Design Strategies for Welding-Based Additive Manufacturing: A Review of Topology and Lattice Optimisation Approaches

Abstract

Topology optimisation and lattice design constitute key enablers in the transition towards high-performance and resource-efficient engineering, particularly within the framework of additive manufacturing and welding-based deposition processes. The increasing integration of arc-based technologies, such as Wire Arc Additive Manufacturing, has strengthened the relevance of these methodologies by enabling the fabrication of large-scale, structurally efficient components with controlled material distribution and mechanical performance. These design strategies provide unique opportunities to achieve lightweight structures, functionally graded behaviour, and tailored properties beyond the limitations imposed by conventional manufacturing and joining techniques. The growing demand for functionally efficient components in sectors such as aerospace, biomedical, and automotive engineering continues to drive the adoption of these approaches, where both material efficiency and structural integrity under welding-induced thermal effects are critical. This chapter introduces the fundamentals of topology optimisation and functionally graded lattice architectures, describes their integration into advanced design and manufacturing workflows, including welding-based additive processes, and presents selected case studies that demonstrate their practical impact. Finally, emerging strategies based on generative design and artificial intelligence are discussed as key drivers for the automated and process-aware optimisation of future additively manufactured and welded structures.

1. Introduction

For more than a century, welding has been a fundamental technology for joining metals, but recent developments have transformed it into an advanced scientific discipline that combines metallurgy, automation, artificial intelligence (AI), and additive manufacturing. Traditionally, welding was based on thermal fusion processes such as Shielded Metal Arc Welding (SMAW), Gas Metal Arc Welding (GMAW), and Gas Tungsten Arc Welding (GTAW), which enabled major advances in infrastructure, transport, and energy. However, today’s demands for efficiency, precision, and sustainability have driven a new generation of techniques and materials.

In recent years, welding research has reached an unprecedented level. According to a global analysis of technological trends, the most significant improvements fall into three main areas: processes (45%), equipment (30%), and materials (25%), which have made it possible to reduce welding time by 20%, decrease heat-affected zones by 15%, and increased the strength of joints by up to 10% [1].

One of the most notable advances has been in aluminium alloys, which are widely used in aerospace due to their high specific strength. However, conventional fusion welding of these alloys often suffers from hot cracking and degradation of mechanical properties. In 2024, a Zr-core/Al-shell composite wire combined with a hybrid laser–oscillating arc process was reported to suppress cracking in AA2024 welds and to achieve joint strengths comparable to solid-state friction welding [2].

In parallel, friction stir welding (FSW) has continued to demonstrate high joint quality in aluminium alloys. For AA2024-T3, recent studies report that FSW joints can achieve markedly higher mechanical performance than conventional MIG/TIG welds, together with improved hardness profiles and microstructural integrity [3].

In the context of Industry 4.0, welding has become integrated into cyber–physical systems with real-time monitoring using computer vision and optical sensors that automatically adjust the parameters of the electric arc to optimise the quality of the joint. These technologies lay the foundations for the convergence between traditional welding and additive metal manufacturing (AM), where one emerging technique stands out: Wire Arc Additive Manufacturing (WAAM).

WAAM uses an electric arc as a heat source and a metal wire as a filler material to build metal structures layer-by-layer, combining the principles of welding and 3D printing. This process offers high deposition rates (on the order of kilogrammes per hour), material utilisation of over 90%, and a drastic reduction in costs compared to methods based on metal powder. In addition, it allows the manufacture of large components in materials such as steels, titanium, and aluminium alloys, with precise control of the microstructure and residual stresses [4].

Figure 1 reports annual publication counts retrieved from Scopus and Web of Science for two reproducible queries—(i) welding and (ii) (“waam” OR “DED-ARC”)—over 2020–2025. The figure was built by running the searches in each database, filtering results by publication year, and recording the number of hits returned for each year. In Scopus, the search was performed in “Article title, Abstract, Keywords” (advanced query equivalent: TITLE-ABS-KEY (welding) and TITLE-ABS-KEY (“WAAM” OR “DED-ARC”)). In Web of Science, the search was performed in “Topic” (advanced query: TS = (welding) and TS = (“WAAM” OR “DED-ARC”)). Across both databases, the welding literature remains much larger in absolute volume and shows year-to-year variability, whereas the WAAM/DED-ARC query exhibits a clear and consistent growth trend from 2020 to 2025, indicating a rapidly expanding research focus on arc-based additive manufacturing. To improve readability, the two series are plotted using high-contrast colours and separate y-axes to account for the different magnitudes.

