouraging energy recovery and/or waste reduction at source by improving the used material quality (up-cycling). This decision support model is constructed mainly based on the cultural requirement and local context of a region and synergize the geographic, environmental, social capital and economics aspects in order to fulfill the needs of the respective region and its society. The method employed in this model is not solely a new developed model, but also an advanced model in material flow analysis (STAN), and life cycle assessment on solid waste system (EASEWASTE) and Geographic Information System (GIS). At the same time the model also assists the stakeholders in improving the environmental quality and the public health by promoting waste separation at source and reducing the greenhouse gas emission potential from waste sector.

Many severe ship accidents in the past were caused by a large water ingress followed by the progressive flooding of the watertight integrity of the ships and finally resulted in the sinking and total loss of the vessels. These accidents show a high demand for a better understanding of the flooding of ships with the help of numerical methods. This would allow to avoid such accidents in the future and to find new designs with an increased capability to withstand flooding. The circumstances of the accidents, which involve flooding are difficult to understand due to the complexity of the inner subdivision and the large number of resulting flooding paths. Hence, a fast numerical simulation method is developed in this thesis to analyse and predict such flooding events. The validation of the method comprises the comparison with experimental results from three test cases of a model test and the re-investigation of three severe ship accidents, which have been carefully investigated before: The accident of the European Gateway, the Heraklion and the Estonia. The results obtained with the new method in these applications are very promising. A quasi-static approach in the time domain is chosen to compute the flooding of the inner subdivision and the resulting equilibrium floating position at each intermediate time step. The flooding paths are modelled with the help of directed graphs. The water fluxes through the openings are computed by a hydraulic model based on the Bernoulli equation. Large and partly flooded holes are taken into account, as well as conditional openings like breaking windows or the flooding through completely filled compartments. The effect of air compression can be taken into account as well if this is required for a certain case. After the determination of the new flood water distribution and the corresponding filling levels in each compartment, the resulting equilibrium floating position is computed. The simulation ends, either if the intermediate floating position does not change any more or if all buoyancy reserve is lost and the ship starts to submerge below the sea surface. The method is developed within the ship design system E4, which ensures a consistent data model and allows the direct coupling with existing and new methods. All essential effects are taken into account without unnecessarily increasing the complexity of the model as this would lower the performance of the simulation method. The method enables the user to study many flooding scenarios within a short period, for example to investigate the influence of watertight doors that were left open.

In this thesis, we present a novel approach for the automatic generation of rich semantic metadata based on surface-level information. For the extraction of the required surface-level information, state-of-the-art analysis tools are used. The approach exploits a logic-based formalism as the foundation for knowledge representation and reasoning. To develop a declarative approach, we formalize a multimedia interpretation algorithm that exploits formal inference services offered by a state-of-the-art reasoner. Furthermore, we present the semantic interpretation engine, a software system that implements the logic-based multimedia interpretation approach, and evaluate it through experimental studies.

In this thesis, a low-level RF (LLRF) system for the novel polarizable transverse deflecting structure PolariX TDS is developed, installed, and commissioned at the FLASHForward beamline at DESY. The PolariX TDS operates in the X-band at 12 GHz and requires an LLRF system capable of measuring and processing signals in this frequency band. Using an approach to reuse the existing S-band LLRF system that can process signals at 3 GHz, new hardware is being developed that converts RF signals between 12 GHz and 3 GHz. The underlying components of this conversion hardware are RF mixers, filters, amplifiers, and frequency multipliers. Various candidates for all these components are characterized and compared in terms of their electrical characteristics. Since the final system performance depends heavily on the phase noise, the main focus of the characterization will be on this. With the components selected, printed circuit boards are designed, simulated, produced and packaged in hermetically sealed housings before being mounted in 19" modules. After completion, the modules are also characterized in the laboratory. As a final step, the prototype of the PolariX TDS will be installed in the FLASHForward beamline together with the developed LLRF system. After successful commissioning, first measurements will be performed in the running accelerator. To analyze the short-term behaviour, possible procedures to eliminate the long-term effects will be discussed. The actuator chain, consisting of LLRF system, klystron and its preamplifier, is measured and characterized. The results show some discrepancy between expected and measured values and an analysis of possible sources of the problem is performed. Beam-based measurements are used to analyze the relation between the RF and the beam in terms of phase and amplitude noise. These measurements show good performance of the overall system in terms of short term phase stability in general. However, small deviations from the expected behavior are seen, analyzed, and procedures to correct them are proposed. Analyses of the long-term behavior show a large difference between the RF and beam-based measurements, which causes are still subject of research. Although not as important as the phase noise, the amplitude noise is also investigated and similar to the long term behavior, the measured RF shows better performance than the beam based measurements.

