Additive manufacturing: Challenges,trends,and applications

Abstract

Additive manufacturing is a recent trend in production processes owing to its many benefits. It can be defined as the process of producing parts through the deposition of material in a layer-by-layer fashion. It has been a topic of intense study and review by many researchers. In this work, a comprehensive review pertaining to additive manufacturing has been accomplished. The evolution of additive manufacturing as a prominent technology and its various phases are discussed. The importance of part orientation, build time estimation, and cost computation has also been reviewed. The remarkable aspect of this work is the identification of problems associated with different additive manufacturing methods. Because of the imperfections in additive manufacturing, its hybridization with other methods, such as subtractive manufacturing, has been emphasized. This review will help readers understand the different aspects of additive manufacturing and explore new avenues for future research.

Manufacturing processes

The evolution of industries depends on innovative and cutting-edge research activities associated with manufacturing processes, materials, and product design. In addition to the customary demands of low price and best quality, the market competition in current production industries is linked to requirements for products that are intricate, possess shorter life cycles, exhibit shorter delivery times, involve customization, and require less skilled workers. In fact, the current breed of products is very complicated and challenging to design. Accordingly, there is a strong incentive toward the design, development, and implementation of new and ingenious manufacturing processes. Manufacturing processes can be categorized into five categories, namely, subtractive, additive, joining, dividing, and transformative, as shown in Figure 1. ¹

• Subtractive technology can be defined as a method in which layers of materials are removed to produce a desired geometry. Over the last 20 years, subtractive technologies have undergone a tremendous change. The introduction of three-dimensional (3D) complex surface modeling software has replaced traditional code generation, such as G and M codes. Unlike the computer numerical control (CNC) machines of the 1940s, the contemporary lineage of CNC machines is highly automated based upon the integration of computer-aided design (CAD)/computer-aided manufacturing (CAM) systems.

• Additive technology is based on the addition of material layers to create a desired workpiece shape.

• Joining technology such as welding consists of physically joining two or more workpieces together to produce the required shape.

• Dividing technology such as sawing is the opposite of the joining process.

• Transformative technology, for example, forming, heat treatment, and cryogenic cooling, uses a single workpiece to fabricate another workpiece, keeping the mass unchanged.

Figure 1.

Classification of manufacturing processes. ¹

Additive manufacturing

Additive manufacturing (AM) can be described as a technique of blending materials by either fusion, binding, or solidifying materials such as liquid resin and powders. It builds part in a layer-by-layer fashion using 3D CAD modeling. The terminologies such as 3D printing (3DP), rapid prototyping (RP), direct digital manufacturing (DDM), rapid manufacturing (RM), and solid freeform fabrication (SFF) can be used to describe AM processes. AM processes fabricate components using 3D computer data or Standard Tessellation Language (STL) files, which contain information regarding the geometry of the object. AM is very useful when low production volumes, high design complexity, and frequent design changes are required. It offers the possibility to produce complex parts by overcoming the design constraints of traditional manufacturing methods. Although, AM has many benefits, its applications are still limited because of its low accuracy and long build times compared to CNC machines. It does not have the same constraints as CNC machining because it segregates the part in cross sections with a resolution equal to that of the process. ² Nevertheless, the accuracy and build time can be improved by employing suitable part orientation. Optimized part orientation can enhance the accuracy and diminish the building time and support volume, which in turn minimizes the part production cost. Moreover, AM in contrast to conventional production processes consists of additional controllable process parameters and higher active interaction between the material properties and process parameters. There are different kinds of AM processes depending on the material preparation, layer generation technique, phase change phenomenon, material type, and application requirements. An AM process involves mainly three phases, namely, the design phase, the processing phase, and the testing phase as shown in Figure 2. ³

Figure 2.

Phases in AM process. ³

Trends in AM

AM was first demonstrated in the 1980s by Kodama, who published an article titled “Three-Dimensional Data Display by Automatic Preparation of a Three-Dimensional Model.” ² The emergence of AM took place in 1987 with STL by Chuck Hull (courtesy: 3D Systems), a technique which solidifies thin layers of ultraviolet (UV) light-sensitive liquid polymer using a laser. It can successfully manufacture complex and customized products with fewer skilled workers, shorter delivery times, and shorter product life cycles. The chart in Figure 3 describes an increase in the number of patents associated with AM from 1982 until 2012. ⁴ The technology based on selective laser sintering (SLS) was first patented in 1988 by Carl Deckard at the University of Texas, followed by fused deposition modeling (FDM) in 1989 by Scott Crump, co-founder of Strasays Inc. It can be asserted that the academic project in 2005 by Dr Adrian Bowyler at the University of Bath, England was one of the major breakthroughs in the area of 3DP. They developed a self-replicating fabricating system, known as Rep Rap, capable of manufacturing its own components. Since then, a number of ventures, such as Makerbot (2009), Form Labs (2011), and HP Fusion Jet 3D printer (2016), have come up.

Figure 3.

Yearly comparison in the number of granted patents. ⁴

It can be observed that the emergence of AM technology commenced in the 1980s for generating 3D objects from CAD files. Each of the subsequent year, as shown in Figure 3, either new AM principle was developed or there was an improvement of the existing one. It can be concluded that AM is certain to transform and revolutionize manufacturing in the future Industry 4.0. The demand of AM has expanded from automobile and aircraft to the manufacturing of houses as well as medical, food, and space industries. The National Aeronautics and Space Administration (NASA) has been experimenting to achieve effective and efficient 3DP in Zero-G technology. The researchers have been working to produce 3D printed particles in micron level or lesser, to be useful in the production of smaller electronics and batteries. Even, there is a 3D printer on the market (courtesy: Photonic Professional GT) that can manufacture parts no wider than a human hair. There have been significant improvements since its inception, in terms of innovative technologies (such as hybridization), materials (e.g. gamma TiAl), products, and businesses in AM.

Classification

AM processes or machines can be classified based on machine dimension, nozzle dimension, speed of the nozzle, and workspace dimensions. AM can be categorized in numerous ways based on the functional framework of the material. Although the methods of classification can also include the patterning energy, the technique of generating primitive geometry, the nature of used materials, and the support procedure.^5,6 However, in broad sense, AM processes can be summarized and classified according to the type of material used. Figure 4 summarizes the existing methods for AM based on the type of material.

Figure 4.

Classification of AM processes depending on the state of raw material.^7,8

Solid-based AM

The AM technologies in which input raw material is in solid state have been discussed in this sub-section. Among so many existing solid-based AM technologies, FDM, freeze-form extrusion fabrication (FEF), laminated object manufacturing (LOM), and direct ink writing (DIW) are the most popular.

FDM

The working of FDM as shown in Figure 5 is based on the principle of layered manufacturing technology. ⁹ In this technology, the plastic raw materials (filaments) are extruded through the nozzle, which is heated to melt the material. The nozzle head moves according to the tool path, which is generated for each layer.

Figure 5.

Schematic of FDM. ¹⁰

FEF

This process was developed at the University of Missouri–Rolla, Rolla, Missouri, ¹¹ and it worked by extrusion of an aqueous ceramic paste. In this technology, the paste was extruded layer-by-layer into a build chamber, which was kept below room temperature in order to achieve the freezing of the paste. It was an environment-friendly approach because only water was utilized as the binding media. Solid loading as high as 60 vol.% had been achieved with this method using aluminum oxide (Al203). Figure 6 shows the schematic FEF process.