In this context of accelerated evolution of welding towards intelligent, automated processes that are fully integrated into digital manufacturing environments, it is necessary to place these emerging technologies within a broader framework of industrial design and production transformation. The convergence between techniques such as Wire Arc Additive Manufacturing and the principles of additive manufacturing not only redefines the traditional limits of material joining, but also drives a profound change in the way engineering structures are conceived, optimised, and materialised. Thus, the ability to manufacture components of great geometric complexity, with microstructural control and high material efficiency, connects directly with the new design paradigms oriented towards lightness, multifunctionality, and sustainability that characterise modern additive manufacturing and advanced structural optimisation methodologies.

Metal Additive Manufacturing

In recent years, the quest to find lightweight and efficient structures has become a fundamental objective in the fields of mechanical, aerospace, biomedical, and product development engineering [5]. Interest in reducing material consumption, optimising structural performance, and reducing environmental footprint has also led to the creation of new design and manufacturing methodologies [6].

The evolution of additive manufacturing in industry has been marked by technological advances and growing adoption in various applications, positioning it as a technology that has transformed structural design. It is one of the fundamental pillars of the well-known 5th Industrial Revolution, which has brought about a change in both the technological and social paradigms [7,8].

Compared to conventional processes, this technique offers intrinsic advantages such as geometric freedom, which allows complex and multifunctional structures to be created that were not previously feasible. This technique reduces costs, development times, and waste, while offering unprecedented flexibility in product customisation and optimisation [9].

One of the most notable applications is the lattice-type structure, which is characterised by its mass optimisation and high strength [10]. This configuration allows for the strategic distribution of material and loads, creating highly rigid components that absorb energy and tend to adapt to different loads [11,12].

The importance of additive manufacturing has increased through recent research based on topological optimisation, generative design, and multiphysics modelling, which integrates structural strength, thermal conduction, and even vibration damping into a single component [13]. However, traditional design methodologies continue to limit the full adoption of these approaches, as they are not 100% adapted to new technological capabilities or to the demands of sustainability and material efficiency [14].

Additive manufacturing not only represents innovation in the production process, but also a change in the conception of structures, directing research towards adaptive, efficient, and sustainable design [15]. Despite progress in the optimisation of structures and the application of additive manufacturing, there are still limitations derived from traditional design that prevent its full potential from being realised [16].

Traditional design methods are based on simple geometries and uniform material distributions, which prioritise simplicity in manufacturing and load control [17]. These are effective in industrial contexts, but we live in a fast-paced and constantly changing era where these approaches often do not meet the demands of material optimisation, sustainability, and multifunctionality [18]. On the other hand, traditional design tends to separate structural and functional roles, limiting the creation of components that combine rigidity, energy absorption, thermal management, or cushioning [18]. In this context, welding-based processes and WAAM/DED also provide a distinctive route to multifunctionality through multi-material and dissimilar metal architectures—where tailored material placement and engineered interfaces enable property gradients and localised functionality—supported by recent studies on dissimilar metal interface evolution in fusion welding, bi-metallic WA-DED design strategies, and steel–bronze bimetal additive manufacturing with quantified interface and tensile behaviour [19,20,21].

In contrast, additive manufacturing offers the possibility of producing bespoke physical objects and functional prototypes, supporting just-in-time production and limited-series manufacturing to produce small quantities of products in a cost-effective manner. Moreover, it enhances product customisation and contributes to the optimisation of efficient and sustainable value chains. By incorporating successive layers of material, it allows complex parts to be created at a lower cost for industries with very specific needs. Configurations such as lattice structures allow for minimal periodic surfaces (TPMSs) or density gradients to develop hierarchical, lightweight, and adaptive structures [10]. However, current CAD design tools and simulation models still have limitations in representing and analysing the complexity of these geometries, especially when multiphysical or anisotropic behaviours derived from the manufacturing process are involved [22].

Overcoming these limitations requires adopting more comprehensive design strategies, such as AI-assisted topographical optimisation or generative design specific to additive manufacturing, which allows structural, thermal, and functional criteria to be balanced simultaneously [23].

Therefore, the main problem lies in the gap between the technological capabilities of additive manufacturing and design methods that are still dominated by traditional paradigms. In order to narrow this gap, it is necessary to rethink the principles of structural design, orienting them towards a more global vision, where material efficiency, multifunctionality, and sustainability are fundamental pillars in the development of future advanced structures [24].

Accordingly, this review establishes an integrative framework that connects topology optimisation, structural metamaterials, and lattice architectures as complementary pillars of advanced design for additive manufacturing (AM). Within this framework, efficient material distribution, functional customisation, and morphological adaptation are examined as converging design strategies that extend beyond the capabilities of conventional manufacturing. The article is structured around three main axes: (i) topological optimisation as a methodology for generating lightweight and mechanically efficient configurations through optimal material distribution in a given design domain; (ii) metamaterials and lattice structures as microarchitectural extensions of topological design enabling local tailoring of mechanical, thermal, and dynamic properties; and (iii) the integration of these approaches into AM workflows, with particular attention to Design for Additive Manufacturing (DfAM) and the explicit consideration of process-related constraints and opportunities from the earliest design stages. Throughout the review, emphasis is placed on the evolution towards multi-scale and multifunctional design strategies supported by computational modelling, simulation, and experimental validation, as well as on emerging perspectives that arise from the convergence of topological optimisation, generative design, and artificial intelligence for the intelligent automation of the design process.

