Influence of Material‐Related Aspects of Additive and Subtractive Ti‐6AI‐4V Manufacturing on Energy Demand and Carbon Dioxide Emissions

The additive manufacturing of metal parts represents a promising process that could be used alongside traditional manufacturing methods. The research scenario in this field is still largely unexplored, as far as the technological solutions adopted to integrate different processes are concerned and in terms of environmental and economic impact assessment. In this article, an electron beam melting (EBM) process and a machining process have been analyzed and compared using a cradle&#8208;to&#8208;grave life cycle&#8211;based approach. The production of components made of the Ti&#8208;6Al&#8208;4V alloy has been assumed as a case study. The proposed methodology is able to account for all of the main factors of influence on energy demand and carbon dioxide emissions when the component shape is varied. The results prove that, besides the direct energy intensity of the manufacturing processes, the impacts related to material usage are usually dominant. Therefore, when complex geometries have to be manufactured, the additive manufacturing approach could be the best strategy, if it enables a larger amount of material savings than conventional machining. Vice versa, when a small amount of material has to be machined off, the high energy intensity of an EBM process has a negative effect on the performance of the process.


Introduction
Manufacturing accounts for about 98% of the total direct carbon dioxide (CO 2 ) emissions from the industrial sector and is responsible for around 35% of the global electricity use and over one quarter of the primary resource extraction (Fischedick et al. 2014;UNEP 2011). Gutowski and colleagues (2013a) analyzed CO 2 manufacturing emissions throughout the world over the last few decades. The observed trends revealed that CO 2 emissions gradually increased from 1970 to 2002, and then a relatively sharp rise appeared (mainly attributed to the increased production activity of developing and emerging nations). In short, the total global CO 2 emissions attributed to manufacturing continued along an upward slope over the