This thesis deals with selecting, implementing, testing, and evaluating numerical methods to support large construction projects’ construction management. The methods are bundled in a tool, which uses real-time construction process data measured on the construction site to compute project time and cost forecasts. These predictions are then combined in a numerical optimization algorithm to determine the Pareto optimal number and operating time of construction site machinery.

Substitution of conventional metals by lightweight materials is inevitable. However, employment of multi-materials in a structure presents a challenge and choosing the right materials, design, and manufacturing technique is essential in the development of any engineering structural application. New joining and additive manufacturing techniques complement the successful integration of materials, design, and production. Conventional joining methods, such as adhesive bonding and mechanical fastening, face technological limitations when used to join dissimilar materials, such as metals and composites. The relatively long curing time of an adhesive is also a significant drawback for adhesive bonding. In recent years, new joining techniques have been investigated to develop hybrid joints that overcome the limitations of traditional joining methods. There is an increased interest in recent advances in the field of additive manufacturing (AM). These offer the flexibility to produce complex geometric parts, such as sandwich structures with AM honeycomb cores. Combining the principles of joining and polymeric AM is the main aim of the present doctoral thesis, which introduces AddJoining, a new technique that was co-invented by the author (patent application by HZG, DE 102016121267A1, 2018) to contribute to the manufacturing options for joining multi-material parts. The process was inspired by additive manufacturing and joining technology principles, and this new technique uses polymer 3D printing, e.g., fused filament fabrication, FFF (also known as fused deposition modeling, FDM), to add layers of polymer or composite to a metal substrate. The AddJoining process has a potential to produce structures with geometric flexibility, such as honeycomb cores. As an early phase of this technology this PhD work was devised to understand and develop the fundamentals of the AddJoining process by joining transportation grade lightweight aluminum 2024-T3 with a combination of unreinforced polyamide 6 (PA6) and carbon-fiber reinforced polyamide 6 (CF-PA6). To understand the process-induced changes in the selected material combination (aluminum 2024-T3 and the composite [PA6/CF-PA6]4S), three combinations of process parameters were selected. The optimized process parameters with coating temperature (CT) at 229 °C and heating bed temperature (HBT) at 77 °C were kept constant. Moreover, the optimized HBT was fixed and the minimum and maximum values of CT selected at 20 °C and 150 °C, respectively. Intimate contact of polymer-to-polymer was promoted purely by temperature, which led to strong adhesion and influenced polymer bond formation. From a microstructural aspect, it was shown that strengthening occurred close to the interface of the aluminum 2024-T3 by reprecipitation of particles once solubilized, due to exposure to high processing temperatures. Within the composite [PA6/CF-PA6]4S the process induced a global modification of the polymer, the effect of HBT slightly changed the thermal properties of the composite part. However, to better understand the local changes on the polymer, a stepwise approach was carried out to combine the phase-identification of the polymer and nanohardness investigations. The results offered a more comprehensive understanding of the local variations in material properties by means of the phase-changes in each layer while it indicates the presence of two polymorphs of α-PA6 and γ-PA6. Thus, it is associated with the presence of a thermal history of AddJoining, which can affect the crystallinity locally and decrease the hardness. From the hybrid joint mechanical performance, the influence of heating bed temperature on secondary bending could be an indication of low stiffness in the composite part caused by a variation in its crystal structure. This indicates that changes to PA6 phases could have led to a variation in hybrid joint’s ductility and strength. The mechanical performance of AddJoining hybrid joints was assessed by a wide range of mechanical tests. An interfacial intralaminar failure mode was observed with quasi-static loading. Compared to adhesively bonded joints there was a significant increase in ultimate lap shear force (ULSF). In addition, the S-N curves obtained from fatigue testing indicated outstanding results and the hybrid joints produced reached their fatigue limits (106 cycles) with loading levels corresponding to 30 % of ULSF. By monitoring stiffness degradation, it was seen that damage evolution was dominated mostly by fiber rupture throughout fatigue life due to the high stiffness stability of hybrid joints. To summarize, this PhD work has been successful in fulfilling its objectives, namely describe the AddJoining process fundamentals and mechanical behavior. Moreover, this work shows the potential for the AddJoining technique in the manufacturing of future complex multi-material parts for structural applications with tailored design aspects to improve damage tolerance and load-bearing carrying.

