Wire and Arc Additive Manufacturing (WAAM) emerged as a manufacturing process for large scale structures with extensive form and design freedom. WAAM can be fully exploited once the relation between the transient thermal history and its relation to microstructure development and resultant mechanical properties is established. This relation can be further used for computational design tools such as Topology Optimization. This paper presents a model to predict the relation between the thermal history and solid-state phase transformations in a widely applicable High Strength Low Alloy steel ER110S-G. The transient thermal history of parts manufactured by WAAM is modelled using finite element analysis. The modelled thermal history is validated with thermocouple measurements. Our results show that a critical cooling cycle is responsible for the solid-state phase transformation in an AM part. The cooling rate of this particular cooling cycle is superimposed onto an experimentally constructed Continuous Cooling Transformation (CCT) diagram to determine the local solid-state phase fractions. The predicted phase fractions in three wall samples with different design and processing conditions of AM parts are used to predict the hardness. The predicted hardness is 10% higher than the measured hardness of AM samples. The effect of tempering is also considered in the model through JMAK equation. The results show that the tempering is caused in regions with high martensite content and it lowers the hardness by 8 − 10%. The micrographs of the AM parts show that the microstructural features are same for the AM parts with similar critical cooling rates.

In metal Additive Manufacturing (AM), the deposited material is subjected to a series of heating and cooling cycles. The locally occurring temperature extremes and cooling rates determine solid-state phase fractions, material microstructure, texture, and ultimately the local material properties. As the shape of a part determines the local thermal history during AM, this offers an opportunity to influence these material properties through design. In this paper, we present a way to obtain desired properties by controlling the local thermal history. This is achieved through topology optimization of the printed part while considering its entire transient thermal history. As an example of this approach, this work focuses on high strength low alloy steels, where resulting phase fractions significantly influence mechanical properties such as yield strength and ductility. These solid-state phase fractions depend on cooling rates in a particular critical temperature range. The phase composition and hence the local yield strength in target regions can be controlled by constraining the cooling time in this range. Numerical examples illustrate the capability of the proposed approach in adapting part designs to achieve various desired material properties.

A remarkable elastic anisotropy in plates of austenitic stainless steel produced by the Wire and Arc Additive Manufacturing process is recently reported. The Young's modulus depends on the angle of orientation with respect to the material deposition direction. Here, for the first time, this anisotropy is exploited to maximize structural stiffness by simultaneously optimizing the structural design layout and the local deposition path direction for WAAM. The results obtained indicate deposition that is commonly preferred along the load-path directions for WAAM is sub-optimal and stiffness can be increased at least 53% upon optimizing the deposition directions.

Topology optimization typically generates designs that exhibit significant geometrical complexity, which can pose difficulties for manufacturing and assembly. The number of occurrences of an important design feature, in particular intersections, increases with geometrical complexity. Intersections are essential for load transfer in many engineering structures. For certain upcoming manufacturing processes, such as direct metal deposition, the size of an intersection plays a role. During metal deposition, slim intersections are more prone to manufacturing defects than bulkier ones. In this study, a computationally tractable methodology is proposed to both control occurrence and size of intersections in topology optimization. To identify intersections, a stress-based quantity is proposed, denoted as Intersection Indicator. This quantity is based on the local degree of multi-axiality of the stress state, and identifies material points at intersections. The proposed intersection indicator can identify intersections in both single as well as multi-load case problems. To detect the relative size of intersections, the average density in the vicinity of an intersection is used to penalize or promote intersection sizes of interest. The corresponding sensitivity analysis involves solving a set of adjoint equations for each load case. Numerical 2D experiments demonstrate a controllable reduction of penalized slim intersections compared to the designs obtained from conventional compliance minimization. The overall geometrical complexity of the design is reduced due to the promotion of bulkier intersections which leads to an increase in compliance. The designs obtained are more suitable for manufacturing processes such as direct metal deposition.

Composite materials are finding increasing application, for example in commercial aircraft. Traditionally fiber angles are constant in a single layer. Currently, so called variable stiffness panels with steered fibres, where the angle is changing within a layer are investigated. These panels are usually manufactured using automated fibre placement machines. Since the fibre angle is changing, and the tow paths are shifted as a whole in a single direction, gaps and/or overlaps between consecutive tows are created. This paper explores the effect of these gaps on the stiffness and buckling load of variable stiffness panels. A methodology is presented using homogenization to account for the gaps in a computational efficient way. The result shows that the stiffness results are on conservative side and are within 5% accuracy. However, the buckling results are on the unconservative side. The computational cost of the pre-processing of the proposed method is 45 times lower than the cost of the defect-ply method presented in the literature.