Figure 6.

FEF process. ¹¹

LOM

In LOM process, the metal, paper or a form of polymer in sheet form can be employed as a raw material. The sheet materials are joined together by ultrasonic consolidation and then machined into desired shape. In case of paper as raw materials, the paper sheets are glued together by adhesive and cut in shape by sharp blades. Moreover, the polymer sheets are joined together with the help of heat and pressure.^7,12 The advantages of LOM include high surface finish, low material requirement, lesser machine and process costs, and high strength (Figure 7). ¹³

Figure 7.

Schematic of LOM process.^12,13

DIW

The DIW (also known as “Robocasting,”“Robot-Assisted Deposition,”“Direct Write Fabrication”) was first appeared in the mid-1990s at Sandia National Laboratories.^14,15 In this technology, a pseudo-plastic paste mixed with ceramic particles was extruded to fabricate a part. The movement of the nozzle was controlled through the instructions generated by the software using the information from the CAD model. It is very similar to FDM, but its principle depends on the pseudo-plasticity, in contrary to solidification in order to maintain parts’ shape. The nozzles as fine as 30 μm are required with ceramic pastes ¹⁶ and 1 μm with polymer pastes. The larger nozzles (>1 mm) are also compatible with the process in order to build parts faster than other AM processes (Figure 8).

Figure 8.

Schematic of DIW process.^14,15

Liquid-based AM

In liquid-based AM technologies, the input raw material is in liquid state. Among so many existing liquid-based AM technologies, stereolithography (SLA), multi-jet modeling (MJM), rapid freeze prototyping (RFP), and digital light processing (DLP) are the most prominent.

SLA

In SLA, a vat of liquid photopolymer resin is utilized to build parts in a layer-by-layer fashion. This part is subsequently hardened or cured using the UV light. The building platform moves down the object as the layers are being made after each new layer is cured. The principle of SLA can be seen in Figure 9. The fabrication of part in SLA technologies requires supporting structures to attach the part on the building platform and hold the object as it floats in the liquid resin. ¹² No special material is required for support structures. The support structures are fabricated using the same material as that of the actual part.

Figure 9.

Vat Photo polymerization process scheme. ¹²

MJM

This process employs multiple nozzle jets to supply a UV curable polymer or wax as required. When the printed polymer gets out of the head, the UV lamp flashes to cure the material. The MJM prints parts on a moving platform that moves down after the current layer is cured in order to continue the process. It is a cost-effective technique because it can fabricate parts in a lesser amount of time. The MJM process is safe and quiet and therefore can be used for printing polymeric parts in the laboratory environment. However, in this process, the strength of parts and part quality are relatively poor as well as the bottom surface is most often printed parts quality; the part strength is relatively poor and the bottom surface is often indented, rough, or uneven (Figure 10). ¹⁷

Figure 10.

MJM process. ¹⁸

RFP

This process fabricates part by selectively depositing and freezing the water or brine in a layer-by-layer manner. Two types of techniques such as continuous deposition and drop-on-demand deposition (Figure 11) can be utilized to deposit water. ¹⁹ The building environment is kept at a low temperature, well below the water’s freezing point. In this process, the pure water or brine can be used as the support material which is ejected from the nozzle and deposited on the previously solidified ice surface. The nozzles and feeding pipes are maintained at a particular temperature to pre-cool water near to its freezing point as well as kept them in the liquid state to allow material flow freely and smoothly.

Figure 11.

Principle of RFP.^19,20

DLP

The DLP was initially came into existence in 1987 by “Texas Instruments.” This process is quite similar to SLA in the sense that it also utilizes photopolymers. The primary difference lies in the light source. In DLP, a conventional light source (an arc lamp) and a projector (DLP projector) are utilized as shown in Figure 12. One of the benefits of DLP over SLA is that a shallow tub of resin is needed to carry out the process, causing less wastage and lower running costs. ²¹

Figure 12.

Principle of DLP. ²²

Powder based AM

In powder-based AM technologies, the input raw material is in powder state. Among so many existing powder-based AM technologies, three-dimensional printing (3DP), electron beam melting (EBM), SLS, selective laser melting (SLM), laser-engineered net shaping (LENS), and pro metal and laser metal deposition (LMD) are the most extrusive.

3DP

In 3DP, a liquid binder is utilized to bond the powder materials. The powder layer is spread on the building platform and the binder liquid is extruded through nozzles depending on the cross section of the 3D model. This process does not require additional support structure, instead the support is provided by the unbound powder particles (Figure 13). ¹²

Figure 13.

Schematic of 3DP. ¹²

EBM

EBM technology is a relatively new AM process with majority of applications in medical and aerospace industries.^23–25 It was a technology patented by ARCAM for 3DP primarily. As shown in Figure 14, the electrons are emitted from a filament, which is heated at above 2500°C. These high-energy electrons are accelerated through the anode. The magnetic field of lenses focuses the beam and electromagnetically scans the layers. The powder particles are fed by gravity from hoppers and are raked into layers across the building table. The entire process is carried out under vacuum, which eliminates impurities and yields high strength.^23,26,27

Figure 14.

Schematic diagram of electron beam melting process.^23,26,27

SLS

In SLS, as shown in Figure 15, the laser beam is used to sinter or melt the spread layer powder particles. The energy beam selectively sinters or melts the powder particles through scanning the layers on the building platform. Once the layer is scanned, the building platform is lowered and the new powder layer is spread on top, and the process is repeated until the part is finished. ²⁸

Figure 15.

Selective laser sintering system. ¹²

SLM

The part is fabricated layer-by-layer through selectively melting the metal powder using a laser in SLM. The mechanical properties of parts manufactured by SLM are nearer to those produced through conventional manufacturing techniques.^29,30 Indeed, the previous works have asserted that parts with densities up to 99.9% can be produced using SLM. ³⁰ Although this technology possesses a great potential, however, the implementation in industries is limited due to several obstacles.^31,32 One of the primary hindrances is the presence of high residual stresses and large deformations in the part. The residual stresses significantly contribute to the formation of thermal cracks.^33,34 Moreover, it is difficult to maintain consistent quality during the build-up process, thus causing varying porosity within the part (Figure 16). ³⁵

Figure 16.

Schematic overview of an SLM machine setup. ³⁶

LENS

The LENS technique was developed by Sandia National Laboratories and commercialized by Optomec Inc (Albuquerque, NM, USA) since 1997. ¹⁷ It can be identified as laser cladding process, in which a laser is employed as a heating source to melt the powder particles to be cladded onto a substrate. ³⁷ The materials that can be processed by LENS may include steel, aluminum, titanium alloys, nickel-based alloys, and metal matrix composites. ³⁸ The extremely rapid cooling in this process generates a fine-grained microstructure, producing a high tensile strength and high ductility (Figure 17). ³⁹

Figure 17.

Laser-engineered net shaping principal process scheme. ⁴⁰

Pro metal

Pro metal can be defined as a 3DP process to build injection tools and dies. It is a powder-based process in which mainly stainless steel is used as the raw material. The printing process occurs when a liquid binder is sprayed on to the steel powder. The powder is located in a powder bed which is controlled with the help of pistons. After fabrication of each layer, the bed lowers down and the feed piston provides the material for the subsequent layers. In case of a mold, no post-processing is needed; however, the building of a functional part requires sintering, infiltration, and finishing processes.^41–43

LMD

In this technology, the 3DP head is generally connected to a robotic arm, which comprises nozzles to deposit metal powder on the building platform. The laser beam provides the required energy and melts the powder particles to fabricate the desired geometry (Figure 18). ¹²

Figure 18.

