Pistons for reciprocating compressors for industrial applications are often made of specialised materials. These prove to have problems with manufacture due to the high quality and short production times needed in combination with a low production volume per design. Additive manufacturing, specifically selective laser melting, could solve the production problems, provided that the material retains the needed mechanical properties. Ti6Al4V is the most appropriate material for this application. The most important mechanical properties for the application are the fatigue limit and the stress intensity factor, which are not well established properties for printed materials. For this reason fatigue limit and stress intensity factor tests were performed for both stress relieved and hot-isostatic pressed test pieces on longitudinal and transverse directions. Strength, toughness and fatigue limit is higher in hot-isostatic pressed test pieces of Ti6Al4V, and these are proven to be appropriate for application in compressor pistons. However the fatigue limit of stress relieved Ti6Al4V is lower, anisotropic, and has more scatter, and as such is insufficient for the application, which is due to deleteriously oriented microstructure and the presence of porosities. This can be solved by changing the printing parameters – laser power, cooling rate or heat treatment – although the exact combination of parameters for optimised values is not known and will be part-specific.

Artificial intelligence has a strong need for faster and more energy-efficient solutions, especially for computation performed at the sensor edge. On-chip photonic neural networks (PNNs) offer a promising solution for high speeds and energy efficiency. A less explored side of PNNs is their application to time-series data, which is often the case for real-world sensor applications. While PNNs promise high speeds and energy-efficient solutions, no good use cases have been proposed. This report will first review the state of the art of PNNs. It will be seen that to solve time-series tasks, photonic continuous-time recurrent neural networks (CTRNNs) are required. The dynamics of CTRNNs are thoroughly explored to leverage obtained insights to recommend novel practical applications. This is done through simple examples, and applied to a real machine learning task. It was seen that on a real classification task the network learned two distinct fixed points corresponding to the classes. A link between the time constant of the continuous-time neurons and the temporal dynamics of the task is also found. Two general directions for novel applications are then proposed. Firstly, photonic PNNs can be slowed down to match the task. Opto-electronic PNNs allow for more control of the time constant, and on-chip photonic filter neurons are suggested. Secondly, the high speeds of photonic neural networks can also be directly leveraged. Extremely fast convergence of photonic CTRNNs can be utilized for Modern Hopfield networks, pathfinding algorithms, and time-dependent optimization problems such as for example Model Predictive Control.

Artificial intelligence has a strong need for faster and more energy-efficient solutions, especially for computation performed at the sensor edge. On-chip photonic neural networks (PNNs) offer a promising solution for high speeds and energy efficiency. A less explored side of PNNs is their application to time-series data, which is often the case for real-world sensor applications. While PNNs promise high speeds and energy-efficient solutions, no good use cases have been proposed. This report will first review the state of the art of PNNs. It will be seen that to solve time-series tasks, photonic continuous-time recurrent neural networks (CTRNNs) are required. The dynamics of CTRNNs are thoroughly explored to leverage obtained insights to recommend novel practical applications. This is done through simple examples, and applied to a real machine learning task. It was seen that on a real classification task the network learned two distinct fixed points corresponding to the classes. A link between the time constant of the continuous-time neurons and the temporal dynamics of the task is also found. Two general directions for novel applications are then proposed. Firstly, photonic PNNs can be slowed down to match the task. Opto-electronic PNNs allow for more control of the time constant, and on-chip photonic filter neurons are suggested. Secondly, the high speeds of photonic neural networks can also be directly leveraged. Extremely fast convergence of photonic CTRNNs can be utilized for Modern Hopfield networks, pathfinding algorithms, and time-dependent optimization problems such as for example Model Predictive Control.

