Perspective Chapter: Design Considerations for Additive Manufacturing

Abstract

The Additive Manufacturing (AM) technology, which was known as rapid prototyping referring to its original usage in prototype production, is progressing fairly well toward full-scale manufacturing of final parts with improvements in precision, strength, and speed. The technology is no wonder a revolution in manufacturing as it allows significant advantages over traditional part making especially for often preferred complex free-form geometries. The AM in theory should allow the production of a simple cube or complex structures of the same volume with the same time and effort. However, this is not always the case; realizing the infinitely design flexible capability of AM with more functional and lighter parts is not a trivial task requiring many iterations and part-specific design considerations such as support structures, part orientations, and interdependent process parameters. Although traditional Design for Manufacturing (DFM) concept has been well documented, Design for Additive Manufacturing (DfAM) is still far from reaching satisfactory levels and correct designs are usually obtained with experience-based heuristic approaches. Here we aim to extract the very best practices in DfAM approaches from open literature and offer a generic guide for engineers designing for AM.

1. Introduction

The Additive Manufacturing popularly known as 3D printing is broadly defined by ASTM as a “process of joining materials to make objects from 3D model data” [1]. This definition of rather new manufacturing process essentially refers to a set of technologies in which parts are produced via layer-by-layer material addition as opposed to subtractive methods based on part removal. Hence physical prototypes are easily and cost-effectively produced directly from digital model data without any tooling. The process begins with extracting 3D CAD data in.STL format used in slicer software. The most commonly used. STL (standard triangle language or surface tessellation language) is a file format in which geometries are expressed in finite number of adjoining triangular facets. STL file does not contain any accommodate surface specific data; for example, to identify characteristics such as allowable roughness, intended fit, color, and other surface functions. However, thanks to the user-definable data in binary STL format, relevant surface data can be defined informally [2]. Other data formats such as Polygon File Format (PLY), Additive Manufacturing Format (AMF), and 3D Manufacturing Format (3MF) have been developed to overcome these limitations in STL. The density of these triangles affects the geometrical resolution for curved surfaces (Figure 1). For higher resolution, triangulation elements should be denser in complex and curved surfaces, while they can be coarser on planar surfaces.

Figure 1.

For higher resolution, triangulation elements should be denser in complex and curved surfaces (a) No need to increase triangular elements, (b) cylindrical part resolution is enhanced with more triangular elements.

The process steps of the CAD model preparation for additively manufacturing are presented in Figure 2. First of all, the model is initially triangulated depending on the feature complexity described in Figure 1. Secondly, the STL model is cut into stack of flat layers by using slicer software (Figure 2(c)). These layers describe linear movements (toolpaths) of the extruder, laser or equivalent depending on the additive manufacturing method used. Printing parameters such as layer thickness, printing speed, build orientation, support structures are also determined in the slicer software. At last, since the printing of solid objects requires large amount of materials and time, slicer can automatically convert solid volumes to porous structure with specific infill density (Figure 2(d)). These parameters also directly affect the accuracy, quality, and strength of the part [3].

Figure 2.

Additive manufacturing process steps (a). Stl CAD model, (b) triangulation, (c) slicing, (d) print preview showing infill density, pattern, and toolpath.

The origin of additive manufacturing is based on the patent on stereolithography in the 1951 [4]. In terms of device, with the patent taken by Charles Hull in the 1980s, the first company to commercialize AM technology with the Stereolithography apparatus was founded [5]. Many different additive manufacturing methods were developed in the following periods. Some of these methods can be listed as follows; Selective Laser Sintering/Melting (SLS/SLM) method in which powder material is laid on a tray and sintered/melted with laser or electron beam, Stereolithography Apparatus (SLA) method in which the photopolymer resin is cured with UV or visible light, or Fused Deposition Modeling (FDM) in which thermoplastic filament is passed through a preheated nozzle and deposited in semi-melted state [6]. Other additive manufacturing methods, layer thickness ranges, materials costs, and usable materials are given in Figure 3.

Figure 3.

Layer thickness versus material price chart for AM technologies (selective laser sintering/melting (SLS/SLM), Stereolithography (SLA), fused deposition modeling (FDM), binder jetting (BJ), direct energy deposition (DED), multijet printing (MJ)). The colored circles show the materials that can be used in the related technology [data in figure taken from [7, 8, 9]].