Laser metal deposition. ¹²

It has been observed in the literature that many researchers and academicians have primarily adopted AM classification based on a functional framework. This includes the categories of material, AM technologies, AM material preparation, layer creation technique, phase changes in AM, patterning energy, phenomena of creating primitive geometry, support mechanisms, and AM applications as shown in Table 1. Table 1 shows that the main classifications of AM depend on the material used and the applied technology. The preparation of these materials before actual fabrication varies. The layer creation phenomenon can also vary depending on the technological methods used. After the creation of a layer, the phases can also be classified as full melting, partial melting, or solidification phases. The application of these parts, such as for prototyping or final product, is different, depending on the technology used.

Table 1.

Different AM methods and their characteristics. ⁷

State of starting material	Process	Material preparation	Layer creation technique	Phase change	Typical materials	Applications
Liquid	SLA	Liquid resin in a vat	Laser scanning/light projection	Photopolymerization	UV curable resin and ceramic suspension	Prototypes, casting patterns, and soft tooling
	MJM	Liquid polymer in jet	Ink-jet printing	Cooling and photopolymerization	UV curable acrylic plastic and wax	Prototypes and casting patterns
	RFP	Liquid droplet in nozzle	On-demand droplet deposition	Solidification byfreezing	Water	Prototypes and casting patterns
Filament/paste	FDM	Filament melted in nozzle	Continuous extrusion and deposition	Solidification by cooling	Thermoplastics and waxes	Prototypes and casting patterns
	Robocasting	Paste in nozzle	Continuous extrusion	–	Ceramic paste	Functional parts
	FEF	Paste in nozzle	Continuous extrusion	Solidification by freezing	Ceramic paste	Functional parts
Powder	SLS	Powder in bed	Laser scanning	Partial melting	Thermoplastics, waxes, metal powder, and ceramic powder	Prototypes, casting patterns, and metal and ceramic preforms (to be sintered and infiltrated)
	SLM	Powder in bed	Laser scanning	Full melting	Metal	Tooling and functional parts
	EBM	Powder in bed	Electron beam scanning	Full melting	Metal	Tooling and functional parts
	LMD	Powder injection through nozzle	On-demand powder injection and melted by laser	Full melting	Metal	Tooling, metal part repair, and functional parts
	3DP	Powder in bed	Drop-on-demand binder printing	–	Polymer, metal, ceramic, and other powders	Prototypes, casting shells, and tooling
Solid sheet	LOM	Laser cutting	Feeding and binding of sheets with adhesives	–	Paper, plastic, and metal	Prototypes and casting models

SLA: stereolithography; MJM: multi-jet modeling; RFP: rapid freeze prototyping; FDM: fused deposition modeling; FEF: freeze-form extrusion fabrication; SLS: selective laser sintering; SLM: selective laser melting; EBM: electron beam melting; LMD: laser metal deposition; 3DP: three-dimensional printing; LOM: laminated object manufacturing; AM: additive manufacturing; UV: ultraviolet.

The design tools for AM can be organized in four categories. ⁴⁴ Three categories, namely, point cloud processing, solid modeling (CAD) systems, and process-oriented design software, allow the modeling, visualization, meshing, and conversion of 3D model into STL files. The STL files have issues associated with handling interior volume to generate internal structures. Therefore, a fourth type of design tool was introduced which was further divided into two sub-categories: topology-optimization-based software and cellular structure design software. There have been several unresolved problems associated with the characterization of part properties (i.e. stiffness) and specification of material distribution.^45,46 A four-step procedure that considered the manufacturing capabilities and constraints of AM was developed by Vayre et al. ⁴⁷ to investigate the design process. The different steps can be identified as design methodology, design analysis, tuning up, and validation of final geometry. For an effective selection of production strategies in AM techniques, Achillas et al. ⁴⁸ proposed a decision-making framework. This framework integrated a multi-criteria decision aid (MCDA) and data envelopment analysis (DEA) to assist manufacturers in the selection of optimal production techniques among several alternatives.

Because CNC machining possesses several limitations associated with shape complexity, many research activities have concentrated on shape complexity in AM using non-metallic prototypes, rather than functionality. The domain of applications in metallic prototypes has been growing at a rapid rate. ⁴⁹ A lot of effort has been focused on AM of metallic objects based on functional classification. ⁵⁰ Figure 19 summarizes some of the current techniques for AM of metallic components based on functionality.

Figure 19.

Prevailing techniques for additive manufacturing of metallic products. ⁵⁰

Owing to its numerous benefits, AM of metal components has been in great demand. With metal AM processes, the industries and manufacturing sectors have completely revolutionized. The shapes and designs that were not earlier possible are now feasible with the emergence of AM. The structures are stronger due to the minimization of welded (or joining) components. However, the AM technology still needs better understanding, frequent enhancement, as well as upgradation in order to meet demanding customers and competitive market. Indeed, the researchers have been exploring various aspects of metal AM. For example, R Martukanitz et al. ⁵¹ presented an integrated computational system for simulation of AM metallic components, including process and material models directed at energy deposition and powder bed fusion. He found that the complexity of the part geometry and build path were the main challenges in establishing and preserving an efficient computational mesh, and it was exasperated for simulation of large structures. According to Thompson et al., ⁵² the variation in thermal energy produces anisotropic changes in the part, which in turn influence mechanical properties, particularly tensile and fatigue behavior. Therefore, Shamsaei et al. ⁵³ studied the application of AM in Nitinol (nickel–titanium alloy) parts for medical applications addressing the challenges associated with fatigue life modeling and prediction. These complexities were mainly caused due to the presence of several process parameters, including laser power, transverse speed, powder feed rate, layer thickness, scanning strategy, and build orientation. ⁵² He et al. ⁵⁴ studied preheating of Ti-6Al-4 V powder in full form (EBM) and concluded that partially or completely melted small particles played an important role as binders to bond the majority of large particles together. The bond held the particles together, withstood impact from electrons, and prevented the spheroidization effect on the part surface. The mixing of different powder materials was studied by Ono et al. ⁵⁵ for manufacturing parts with new properties intermediate to their constituent materials. A greater heat source was used to provide high energy in laser cladding as well as accelerate the deposition speed and reduce production time. This additional energy, however, affected the accuracy of the layer deposition (Qian et al. ⁵⁶ ). A major area for future development in metal AM is surface roughness and reduction of production costs.