2. Fundamentals of Topology Optimisation

Topology optimisation (TO) is one of the most powerful tools in computer-aided structural and functional design. Its objective is to determine the best possible distribution of material in a given domain in order to maximise the performance of a structure—for example, its rigidity, stability, or thermal efficiency—while simultaneously minimising the use of resources. Unlike shape or sizing optimisation, TO operates at a conceptual level, enabling the automatic generation of innovative structural configurations that are not easily found through designer intuition or traditional trial-and-error methods [25].

Since its consolidation in the 1980s, the discipline has evolved significantly, driven by advances in numerical calculation, finite element methods, and computational capacity. The mathematical formulation of these problems is based on an objective function that represents the overall performance of the structure, together with a set of physical constraints (equilibrium, volume, stresses, natural frequencies, etc.) and design parameters that define the distribution of material. This approach provides a systematic framework for exploring the design space, where decisions are no longer limited to adjusting dimensions, but to creating or eliminating regions of material according to structural efficiency criteria.

The current relevance of topological optimisation has multiplied with the expansion of additive manufacturing (AM). The geometric freedom inherent in AM technologies—capable of producing complex geometries and functional gradation—allows the results obtained by optimisation algorithms to be directly materialised. A quantitative picture of mass-saving potential in welding-based AM can be drawn by separating applications by scale. For large-scale structural parts (e.g., metre-scale beams), topology optimisation tailored to wire-arc DED has been reported to deliver 38–63% material savings for fully optimised designs, while hybrid concepts (conventional profiles locally reinforced by wire-arc DED) typically achieve 30–52% savings, maintaining a better balance between performance and manufacturability [26]. In civil/architectural steel connections, topology-optimised DED-Arc joints have been reported with up to ~70% mass reduction compared to conventional joint geometries (case study on gridshell joints), while still meeting structural requirements [27]. For relatively small WAAM/DED-Arc parts, achievable lightweighting is typically constrained by the minimum wall/strut thickness imposed by bead size and by the need to preserve machining allowances on functional surfaces. Nevertheless, a WAAM-oriented case study based on the GE Aircraft Engine Bracket Challenge reported a WAAM-feasible redesign that reduced the component volume to 21% of the original (≈79% volume reduction) while keeping the functional interfaces unchanged; the authors also note that this example was primarily a geometric optimisation and did not yet enforce stiffness/rigidity constraints [28].

However, the relationship between computational design and manufacturability poses new challenges: minimum printable thicknesses, control of draft angles, material anisotropy, and internal accessibility, among others. In this context, Design for Additive Manufacturing (DfAM) emerges as a complementary framework that integrates manufacturing criteria into the optimisation process itself.

Therefore, understanding the fundamentals of topological optimisation involves not only mastering its mathematical formulation and the computational methods that underpin it, but also interpreting its results under the constraints and opportunities of real manufacturing processes. The following sections address these three pillars progressively: the conceptual formulation of the problem and the most commonly used objective functions (Section 2.1), classical computational approaches and their hybrid evolutions (Section 2.2), and finally, the challenges of interpretation and manufacturability in the context of additive manufacturing (Section 2.3).

2.1. Conceptual Foundations

Topological optimisation can be formulated as the problem of finding the material distribution $\rho$over the design domain $\mathsf{\Omega }$ that minimises a given objective under physical and design constraints. A common formulation is compliance minimisation:

$$\begin{array}{cc}\underset{\rho }{\min }J\left(\rho \right)=f^{T}u\hfill & \\ \text{s.t.}\hfill & \\ & K\left(\rho \right)u=f\\ & \underset{\Omega }{\int }\rho \left(x\right)d\Omega \le V^{*}\\ & 0<{\rho }_{min}\le \rho \left(x\right)\le 1\end{array}$$

(1)

where $u$is the displacement vector, $f$ is the external load vector, and $K\left(\rho \right)$ is the global stiffness matrix resulting from the material distribution $\rho$. $V^{*}$ is the maximum allowed material volume in the design domain, which controls the mass/volume trade-off by restricting how much material the optimiser can place. In the continuous setting, the strain field is $\epsilon (u)={\displaystyle \frac{1}{2}}(\nabla u+\nabla u^{\mathsf{T}})$, and the constitutive relation is $\sigma =C\left(\rho \right):\epsilon \left(u\right)$, where $C\left(\rho \right)$ is the interpolated constitutive tensor (e.g., SIMP interpolation).