1.1 When to consider AM

Additive manufacturing (AM) methods currently have limited implementation in the manufacturing of final products. However, this is expected to increase with progressive improvements in printing quality, obtaining good surfaces, and decreasing the costs. AM technologies inherently offer design flexibility so that functional, lightweight parts can be produced quickly. The comparison of different production methods in terms of production properties is shown in Table 1.

Table 1.

The comparison of different production methods in terms of production properties [2, 3, 10, 11].

As can be understood from Table 1, AM is the most efficient method for the rapid production of specific or assembled parts. While Computer Numerical Control (CNC) machining is used mostly to produce metal parts and Injection Molding (IM) is only used to produce plastic parts, various materials can be used depending on different AM methods as seen Figure 3. CNC machining is better than other methods for producing durable parts with high precision. However, when manufacturing with CNC, expert knowledge is needed to determine the tool paths carefully. AM enables design-driven manufacturing process, and there is less need for expert knowledge due to limited restrictions. IM is the most preferred method for mass production. Although the initial investment and mold design costs are high, it offers low unit cost per part. However, IM requires very expensive special molds to produce parts with reduced weight, which is not preferred in this specialized method for cost-effective mass production.

More capable and cost-effective printers are being introduced to the market as a result of rapid development of the 3D printing industry [12]. At the same time, there is an increasing demand for special materials that can fulfill the required properties of special parts [13]. However, in order to establish AM methods with high production cost and slow production rate compared to mass production methods, it is necessary to maximize their benefits. The points in which AM can be considered as the most efficient production method to be preferred can be listed as follows [14, 15].

• Rapid prototyping: AM speeds up product development by enabling a wide variety of prototypes to be created quickly and inexpensively compared to traditional methods.

• Customization: With the design flexibility of AM methods, functional parts tailored to specific needs can be easily produced.

• Part flexibility: Almost any geometric form that maintains stability while reducing the object’s weight can be designed and produced using AM.

• Production flexibility: AM does not require costly and/or time-consuming setup processes to manufacture different products. Thus, enabling more streamlined supply chains with on-demand production.

• Assembly: AM allows for the consolidation of an assembly into a single part as opposed to producing numerous pieces and assembling them later.

• Part reliability: Strict production tolerances and tightly controlled assembly procedures are often needed for small components, those with complicated elements, or those with small moving parts. AM techniques aid in lowering component defect rates and enhancing part reliability.

• Material/energy savings: Compared to conventional manufacturing, AM reduces energy use by using less materials and eliminating steps in the manufacturing process. Additionally, AM uses only as much material as needed to create that part, ensuring minimal waste.

• Storage and inventory costs: In AM, components are printed according to demand, thus eliminating the need to keep excess inventory stock and associated transportation costs.

Part manufacturing with AM methods has gained great momentum especially in sectors such as space, aviation, and automotive where part weight and functionality are important [16]. Although AM technologies offer design flexibility, the strength, and surface properties of the part vary according to printing parameters and orientation. Therefore, there are some factors to be considered in the design of the part to be produced by AM manufacturing. This design approach, which takes into account the limitations and unique capabilities of AM, is called DfAM.

2. Design for additive manufacturing

Design for additive manufacturing (DfAM) is the process of product design that uses AM’s unique capabilities. DFAM also considers manufacturing constraints and design criteria, which included creativity, applicability, economics, and esthetics of the design [17]. The designer sets the defining and describing the characteristics of the part, known as form, fit, and function referred to as F3 or FFF, design goals, and design validation criteria. Form is the geometric structure of the part. Properties such as shape, size, mass and how the part is seen are evaluated under the form. Fit is the physical relationship of the parts with each other. The positions, assemblies, and tolerances of the parts concerning each other are evaluated within this scope. The function defines the design and intended use of the product. A product can be customized to be used for multiple purposes. So, it is often unnecessary to manufacturing parts designed for traditional manufacturing using the AM method without any changes in their design. Because it will not take advantage of the design flexibility that AM offers, and therefore, the manufacturing cost will likely be more expensive. Thus, to use AM’s unique advantages, traditional design must be adapted or redesigned for additive manufacturing [18]. In Table 2, comparisons of Replicate with AM, Adapt for AM, and Design for AM belonging to lower wishbone used in cars are given.