State of the Art
The growing interest in quantifying the CO 2 footprint of processes has led to the development of a methodology for the systematic analysis and improvement of the manufacturing unit process life cycle inventory (UPLCI) ). The need for environmental impact analyses becomes more and more urgent if the increasing success of additive manufacturing (AM) approaches is taken into account. AM could be used directly to produce a final item for sale or use. Powder bed fusion processes have been developed to manufacture titanium alloys, stainless steel, and aluminum alloys (Mower and Long 2016). AM processes might enable the obtainment of a fully dense component (as mentioned by Elahinia et al. [2012], Facchini et al. [2009], and Tammas-Williams et al. [2015] for electron beam melting [EBM] processes), whose mechanical properties could be comparable with those of components achieved by conventional manufacturing. In this context, Wang and colleagues (2016) and colleagues (2009a, 2009b) have shown interesting results for the EBM of Ti-6Al-4V.
Three fundamental approaches can be followed for metal shaping: mass conserving, subtractive, and additive processes. The environmental impact of each manufacturing approach should be explored in detail in order to identify the potential of all the emerging processes. In order to tackle this challenge, manufacturing scientists have turned their attention toward environmental impact assessments over the last few years, and a clear picture of the so-far developed researches has been provided by Haapala and colleagues (2013) and Duflou and colleagues (2012). However, the route toward achieving complete knowledge of the environmental performance of metal-related manufacturing processes is still long. Currently, sustainability analyses predominantly focus on material removing processes. Several articles have either focused on the effects of process parameters (such as those of Dahmus and Gutowski [2004], Diaz et al. [2011], and Kara and Li [2011]), on machine tool architectures (Behrendt et al. 2012), or debated the demands of different processes (Gutowski et al. 2006). Much less attention has been paid to forming processes. The available studies concern LCI guidelines for hot forging processes (Buis et al. 2013) and the environmental performance of sheet metal forming processes (Santos et al. 2011;Ingarao et al. 2014).
As far as AM is concerned, a state of the art on the sustainability analyses of AM processes can be found in the article by Huang and colleagues (2016). The specific electric consumption of a large number of AM processes has also been reported in Yoon and colleagues (2015) and in Le Bourhis and colleagues (2013). So far, most of the studies have focused on polymeric materials Mognol et al. 2006). Only a few researches have been conducted on the environmental performance of AM processes for metal parts. Duflou and colleagues (2011) proposed a unit process LCI for a selective laser melting (SLM) process. The researchers also pointed out impact reduction opportunities in that article. Baumers and colleagues (2011) analyzed the effect of capacity utilization (i.e., the utilization of the build volume) on the energy efficiency of different metal-based AM processes. The same research group proposed a cost and energy accounting model for the AM of stainless steel. The effects of the product demand were also analyzed (Baumers et al. 2013). Moreover, the influence of shape complexity on the specific energy consumption of EBM processes has recently been studied (Baumers et al. 2017). Le Bourhis and colleagues (2013) applied a life cycle assessment (LCA)-based methodology to direct additive laser manufacturing (DALM), in which electric energy, fluids, and raw material consumption were all accounted for.
However, in order to evaluate the actual environmental impact of a given manufacturing approach, the whole life cycle of the product should be analyzed. In particular, given that the three fundamental manufacturing approaches, besides adopting different setups and machine tools, also use different kinds and amounts of material, the material flow is expected to have an important effect on the performance of the environmental process. Additionally, comparative assessments are of utmost importance in order to properly label and sort the manufacturing processes from a sustainability point of view. A first comparative analysis was provided by Morrow and colleagues (2007). They produced a quantitative estimate of the energy consumption and emissions associated with the production of molds and dies by laser-based direct metal deposition and computer numerical control milling. Another comparison of different approaches was developed by Serres and colleagues (2011): The researchers compared the direct additive laser manufacturing process (Construction Laser Additive Directe; CLAD R ) with conventional machining for the production of components made of Ti-6Al-4V. The researchers developed a full LCA analysis and proved that AM leads to an environmental impact (the Ecoscore from Eco-Indicator 99) reduction of as much as 70%. Huang and colleagues (2016) applied the comparative analysis to five case studies pertaining to aircraft components made of metallic materials by EBM, SLS, and direct metal laser sintering (DMLS). It was found that the energy savings were primarily attributed to the reductions in resource production and energy use, which, in turn, were attributable to the lower buy-to-fly ratios (given that the AM approaches used less material than the conventional processes). Ingarao and colleagues (2015) conducted an environmental comparison of a hot extrusion process (bulk forming process) and a machining process for a simple-shaped aluminum component. The results revealed that workpiece material usage was the dominant factor of influence in terms of environmental impact, even when recycling credits were considered, for different geometries (Ingarao et al. 2016). In this context, another decision-support model has recently been proposed by Watson and Taminger (2016), who established a computational model to determine when AM is preferable (in terms of energy efficiency) over subtractive machining. The three product shapes to be manufactured by additive and subtractive processes.

Aim of the Article
Some guidelines for future research have emerged from the literature review. Overall, the environmental footprint of discrete part manufacturing processes still needs to be fully understood. Moreover, there is still a need to standardize international research efforts in order to make the obtained results comparable. Currently, there is a lack of studies on the quantification of energy use (throughout the whole life cycle) and on the greenhouse gas emission implications of additively manufactured components, and the conducted research has mostly been concentrated on polymeric materials. Besides the direct energy intensity of AM processes, material-related environmental impacts should be included when cradle-to-grave boundaries are considered. Finally, comparative analyses are able to point out the actual environmental impact of different manufacturing approaches. However, these kinds of studies have rarely been applied.
The present article has the aim of partially filling the aforementioned knowledge gap. An AM process (EBM) and a conventional machining process have been analyzed, integrated, and compared for the production of components made of Ti-6Al-4V. A complete methodology, which considers all the factors of influence necessary to compile an energy and CO 2 LCI, has been implemented. Impacts related to material usage have been accounted for in a cradle-to-grave approach, and material recycling has been identified as the end-of-life (EoL) scenario. The proposed methodology also includes the process scrap and postprocessing (finish machining) operations for the EBM process. The product shape has been varied in order to analyze the effect of such a parameter on the process footprint. Summing up, the article has the aim of developing a quantitative methodology to assess the primary energy demand and the CO 2 emissions of additive and subtractive manufacturing processes. Such a tool can be expected to be useful to detect the production scenarios that make additive processes preferable (from the environmental viewpoint) over conventional manufacturing approaches.