The most common objective functions include minimising compliance (maximising stiffness), minimising mass, or controlling the first natural frequency. Multiphysical problems can also be considered, combining thermal, mechanical, or vibration criteria [29,30].

2.2. Computational Approaches

The most widely used methods for solving topological optimisation problems include SIMP, ESO/BESO, Level-Set, and MMA, better described in

Table 1. Overview of principal topology optimisation methods and their defining characteristics.

Method	Characteristic
SIMP (Solid Isotropic Material with Penalisation)	Interpolates the stiffness of each element as $E\left(\rho \right)={\rho }^{\mathrm{p}}{E}_0$, penalising intermediate solutions. Uses filters and projections to control the size of structural elements [31].
ESO/BESO (Evolutionary Structural Optimisation)	Removes or reincorporates material based on deformation energy. Generates sharp boundaries, but is highly dependent on removal parameters [32,33].
Level-Set	Represents the boundary using an implicit function $\phi \left(x\right)=0$. Allows precise geometric control and smooth curvatures, although it requires additional techniques to create or remove holes [34].
MMA (Method of Moving Asymptotes)	A robust optimisation method for multiple non-linear constraints, widely used in combination with SIMP [35]

. Each is based on different strategies for representing geometry and updating design variables.

Recent developments combine hybrid methods (e.g., SIMP + Level-Set) and multi-scale strategies that integrate macrostructure and microstructure optimisation. Specific additive manufacturing constraints such as self-support, critical printing angles, or minimisation of support volume are also incorporated [36,37,38].

2.3. Design Interpretation and Manufacturability

One of the main challenges of topological optimisation is translating numerical results into manufacturable geometries, especially through additive manufacturing (AM). The algorithm outputs may present elements that are too fine, inaccessible cavities, or angles that are unsuitable for the printing process.

The main aspects are shown in Figure 2 and described below:

• Characteristic scale and minimum thickness. Ensure that minimum wall thickness, strut diameter, and feature radii exceed the process-specific printable limit. In welding-based AM/DED, this limit is largely governed by bead width/height, melt pool stability, and achievable dimensional control; thin members may also be prone to local overheating, distortion, or collapse during multi-pass deposition.
• Overhangs and supports. Identify surfaces with critical inclinations and redesign them to be self-supporting where possible. For powder bed processes, this is typically governed by overhang angles and support removal, whereas for WAAM/DED it is additionally constrained by bead wetting behaviour, gravity-driven sagging, and the need for continuous, stable deposition paths (often favouring gradual slopes and fillets rather than sharp unsupported overhangs).
• Internal accessibility. Guarantee access for powder evacuation (PBF) and, more generally, for inspection, finishing, and machining. In addition to avoiding trapped powder, internal channels or cavities should be designed to allow tool access for post-processing, as well as sensor/inspection access when qualification is required.
• Material anisotropy. Account for direction-dependent properties arising from layerwise deposition, thermal gradients, and solidification texture. This requires selecting build orientation and deposition trajectories consistent with principal load paths, and considering how thermal history and inter-layer bonding may influence strength and especially fatigue performance.
• Post-processing. Anticipate post-processing from the design stage by reserving machining allowances on datum and functional surfaces, defining support removal/finishing strategies, and planning heat treatments (e.g., stress relief) where needed. For welding-based AM, post-processing is often essential to achieve tolerances and to mitigate residual stresses and surface roughness.

Design for Additive Manufacturing (DfAM) integrates these factors from the initial optimisation phase, allowing the generation of self-supporting structures that are robust against anisotropy and require fewer supports [39,40].

2.4. Lattice Structures

In recent years, topological optimisation has transcended its classic application in macroscopic structures to integrate with multi-scale design approaches, in which material distribution is controlled not only at the global level—defining where to place or remove material—but also at the local level, modulating the internal microarchitecture of the component. Within this context, lattice structures and Functionally Graded Lattice Structures (FGLSs) represent a natural extension of the principles of topological optimisation.

These allow the stiffness, density, or energy absorption capacity to be adjusted according to the load conditions or local functional requirements, while maintaining structural continuity compatible with additive manufacturing [41,42]. The integration of these structures into the optimised design flow opens up the possibility of combining the material efficiency of topological optimisation with the mechanical and energy customisation of graded lattices, generating components with superior performance and adaptive behaviour.

They are based on the periodic repetition of three-dimensional unit cells, which can adopt open (truss-type) or continuous (TPMS-type, Triply Periodic Minimal Surfaces) geometries, giving rise to architectural materials with mechanical and thermal behaviour that can be tuned according to their application [43].