Part orientation in AM

Suitable orientation of the part is crucial in AM for improving the Geometrical and Dimensional Tolerance (G&DT), reducing build times and minimizing support volume and part production costs. The part orientation affects surface accuracy, builds height, and supports volume. This criterion, along with the production cost, has been the main research area for many years.^57–59 The work of Cheng et al. ⁶⁰ offered a classification based on different types of surfaces using a weighted value to acquire high surface accuracy. Two different objective functions were presented for selecting the optimal part orientations. The objective function of accuracy was formulated by the multiplication of each surface area type with the given weight and the maximum value of the objective function that provided the optimal orientation. If different orientations resulted in the same value for the accuracy, then the secondary objective function, which was minimum build time, was considered. Wenbin et al. ⁶¹ introduced a staircase effect simulation of a layered part build utilizing the finite element method (FEM). The staircase effect has been recognized as one of the most important factors, affecting the part accuracy, and it is mainly caused by an inclined or freeform surface. A prediction model was presented to study the shrinkage phenomena in photopolymerization during AM. In addition, it was found that when layer thickness increased from 0.2 to 2.0 mm, the staircase amount increased by 300%. The study by Frank and Fadel ⁶² proposed an expert system that considered many geometric features such as planes, cylinders, round corner surfaces, thin surfaces, and overhang surfaces. The optimal orientation was selected based on the surface quality and the support structure requirement. A multi-criteria decision-analysis method employing user-specified weights was presented. Surface roughness, build time, and production costs were the criteria used for identifying the optimal part orientation. The four different AM processes used to determine the optimal part orientation were SLA, FDM, SLS, and LOM. ⁶³ Danjou and Köhler ⁶⁴ proposed a genetic algorithm (GA) to optimize the build orientation for simultaneously accomplishing the four different objectives. Different objectives were integrated using the sum-weighted method to provide a single-objective function for optimization. Parts were rotated systematically with a 1° angle interval and subsequently each resultant orientation was evaluated. Padhye and Deb ⁶⁵ performed a multi-objective optimization for minimizing surface roughness (Ra) and build time (T) in SLS process. The minimization of Ra and T was accomplished through two renowned optimizers, multi-objective GA (non-dominated sorting genetic algorithm-II (NSGA-II)) and multi-objective particle swarm optimizers (MOPSO). Decision-making schemes were also employed to optimize the build orientation from the non-dominated solutions and two-dimensional Pareto front, according to user preference. Zhang and Li ⁶⁶ presented a new procedure to optimize build direction leading to minimized volumetric error in RP processes. To scrutinize the overall directional space (generate alternative orientations), a unit sphere was systematically discretized to represent the potential directions in a 3D space. Afterwards, each facet including the STL geometric model was individually mapped onto the discretized unit sphere as a great circle to represent the optimal directions for that facet. The orientation optimization problem for multi-part production in AM was defined and analyzed. It was a variant of the multi-objective traveling salesman problem (MOTSP), which is a typical non-deterministic polynomial (NP)-hard problem. A two-step solution was developed in this article. The solution utilized an AM feature–based alternative orientation generation technique to minimize the solution space and adopted a GA-based multi-attributes decision-making (MADM) to overcome issues associated with other optimization methods while computing numerous objective optimization problems. ⁶⁷ MATLAB was employed to formulate an unconstrained optimization algorithm for predicting the optimal orientation and minimizing support volume in SLM or direct metal laser sintering (DMLS) additive layer manufacturing. The methodology was tested on three different geometries to improve the standard orientations recommended by the pre-processing software. The software did not always provide the best results as compared to the build positioning suggested by experienced operators. Although this code enabled dependable and efficient builds to be achieved by inexperienced users, incorporation of the manufacturer’s expertise would further enhance its efficacy for future developments. ⁶⁸

Build time estimation in AM

In general, the build time for a single component or an assembly relies on the printing speed, part size, layer thickness, and build orientation. The fabrication time primarily depends on the object’s height in the z-direction. An object with greater height will have a longer processing time (printing time), independent of the printing process. Therefore, designing for low overall build height is crucial for build time reduction. ⁶⁹ A simple and easy-to-use parametric approach was presented by Cheng et al. ⁶⁰ to estimate the build time for SLA. They assumed that the total build time had a directly proportional relation to the number of layers. Therefore, the time needed to complete each layer was assumed as constant. Xu et al. ⁷⁰ proposed a parametric estimator that expressed the build time in a more advanced way. The method established a mathematical model that could be used to compute the build time, TBuild. The build time was computed as the summation of the time required for processing each of the n layers and the deposition of volume for the producing parts. Build time estimation approaches have typically been designed for specific AM technology variants. Chen and Sullivan ⁷¹ presented an approach with detailed analysis for the time estimation of SLA. This approach was used to predict the build time by considering the different laser scanning factors, such as scan speed, laser power, spot size, and scanning strategy. Furthermore, this estimator was used to predict the time spent in vertical platform dip, re-raise, and wiper activity and to predict unspecified hardware delays. It is worth noting that the build time estimator proposed by Chen and Sullivan did not employ data on the real application. Giannatsis et al. ⁷² presented a similar estimator for SLA, based on variables such as contour length, hatch area, scan speed, and resin properties. This estimator was used with data on part geometry and they proposed two different implementations of the estimator. The first variant investigated the data on part geometry in the STL file format. The STL file represents the standard file format for the exchange of part data among AM users and for communication with other AM technologies. The second implementation ⁷³ used presliced data in the Common Layer Interface (CLI) format to represent individual layers and yielded more accurate results. They also developed a model for computing the build time as the sum of the time utilized to deposit each layer, and it was corresponding to the successive addition of layers of geometry in additive processes. Byun and Lee ⁶³ discussed the build time estimation in more detail including the AM technology variants SLA and FDM. They also proposed a similar framework of relationships governing build time as that introduced by Giannatsis et al. ⁷² Regarding the variation of these relationships for various technology variables, Byun and Lee discussed the component of time consumption for both SLA and FDM. For instance, the time needed to scan the interior area was controlled by different factors for laser-based systems in contrast to systems based on heated extrusion nozzle, such as FDM. Furthermore, Byun and Lee investigated effects of various operating principles on the time needed to generate support structures. The identification of main factors such as volume, size of the bounding box for computing the LS build time, including part complexity, was carried out by Ruffo et al. ⁷³ According to Ruffo et al., the components of total build time were total scanning time, total recoating time, and time spent in build preparation and post-processing. Another parametric approach to estimate SLA build time was proposed by Campbell et al. ⁷⁴ They investigated the feasibility of employing geometrical primitives, such as spheres, cylinders, and cones, in build time estimation in case the actual part geometry was not available in the STL format. The parametric model by Campbell et al. computed the scan time per layer using the area cured per layer divided by the hatch spacing and then multiplied by the scan velocity. It is worth noting that this approach assumed a constant recoating time and did not differentiate between contour scanning and hatch scanning. Campbell et al. stated that hatch-scanning time in SLA was directly proportional to the cross-sectional area and inversely proportional to the laser output. Furthermore, they conducted a comparison between the accuracy of their estimator and the commercial software package Magics RP (Version 6.3). They observed a mean percentage error of 3.8%. Munguía and Javier ⁷⁵ applied concepts of artificial neural network (ANN) and estimated build time for AM technology variants LS and SLM. The author assumed ANN as a non-parametric methodology utilized to fit curves through data in the absence of a predetermined function having free parameters. This model was able to determine hidden functional relationships. The author also mentioned that the unavailability of build time estimators in metallic AM technology variants might be due to competitive behavior exhibited by AM control software authors or by the characteristics of post-processing metal parts. Other interesting results in the application of build time estimators were provided by Ruffo et al., ⁷³ reporting mean errors of 14.98% and 22.68%, respectively. Choi and Samavedam ⁷⁶ introduced a specialized time estimator that was designed for LS. This estimator carried out detailed analysis, depending on the categorization suggested by Di Angelo and Di Stefano. ⁷⁷ Choi and Samavedam’s approach was based on variables, including the laser scan speed and laser scan distance. It was also proved that the build time can be reduced through part orientation that can minimize Z-height. Ruffo et al. ⁷³ concluded that minimization in Z-height might not provide the shortest build duration in case the available build volume is completely utilized. Di Angelo and Di Stefano ⁷⁷ proposed a different and versatile ANN-based build time estimator that can be used for various AM technologies. The implementation of this estimator resulted in a mean estimation error of 12%. In general, the studies involving build time estimation can be classified into two groups: parametric approaches^60,70,73,74 and detailed-analysis techniques.^63,71,72,76