Recent advancements in additive manufacturing have led to significant progress in the field of metamaterials, wherein the introduction of microscopic features affects the material properties on a macroscale. Common examples of these materials are truss-based and plate-based structures. However, these lattices bear inherent susceptibility to stress concentration points, which undermine their overall performance. Incorporating smooth surfaces in the material design can be a promising strategy for solving that issue. One noteworthy class of smooth metamaterials originates from the topologies formed in the process of spinodal decomposition. For small fluctuations, these structures can be described mathematically using Gaussian random fields (GRF) stemming from the superposition of standing waves. Recent research work has led to the development of a subtype of these structures, termed spinodoids. Spinodoids are anisotropic structures formed through a biased sampling of wave direction vectors encompassing the underlying GRF. The development of spinodoid topologies has allowed for extensive design exploration and property-structure investigations, important in the context of many potential future applications, such as bone biomaterials design. Until now, the majority of studies performed on spinodoid metamaterials were limited to the elastic regime. However, many materials, especially biological tissues such as bone, rarely exhibit purely elastic behaviour. Thus, exploring other material regimes is of great interest. This work attempts a finite element analysis of three different types of spinodoid structures: lamellar, cubic, and columnar. These structures' properties are based on a generalized Maxwell model for cortical bone. The study explores how tuning the design parameters influences their mechanical behaviour in a viscoelastic regime.

Additive manufacturing (AM) is quickly becoming one of the more popular methods to manufacture components made of Ti-6Al-4V in the aerospace and automobile industry due to its flexibility in producing complex geometries and reducing tooling costs. As the world of additive manufacturing is still relatively young, there is no globally accepted method for the certification of AM parts, allowing freedom to develop procedures to utilize the advantages of AM. To greatly reduce material wastage, the use of smaller specimens has been explored to characterize the mechanical properties of the material. This study aims to analyze the influence of test specimen size on the near threshold fatigue properties of Ti-6Al-4V manufactured by conventional processes and Laser powder bed fusion (LPBF) . Two different specimen sizes were utilized, the standard size and the sub-sized specimens. The effect of two heat treatments, stress relief (SR) and annealing (AN) on the LPBF material on the fatigue properties was also studied. In addition, the effect of anisotropy on the fatigue threshold value was also touched upon in this study. A fatigue test setup for the sub-sized samples was designed and a constant Kmax test was employed to characterize the threshold properties. The direct current potential drop (DCPD) method was employed to measure the crack length during the test. The fatigue threshold of the SR microstructure did not show any effect due to sample size. This was attributed to the partly intergranular crack propagation observed playing a prominent role in the fatigue threshold. The conventional and annealed samples witnessed a drop in threshold values of about 0.5 MPa√m when sub-sized samples were used. It was observed that the sub-sized samples had a greater crack depth at threshold leading to increased constraint at the crack tip and hence smaller degree of crack closure explaining the drop in threshold values. The current study has thus established successfully the feasibility of using sub-sized specimens to characterize the near threshold fatigue properties of conventionally and additively manufactured Ti-6Al-4V.

In order to scale the planar flow casting (PFC) process to industrial levels a Python model is created. The model which is based on a combination of empirical and theoretical equations, is used to explore the limits of the process. The model is independently verified with a Comsol simulation, which is extended with solidification and heat transfer physics. The model is then used as a bases for a system and control simulation of the complete planar flow casting system. Extra elements are added, such as thermal expansion, out of roundness of the wheel, crucible and actuator mechanics, and sensor behaviour. The controllability of the system using a PID controller is presented. With this knowledge, the design requirements for the adjustment of the existing PFC machine were defined, qualitatively as well as quantitatively, for the nozzle, pressure system and gap sensor. Following this, the microstructure of the material produced by the free flow casting (FFC) process is analysed. The microstructure is measured using XRD, SEM and EDS techniques. These measurements are then compared with a model based on the Scheil equation and a kinetic phase diagram, developed for the aluminium silicon alloy. The microstructure found consisted of primary silicon particles in which a significant amount of aluminium was dissolved, as predicted by the material model. The silicon particles are dispersed within an aluminium silicon eutectic matrix with rounded needle like features. The silicon particle size is in the order of 5 μm, whilst the feature size in the eutectic is in the order of 100 nm.