Table 2.

Comparison of AM design methods for lower wishbone used in car suspensions [19].

In DfAM, mechanical properties are improved by decreasing stress concentrations as well as reducing weight and therefore cost as can be seen in Table 2. The designer’s strategy to reduce these stress concentrations is based on experience and knowledge. However, computer and artificial-intelligence-based iterative design processes such as topology optimization [20] and generative design [21] used to obtain the right-hand side model in Table 2 can be helpful methods in DfAM. These complex models, which are impossible to manufacture with traditional methods, can be produced fast, cost-efficient and with advanced mechanical properties using AM’s unique production capabilities.

2.1 Fundamental principles of DfAM

While AM seems to offer unlimited design flexibility, there are some factors to consider in the design process for AM. Before starting to design with AM, the suitability of the product for 3D printing should be evaluated. Key factors to consider in the design process for AM by the designer are as follows.

2.1.1 Printing limits

Although the printing volume in 3D printers varies according to the printing method and models, it is generally 16,000 cm³ (250x250x250 mm) or less. For larger parts, special printers can be used, or modular parts should be printed and assembled afterward. Besides, traditional methods are more cost-effective and faster in the production of large block parts with low complexity. In Figure 4, depending on the volume and complexity of the part, additive manufacturing or traditional manufacturing regions that can be preferred regions are given. The designer should determine the manufacturing method according to the volume and complexity of the existing part. Or the existing design can be updated to take advantage of AM. Thus, low-volume parts can be redesigned and produced easily with AM even if they have complex shape.

Figure 4.

Additive manufacturing (AM) or traditional manufacturing (TM) regions can be preferred depending on the volume and complexity of the part [adapted from [2, 22, 23]].

2.1.2 Printer parameters

While designing for AM, the method and process parameters to be used should be well known. Because parameters such as laser intensity, layer thicknesses, and scan pattern can determine the limitations or capabilities of the method. In Figure 5, the effects of the process parameters on the surface roughness of the AlSi10Mg part produced by the SLM method are given [24]. Here, the surface roughness is negatively affected by the increase in scanning speed, hatch gap, and power, while the increasing energy density increases the surface quality. Besides, the accuracy and quality of the part also depend on the minimum layer thickness and element size. The minimum element size may be related to the printer used, as well as other factors due to the printing method. For example, in the powder bed fusion (PBF) method powder around the part gets partially fused in the part surface, and it affects the surface quality and part tolerance [25, 26]. Hence, holes with high thickness and small diameter may not be printed in PBF [3]. To avoid this situation, the area where the hole is located can be redesigned to require minimum melting or sintering. In addition, small holes with high accuracy can be obtained by changing factors such as layer thickness, printing direction, powder diameter, and laser power.

Figure 5.

The variation of the surface roughness of the AlSi10Mg sample printed by SLM method according to (a) laser power (W), (b) energy density (J/mm³), (c) hatch spacing (mm), and (d) scan speed (mm/s) [24].

2.1.3 Affordability

Traditional or other unconventional manufacturing methods can produce coarse parts faster and cost-effectively than AM. For example, considering that the part shown in Figure 6 does not have red regions, it will be faster and cheaper to produce with 3-axis CNC machining than AM. For this reason, it will be more advantageous to produce parts that do not require changes in design and do not have strategic importance, using traditional methods. In other words, while it is affordable to produce the non-functional reducer cover by casting method, AM is more affordable to produce complex engine block with microfluidic channels that significantly increase efficiency.

Figure 6.

3-Axis CNC machining schematic. Unreachable areas and holes can be machined, respectively, by increasing the number of axes and using a small diameter tool.

Designs for traditional manufacturing (DfM) can be upgraded with the DfAM approach, making them lighter, more functional, and more affordable. Furthermore, depending on the number of the piece to be manufactured, the affordable method may differ. For example, injection mold cost is quite expensive and requires a long lead time. So, it should be preferred if it is desired to produce hundreds of thousands of repetitive parts. The unit price chart for AM and IM depending on the production volume can be seen in Figure 7.

Figure 7.