Cost estimation in AM

There are two categories motivating the examination of AM costs. The first category involves comparison of AM with other conventional processes, namely, machining and injection molding (IM). The objective is identification of scenarios and circumstances in which AM is cost effective. The other category includes determining the resources that are used at different steps in AM. This category determines information about the resources being used and their overall utilization. Recently, cost models have been developed for various AM technologies. This section has discussed leading cost models in order to understand their applications. The study of Son ⁷⁸ developed a useful classification system for various costs associated with the use of AM machinery. In the article, the costs were divided into two types: ill-structured costs reflecting items, such as error prevention, build failure, machine setup, idleness, waiting time, and inventory, and well-structured costs involving all costs related to materials, machines, labor, and overhead. The work cited in this section focuses on the analysis of well-structured costs emerging from AM and conventional processes. Hopkins and Dickens ⁷⁹ developed a model for comparing AM processes such as SLS, FDM, and SLA with traditional processing methods such as IM. The model was used for various part designs and batch sizes to determine the individual unit costs and to accurately compute break-even points. The three following elements were provided for the breakdown of costs: material costs, personnel costs, and machine costs. In this research, the model computed the material costs of SLS, excluding the recycling of the powder. Moreover, post-processing operations, such as surface treatment, were not included during the computations. Similarly, the cost model developed by Ruffo et al. ⁸⁰ was used for the SLM process. They computed the cost of small to medium batch sizes. Based on the model of Hopkins and Dickens, ⁷⁹ the authors introduced their model as a complete cost model. The proposed cost model assumed the total build cost as the sum of all direct raw material costs and the indirect costs. The direct costs were obtained by multiplying the mass of deposited material with the cost of the raw material. The indirect costs were calculated by multiplying the total build time (sum of times to laser scan the section, add layers of powder, heat the bed before scanning and cool down slowly after scanning, add layers of powder, and waiting time to attain the right temperature) with an indirect cost rate. In the work of Allen, ⁸¹ relative economics of producing a near net shape by AM were compared with conventional machines. An analogy was suggested to compute cost of AM material required to accomplish a 30% savings over traditional processes. It was observed that AM was commercially applicable and economical for parts with a buy-to-fly ratio of about 12:1. Sahebrao Ingole et al. ⁸² improved the tooling and made it cheaper and faster with the use of FDM process. A cost model was used to evaluate the investment or magnitude of the process change. The cost model included costs related to material, machine, labor, pre-processing, and post-processing. In the computation, the individual cost items were associated with dimensional homogeneous equations that included the advantages of using RP in the computation. Two different processes for plastic part fabrication, that is, conventional IM and RM, were compared and evaluated to study the economic aspect and the geometric possibilities in RM. From an extended literature review, the redesign guidelines and the cost model were identified and then applied to a component selected based on its shape complexity. The economic and geometric differences between RM and IM were considered. This article demonstrated that RM combined with redesign could be an economically competitive and appropriate choice in medium batch production of plastic parts. ⁸³ The study of Lindemann et al. ⁸⁴ introduced a time-driven activity-based approach to design the cost model. All activities were divided into the following process steps: preparatory activities, printing processes, and post-processing in order to calculate the cost. Atzeni and Salmi ⁸⁵ presented a correlation between two processes used for metal part production, namely, the conventional high-pressure die-casting process and the DMLS additive process. The comparison was done from an economic point of view and from consideration of the geometric possibilities of AM. This article demonstrated the cost effectiveness and suitability of the additive process in comparison to conventional high-pressure die-casting processes for medium volume production of metal parts. A novel estimator combining energy consumption, build time, and production cost was provided for the EOSINT M270 DMLS system. ⁸⁶ It was concluded that the variety and the quantity of parts being produced and the ability to utilize available machine capacity could influence process effectiveness, both in terms of energy and finances. The costs of AM were categorized into the following four main groups by Gibson et al.: ⁸⁷ material, machine, production, and labor costs. The sum of these costs represented the total unit cost. The work of Schröder et al. ⁸⁸ presented AM business model that evaluated process costs of AM technologies and a sensitivity analysis on the output of their Excel tool. In their review, they analyzed the main business models developed for the different AM processes. The authors proposed the following requirements that needed to be included in the new cost model: recycling and waste material, support structures, printing time, maximum number of products that can be printed simultaneously, level of complexity, post-processing time, and quality management methods for the protection and monitoring of product and process quality. Baumers et al. ⁸⁹ developed a cost model to compute the overall cost involved in EBM and DMLS. Currently, this research has focused on well-structured costs, which accounts for a significant portion of AM. Moreover, this research has only considered the fabrication of single parts, neglecting mass production as well as other costs such as transportation and inventory. To understand the cost difference between AM and other techniques, it is important to include the costs from raw material extraction, through production to the sale of the final product. As a result, Thomas ⁹⁰ developed a method for understanding and examining the societal costs and benefits associated with the AM process from both resource consumption and financial point of view.

Impact on manufacturing

There are many AM applications including lightweight products for the aerospace, automotive, medical, architectural modeling, and energy industries. These include applications where low volume production, high design complexity, and the ability to change designs frequently are needed. However, AM still needs significant development with regard to design, material, novel techniques, and machines. ⁹¹

Aerospace and automotive industry

Aerospace has shown interest in these technologies because of its ability to fabricate metal parts directly using materials such as titanium (suitable for aircrafts) and the ability to fabricate complex, high-performance products easily without any tooling. ⁹²

Ding et al. ⁹³ investigated an aircraft component with thin-walled structures of varying thickness and comprising several crossings using an arc welding–based AM process (Figure 20).

Figure 20.

(a) Process of post-machining. (b) Final finished part with the desired dimensional accuracy. ⁹³

Finally, as reported by Raja et al., ⁹⁴ AM technologies increased the productivity of the aerospace sector in the following ways:

• Decline in product lead time by up to 30%–70%.

• Lessening non-recurring product costs by up to 45%.

• 30%–35% abatement in manufacturing costs for low-volume parts.

Medical industry

With AM, complex anatomical parts can be fabricated directly from scanned data (computed tomography images) which provide better visualization of specific anatomies. It also assists in precise pre-surgical planning, aids surgeons and medical students to practically re-enact various surgical procedures, and acts as a communication tool between surgeons and patients.^95–97 AM was used for the separation of Siamese twins ⁹⁸ by performing precise pre-surgical planning on AM-built models, as shown in Figure 21.

Figure 21.

AM medical applications.^98,99

Architectural and jewelry industry

Architectural and jewelry models require a high degree of manual effort and time because of the complex shapes required to implement artistic ideas in physical form. Much better resolution is possible with AM than is available from other processes for addressing these complex models. ¹⁰⁰ Models such as those in Figure 22 can only be fabricated by AM processes.

Figure 22.

Artistic models fabricated by AM technology. ¹⁰¹

Impact on environment

There are several ways that AM can impact the environment, but most researchers have focused primarily on energy consumption and environmental hazards.