Restoration of the normal mandibular form and function is attempted with reconstructive surgery. The current standard procedure to restore continuity defects of the mandible involves free tissue transfer with an autologous bone flap. Even though the success rates are high, several critical drawbacks have been associated with this procedure, including severe donor site morbidity, long hospital stay and re­covery process, need for high surgical expertise, insufficient bone graft height, and mechanical failure of the plating system. A systematic approach for the designing and testing of patient­-specific implants used as an alternative to the free flap procedure appears to be still lacking. The goal of this thesis project was to develop a semi­automatic workflow for designing patient­ specific mandibular reconstruction implants and assess the effect of topology optimization on the biome­chanical performance of the implant using the finite element analysis (FEA) and experimental validation. Using the proposed workflow, a fully porous implant (lattice implant) and a topology optimized implant (TO implant) were designed according to the anatomy of a synthetic mandible analog subjected to lateral resection, and were additively manufactured using selective laser melting. Both implants were designed in the shape of a cage to allow for the insertion of a bone graft and the integration of dental implants. The mechanical performance of the reconstructed mandibles was predicted with computa­tional FEA and evaluated through quasi­-static and cyclic experimental testing. Digital image correlation was used to validate the finite element model through principal strain field comparison. An excellent fit between the implants and mandibles was established, indicating the capability of the workflow to develop customized implants with accurate dimensions. The results obtained with FEA were in agreement with the DIC measurements and quasi-­static testing results. No significant differences (P < 0.01) in mean stiffness, mean ultimate load, and mean ultimate displacement were found between the non-­implanted control mandibles, lattice­-implanted mandibles, and TO-­implanted mandibles during quasi-­static testing. No implant failure was observed during static nor cyclic testing at loads substantially higher than the average maximum biting force of healthy individuals, indicating the high resistance of the implant designs to mechanical failure. Yet, the lattice implant would likely be preferred over the TO implant for clinical application due to its lower weight (16.5%), higher porosity (17.4%), and shorter workflow time (633.3%). The workflow proposed in this study may offer surgeons and medical engineers the tools to sys­tematically design and evaluate patient­-specific reconstruction implants. This would result in more cost­-effective and time­-effective pre-­surgical planning and result in implant designs that can minimize morbidity and maximize aesthetic and functional outcomes.

Hall spars, a leading innovator in the composite mast-building industry with a long history of successful projects, provided a challenge which is the inspiration of this thesis. This thesis aims to contribute to the challenge: Joining techniques for the internal bonding of carbon fibre-reinforced polymer components into carbon fibre-reinforced polymer structures. ”Internal bonding” can also be called ”bonding. Currently, the epoxy adhesive film is used for these types of joints. Adhesive co-bonding raises challenges due to the limited access to the mast. Challenges concerning surface preparation, low mechanical strength, reliability and labour intensity. Other techniques will be reviewed for the internally bonding limited access hollow tubes. The aim is that no new tooling is required, and the currently used materials and process parameters can be used. The thermoset resin used is an epoxy carbon fibre prepreg. According to the literature study, potential joining techniques for this application are joining with partially cured thermosets and joining with fusion welding of thermoset composites. Differential scanning calorimetry measurements are performed to analyse the thermoset resin. Material characterisation is required to apply the joining methods with the currently used material and process parameters of Hall Spars. Process parameters for the curing model, and glass transition temperature model are derived. Using this material model allows creating the insight that partial cured joining with the thermoset resin of Hall Spars is not feasible. The structural integrity of the joined parts can not be guaranteed in the curing cycle. Polyetherimide is used as thermoplastic material to apply the fusion welding technique on the thermoset. This thermoplastic film is co-cured with the thermoset and used as a coupling layer to join the two thermoset parts. Performing an interphase analysis with the scanning electron microscope indicates interphase formation at the PEI/epoxy interface. The higher the isothermal curing temperature, the thicker this interphase. Gelation causes the interphase formation to slow down. The interphase formation at the interface is not been observed as expected based on the literature review. This required a test if the interphase formation would have occurred at a smaller scale than the scanning electron microscopy could observe. This can be achieved by performing single-lap shear experiments. These joints are processed according to Hall Spars process parameters. Joints created with the fusion coupling technique with the PEI film show promising results. However, due to high void content, a lower ultimate lap shear strength is observed. Therefore it is advised to investigate a curing cycle with a second dwell phase. This changes the viscosity profile of the resin. Therefore allows the trapped air at the PEI/epoxy interface to leave the joint. This could potentially lead to more reliable joints.