The transition from municipal waste landfilling towards higher recycling quotas is supported by the application of thermal treatments, especially effective for mixed waste, sorting residues and unrecyclable fractions. Thus, increased amounts of incineration residues such as bottom ash are expected, and innovative solutions for their circularity are needed. This thesis identifies limitations on bottom ash treatments and circularity in Europe. Furthermore, the material recovery potential for bottom ash mineral fractions is assessed and a novel treatment for their refinement is developed and tested. The results are contextualised in a plant concept aimed to increase the total material recovery rate of incineration bottom ash.

Fluidized bed spray granulation is a process that facilities particle size enlargement by injecting solids in the form of a suspension, solution or melt. The produced particles are often of high value due to their highly specific, functionalized nature. This work provides a method for utilizing particle-scale simulations method like the coupled Computational Fluid Dynamics-Discrete Element Method (CFD-DEM) to predict the properties of particles produced in scaled-up apparatuses based on laboratory-scale experiments by correlating the conditions that particle experience to the properties they develop. Furthermore, a method for calibrating DEM models of weakly wetted particle systems by utilizing a novel, probabilistic liquid bridge state model is proposed.

The present work is concerned with the partitioned solution of the multifield problem arising from a hierarchical modeling approach to cardiovascular fluid-structure interaction. Different strategies to couple the participating field solvers are investigated in detail. This includes staggered and parallel coupling algorithms as well as different methods for convergence acceleration, spatial interpolation and temporal extrapolation of coupling quantities as well as convergence criteria. In the developed modeling and simulation approach, a fully resolved model of a segment of the arterial network is coupled to reduced order models in order to account for the in uence of the surrounding. The resulting problem is solved using five specialized field solvers, namely a fluid and a structural solver for the three-dimensional fluid-structure interaction problem, a one-dimensional blood flow solver for the surrounding vessel network, a solver for diffrent types of windkessel models used to obtain physiological boundary conditions at the distal ends of the one- and three-dimensional models, and a solver for an elastic foundation that describes the surrounding tissue. The applicability of the solution approach is demonstrated in terms of several exemplary applications including studies of idealized and patient specific end-to-side anastomoses of bypass grafts. They are known to be prone to the development of intimal hyperplasia, i.e. a thickening of the vessel wall that may lead to occlusions in the anastomosis region. There is experimental evidence that hemodynamic quantities such as the wall shear stress promote the progression of this secondary disease. Cardiovascular FSI simulation are therefore of great interest and can aid in surgical planning and optimization of anastomoses shapes and graft materials.

This work presents a novel decentralised approach for the resource allocation in Peer-to-Peer (P2P) networks. By decomposing a global optimisation problem distributed algorithms are derived and their efficiency, fairness and stability are shown. The model is based on resource pricing for TCP/IP. Therefore, also P2P over IP networks are studied as cross-layer optimisation problem.

A generic process to evaluate new technologies and concepts within the air transportation system is developed. Life cycle cost-benefit methods for the economic assessment from the aircraft operator's and manufacturer's point of view are in the center of attention. As a proof of concept, the concept of staging long-range flights for, primarily, fuel saving purposes is chosen. The so-called intermediate stop operations (ISO) concept is processed through all steps of the introduced approach, with a focus on the system analysis. This includes a fuel analysis, cost-benefit analysis, and a safety as well as an airline network compatibility analysis. As a result, a comprehensive picture of ISO for further decision making is provided.

This thesis introduces a method based on program transformation for developing generic software libraries. More precisely, the signatures of generic functions, i.e., functions parameterised on types, are transformed from a functional language to an object-oriented language. Type level functional constructs, such as higher-order functions and type constructors, are mapped to type parameters specified using a mechanism called concepts. The transformation is independent from particular languages and can be formally described.