Additive Manufacturing (AM) is gaining popularity in aerospace and automotive industries. This is a versatile manufacturing process, where highly complex structures are fabricated and together with topology optimization, a powerful design tool, it shares the property of providing a very large freedom in geometrical form. The main focus of this work is to introduce new developments of Topology Optimization (TO) for metal AM. The thesis consists of two parts. The first part introduces background and theory, where TO and adjoint sensitivity analysis are described. Furthermore, methodology used to identify surface layer and high-cycle fatigue are introduced. In the second part, three papers are appended, where the first paper presents the treatment of surface layer effects, while the second and third papers provide high-cycle fatigue constraint formulations. In Paper I, a TO method is introduced to account for surface layer effects, where different material properties are assigned to bulk and surface regions. In metal AM, the fabricated components in as-built surface conditions significantly affect mechanical properties, particularly fatigue properties. Furthermore, the components are generally in-homogeneous and have different microstructures in bulk regions compared to surface regions. We implement two density filters to account for surface effects, where the width of the surface layer is controlled by the second filter radius. 2-D and 3-D numerical examples are treated, where the structural stiffness is maximized for a limited mass. For Papers II and III, a high-cycle fatigue constraint is implemented in TO. A continuous-time approach is used to predict fatigue-damage. The model uses a moving endurance surface and the development of damage occurs only if the stress state lies outside the endurance surface. The model is applicable not only for isotropic materials (Paper II) but also for transversely isotropic material properties (Paper III). It is capable of handling arbitrary load histories, including non-proportional loads. The anisotropic model is applicable for additive manufacturing processes, where transverse isotropic properties are manifested not only in constitutive elastic response but also in fatigue properties. Two optimization problems are solved: In the first problem the structural mass is minimized subject to a fatigue constraint while the second problem deals with stiffness maximization subjected to a fatigue constraint and mass constraint. Several numerical examples are tested with arbitrary load histories.

Since Additive Manufacturing (AM) techniques allow the manufacture of complex-shaped structures the combination of lightweight construction, topology optimization, and AM is of significant interest. Besides the established continuum topology optimization methods, less attention is paid to algorithm-driven optimization based on linear optimization, which can also be used for topology optimization of truss-like structures. To overcome this shortcoming, we combined linear optimization, Computer-Aided Design (CAD), numerical shape optimization, and numerical simulation into an algorithm-driven product design process for additively manufactured truss-like structures. With our Ansys SpaceClaim add-in construcTOR, which is capable of obtaining ready-for-machine-interpretation CAD data of truss-like structures out of raw mathematical optimization data, the high performance of (heuristic-based) optimization algorithms implemented in linear programming software is now available to the CAD community.

The papers in this volume focus on the following topics: design optimization and inverse problems, numerical optimization techniques,efficient analysis and reanalysis techniques, sensitivity analysis and industrial applications. The conference EngOpt brings together engineers, applied mathematicians and computer scientists working on research, development and practical application of optimization methods in all engineering disciplines and applied sciences.

In recent decades, the development of computer-controlled manufacturing by adding materiallayer by layer, called Additive Manufacturing (AM), has developed at a rapid pace. The technologyadds possibilities to the manufacturing of geometries that are not possible, or at leastnot economically feasible, to manufacture by more conventional manufacturing methods. AMcomes with the idea that complexity is free, meaning that complex geometries are as expensiveto manufacture as simple geometries. This is partly true, but there remain several design rulesthat needs to be considered before manufacturing. The research field Design for Additive Manufacturing(DfAM) consists of research that aims to take advantage of the possibilities of AMwhile considering the limitations of the technique. Computer Aided technologies (CAx) is the name of the usage of methods and software thataim to support a digital product development process. CAx includes software and methodsfor design, the evaluation of designs, manufacturing support, and other things. The commongoal with all CAx disciplines is to achieve better products at a lower cost and with a shorterdevelopment time. The work presented in this thesis bridges DfAM with CAx with the aim of achieving designautomation for AM. The work reviews the current DfAM process and proposes a new integratedDfAM process that considers the functionality and manufacturing of components. Selectedparts of the proposed process are implemented in a case study in order to evaluate theproposed process. In addition, a tool that supports part of the design process is developed. The proposed design process implements Multidisciplinary Design Optimization (MDO) witha parametric CAD model that is evaluated from functional and manufacturing perspectives. Inthe implementation, a structural component is designed using the MDO framework, which includesComputer Aided Engineering (CAE) models for structural evaluation, the calculation ofweight, and how much support material that needs to be added during manufacturing. Thecomponent is optimized for the reduction of weight and minimization of support material,while the stress levels in the component are constrained. The developed tool uses methodsfor high level Parametric CAD modelling to simplify the creation of parametric CAD modelsbased on Topology Optimization (TO) results. The work concludes that the implementation of CAx technologies in the DfAM process enablesa more automated design process with less manual design iterations than traditional DfAM processes.It also discusses and presents directions for further research to achieve a fully automateddesign process for Additive Manufacturing.