High-precision systems often comprise many individual systems, each operating at its own frequency. However, the vibrations from one system can be transmitted to another, which yields a response. In high-tech applications where precision is crucial, such as compliant transmission systems, these responses are undesired as they result in a loss in precision. Topology optimization can be employed to directly include these unwanted vibrations in the design process of compliant mechanisms. However, the current literature only concerns simple beam structures and suffers from issues such as premature convergence and large intermediate-density areas. This thesis aims to employ topology optimization to design compliant transmission systems whilst simultaneously attenuating the effects of unwanted external vibrations in the form of base excitations. This is done by using an objective function capable of minimizing the displacement response of a structure whilst not suffering from the aforementioned issues. The found objective function relies on the principle of global minimization, which minimizes the largest displacements inside the structure resulting from the applied excitation. An extension of the current research is done by only minimizing a subset of the domain, obtaining a localized minimization whilst other areas of the domain are allowed to exhibit larger responses. These two minimization principles are then applied to the design of a compliant inverter mechanism, with local minimization considering two areas of interest: the entire mechanism area and the regions around the input and output of the mechanism. The results show that global minimization is able to obtain discrete results for a large range of frequencies. Local minimization of the mechanism area yields lower displacement responses and, for higher frequencies, resulting topologies with displacement behaviour similar to the principles of vibration absorption and vibration isolation. Decreasing the response area to be minimized only to cover the input and output regions of the mechanism yields inconclusive results in terms of obtaining lower displacement responses compared to local minimization of the mechanism area. A proof of concept for designing a compliant transmission system whilst minimizing the response to harmonic base excitations is established, demonstrating potential benefits for future research in this domain.

Structural design for unconventional aircraft can be quite challenging and could undergo
many iterations until a solution is found for all load-cases such as static sizing, stability and
crashworthiness. Ultimately, all of the design aspect are integrated in the earliest design
changes for which Topology Optimization (TO) could provide the solution. The Flying-V is
one such aircraft with challenging crashworthiness design due to its small sub-floor volume.
However, a suitable formulation for aircraft crashworthiness and TO is yet to be found.

This thesis aimed at incorporating the conventional aircraft-crashworthiness characteristic
of plastic energy dissipation in plastic hinges in the energy based Hybrid Cellular Automata
framework. The HCA principle is to optimize on cellular level, using information of each cell’s
direct neighbors. For the proposed method, the inter-cell stress and strain information is used
to reconstruct the cellular bending energy throughout the non-linear dynamic crash-analysis,
providing greater stability than intra-level bending energy. The bending energy formulation
was based on the assumption of linearly varying stresses and strains in bending. Several
formulations have been formed to manipulate this energy promoting hinge-formation.

The framework has been tested on size optimization of a small-scale slender beam impact
model, representing the final design case. Results show peak force reduction and greater displacements compared to the standard HCA crashworthiness model. However, improvements
can be made such as inclusion of the non-linear stress and strain variation in the bending
energy formulation, to more accurately account for large plastic strain.

Before testing whether the methodology improves the crashworthiness of aircraft, verification
of the voxel-mesh approach was required. An A350-like aircraft has been simplified and
idealized accordingly and compared with its detailed FEM crash-assessment. Simplifications
include modelling only one frame and replacing the cabin and floor with a rigid body to
reduce computational cost. Results show good agreement for the required level of detail of
conceptual design, although it is recommended to idealize the cabin and floor as well. Once verified, the framework has been applied to the entire domain.

Topology optimization is a branch of mechanical engineering in which the topology of a structure is created and optimized to certain conditions and restrictions. In the last few decades, the demand for highly accurate and complex models of these structures has increased and it has a big effect on the computational power needed. To ease the computational load for the dynamical systems one can use model order reduction methods to reduce the size of the models.
Classic Arnoldi is a widely used method for model order reduction (MOR) with topology optimization. In this thesis, we discuss two-sided Arnoldi and IRKA to help find a suitable moment-matching MOR method for topology optimization. These two reduction methods are implemented and improved to create a high fidelity reduced model. For improvements in the accuracy, the use of orthogonalization methods is analysed and discussed as well as including rigid body modes for IRKA and a preconditioner for two-sided Arnoldi. Lastly, a participation factor is discussed and improved to help reduce the model created with two-sided Arnoldi.
In the end, we find that two-sided Arnoldi in combination with the participation factor performs better than IRKA by creating a smaller and more accurate reduced model.