The most representative types are as follows:

• Kelvin and Octet, which are cubic in nature and have a closed structure, exhibiting mainly elastic behaviour and high compressive strength [44].
• Gyroid and other TPMS, based on minimally triply periodic surfaces, characterised by their continuity without intersections and their excellent ability to distribute stresses and facilitate thermal flow [45].
• Chiral and re-entrant, which exhibit auxetic behaviour with a negative Poisson’s ratio, giving them remarkable energy absorption and elastic recovery capabilities after impact [46].

The structural performance of these configurations depends on variables such as relative density, spatial orientation of the cell, internal topology, and base material properties.

In welding-based AM (WAAM/DED-Arc), the selection of lattice topology cannot be based solely on mechanical descriptors (e.g., stiffness-to-weight), because manufacturability is strongly constrained by bead geometry, thermal history, and toolpath accessibility. In practice, the “printability window” is governed by (i) a minimum feature size comparable to bead width/height (which limits strut diameter and node resolution), (ii) the need for stable, continuous deposition paths (penalising frequent start–stop events and highly branched nodes), (iii) overhang sensitivity and local collapse risks in slender members, and (iv) multi-pass thermal cycling and heat accumulation, which can cause local remelting, loss of geometric fidelity, and distortion—effects that become critical in lattices due to their high surface-area-to-volume ratio and thin members.

From a comparative standpoint, truss-based lattices [44] are often more compatible with WAAM/DED when cell sizes and strut diameters are sufficiently large, because they can be decomposed into a limited set of repeated strut directions and relatively simple nodes. Their main limitation is that complex junctions may require many path interruptions, increasing the likelihood of local over-deposition and geometric drift. By contrast, TPMS-based lattices [45] offer smooth stress transfer and reduced stress concentrations due to surface continuity; however, their fabrication in welding-based AM is typically challenged by the need to approximate complex curved surfaces using bead stacking, which may induce staircase effects, local thickness non-uniformity, and high heat input per unit volume. Consequently, TPMS architectures may be more feasible in WAAM/DED when interpreted as thin-wall or shell-like features with controlled layer-by-layer trajectories, or when combined with hybrid machining to recover dimensional accuracy.

Auxetic concepts [46] are attractive for energy absorption, but they are generally more demanding for welding-based AM because they rely on slender ligaments, sharp corners, and local overhangs, which are susceptible to thermal distortion and geometric rounding. For WAAM/DED, these lattices are typically feasible only at coarser scales (larger unit cells and thicker members) or when supported by process strategies such as reduced heat input, strict inter-pass temperature control, and post-deposition finishing.

These considerations motivate a selection logic for welding-based AM in which lattice candidates are screened using process–structure compatibility criteria: (1) feature size margin relative to bead geometry (strut diameter and node radius), (2) path continuity and number of re-ignitions per unit volume, (3) overhang and local self-support capability, (4) sensitivity to thermal cycling (risk of remelting/distortion in thin members), and (5) post-processing requirements (machining access and tolerance recovery). This process-aware screening is particularly relevant for functionally graded lattices, where local density variations should be implemented through robust parameters (e.g., gradual strut thickness changes or cell size scaling) to avoid abrupt transitions that amplify thermal gradients and residual stress concentrations.

2.5. Functionally Graded Lattices

The concept of FGLS arises from the need to design components with non-uniform spatial properties, adapting the internal structure to local load conditions or thermal requirements [47]. In these configurations, parameters such as cell density, pore size, or even material composition vary progressively within the volume of the component, generating a controlled transition between regions of different stiffness or conductivity. Figure 3 shows different types of gradation:

• Uniform, with a constant density distribution;
• Radial Gradient (RG), with variation from the centre to the periphery;
• Horizontal Gradient (HG) and Vertical Gradient (VG), which modify the dimensions or thickness of the struts along a given axis.

These gradations make it possible to reduce stress concentrations, improve energy absorption, or adapt the thermal response of the material. Several studies have shown that a gradual transition in density in metal structures manufactured by DED or SLM can improve fatigue resistance and reduce overall stiffness without compromising structural stability [49]. In biomedical applications, this strategy is used to reproduce elastic modulus gradients similar to bone tissue, promoting osseointegration [50].

From a design methodology perspective, FGLS should not be treated as a separate from topology optimisation, but rather as a continuation of TO across scales. In practice, TO can be used to define the global load paths and the spatial distribution of material at component level, while the lattice gradation acts as the local realisation of that distribution, mapping macro-scale requirements into microarchitecture parameters. This coupling can be implemented by driving the local lattice descriptors (e.g., relative density, strut thickness, unit-cell size, or TPMS wall thickness) with TO outputs such as element-wise density fields, strain-energy density, stress/strain measures, or compliance sensitivities. As a result, gradation becomes a core strategy to translate the TO solution into a manufacturable architecture, enabling smoother transitions, reduced stress concentrations, and improved robustness. For welding-based AM (WAAM/DED), this integrated approach is particularly relevant because thermal history and minimum feature constraints favour continuous, gradual variations (rather than abrupt density jumps), which helps preserve geometric fidelity under multi-pass thermal cycling and reduces distortion-prone discontinuities.