Case Study and the Main Assumptions
A life cycle-based approach has here been proposed to evaluate the integration of AM techniques in conventional production routes based on machining processes. The goal and scope of the study have been to analyze the impact of both addi- tive and subtractive manufacturing strategies in terms of energy demand, CO 2 emissions, and resource depletion. The production of components made of Ti-6Al-4V has been considered. In addition, given that the environmental performance of a manufacturing approach is affected by the type and the amount of the involved materials, an analysis has been designed, in which the component shape has been varied and different solid-to-cavity ratios have been considered. The solid-to-cavity ratio has been defined, according to Morrow and colleagues (2007), as the mass of the final part divided by the mass that would be contained within the bounding volumetric envelope of the part. The study has therefore been conducted to (1) improve knowledge concerning the key factors of influence on the impact of such processes and (2) provide guidelines in which the most energy-and resource-efficient manufacturing strategy is identified as a function of the production scenario. The functional unit, the system boundaries, and the process evaluation metrics are presented hereafter.

Functional Unit
The impact arising from the life cycle of a single component made of a Ti-6Al-4V alloy has been assumed as the basic unit for a process assessment. As shown in figure 1, three different shapes (namely, ID 1, ID 2, and ID 3) were taken into account. The main geometrical features are listed in table 1. The basic idea was to analyze the environmental performance while varying the amount of  process scraps for subtractive approaches (i.e., the machinedoff material) and verify whether additive manufacturing could be a more efficient strategy than traditional subtractive processes.

System Boundaries
A cradle-to-grave system boundary has been adopted, and recycling has been selected as the EoL scenario. The impacts of material production, product manufacturing, transportation, and recycling were evaluated. It was assumed that all the components (made either by a subtractive approach or additive manufacturing plus finish machining) complied with the same product specifications. As a result, it was possible to neglect the differences in environmental impact during the use phase (given that the in-use performance was expected to be the same). Figure 2 shows a sketch of the system boundaries, together with the main considered factors, as well as the energy and resource flows. Differences in feedstock material production and component manufacturing were considered, as pointed out in Life Cycle Inventory. Moreover, transportation impacts due to material shipment from the production site to the manufacturing plant and from the manufacturing plant to the recycling site were also estimated.

Impact Indicators
The energy demand and the CO 2 emissions have been assumed as metrics for the process assessment. The electric energy consumption was converted into (primary) energy source consumption by considering an average efficiency of 34% (Jeswiet and Kara 2008) to account for the energy generation and the transmission losses. The CO 2 emissions attributed to electric energy consumption were computed using the carbon emission signature (CES) method proposed by Jeswiet and Kara (2008) and applying data for the Italian energy mix (IEA 2015).

Life Cycle Inventory
In this section, the main considered factors of influence are explained for each life cycle stage. Because material recycling is considered as the EoL strategy, the material production and the EoL phases are discussed together in Material Usage and Recycling-Related Benefits, for the sake of clarity. It is appropriate to remark that, when combining different energy demand contributions, all the values should be traced back to the same level. Therefore, unless otherwise specified in the text (i.e., for the electric energy consumption), all the presented values (also from literature) are assumed to refer to their primary energy source consumption.