Topology optimization is a valuable tool for the optimization of all kinds of structures. It can create highly efficient but complex designs. This complexity can make these structures challenging to clean. However, cleanability is often a requirement in, for example, the medical or food industries. Currently, no method exists to reduce the complexity and create these cleanable structures using topology optimization. The goal of this project is to create a method that can generate two-dimensional cleanable structures. In this context, cleanability has been defined using two requirements. Every section of the surface must be visible, and no sharp edges can appear on the exterior. These requirements are met by generating a structural and cleanable shell around the design. The shell needs to be optimized to minimize the compliance of the entire structure. The addition of this shell has been achieved by the use of a hybrid method. This method combines two structural optimization methods, a level-set method, and the modified SIMP approach. The level-set method acts as a shape optimization method. Its shape forms a boundary in which the modified SIMP method is used to generate a design. Material is placed at this boundary of the level-set shape to create a shell that encloses the structure. Experiments on several sets of boundary conditions show a successful creation of a shell in every case. The resulting designs are not guaranteed to be cleanable, but a satisfactory result has been achieved in every case by changing some parameters. This method sets the stage for further development toward the application of topology optimization to create cleanable designs.

High performance machines rely on fast moving parts and generally avoid resonance for improved accuracy. To improve the dynamic properties of these high performance machines, their parts are optimized via a lengthy iterative process. Topology optimization for vibrations problems could shorten this time consuming design process and provide a more optimal design compared to the manual iteration process.

In the field of topology optimization for vibration problems there are various methods to solve a given problem. The two most commonly used methods in recent research are the density approach and the level-set approach. The characteristics of the density and level-set approach are well understood in context of topology optimization for vibration problems, however a direct comparison between these two methods has not yet been conducted. Several crucial aspects of topology optimization for vibration problems will be investigated, such as localized eigenmodes, mode multiplicity, grey areas and efficiency for practical applications. Additionally, the applicability of these aspects will be tested in the academic and industrial field to determine their values when applied in industry. This thesis provides an extensive study of various design cases in which the density and level-set topology optimization approaches are compared on their ability to solve vibration problems. These design cases are based on frequently used design cases in literature which are generally seen as benchmark problems.

For this thesis it is opted to have as many similarities between the density and level-set approach as possible, to ensure a fair comparison between the two methods. To accomplish this, the level-set approach uses a density based mapping in combination with material parameter sensitivities and the method of moving asymptotes (MMA) to update the design variables. Furthermore, the level-set function is parameterized with compactly supported radial basis functions (CSRBF). This leaves the difference that the density approach uses element densities as design variables, whereas the level-set approach uses expansion coefficients as design variables.

The design cases indicate that the density approach is versatile as it is able to solve a wide variety of problems. Additionally, there are less parameters, which makes this method easier and faster to work within an industrial setting. Furthermore, the method produces well-performing designs even with more difficult tasks, such as a coarse mesh.
Although occasionally localized eigenmodes occurred whilst using this method, they do not seem to interfere with the final result. Thus, the density approach is less time consuming to setup and needs less tuning of the method specific parameters.
On the other hand, results from the design cases also indicated that the level-set approach is able to produce designs with an improved objective function at the cost of possibly more tuning of method specific parameters. Furthermore, the level-set approach is able to solve all the design cases without the occurrence of localized eigenmodes. Although the level-set approach is less optimal for coarse meshes, it outperforms the density approach at more refined mesh sizes. Additionally, it features a crisp geometry description by the zero level-set contour. Thus, the level-set approach is able to produce more optimal designs without the occurrence of local eigenmodes at the cost of more complexity and possibly more tuning of the method specific parameters.

To conclude, both approaches have unique properties to be able to solve vibration problems. The density approach is more applicable as a standard approach in an industrial setting due to it being more robust and the method is less time consuming. However, the level-set approach should be opted for more complex vibration problems due to the crisp geometry definition of complex geometric features and its ability to outperform the density approach.
A practical application has been solved with an optimization run, where the use of a set of predefined parameters that solved the benchmark cases has been used. Additionally, an optimization run where all parameters are optimized for the specific design case was performed to see the ultimate performance. The level-set approach was able to outperform the density approach in the predefined parameter case, whereas the ultimate performance case gave usable results for both methods. The differences came down to a more improved objective function for the density approach, or a more simplistic and lighter design for the level-set approach.