Energy consumption

Concerning energy consumption, a comparison of various manufacturing parameters was studied by Mognol et al. ¹⁰² for the following three systems: FDM 3000 (Stratasys), EOSINT M250 Xtended (EOS), and Thermojet (3DS). A set of parameters for reducing electrical energy consumption was identified. Another study by Luo et al. ¹⁰³ compared energy consumption rates (kW h/kg of part geometry) for LS, FDM, and SLA. The output of this study revealed that energy consumption was entirely dependent on the duration of the job. Another study on SLS by Telenko and Seepersad ¹⁰⁴ cited the difficulties in determining a specific energy consumption constant as a result of the variation in the part height and material density. In conclusion, a research was presented by Sreenivasan and Bourell ¹⁰⁵ that proposed the accepted value of energy consumption as 52.2 MJ/kg.

Environmental hazards

The environmental hazards of the AM industry are primarily related to the chemicals (solvents) used in support removal. These chemicals may cause skin reactions, eye irritation, and allergies. Therefore, need for materials standard in the AM industry still remains a topic of extensive research. ¹⁰⁶ The environment hazard effects of different chemicals used in various AM processes are shown in Table 2.^106–114

Table 2.

Impact of different chemicals used in AM. ¹⁰⁶

AM process	Chemical/solvent	Emissions	Hazards of usage	Biodegradability
SLA	Propylene carbonate	CO2, CO, SOx	Low system toxicity was found in rats	Readily biodegradable (more than 80% degraded in 10 days)
	Urethane resins		Too much ingestion may lead to vomiting	Not found to be dangerous to the environment
	Tripropylene glycol		Slight irritation after eye contact. No absorption or irritation to the skin	Can be biodegraded by 50% in just 8.7 days and by 81.9% over a 28-day test period
	Isopropanol		Irritation and burning sensation in eyes and sometimes corneal injuries; irritation and soreness on the skin and prolonged exposure may cause dermatitis	Has a potential to acutely decrease oxygen from aqueous systems
SLS	Polyamide resin	CO2	No serious hazards were found during handling or exposure to this chemical	Form inflammable mixture with some chemicals or long exposure to air
	Acrylonitrile butadiene styrene		Molten plastic likely to cause lethal burns and processing fumes may lead to eye irritation and choking of the respiratory tract	Since it is insoluble in water, its eco-toxicity is low
LENS	Photopolymers	CO2, CO, SOx	Inhalation may cause ulcers and burning in throat and coughing; contact with skin and eyes causes redness, irritation, and swelling	No hazardous decomposition products
FDM	Propylene glycol monomethylether	CO2, CO, SOx, PMc, NOx	Irritation in eyes, skin, nose, throat; headache, nausea, dizziness, drowsiness, incoordination, vomiting, and diarrhea	No hazardous decomposition products

SLA: stereolithography; SLS: selective laser sintering; FDM: fused deposition modeling; LNES: laser-engineered net shaping; AM: additive manufacturing.

Problems in additive manufacturing

A correlation of different AM technologies with reference to their strengths and weaknesses is shown in Table 3. It is observed that none of these technologies are ideal in every dimension. For example, the energy consumption in FDM is low compared to EBM; however, EBM surpasses FDM in terms of high material strength. Certainly, it is not feasible to compare two entirely different AM processes such as EBM and FDM. The objective here is to emphasize the selection of appropriate AM technique depending on the application requirement. For instance, in some manufacturing applications or machines, even parts with low strength are sufficient to achieve desired mechanism. In that case, it is always preferable to employ low-cost FDM part. Similarly, in medical industries, plastic parts (FDM) cannot be used as implants. Therefore, EBM parts, irrespective of their cost or energy consumption, have to utilized. The greatest benefit that can be associated with AM is the ability to manufacture a wide variety of very complex parts. However, the limitations of AM, such as poor accuracy, low surface quality, and low speed, can be overcome when AM is combined with other manufacturing processes such as subtractive manufacturing. This concept led to the development of hybrid manufacturing (HM) processes where different AM methods are combined or subtractive methods are added. A discussion on HM processes is presented in the following section.

Table 3.

Comparison of different AM technologies.

AM process	Surface finish	Support structure	Temperature	Energy consumption	Material variety	Capability of multiple materials	Material resolution	Production rate	Product properties	Technology cost	Technology maintenance	Technology knowledge
FDM	Good accuracy	Need support structure	Low processing temperature	Low	Polymer	Fair	Low	Low speed	Fragile parts	Low cost	Polymer can be trapped inside the pores	High knowledge needed
Robocasting	Good accuracy	Need support structure	Low processing temperature	Low	Polymers, ceramics, and metals	Good	Middle	Low speed	Fragile parts	Low cost	Polymer can be trapped inside the pores	High knowledge needed
SLS	Limited accuracy	No support structure	High processing temperature	High	Metals, ceramics, and polymers	Fair	Low	Low speed	High mechanical properties	High cost	Powder can be trapped inside the body	High knowledge needed
SLM	High accuracy	Need support structure	High processing temperature	High	Metals	Fair	High	High speed	High mechanical properties	High cost	Powder can be trapped inside the pores	High knowledge needed
EBM	High accuracy	No support structure	High processing temperature	High	Metals	Fair	High	Low speed	High mechanical properties	High cost	Powder can be trapped inside the body	High knowledge needed
LENS	Low accuracy and post-processing required	No support structure	High processing temperature	High	Metal	Fair	High	High speed	High mechanical properties	High cost	Powder can be trapped inside the pores	High knowledge needed
LMD	High accuracy	No support structure	High processing temperature	High	Metals	Fair	High	High speed	High mechanical properties	High cost	Powder can be trapped inside the body	High knowledge needed
3D powder printing	Post-processing required	No support structure	High processing temperature	High	Metals, ceramics, and polymers			High speed	Fragile parts	Low cost	Powder can be trapped inside the pores	High knowledge needed
3D plotting	No post-processing required	No support structure	Low processing temperature	Low	Biomaterials	Very good	High	Middle speed	Fragile parts	High cost	Biomaterial can be trapped inside the pores	High knowledge needed
Ultrasonic consolidation	High accuracy	No support structure	Low processing temperature	Low	Metals like aluminum	Good	Low	Middle speed	High mechanical properties	High cost	–	High knowledge needed
Stereroli-thography	High accuracy	Photopolymer needed and support structure needed	Very low processing temperature	Very low	Reactive resin	Good	Low	Middle speed	Fragile parts	High cost	–	High knowledge needed
Digital light processing	High accuracy	Photopolymer needed and support structure needed	Very low processing temperature	Very low	Reactive resin	Good	Low	Middle speed	Fragile parts	Low cost	–	High knowledge needed
LOM	Fine accuracy	Need support structure	Low processing temperature	Low	Paper, polymers, and composites	Fair	Low	High speed	Fragile parts	High cost	–	High knowledge needed

FDM: fused deposition modeling; FEF: freeze-form extrusion fabrication; SLS: selective laser sintering; SLM: selective laser melting; EBM: electron beam melting; LMD: laser metal deposition; 3D: three-dimensional; LOM: laminated object manufacturing; AM: additive manufacturing; LENS: laser-engineered net shaping.