Unit cost chart for additive manufacturing (AM) and injection molding (IM) depending on production volume. Mold manufacturing costs in IM can be hundreds of thousands of dollars. But then millions of parts can be mass produced. As the number of produced parts increases, the cost per part will decrease as the initial setup cost will be shared. In AM, the volume of production does not affect the cost per unit. Because there is no mold and lead time cost in AM [adapted from [27]].

The prices of metal AM machine can up to couple of million dollars. These devices are expected to pay off within 2 years. Assuming with a very optimistic estimate that devices are operating for % 80 of the year, which means that the machine produces parts by working non-stop. Because the 20% loss is not caused by the device being in standby mode, but by maintenance, machine cleaning, preparation, preheating time, parts removal stages. In this case, the hourly operating cost of a metal AM device with an initial investment cost of $ 1 million is calculated as $ 71.3 from the formula [3].

$$\mathit{\text{Machine hourly}}\text{operating}\mathit{\text{costs}}=\frac{\text{initial investment cost}+\text{interest}\left(\text{if}\mathrm{any}\right)}{\mathit{\text{payback period}}x\%\mathit{\text{running time}}x\mathit{\text{yearly hours}}\left(8765h\right)}$$

Material costs are shown in Figure 3. Material losses such as powder used in the support structure and partially sintered powder also affect this cost. Hence, with DfAM, the solid volume should be reduced without changing the functionality of the part. Furthermore, designs that do not require a support structure or in which the support structure is a feature of the part are among the most important goals of DfAM. Hourly operating cost also increases in post-processes such as heat treatment, removal of the support structure, and ensuring good surface quality. Moreover, pre-processes such as preparing the printing chamber atmosphere, loading the feedstock, and powder handling also increase the operating cost. According to the 2018 Wohlers report [16], %30–45 of the part costs in AM are due to pre- and post-processing. This means that the hourly operating cost can reach up to $ 100.

2.1.4 Printing time

Although the length of the printing time does not mean the design is good or poor, it is very important to minimize costs. For example, positioning the long part of the piece in the vertical (printing) direction increases the number of layers and thus the number of re-coating. The recoater time varies between 4 and 15 s per layer depending on the metal printer used [3]. This means there may be a 300 min (for 50 μm layer thickness and 10 s recoater time per layer) recoating time difference depending on the positioning of the rectangular prism of 100x10x10 mm lengths. This corresponds to $ 356 in terms of hourly machine operating cost calculated at the above. Besides, the printing chamber must be filled with inert gases (argon, nitrogen, or a vacuum) and heated before the printing process in metal printers. The process of removing oxygen, known as purging, from the print volume and preheating of the build chamber can take several hours. For this reason, printing many parts at the same time and with the long edges on the horizontal axis should be preferred in terms of printing time.

Contrary to traditional methods in DfAM, large masses should be avoided as adding material is costly and time-consuming. Moreover, material accumulation leads to higher residual stresses due to high energy input. One of the frequently used methods to reduce the printing time, residual stresses, and the amount of material used is to create gaps in the part. As seen Figure 8, in order to create these gaps, contour lines that will form the outer surface of the part are printed first. Then the inside of the piece is scanned in different patterns and infill rates. As the part has a hollower structure, the amount of material used, printing time, and part weight decrease, but the part strength also decreases. Design strategies that do not affect the strength of the part considerably but maximize the amount of hollow should be developed with DfAM.

Figure 8.

Printing strategy commonly used in AM methods. First, the outer contour is determined and then the interior is scanned at various infill rates. The low-density hatch pattern provides fast printing of light parts while reducing mechanical properties.

It can be used in shell other than hatch pattern to reduce part weight, amount of material used, and printing time. Shelling means that bulk of the material is extracted from interior and only a certain wall thickness remains (Figure 9). Thus, the material used and the scanning distance can be reduced by up to 95% [3]. During the shelling process in DfAM, orifice should be added to remove the materials remaining in the interior of part as support.

Figure 9.

(a) Solid geometry with complex channel structure, (b) shell geometry with specific wall thickness.

Reducing print time is one of DfAM’s priorities. However, design decisions made to reduce printing time such as part orientation also have a great impact on mechanical properties, surface finish, geometry accuracy, and support structure of part. Because of these conflicting effects, the designer must set design priorities. In addition, these contradictions can be minimized with different design methodologies such as TRIZ, which offers possible solutions for these contradictions.