Historically, contact analysis has been an important design tool for various applications such as vehicle crash analysis or manufacturing. It is also becoming increasingly more relevant in newer fields such compliant mechanism or metamaterial design. Even though contact has been a topic of discussion for decades, there are still challenges that need to be solved before a truly general contact analysis and optimization framework can be developed. As a milestone towards such framework, this project builds upon an existing novel linear contact analysis and a geometry immersion methodology developed in Interface-enriched Generalized Finite Element Method (IGFEM), to extend its capabilities to support large deformations (geometric non-linearities). Furthermore, a solution is presented to overcome void element over-stiffening in immersed large deformation contact analyses.

As a type of powder-bed-based Additive Manufacturing (AM), Selective Laser Melting (SLM) is widely used for building metallic lattice structures. However, during SLM process, geometrical imperfections and defects, such as strut over-sizing or under-sizing, typically exist in components due to overheating, which strongly influence their mechanical response. Therefore, it is common that the as-built lattice has discrepancies from the designed lattice structure in mechanical properties. The main aim of this study is to simulate the transient temperature field of lattices during SLM process to predict the possible manufacturing issues for the improvement of product quality. Based on a technique, the moving grid method, growing struts of a lattice whilst it being built by SLM can be regarded as moving domain problems, and thermal process of a building lattice can be simulated numerically. The primary contribution of this study is to evaluate the effect of process parameters, including laser power, energy density and material deposition rate, on thermal evolution of a lattice during SLM. Another contribution is related to the comparison of transient temperature field between different lattice architectures. Analyzing the influences of various factors is important to recognize the thermal evolution of a building lattice. The results reported in this study offer a reference for quality improvement of produced parts, to realize that both geometry and mechanical properties of the as-built lattices as close as possible to their as-designed counterparts.

Topology optimization has seen increased interest with the development of additive manufacturing (AM) as a manufacturing method, because of its ability to utilize the geometric complexity that AM offers. However AM still imposes some restrictions on the design, most notably on its minimum feature size, overhang, and enclosed voids. Enclosed voids are problematic because for most AM methods it is impossible to remove supports from them, additionally for powder-based AM these enclosed voids trap unmelted powder adding parasitic mass. In this thesis, two new methods have been investigated to alleviate this issue, the eigenfrequency method and the flood fill method. The eigenfrequency method utilizes the eigenfrequency of an inverted density field, which changes enclosed voids to floating masses. These floating masses have rigid body modes, which can then be prevented by applying a minimum eigenfrequency constraint. The eigenfrequency method encountered a large problem in the form of intermediate densities, which were used to satisfy the eigenfrequency constraint instead of the elimination of enclosed voids. The flood fill method utilizes a modified flood fill algorithm, which is a filter with a resulting density field where every enclosed void element is changed to solid. The flood fill method did successfully eliminate enclosed voids in both 2D and 3D problems, at the cost of very little additional computational effort. It also allows for direct control over the location, amount, and size of the void elimination features by varying boundary elements, running additional flood fills, and morphologic operators, respectively.

Standard treatment methods for prostate cancer often result in side-effects because of damage to the surrounding tissue. Magnetic resonance imaging (MRI) guided focal laser ablation to treat prostate cancer reduces the risk of adverse effects by preserving noncancerous tissue. Prostate cancer diagnosis and focal laser ablation treatment both require the insertion of a needle for biopsy and optical fibre positioning. The insertion of needles in soft tissues causes tissue motion and deformation, resulting in tissue damage. In this study, we propose an MRI-ready actuation system for a needle that can be inserted into tissue with a zero external push force and without buckling. The zero external push force is achieved by moving parallel needle segments in a reciprocating manner. The actuation unit’s design inspired by the click-pen mechanism actuates the reciprocating motion of six parallel needle segments by solely a discrete manual translating actuation as its input. A prototype, called the Ovipositor MRI-Needle, was built using 3D printed parts and six 0.25-mm diameter Nitinol rods. Experimental validation of the Ovipositor MRI-Needle in ex vivo human prostate tissue inside a preclinical MRI scanner showed that the needle could advance in three out of five measurements through the tissue. The Ovipositor MRI-Needle is a step forward in the direction of developing a self-propelling needle for MRI-guided transperineal laser ablation to treat prostate cancer.