HM processes

HM can be carried out either in parallel or in a serial manner depending on the requirements. These processes can significantly improve tool life, material removal rate, and dimensional and geometric accuracy and reduce production times. The primary objective of hybridization is overcoming the limitations associated with individual processes. For example, in hybrid AM, multiple techniques such as mixing of different materials during deposition and melting of deposit material at various temperature conditions can be used to overcome various limitations. The hybrid processes can be categorized in terms of four hybrid and three sub-hybrid types as shown in Figure 23. HM can be described as an approach that integrates more than one manufacturing operation from different manufacturing technologies. In contrast, sub-hybrid manufacturing is the combination of operations of a single manufacturing technology. The sub-hybrid classification is intended to overcome geometrical issues.

Figure 23.

Categorization of hybrid processes. ¹¹⁵

The hybrid process can also be classified depending on the raw material used (such as composite products), ¹¹⁵ working principle (such as laser-assisted milling and turning), ¹¹⁶ or form of energy. ¹¹⁷ Similarly, compared to individual processes, the combination of different manufacturing processes has opened new application areas for producing low-cost components. ¹¹⁶ The International Academy for Production Engineering (CIRP) proposed the following definitions for defining hybrid processes: ¹¹⁸

1. Open definition. An HM process integrates two or more manufacturing methods into a new distinct process whereby the benefits of each individual process can be utilized synergistically.

2. Narrow definition. Hybrid methods involve a concurrent action of various (chemical, physical, and controlled) processing principles in the identical processing zone.

A hybrid process is defined as a practice where two or more manufacturing procedures from different manufacturing technologies are coalesced. The benefits of combining multiple techniques such as mixing of different materials during deposition, melting at different heat conditions, and deposition of discrete materials have been reviewed by Z Zhu et al. ¹ Du et al. ¹¹⁹ presented the benefits of combining additive and subtractive manufacturing on geometric and dimensional accuracy, density, microstructure, and hardness of the parts. To address the needs of HM, it is important to understand the principles of every individual process and identify existing hybrid approaches enabling the identification of the requirements and constraints. The combination of additive and subtractive manufacturing processes may pose difficulties from the geometrical standpoint because hybridization affects the fabrication of functional parts with complicated shapes. ¹²⁰ A subtractive process such as CNC can be combined with additive processes to avoid limitations like tool accessibility, low accuracy, surface quality, and low build speed. ¹²¹ For example, to achieve a required accuracy, CNC can be employed to mill the cladding of the surface on a part fabricated by melting a mixed powder (Fe, Ni, and Cr) by a laser. ¹²² Laser cladding was consolidated with a five-axis CNC milling machine for building two-dimensional feature on a rotating workstation. It eliminated the need for supporting materials and reduced the build time.^123,124 Karunakaran et al. ⁵⁰ also used three-axis milling to face mill build slices of metal inert-active gas welding for improving the geometrical and dimensional accuracy of the product. In this hybrid process, the layers were deposited and followed by face milling until the entire part was produced. The milling of each layer helped to reduce the machining time. ⁸ Karunakaran et al. ¹²⁵ also retrofitted a CNC machine with arc weld deposition. The arc weld deposition (additive process) was used for building near net shapes, while the CNC machine (subtractive) was utilized for finishing operation on the same station. The approach resulted in time and cost savings. Special software was developed to generate the CNC program for this process. A hybrid additive–subtractive manufacturing process, named as hybrid plasma deposition and milling (HPDM), was introduced by Xiong et al. ¹²⁶ This process was used in a layer-by-layer fabrication of near net shape. The milling operation was optimized to improve the accuracy and surface finish of the fabricated part. Similarly, a hybrid subtractive–additive welding step was developed to produce a microfluidic system capable of grading different sized micro-particles. ¹²⁷ In this work, a femtosecond laser for additive and subtractive processes was also implemented in a single machine. Laser ablation was initially used to machine channels in polymethyl methacrylate (PMMA) and glass substrates. The 3D laser lithography was then used to integrate microgrooves into the channels. Finally, the grooves were sealed through laser welding. Karunakaran et al. ⁵⁰ combined arc welding with a CNC milling machine. The near net shape was first built using weld deposition, and then, the surface finish was improved through CNC milling. Customized software was developed to create the tool path for near-net-shape building. The cost and time were saved through elimination of rough machining and a reduction in the NC programming effort.

Hu and Lee ¹²⁸ proposed a concave edge-based approach for part segmentation into layers, depending on the total number of layers, layer thickness, and tool accessibility. As a result, the complete part was divided into subparts, which eliminated underside edges during machining. Liou et al. ¹²⁹ and Ruan et al. ¹³⁰ proposed a process planning approach for LS and CNC machining to divide parts into a variable layer thickness and generated tool paths for avoiding tool collisions. Jiang et al. ¹³¹ presented a part-decomposing algorithm based on STL solid model using shape deposition manufacturing (SDM). This approach, which was formerly developed at Carnegie Mellon University in early 1990s, adopted the technologies of both additive and subtractive machining.^132,133 The algorithm could decompose STL models efficiently, but it lacked evidence of reliability and needed efficacy improvements. Kerbrat et al. ¹³⁴ also introduced a design for manufacturing (DFM) approach incorporating a hybrid modular vision that combined additive processes with a traditional high-speed machining (HSM) subtractive process. A hybrid modular design methodology was created that integrated manufacturability issues at the design stage. The approach was tested on industrial products for the automotive industry. J Liu and AC To ¹³⁵ proposed a topology optimization method for hybrid additive–subtractive process design. The topology design was achieved through AM and the shape-preserved boundary segments were treated by post-machining. A hybrid additive and subtractive manufacturing technique developed by Ikonomov and Rodriguez ¹³⁶ can be used to produce shaped cavities for casted metallic objects. RP infrared light sintering and machineable mold material machining (MMM) technologies were used for the synchronize building and machining of carbon-shell sand molds. In this approach, the MMM process was used for machining each layer before the next layer was built. El Kashouty et al. ¹³⁷ considered the use of a hybrid AM and subtractive manufacturing to manufacture an insert for the tooling of a vehicle headlight adjuster clip. SLM technology and CNC milling were used to improve the mechanical properties, chemical composition, surface quality, and dimensional precision. Song and Park ¹³⁸ developed a process by combining additive and subtractive techniques and named it as 3D welding and milling. To achieve hybridization, welding was used as an additive component, while milling as a subtractive technique. The authors developed two alternative strategies to manufacture the parts with different materials for this process. The first strategy used two independent welding guns for providing different materials on the boundary and inside layer. The following strategy fabricated exclusively the shell structure for the part and subsequently filled it with low-melting alloys. Finally, the milling head was employed to enhance the accuracy and the surface quality. Ruan et al. ¹³⁰ presented a method for integrating the multi-axis deposition AM process with the machining process that provided an automatic sequence and tool path for both processes. In addition, the switchover between the two processes was optimized to obtain procedural accuracy and speed.