2.1.5 Print orientation

The orientation of the part to the build plate significantly affects material consumption and printing time. Support structures may vary depending on the print orientation. Printing more support structures means extra material, time, and energy consumption (Table 3). Additionally, post-process is required to remove the support structure after printing sequences. This increases labor and cost. The support structure not only increases the printing time and cost, but also dramatically reduces the quality of the part surfaces it contacts. For this reason, DfAM aims to design support free parts or positioned on the build plate requiring minimal support.

Table 3.

The effect of print orientation on build time, material consumption, and price for the part to be produced with the FDM method.

Another issue that should be considered in printing orientation is holes and cylindrical structures. If the holes and cylindrical structures are printed on the horizontal axis, support structure and stair-step effect will be formed and the geometry will shift from circular to slightly elliptical (Figure 10). Besides, the height of the part in the z-axis increases the number of layers. The printing time will be longer as the number of recoating will increase depending on the layer thickness. Moreover, large horizontal surfaces in a single plane result in high residual stresses.

Figure 10.

Hole and cylindrical feature quality change depending on printing orientation. If the circular elements are perpendicular to the printing axis, elliptical structure and stair step are seen. Also, poor tolerance is observed if the support structures contact with the circular elements.

Print orientation affects surface quality regardless of support structure. In the SLM method, as the build angle increases, the surface roughness decreases (Figure 11) [25]. However, when the length in the Z direction increases, since more re-coating is required, the printing time and cost will increase. Surface quality can be improved further by changing other factors such as laser power, scanning speed, hatch spacing, laser focus diameter, and shield gases in the environment. In addition to printing parameters, surface quality can also be improved by post-processes such as surface remelting [28], shoot peening electropolishing, chemical polishing, grinding, and sand blasting [29].

Figure 11.

(a) Building orientations for stainless steel 316 L printed using the SLM method and (b) average roughness (R_a) of inclined surfaces [25].

2.1.6 Part strength

The support structure is not the only factor to be considered when orienting the part to the build plate. The strength, material properties, and surface quality of the part also vary according to the printing orientation. Anisotropy, one of the biggest drawbacks of layered production, occurs in the vertical printing direction and significantly affects the strength. If the part will be exposed to axial loads, the part must be horizontally oriented to stand these loads (Figure 12). Although anisotropy is seen in all AM methods, it is most effective in FDM method [30].

Figure 12.

Anisotropic property for FDM method due to printing orientation. Due to the gaps between the layers formed in the vertical (printing) direction of the part, they are more loosely bonded to each other. Thus, a smaller force causes the breakage. Since there is continuity in the horizontal direction, greater forces can be handled.

While it is largely eliminated in the PBF method with post-processes such as hot isostatic pressing (HIP). Apart from anisotropy, porosity and micro cracks that reduce the strength of the metal printed part can be reduced with the HIP. Also, residual stresses affect the strength of the metal AM parts. Residual stresses arise with the effect of shear forces between layers due to sudden heating and rapid cooling, which is inherent in laser melting. In regions where excess material is fused, there is more residual stress due to the greater thermal input. High residual stress areas are prone to distortion. Therefore, heat treatment such as residual stress relief should be implemented after the printing process. Or printing strategies that reduce residual stress such as scan patterns should be specified in DfAM stage. Ribs can be used to increase strength while preventing excess material in a certain area. Thus, while preventing residual stresses and distortions, the amount of material used and printing time are reduced at the same time. Figure 13 shows the rib structure used to increase the strength of the shell geometry.

Figure 13.

Rib structure used to increase the strength of the shell geometry. Orifices should be used for the evacuation of excess powder. Adding fillets to the junction areas of different geometries provides stress distribution and extra reinforcement.

3. Design to minimize post-processing

Post-processes may be required for AM to meet features such as microstructure, surface roughness, residual stresses, porosity, and dimensional tolerance. Two types of post processes, thermal and mechanical, are encountered. Heat treatments are performed to improve the microstructure, residual stresses, and porosity of AM parts, while mechanical processes are aimed to enhance the surface properties and dimensional tolerances of the parts. Post-processing steps for products produced with metal AM methods are given in Figure 14.

Figure 14.

Standard post-processing steps for metal AM products.