The work presented in this report is an attempt to allow density-based topology optimization algorithms to take into account machining costs when optimizing. To this end, and estimation model for the cost of 2.5-axis CNC milling was developed based on the principles of Design for Manufacturing.
This estimation model was implemented in a generic topology optimization algorithm in five steps of complexity. The behaviour of the model was evaluated at each step by means of a set of experiments, showing how the addition of the cost estimation influences the optimization results.
The most promising results from the experiments are validated with Autodesk Fusion 360, with which the designs were programmed to be machined on a CNC mill. The machining times from this software were used to calculate the actual machining costs and these were compared to the machining costs estimated by the optimizer.
Two formulations of the method were found to be useful for estimating the machining cost of designs. A simple formulation that uses the shape factor and the differentiation between internal and external pockets to estimate the cost resulted in a cost saving of 13% at the expensive of a compliance increase of 9%. The more complex formulation uses multiple pocket domains to evaluate pocket-specific properties. This allowed a more accurate estimation of the costs, but did not result in a better performing optimization. The cost saving was comparable at 12%, but the compliance increased by 16%.
It was concluded that there is a use for both methods. The simple formulation allows to find cheaper designs, but does not allow much control in the cost estimation. The complex formulation however can be used to fine-tune the method for a specific situation, which could enable it to perform better that the simple formulation.

There is a constant drive to make the next generation machines in the semiconductor industry more precise and faster. For this machines with a high repeatability, good dynamics and long lifetime are needed. Compliant mechanisms are suitable candidates to be used in this kind of machines because they can be manufactured monolithically, don’t wear out over time and do not suffer from backlash which makes them ideal for precision mechanisms. However, in these machines parasitic motions and cross-axis couplings are present. These unwanted motions reduce the precision and increases the control complexity respectively.

Strategies presented in literature to compensate for unwanted motions are summarized in the first part of this report. These strategies are evaluated and by combining two promising strategies a new compensation strategy is proposed. The second part of this report focusses on this new strategy. Using this new strategy a constant stiffness linear joint, a near zero parasitic motion translational guide with well constraint uncontrollable masses and a decoupled 2-DoF mechanism are synthesized. All these case studies needed to have a large range of motion because many effective strategies for small range of motion mechanisms are available in literature.

As of writing this thesis little research has been done in truss layout optimization for multiple design alternatives. A hybrid scheme is proposed in which the fast gradient-based search of the Ground Structure Method (GSM) is employed for the search of optimal topology and size, while a population-based meta-heuristic algorithm explores the non-convex parameter space of geometry optimization. The scheme works by employing a meta-heuristic algorithm to optimize the nodal coordinates of a truss structure where for each iteration and each member of the population a small-scale (i.e. problems with a small number of degrees of freedom) GSM is performed, in order to obtain the optimal topology for the given nodal locations. Three variants of the scheme based on three different meta-heuristic algorithms are developed: Firstly, an Artificial Bee Colony (ABC) and Particle Swarm Optimization (PSO) variant are developed in combination with an original topology identification method, to extract design alternatives from the solutions found throughout the optimization process. The developed topology identification method identifies different design alternatives based on topological differences by making use of existing graph isomorphism testing algorithms from graph theory. Additionally, the topology identification method employs a multi-step filtering process to prevent members which are non-critical to the performance of the structure to influence the design alternative selection process. Finally, a Multi-Species Particle Swarm optimization (MSPSO) variant is developed without the need for a topology identification method. For all methods convergence speed in terms of number of iterations, topological variety and computation time per iteration have been evaluated and are compared. Of the developed methods, the ABC variant converges fastest towards a single good solution, however the topological variety is lacking. The slower converging MSPSO variant produces solutions with moderate topological variety, as well as reasonable material volumes. The PSO variant requires the least computation time per iteration of the developed methods. Its produced topological variety is closer to that of the ABC than the MSPSO variant and it has a slightly faster convergence speed than the MSPSO variant. Direct usage of the MOSEK API is made instead of the more commonly used CVXPY API which reduces GSM problem setup times for small-scale (28 degrees of freedom) problems by a factor of 13. Computation times for the hybrid method, from start to finish, for 3D structures with 4 to 8 movable nodes (12 to 24 geometric degrees of freedom) range from 15 to 50 seconds on standard desktop PC hardware. Because of the multiple design alternative nature of the hybrid method, and consequently the end-user does not require fast back-to-back optimization runs, these computation times are deemed acceptable. It is concluded that while in its current state the methods based on the hybrid scheme are unlikely to be suitable for usage in practice, further developments in methods to distinguish design alternatives could make the hybrid scheme, in particular the ABC variant, useful in the design of more material efficient truss structures.