Ceramic matrix composites (CMCs) are advanced materials that consist of a ceramic matrix reinforced with a high-strength, high-stiffness material, such as carbon fibers. They offer excellent thermal and chemical stability while exhibiting low weight and exceptional mechanical properties. A novel CMC material is the C/C-SiC produced with 2/2 twill weave fabric. It consists of a carbon fiber-reinforced carbon (C) and silicon carbide (SiC) matrix. In this study, a macroscopic non-linear constitutive model accounting for the damage-induced plasticity is proposed for the 2/2 twill weave C/C-SiC composite. In the context of this thesis, a computational model is developed, based on the framework of continuum damage mechanics and general plasticity theory. A potential function inspired by the Tsai-Wu criterion combined with a damage model is used to predict the strain and damage evolution. An exponential damage evolution law is introduced while the coupling of different damage modes is also considered. Moreover, an experimental investigation on the macroscopic mechanical behavior and damage mechanisms of C/C-SiC under in-plane onand off-axis loading conditions is performed. Specimens with 0𝑜, 30𝑜 and 45𝑜 on- and off-axis angles were manufactured and tested under monotonic and cyclic tensile and compression loads. Furthermore, the microstructure of the pristine material and the fracture surfaces of the tested specimens are studied through scanning electron microscopy (SEM). A Bayesian optimization algorithm is finally used to optimize simultaneously the different material parameters based on the experimental test data. The predicted stress-strain curves are in good agreement with the experimental curves, especially in the case of monotonic tensile loading. Both damage initiation and evolution are predicted accurately by the chosen laws and coupling functions. Moreover, the combination of the Tsai-Wu criterion with a damage evolution law is proven to predict the ultimate strength well. Fiber pull-out is observed in tension, while interlaminar and translaminar cracks in compression. This study thus provides an accurate constitutive model, a complete mechanical characterization of the in-plane behavior and a better understanding of the fracture mechanisms of C/C-SiC.

Pressurized liquid Tin finds application in the generation of Extreme Ultra-Violet light for semiconductor lithography. In order to improve the throughput of the lithography systems, tin must be pressurized to higher levels, and in turn, new pressurization methods are needed. A phase change tin pump is an innovative system that pressurizes and pumps liquid Tin by harnessing the expansion and contraction during phase changes, without the need for any moving parts. The pump needs to pressurize liquid tin up to 2000 bars, with a pumping capacity of 4 ml/hr. Since this system relies heavily on control over the temperatures of tin, this study is set up to address the thermal constraints in the system by investigating three aspects of temperature distribution in the system. Firstly, the heaters in the pump are placed at discrete locations, but the working volume is continuous. Thus, it is challenging to define a temperature control function that can facilitate uniform melting and continuous flow of tin. The relation between rate of heat input to the pump and the rate of heat transfer in tin is estimated using an analytical model. From the analytical model, it is found that heating rates of the order of 0.1 K/s are required in order to melt tin in a reasonably uniform fashion over a zone length of 5 mm. Secondly, the number of heaters are limited, and it is hard to achieve precise control over the temperature of tin at any given location. In order to establish a good basic control, the free design parameters are optimized so that a steady state gradient of 50 K is achieved between solid (200°C) and liquid (250°C) tin in the working volume. This is done by evaluating the thermal profile of the system for different combinations of the design variables, using Finite Element Analysis. The two objectives of this optimization problem (maximum temperature gain and minimum crosstalk) are seen to have contrasting requirements of the design variables. An optimal combination of the variables is found such that a gradient of 50 K is possible, but with a little trade-off on both the objectives. Thirdly, a direct measurement of temperature of tin inside the pump is not feasible, and tin temperatures are estimated analytically. The accuracy of estimation is impacted by changes in local temperatures due to the non-linear properties of tin like absorption/release of latent heat, pressure-dependent melting point. The effect of non-linear tin properties on local temperature distribution is studied by setting up a finite difference model. It is seen that the absorption of latent heat during melting of tin results in a temperature that is 12 K lower than what would have been without the effect of latent heat.