The selection of an appropriate fabrication technique is also a challenging task. A DFM approach was proposed by Karunakaran et al. ¹³⁹ to analyze features during the product design stage and subsequently select the subtractive (machining) or AM (such as SLS or laser deposition) approaches based on feature complexity. Similarly, Townsend and Urbanic ¹⁴⁰ proposed an analytic hierarchy process (AHP) model for process selection by matching the part specifications to process capabilities and then mapping to process selection. They studied hybrid design and manufacturing approaches for AM (FDM) and subtractive manufacturing (CNC machining). In their study, a teleological system was implemented to fine-tune part functionality with geometry into units. A simulation tool was employed by Hansel et al. ¹⁴¹ to select the suitable nozzle geometry for ensuring the best particle distribution in a hybrid on additive–subtractive hybrid machine (directed energy deposition (DED) on a hybrid platform). Taguchi analysis was used to determine the optimal process parameter depending on the requirements of porosity, tensile strength, hardness, and microstructure. ¹⁴¹ A novel algorithm associated with optimized manufacturing operation sequence (OMOS) software for feature recognition in HM (subtractive and additive) has been developed by Homar and Pušavec. ¹⁴² The algorithm analyzed the geometry of CAD models and determined parts and features of the CAD model that needed to be manufactured by additive and subtractive manufacturing based on the cost and the quality of the produced parts. The benefits of developing an HM cell using the combination of an AM metal system (Concept Laser M2 Cusing) and a subtractive process (a five-axis milling machine) were investigated by Boivie et al. ¹⁴³ The objective was the manufacturing of injection-molded tool inserts, incorporated with appropriate cooling channels. The clamping, positioning, and control systems for additive and subtractive machines were developed to minimize the number of process steps and auxiliary operations. ¹⁴⁴ To study the economics of HM, Manogharan et al. presented cost models for a hybrid additive–subtractive manufacturing process. The cost models were developed to examine the effect of production volume, batch size, material and operation costs, material machinability, and build time on the unit costs associated with the engineering requirements. The results showed that batch size, build time, and build cost were the dominant cost factors affecting the unit cost produced using AM and SM methods. ¹⁴⁵ Newman et al. ¹⁴⁶ proposed a framework called iAtractive for combining additive and subtractive technologies and inspection operations in hybrid systems. The iAtractive system provided the process planning and the Re-Plan of the hybrid system operations. The concept of hybridization brings a host of benefits to AM allowing manufacturers to produce more complicated parts at lower cost.

AM challenges

Certainly, there are numerous benefits, such as design flexibility, ability to print complex structures, ease of use, and product customization, that can be associated with AM. However, AM technology has still not matured enough so that it can be employed in real world applications. There have been drawbacks and challenges that need investigation as well as advanced technological development. The limit on the part size, anisotropic mechanical properties, building of overhang surfaces, high costs, low manufacturing efficiency, poor accuracy, warping, pillowing, stringing, gaps in the top layers, under-extrusion, layer misalignment, over-extrusion, elephant foot, mass production and limitation in the use materials are the challenges that need further analysis and exploration.^147,148 Some of the limitations and challenges related to AM are described as follows.

Void formation

The void formation between subsequent layers of AM parts is one of the major drawbacks. This kind of problem occurs due to reduced bonding between layers, thus causing inferior mechanical performance.^147,148 For example, extrusion-based AM technologies such as FDM results in void formation between the fabricated layers, thus inducing anisotropic mechanical properties^148,149 and delamination. ¹⁵⁰ Indeed, the amount of the porosity induced by void formation typically depends on the type of AM process and the material used. Hence, to minimize the effect of void formation between subsequent layers, Paul et al. ¹⁵¹ evaluated the effect of nozzle geometry on the void formation between subsequent layers. The results of their study showed that the performance of rectangular nozzles was better than cylindrical nozzles.

Stair-stepping

One of the biggest challenges in AM process is the appearance of staircase effect or layering error in the fabricated parts. This kind of error is insignificant for internal fabricated surfaces; however, it substantially affects the quality of external surfaces. Although, many methods (post-processing) like sand sintering can be employed to minimize or get rid of this defect ¹⁵² but they also increases the time and cost of the overall process.

Anisotropic in mechanical properties and microstructure

Another challenge that can be observed with AM is the existence of anisotropy in microstructure and mechanical properties. AM technologies produce parts in a layer-by-layer fashion by curing the photo resin, melting the filament or melting the powder bed, resulting in the generation of thermal gradient. The AM parts often results in different microstructure and mechanical properties along build direction and the other directions.^153,154 For example, the plates manufactured by FDM technology possess better strength in x, y direction as compared to the strength in z-direction (build direction).^155,156

Small build volume

The user of AM technologies is also dealing with the challenge of small build volume. It is considered as one of the main disadvantages of AM technology. Generally, the large parts are scaled down or cut to subparts, which adds time and effort. Moreover, the scaling down of the model in most cases is not feasible and effective. The assembly of subparts after scaling down possesses lesser strength if adhesives are used or become bulky in case mechanical fasteners are employed. ¹⁵⁵ Until now, AM has not been successful for large-scale industrial applications. ¹⁵⁷

Fabrication of weapons or drugs for crime purposes

As a result of its capability to fabricate complex structures, AM can also be employed to produce weapons that can be used for crime purposes. This aspect of AM has also limited its spread in some countries.

Compliance with Food & Drug Administration safety standards

The medical devices and implants as well as the food that are fabricated and produced by AM must comply with the regulations outlined by local Food & Drug Administration (FDA). Since the AM technologies are relatively new to the medical industry, FDA is still working on the design of distinct rules and regulations. Moreover, the existing regulations are complex and difficult to implement; as a result, companies are hesitant to use AM in healthcare sector.

Conclusion

Advancements in manufacturing industry depend on leading edge research associated with manufacturing processes, materials, and product design. As product complexity increases, there is a need for new and innovative manufacturing processes. AM is a recent trend in production processes because of the many benefits it provides as well as the challenges it has to overcome. It has been subjected to intensive investigation and in-depth review by the research community. In this work, an exhaustive review related to AM has been carried out. The importance of part orientation, build time estimation, and cost computation have been reviewed in detail. The most important feature of this work is the identification of the problems associated with different AM methods. Among the primary AM challenges, part size limitation, anisotropic mechanical properties, building of overhang surfaces, high costs, poor accuracy, warping, layer misalignment, mass production, and limitation in the use materials need further research and investigation.

Based on this review, the various aspects of AM technology can be summarized as follows. It can be asserted that the selection of suitable part orientation is crucial in AM. It helps to improve geometrical and dimensional error, reduce build time, and minimize support volume and part production costs. The staircase effect has been identified as the most important factor affecting the part accuracy. Indeed, the staircase effect is directly proportional to the layer thickness. It has also been observed that build time, which possess greater significance in improving productivity, depends on machine speed, part size, layer thickness, and build orientation.

It has been noticed that there is a need for the development of an effective and ubiquitous cost model for the AM. Meanwhile, the costs of AM can be categorized in terms of material, machine, manufacturing, and labor costs. The summation of these costs represents the overall unit cost. It is important to incorporate the following requirements in the new cost model: reused or wasted material, support design and arrangement, build time, maximum possible number of parts that can be produced simultaneously, level of complexity, post-processing time, and quality management.

Lately, the concept of HM has emerged in order to further expand the applications of AM. The limitations of AM, such as poor accuracy, surface quality, and low speed can be overcome when combined with other manufacturing processes such as subtractive manufacturing. HM can significantly improve tool life, speed up material removal rate, enhance dimensional and geometric accuracy, and reduce manufacturing times. The primary objective of hybridization is to overcome the limitations of the individual processes. In fact, the absence of Computer-Aided Process Planning (CAPP) systems restrict proper utilization of additive–subtractive HM in many industrial applications due to the complexity of designing system inputs at the design phase. Therefore, the development of CAPP for additive–subtractive hybrid systems will encourage further implementation and increase the application of AM in the future.