Tetrahedron is working on a new design for large offshore cranes that can lift higher. This is necessary due to the increasing size of offshore wind turbines, and the inability of the conventional cranes to scale up without needing a larger ship that is carrying them. To compute the forces working on the trusses of the crane structure, Tetrahedron has developed the software program Automatic Crane Engineer (ACE). Right now the program can already optimize the cross-sectional area of the bars, but the optimization should be extended to include the shape of the structure itself.
The goal of this research is therefore to extend ACE with the necessary functionalities to perform shape optimization of the cranes. The research to achieve this goal is structured in two phases. In the first phase, the literature review part, the two research questions have been answered: the formulation of the crane optimization problem is defined and the best optimization methods for this problem have been chosen and implemented in Python. These are the Interior point method as proposed by Byrd, Hribar and Nocedal and the Method of moving asymptotes by Krister Svanberg. As a third method, the Sequential Least Squares Programming method by Dieter Kraft is also tested.
The Method of Moving Asymptotes seems the most promising for the crane optimization problem.

The common practice of designing a ship is to look for ships with similar specifications and alter it to the client's needs. This often leads to structurally redundant and therefore overdimensioned ship designs. Hence, much improvement could be expected from ship structures where optimization algorithms help advise in the design process. In this research, the midsection of a Trailing Suction Hopper Dredger (TSHD) is optimized. A TSHD midsection generally consists of longitudinal stiffened panels and transverse web frames. The typical web frame and longitudinal stiffener layout is optimized, as a weight improvement of this section can have a large effect on the total weight of the ship since it is repeatedly reoccurring.

This research has optimized the midsection of a reference TSHD in two ways; the first step was to perform a shape optimization for the longitudinal stiffener arrangement, which was followed by a topology optimization for the transverse web frame. Both optimization objectives were to minimize mass. The order of optimization follows the hierarchy in which stresses are introduced into the structure; from the plates that make up the hull toward the stiffeners and eventually the web frames. The complete optimization was performed a total of seven times, for seven different web frame spacings ranging from 25% to 175% of the web frame spacing of the reference TSHD.

For the shape optimization, a Simulated Annealing algorithm was used. The reference ship was simplified to be able to parameterize the geometry into eleven panels with T-stiffeners. Each panel has a set of variables that describe its geometry; the plate thickness, number of stiffeners, stiffener web height, flange width and web and flange thicknesses. Although feasible results came out of the optimization, no clear parallel was found when comparing the plates of different web frame spacings. This is due to the fact that it is a high dimensional problem. Although no clear parallels were found, the results were able to cope with all the loads.

The topology optimization was performed with a modified Bi-directional Evolutionary Structural Optimization (MBESO) method. The applied modifications ensure a fast convergence for large topology optimization problems. The new method was first verified by comparing results to two benchmark cases from the original BESO method, which was followed by three examples of common topology optimization benchmarks. Once established that the modified method was capable of reproducing test cases, an aspect ratio analysis was performed to better understand the transmission of stress. After that, the full geometry of the midsection was divided into smaller basic models and the same optimization was carried out in order to help interpret underlying physics of the final results. Finally, the topology optimizations for the seven web frame spacings were performed, resulting in a new orientation of beams. The topology optimization results showed that constructing beams not in an orthogonal way and along the ship hull but rather under various angles could reduce the total mass of the web frame.