2.6. DED-Specific Constraints for TO and Lattice/Graded Lattice Designs

Unlike powder bed fusion, WAAM and other welding-based DED processes build parts through successive bead deposition, inherently involving high heat input and multi-pass thermal cycling [51]. This process signature introduces constraints that directly affect the feasibility and performance of topology-optimised (TO) geometries and lattice-based architectures, particularly functionally graded lattices.

First, heat accumulation and repeated reheating can lead to significant spatial and temporal temperature gradients. For lattice structures, where thin struts and nodes are critical load-bearing features, multi-pass reheating may cause local remelting, geometry drift (thickening or rounding of struts), and loss of designed gradients due to bead overlap and local heat concentration. This is especially relevant for graded lattices, where gradual changes in relative density can be compromised if thermal history is not controlled across regions with different thermal masses [52].

Second, DED is highly sensitive to residual stresses and distortion driven by repeated thermal expansion/contraction [53]. Topology-optimised shapes often contain slender members and asymmetric load paths that can be more prone to warping during deposition. For lattice and graded lattice designs, residual stresses may reduce effective fatigue life and can promote premature buckling of thin struts, particularly in regions subjected to high tensile residual stress states.

Third, welding-based AM imposes geometric/process limits linked to bead morphology and deposition strategy. Minimum achievable feature size is constrained by bead width/height, wetting behaviour, and arc stability, which may conflict with TO-generated fine members and lattice struts [54]. Additionally, deposition requires continuous toolpath planning with feasible torch orientations and access; abrupt changes in direction or unsupported thin features can trigger lack-of-fusion defects or collapse during build.

To improve manufacturability and structural reliability in DED, TO and lattice design workflows should incorporate the following process-aware constraints: (i) minimum strut thickness and node radius consistent with bead dimensions; (ii) thermal management rules (inter-pass temperature limits, dwell times, segmentation strategies, and/or active cooling) to reduce heat accumulation and preserve graded architectures; (iii) distortion-aware optimisation, where constraints on global curvature/warpage or compensation strategies are considered; and (iv) post-deposition treatments (stress relief heat treatment and/or hybrid machining) planned early to restore dimensional accuracy and reduce residual stress effects.

Overall, integrating these DED-specific considerations into TO and lattice design improves the relevance of optimisation outcomes, reduces the gap between numerical designs and manufacturable geometries, and better reflects the process physics governing welding-based additive manufacturing [55].

3. Modelling and Simulation Tools

The design, analysis, and validation of lattice structures requires advanced parametric modelling and finite element analysis (FEA) tools that allow direct relationships to be established between geometry, relative density, and resulting properties [47].

The following are some main platforms commonly used:

• Parametric CAD/CAE environments, such as SolidWorks, Autodesk Fusion 360, Topology, or Rhinoceros + Grasshopper, which allow lattices to be generated by controlling geometric parameters (cell size, angle, strut radius, density, etc.) using custom algorithms or scripts [56].
• Mechanical and thermal simulation using FEA, through software such as Abaqus, ANSYS, or COMSOL Multiphysics, aimed at predicting structural response under real loads, deformations, or heat flows [57].
• Topological and generative optimisation, integrating evolutionary or artificial intelligence algorithms that define the optimal distribution of material based on criteria of stiffness, weight, or energy dissipation, extending the principles of DfAM to the field of lattice design [40].

These tools enable the connection between digital design geometry and the actual limitations of additive manufacturing, especially in DED technologies, where weld accuracy, bead size, and temperature control directly affect geometric resolution and internal pattern reproducibility [58,59].

For welding-based AM, purely mechanical or thermal simulations are often insufficient because the process is governed by a strong thermo–mechanical–microstructural coupling driven by repeated deposition and multi-pass thermal cycling. Temperature fields and cooling rates control melt pool geometry and solidification, which in turn define microstructural features (e.g., grain morphology/texture, phase fractions, and precipitates) and local property gradients. These evolving properties affect the mechanical response, residual stress accumulation, and distortion, particularly in lattice structures where thin struts and nodes experience steep thermal gradients and repeated reheating. Consequently, advanced modelling strategies increasingly combine (i) transient thermal analysis to predict heat accumulation and inter-pass temperature, (ii) thermo-elasto-plastic simulations for residual stress and distortion, and (iii) microstructure/property models (e.g., phase transformation or grain-growth kinetics, and hardness/strength correlations) to capture spatially varying material behaviour. From a DfAM perspective, incorporating this coupling is essential to define process-aware constraints and to improve the reliability of predictions for geometry retention, fatigue performance, and the thermal stability of graded lattice architectures in WAAM/DED.