Bainite steels are in high demand in many application areas owing to their outstanding mechanical properties, mainly due to the presence of a combination of fine bainite plates and retained austenite. Understanding the complicated mechanism of bainite transformation is crucial to creating an optimum design. Researchers have addressed this challenge via computational modelling, where the transformation start temperatures of bainite (Bs) and martensite (Ms) are key indexes when designing highperformance steels and their heat treatments. This thesis aims to explore the potential that data collection has to create a model, using experimental results of the bainite transformation process. A list of metallurgical accepted claims is elaborated to assess the quality of the data. Principal component analysis and clustering techniques are used to identify patterns, most important features and main relationships in the dataset. The results confirm the exponential carbon dependence that bainite and martensite transformation temperatures have, therefore requiring nonlinear models to predict them. Following, regression models and machine learning algorithms based solely on the chemical composition are used to predict Bs and Ms. Train-test split series and cross-validation are used to evaluate the prediction and consistency of each model. The results show that the ensemble learning algorithms outperform the regression techniques. Random forest and gradient boosting decision tree provide excellent Ms prediction on the validation set with R2 values of 0.92 and 0.93. The smaller dataset size adds up to the complexity of bainite transformation, resulting in worse prediction models of Bs, where the random forest and gradient boosting decision tree R2 values are 0.68 and 0.67 respectively. Even though the models are showing signs of learning, the impact that outliers have demonstrates that the data is not good by itself to create a predictive model. The incorporation of microstructural and process parameters would provide significant advances to the models for designing bainitic steels.

Metamaterials derive their properties from microstructure rather than from bulk material properties. This opens property spaces that are difficult, or impossible, to access with traditional methods. However, exploring this vast design space remains challenging because classical techniques can be computationally inefficient. Recent years has seen many successful applications of data-driven methods to this problem. Data-driven models can be used to bypass expensive Finite Element (FE) simulations and experiments by exploiting large datasets. This requires the parameterization of microstructures such that they can be mapped to properties. The forward problem, in which properties of are determined from design parameters, is typically well-posed, which allows straightforward application of machine learning methods. The inverse problem, in which design parameters are identified to match specific properties, is ill-posed because multiple sets of design parameters can produce similar properties. The Voronoi growth method induced by star-shape metrics provides a way to explore a large geometrical design space using simple parameterization. It is used to generate 2D unit cells of void and material pixels with non-trivial topologies. The growth process enforces connected geometry while also allowing for smooth transition between different designs. Using homogenization techniques, we generate a large dataset of design parameters and stiffness properties. Machine learning techniques are first used to model the forward problem. By combining the trained forward model with the inverse model, the inverse problem is rendered well-posed. Both the forward and (deterministic) inverse model show excellent agreement between target and predicted values. The method is generalized beyond the design space by targeting properties of non-growth based structures. We investigate methods to create a stochastic inverse model, that can produce multiple designs to match target properties. Verification is done by comparing tensile tests of 3D printed samples to FE simulations.