To see how the shape optimization result influenced the topology optimization, three studies were carried out where all the surrounding plates had the same thickness, except for one that would have a significant smaller thickness. This showed how the web frame supports the hull plating, but also how the web frame is dependent on the stiffness of certain panels to be able to transfer shear into them.

Finally a comparison was made between the reference ship and the optimized structure. The result was a decrease of 23\% in weight for the midsection. Due to practical production considerations this weight is likely to be higher in practice, however it is a promising start toward a more efficient ship design. The innovative combination of shape optimization combined with the MBESO procedure could help in early design stages where the main components of the construction are defined.

Flex-on-substrate assembly is an increasingly popular electronics assembly type that is based on thermocompression bonding: a combination of high temperature and pressure. Currently, up to two-thirds of the cycle time is spent on heating and cooling. To meet the growing demand for flex-on-substrate production in the future, the heating and cooling steps must be more time-efficient. In this thesis, a bonding module is designed that is heated and cooled by using a 150W halogen lamp as a heating source and a staggered-tube heat exchanger with forced airflow as a cooling system. The design requires understanding of the bonding process and models containing the governing physics. Therefore, finite element models were made for both heating and cooling which were validated by manufacturing a demonstrator and applying a typical thermocompression bonding temperature profile to the heating element. From the experiment it follows that the concept shows potential for further product development but does not yet perform as required. The reflector surface reflectance,which has an important impact on the energy efficiency, was too low in this demonstrator. Furthermore heat leaks away from the heating element which inhibits a fast temperature rise and the complex cooling system geometry leads to manufacturing errors, which cause leaks in the cooling channel. By improving the design on these points, this concept can decrease the current cycle time by more than 50% and keep up with the growing demand for flex-on-substrate assemblies.

Viscoelasticity is a material property that is relevant in a variety of nanoscale materials and interfaces in medicine and industry. Therefore, a method of mechanical quantification has become exceedingly desired. In this thesis the Atomic force microscope (AFM) is applied to accurately characterize the mechanical behavior of viscoelastic samples. The goal is to enhance viscoelastic characterization using the so-called Intermodulation AFM (ImAFM) technique by applying, adapting and improving multiple modelling and optimization methods. In ImAFM force reconstruction is performed by extracting intermodulations around resonance in the cantilever response. These intermodulations present new observables that can be used for characterization. This thesis investigates the potential of this technique in combination with an up-and-coming model describing viscoelastic interaction. A toolbox has been developed for numerical simulations of the model to resemble the experiments. The model has been evaluated in a variety of situations using sensitivity analysis in a large feasibility range, encompassing many complex dynamics. Because of the diversity in model dynamics a global optimization has been performed for experimental reconstruction.

In the past decade, the wind energy subsidies provided by the Dutch government is declining. In order to adopt wind energy as a well-established source of renewable energy, the costs of (offshore) wind energy should be lowered. This can be achieved by increasing the energy production or by lowering the maintenance and initial costs. In search of lowering the costs, the wind industry is pushing towards the boundary limits of design. The use of multi-objective design is needed for exploring those limits, for which the field of control and structural engineering get together. Currently, little research is conducted of combining controller and structural design in the wind turbine industry. The goal of this thesis is to investigate the possibility of optimizing the support structure design and the controller design into one single optimization routine. The method describes the use of nonsmooth H_∞-synthesis. The modelling principles of the turbine makes use of different approaches. A simple finite element model of the wind turbine tower is constructed. The wall thickness of the tower sections is extracted as a tunable parameter, through an affine representation of the Ordinary Differential Equations (ODEs). In this thesis, the Controller Structure Optimization (CSO) for the offshore wind turbine combines the design of a load reducing Rotor Speed Controller (RSC) and wall thickness reduction. In theory, the bandwidth of the RSC is limited by the first tower bending mode. Therefore, the bandwidth of the designed controller is chosen well below the first eigenfrequency. It was found that the limitations do not influence the solution. The limitations of the proposed method are bounded by the weight functions, which impose a limit on the performance. Furthermore, it was found that the CSO framework can minimize the wall thickness of the tower elements, and simultaneously design a suitable RSC for tracking a rotor speed. The results are verified in high-fidelity wind turbine model, where the control-relevant subjects are addressed and specified. It was found that by addressing multi-objectives, we can design a controller with rotor tracking abilities and load reducing properties. Hence, the proposed CSO framework, can simultaneously design controllers and alter the structure and thereby increase the overall performance of the system.