4. Integration into Additive Manufacturing Workflows

Design for Additive Manufacturing (DfAM) represents a paradigm shift from traditional design oriented towards subtractive manufacturing. This approach seeks to exploit the intrinsic capabilities of additive manufacturing (such as geometric freedom, optimisation of material use, and functional integration) while mitigating its limitations related to resolution, anisotropy, and the need for supports. The integration of DfAM into the overall additive manufacturing workflow (Figure 4) allows the conceptual design and optimisation process to be linked to the subsequent stages of manufacturing, control, and post-processing, consolidating an iterative cycle of improvement and validation. This flow ranges from the selection of technology and manufacturing parameters to the characterisation of the physical prototype, establishing a feedback loop that guides redesign and topological optimisation.

4.1. Principles of DfAM

The principles of DfAM are based on the integration of additive manufacturing process conditions into the early stages of design. This includes the definition of geometric constraints (minimum printable thickness, curvature radii, cell size in lattice structures, and internal accessibility), print orientation, and support strategies that ensure stability during the process. Anisotropic effects derived from layer-by-layer deposition and the influence of thermal parameters on the final properties of the material are also considered.

These principles allow the morphology of the component to be adapted to the behaviour of the process, ensuring a consistent relationship between the shape generated by topological optimisation and the actual capabilities of the selected technology. In this regard, it is crucial to identify printable geometric thresholds, such as minimum wall thickness or maximum slope angle, which depend on the technology (e.g., PBF, DED, or WAAM) and the material used [60,61,62].

Likewise, DfAM must incorporate the orientation of the component in relation to the deposition trajectories, as this determines the anisotropic mechanical response of the material [59,63].

The design/redesign phase is supported by integrated CAD and CAM tools that allow not only the geometry to be defined, but also the deposition strategy and manufacturing parameters, such as speed, current, material feed, or inter-pass temperature.

4.2. Workflow Integration

The proposed workflow (Figure 4) articulates the DfAM methodology in a structured sequence that begins with the selection of the technology and culminates in the geometric and mechanical verification of the prototype. This scheme highlights the importance of bidirectional interoperability between digital tools and manufacturing systems, allowing for continuous iteration between design, simulation, manufacturing, and control.

4.3. Technology Selection

The choice of additive process directly determines design constraints, support strategies, and available materials. This selection is made based on criteria such as the required resolution, part volume, expected roughness, or dimensional tolerances, as it can be seen in

Table 2. Comparative analysis of AM technologies and their influence on design decisions, based on data reported by [64].

Technology	Process Principle	Key Design Constraints (DfAM)	Design Opportunities (DfAM)	Typical Impact on Design and Validation
PBF (LPBF/EB-PBF)	Selective melting of powder layers with a laser/e-beam.	Min. wall/strut thickness; overhang limits and supports; residual stress/warpage; build size constraints; powder removal for internal channels/lattices.	Very high geometric complexity; fine lattices/TPMS; internal channels; high-resolution TO outcomes.	High dimensional accuracy but support removal and post-processing often required; properties can be anisotropic; fatigue sensitive to surface/defects → finishing and inspection critical.
DED (Laser-DED/WAAM/EB-DED)	Directed energy creates a melt pool while powder/wire is fed; material is deposited track-by-track.	Lower geometric resolution (bead/track size sets min. features); surface roughness; path planning/access constraints; heat accumulation and multi-pass thermal cycling → residual stress/distortion; tight tolerances require machining allowances.	Large parts and high deposition rates; local reinforcement; repairs and remanufacturing; multi-material/graded transitions; hybrid AM–machining routes.	Near-net-shape builds with significant post-machining; thermo–mechanical history strongly affects microstructure/properties → monitoring + process control important for repeatability and qualification.
Binder Jetting	Binder selectively printed into powder bed; followed by curing and sintering/infiltration.	Shrinkage/distortion during sintering; green strength limits; final density/mechanical properties depend on sintering/infiltration; tolerances driven by post-processing.	No supports; high throughput; complex geometries; potential multi-material (by powder/binder strategy) and low-cost prototyping.	Dimensional change must be compensated in design; mechanical performance depends on densification route; parts often require infiltration/heat treatment and final machining for precision.

.

4.4. Design/Redesign and Application of DfAM

At this stage, topological optimisation algorithms are integrated with CAD/CAM models, generating geometries that are subsequently adjusted according to DfAM principles. The designer must validate that the optimisation results comply with the previously defined geometric constraints and process parameters [65].

4.5. Characterisation and Verification

Once the prototype has been manufactured, a material and process characterisation phase is carried out, including mechanical testing, microstructural analysis, and dimensional accuracy assessment. These results are fed back into the digital model, adjusting both the design conditions and the manufacturing parameters in an iterative optimisation cycle [66].