After the printing process, wire electro-discharge machining (EDM) is commonly used to remove the support structures from the product. There is an electrical discharge between the wire and the part, and using heat from electrical sparks, the support material is removed from the work piece by utilizing wire EDM. Then sandblasting, which is based on spraying abrasive particles onto a surface under high pressure, is used to get rid of unfused or partially fused powder remaining on the part after the support structure is removed. Thermal, mechanical, and residual stresses that may cause distortion can occur in the metal AM printed products due to the rapid heating and cooling cycles during the laser melting process. Thermal methods such as stress relieving (preferred in the treatment of thermal and residual stresses) or annealing (preferred in the treatment of mechanical stresses) can be used to remove such stresses. In these processes, the product is kept in a heated (500–800°C) argon or nitrogen atmosphere for several hours and then slowly cooled. A furnace can be used for this process, as well as the printer can perform the same process in its printing volume. Another most used post-process in metal AM methods to enhance the mechanical properties, increase density, and reduce porosity is Hot Isostatic Pressing (HIP). High-pressure argon gas (up to 2000 bar) and high temperature (up to 2500°C) are applied to the AM product in the chamber in the HIP process [31]. Mechanical machining can be used to increase surface quality and add other features such as holes. Surface finishing can also be improved by non-mechanical processing methods such as Chemical Etching and Electro Polishing. One of the primary goals of DfAM is to minimize these post-processes that cause time, labor, and cost. Support structures must be designed for easy removal. There should be no support structure in interior parts that are impossible to remove. Designs that do not need support structures can be created by using DfAM. The circular hole that requires support in its diameter region needs additional post-process such as machining. Instead of circular geometry, water droplet shape makes is possible to remove support structure. If it is unavoidable, the support structure can be designed as a permanent feature that increases the strength of the part. Designing support structures as thin walls also significantly reduces printing time and the amount of material used (Table 4).

Table 4.

Comparison of solid and shell structure based on build time and cost.

As can be seen in Table 4, compared to the solid model, using a shell structure that may require difficult to remove support geometries significantly reduces parameters such as build time mass and cost. Therefore, support geometries can be designed as a functional element of the part, as shown in Figure 13. The same functional part produced following the DfAM methodology can be produced almost five times faster, 11 times lighter, and twice as cheaper than the solid part. Parts with a high strength-to-weight ratio can be obtained using numerical approaches such as Finite Element Analysis. Additive manufacturing of functional structures with reduced weight, such as lattice structures based on the DfAM methodology, also provides improvements in component performance and reductions in production costs and weights [19, 32].

Although it is very effective to design the support structure that causes poor surface quality as a thin wall as a permanent element of the part, thin and large surfaces are prone to warping in the cooling process. Cooling gaps can be created on these surfaces to minimize warping and also to reduce the material used. Or it can be reinforced with other features such as ribs. Post-processes such as machining and chemical etching are required to improve surface quality that contact to support region. The support structure needs to be positioned in the part region where high precision is not required. It is also important to minimize the subsequent heat treatments applied to enhance the properties of the parts such as microstructure, residual stresses, and porosity. Residual stresses and distortions can be minimized by predicting thermal dispersions during the printing process with some specialized simulation software for production with AM. Part-based optimization of printing parameters or laser remelting during the printing process can be performed in order to improve microstructure quality and reduce the porosity without post-processing [33].

4. Conclusion

Overcoming the basic limitations of traditional production methods with AM methods has revealed the necessity of developing unique design innovations for designers. Thus, designers are looking for new direction to customize products, improve product performance, reduce production and assembly costs, and conceptualize products in general. This chapter has covered an understanding of why and when to use AM, an understanding of DfAM methodology, and design considerations. It should not be forgotten that redesigning the product by considering the limitations and capabilities of AM rather than adapting the existing product for AM in the DfAM process yields more effective results in terms of production performance. The factors need to be considered in the design and a number of design examples to maximize AM benefits are discussed for additive manufacturing technologies. There is no doubt about new concepts will be developed for this approach, which is still at the beginning, and traditional CAD and FEA systems will be adapted to evolve the additive manufacturing capabilities.

Acknowledgments

This work was supported by Gazi University Scientific Research Project Coordination Unit [grant numbers: FGA-2022-75320].

Conflict of interest

The authors declare no conflict of interest.