Hydrostatic bearings are well known for having low friction and low wear in combination with high loading capacity.Traditionally, the running surfaces for a hydrostatic bearing are plane, rigid and smooth, ensuring a parallel lubricating film in the full range of movement. In this paper, we consider a new type of compliant, hydrostatic bearing where the running surfaces have a surface waviness with a large amplitude resulting in a varying curvature for the hydrostatic bearing to follow in its range of movement. Design choices and subsequent modeling of these types of bearings are presented in this work. The introduced bearing concept is based on a compliant whiffletree mechanism able to distribute load over multiple hydro-static slippers. This increases deformability of the complete bearing system. In a case study the limitations of a large deforming hydrostatic bearing are modeled to understand the impacts on performance and implications.

Air bearings are known for their low friction losses, constant viscosity and cleanliness, making them favourable in numerous high-tech applications. The relatively low aerostatic bearing stiffness is the major topic in state-of-the-art research and some recently proposed concepts are able to create infinite stiffness. This results in a strong coupling between the base and flying mass, enabling unwanted propagation of base vibrations. Why not integrate the vibration-isolation and bearing function? To reduce the out of plane vibrations of a flying mass. A previous study shows the possibility of creating a zero-stiffness bearing with the use of multiple aerostatic pads. However, the effect of the dynamic- thin-film stiffness and damping on the transmission of base vibrations is still unknown. The aim of this research is to optimise and validate the zero-stiffness concept and to study the effect of the static and dynamic pad characteristics on the transmissibility function. It is shown that both the stiffness and damping need to be minimised at equal operating fly height to create a stable system. A stable design is found with the use of a parametric optimisation procedure. The design shows a lower transmissibility when compared to a standard aerostatic bearing, although there are some challenges regarding high design sensitivities and displacement non-linearity. Different manufacturing methods are investigated to build a prototype to validate the concept. The measurements prove the validity of the model, although a clear low-stiffness operating point is not found, due to a combination of manufacturing difficulties and high design sensitivities.

Structure Optimization has been an important subject with many applications for centuries. In the last sixty years, numerical optimization has facilitated large advancements in this field. One of the areas in Structure Optimization is Topology Optimization, which is used for Additive Manufacturing purposes. In this thesis we explore Static and Dynamic Topology Optimization. In the optimization problems matrix equations of the type Ax = b, with A sparse and badly conditioned, are accelerated using deflation techniques in addition to preconditioning. We have applied several iterative methods, preconditioners, and deflation types to the topology optimization problems. The static problem concerns compliance optimization of a two-dimensional MBB-beam. The deflation type that reduced the number of iterations the most without introducing large overhead costs was rigid body modes deflation, divided over element squares. It was found that dividing the rigid body modes vectors over density based regions did not reduce the number of iterations. The dynamic problem concerns eigenvalue optimization of a three-dimensional moving wafer stage that is used for laser-printing computer chips. The optimization formulation contains a shifted eigenvalue problem that is solved using model order reduction. In the computations of bases for the reduction matrix equations appear, to which deflation techniques were applied. All deflation types reduced the number of iterations and needed time to solve the matrix equations. The best deflation type was using rigid body modes (RBM) divided over element cubes combined with eigenvectors from the previous iteration. The next best was the same combination without the division over cubes. Using eigenvectors or RBM separately were the least effective deflation types. There is an optimal amount of element cubes to use when dividing the RBM. The tests on a few grid sizes suggested a quadrupling of the amount when the grid size doubles, but more research is needed to really identify a relation between the grid size and optimal amount. When increasing the grid size to a level where parallel computing on a cluster was required, the deflation type using element cubes could not be used due to complications with parallel implementation. Using the deflation type RBM and eigenvectors, reductions by a factor of 1.60 and 1.75 in the total needed time for 150 optimization iterations were achieved for grid sizes 120x120x20 and 180x180x30, respectively. In the first case the objective function of the optimized design converged further in the same amount of iterations when using deflation. future research could include using the promising deflation type RBM in cubes + eigenvectors in parallel computing to obtain an even larger time reduction in the optimization of the wafer stage with large grid sizes.