This integrated workflow consolidates an iterative learning process that links empirical knowledge of the process with parametric design, promoting the maturity of the digital manufacturing model.

4.6. Post-Processing and Validation

Post-processing is an essential step in ensuring the final functionality of the component. It includes machining, heat treatment, removal of support, and surface finishing, which ensure tolerance compliance and dimensional stability [67].

Geometric validation includes comparing the optimised CAD model with the actual geometry using 3D scanning or coordinate metrology techniques, while mechanical validation is based on standardised tests (tensile, compression, hardness, fatigue) that allow the simulation results to be correlated with physical behaviour.

This phase closes the DfAM integration cycle, feeding the results obtained into the iterative redesign and optimisation process. In this way, the digital model is progressively refined, improving both structural performance and process reproducibility.

5. Application Case: Topology Optimisation and DfAM for DED-Arc

Figure 5 illustrates a practical application of the DfAM workflow integrated into a structural redesign process aimed at Directed Energy Deposition by Arc (DED-Arc) [40]. The original component, conceived under conventional machining constraints, was subjected to an iterative process of topology optimisation, followed by a geometry redesign specifically adapted for additive manufacturing through energy deposition.

The workflow comprises three main phases:

• Topology Optimisation (TO Design): initial mass reduction achieved by redistributing material according to principal stress paths, leading to a 61% weight reduction compared to the baseline model.
• Redesign for DED-Arc: geometric adjustment to ensure printability and thermal stability, including the adaptation of overhang angles and minimum bead thickness to the constraints of the DED process.
• DfAM Design for DED-Arc: final morphological reinterpretation incorporating self-supporting features, minimised support volume, and controlled thermal gradients, achieving a total mass reduction of 63% relative to the original design, while maintaining mechanical performance within safe limits validated through FEA simulation.

This case study summarises the essence of the integration of DfAM within the topology optimisation workflow, demonstrating how iterative adaptation of the geometry to manufacturing and structural conditions enables substantial weight reduction while preserving functional integrity. The correlation between mass saving and mechanical reliability validates the efficiency of the approach as a design-for-manufacturing strategy aimed at material efficiency and sustainability.

6. Conclusions

This review consolidates Design for Additive Manufacturing (DfAM) as a practical framework to align design intent with process capabilities. It shows that the success of topology-optimised and lattice-based architectures depends not only on algorithmic performance, but also on manufacturing feasibility (geometrical constraints, process-aware optimisation, and post-processing/validation).

The proposed workflow—linking technology selection, optimisation, fabrication, and mechanical validation—highlights the need for a bidirectional digital–physical loop. Such coupling enables iterative refinement of design rules and improves both performance and manufacturability.

Looking forward, structural innovation in AM is expected to converge toward unified workflows that combine topology optimisation, lattice design, and generative design, enabling multi-objective trade-offs (mechanical, thermal, and environmental) and supporting graded and hybrid architectures responsive to local requirements [68,69]. However, three barriers still limit generalisation and industrial scalability: (i) high computational cost for large-scale/high-resolution models, (ii) limited standardisation for simulation–experiment correlation, and (iii) process variability affecting reproducibility and the definition of transferable DfAM rules [70,71,72]. Addressing these gaps requires tighter integration of monitoring, data analytics, and adaptive feedback.

DfAM also provides a direct lever for sustainability, as design and process choices govern material usage, support volume, energy demand, and downstream operations. Incorporating eco-design principles and circular strategies (e.g., part consolidation, repairability, recyclable alloys, and modularity) enables structural efficiency and environmental impact to be co-optimised within a single workflow [73,74].

Artificial intelligence is likely to accelerate DfAM by enabling automated design generation, predictive modelling, and adaptive process control. Promising directions include AI-assisted topology/lattice synthesis, real-time parameter adjustment using sensing and vision, and early-stage prediction of life-cycle impacts to guide sustainable design decisions [75,76].

Ultimately, several scientific challenges still need to be addressed to make DfAM fully actionable for welding-based AM/WAAM. First, there is a need for integrated WAAM thermo–mechanical (and, where relevant, microstructural) models that remain computationally efficient enough to provide rapid design feedback. Second, closed-loop digital–physical workflows must be enabled, in which in situ monitoring and data analytics inform adaptive path planning and, when necessary, real-time geometry/topology updates. Third, standardised validation and qualification procedures are required to reliably correlate simulations with repeatable mechanical performance—particularly fatigue—for topology-optimised, lattice, and functionally graded architectures. Finally, future design frameworks should support genuine multi-objective optimisation, jointly balancing structural performance, WAAM-specific manufacturability constraints, and sustainability metrics within a unified decision process.