Material Usage and Recycling-Related Benefits
When comparing different technological approaches, such as additive manufacturing (AM) and conventional machining (CM), the different routes for the production of the materials being processed have to be considered (figure 2). First, raw materials have to be extracted and refined in order to obtain the usable, in-stock materials. The energy per unit mass needed to make a material from its ores and feedstock is defined as its embodied energy (Ashby 2013). Then, materials have to be further processed in order to produce the metal powder for AM, or the workpiece for CM. For instance, gas atomization can be applied to produce Ti-6Al-4V powder (Dawes et al. 2015), whereas forming processes are used to create billets, plates, slabs, etc. Postprocessing operations, such as the thermal treatments of the formed parts, should also be included. Overall, the energy necessary for material production (E mat , in MJ/part) can be computed, for each produced part, according to equation (1): where: r m (in kilograms [kg]) is the weight of the material that has to be processed by AM or CM; r E E (in megajoules [MJ]/kg) is the embodied energy necessary to obtain the usable material from ores and feedstock; r E Pr (in MJ/kg) is the energy necessary to further process the material into powder (AM) or a workpiece (CM).
The energy necessary for material production, for both the subtractive and additive manufacturing approaches, should account for the benefits that can be derived from recycling. However, there is no single criterion that is able to account for recycling credits. The substitution method, which considers the impacts on the present climate on the production and supply of the material (cradle-to-gate), and gives a recycling credit for future recyclability (EoL), has been applied in the present article (Hammond and Jones 2010). For materials that do not suffer from losses in their inherent properties, such as metals, the embodied energy (E E , in MJ/kg) is obtained by means of equation (2), in which the fraction of material recycled at the EoL (r) and the embodied impact arising from the recycled material input (E R , in MJ/kg) are included. E V is the embodied energy for the primary production of the material, whereas E D represents the energy involved when scraps are handled. If E D is neglected, equation (2) can be rewritten as equation (3), where the embodied energy savings (E V -E R , in MJ/kg) are directly proportional to the material recyclability. CO 2 emissions can be assessed in a similar manner.

Machining
The weight of the material that has to be processed for CM, or rather of the workpiece, corresponds to the sum of the weight of the part that has to be obtained (m p ) and the weight of the machined chips (m c ), as shown in equation (4).
The process scraps (i.e., the chips obtained from the turning process as by-products), as well as the component (which is disposed of at the end of its first life) are assumed to be recycled. As far as the recyclability of chips and bulk material are concerned, specific data can be obtained for some materials, such as aluminum alloys (Ingarao et al. 2016), but there is a lack of literature information on Ti-6Al-4V. In the present article, the recyclability of the bulk material and chips has been hypothesized to be equal to 0.80 (Mayyas et al. 2012). A forming process to create the workpiece (a 50-millimeter [mm] diameter and 50-mm-long billet) was included in the workpiece material production process. The eco-properties of Ti-based alloys were extracted from Ashby (2013) and are listed in table 2.

Additive Manufacturing
The weight of the powder necessary to properly accomplish the AM process (equation (5)) should include the weight of the part that has to be obtained (m p ), the machining allowance for postprocessing finishing operations (m a ), and the weight of process scraps (m s ).
The weights of the parts (m p ) are listed in table 1, for each case study. EBM is suitable for obtaining free-form parts, and it is successfully used for various applications, such as for the production of components for biomedical implants (Bartolo et al. 2012). However, the achieved surface quality is not usually adequate to meet the strict requirements of the aerospace and automotive industry (Priarone et al. 2012). The arithmetic mean of the roughness of EBM-made parts is between 25 and 35 micrometers. In addition, whereas SLS, DMLS, and selective laser melting (SLM) processes produce parts with dimensional errors of less than 0.1 mm for a 100-mm length, EBM dimensional precision is only half as good (Vayre et al. 2012). Therefore, a further finishing turning operation has been considered for the three case studies. A constant and uniform machining allowance of 1 mm was assumed (Rännar et al. 2007). Figure 3 compares the material usage for CM and AM, as a function of the case study.
Further, the process scraps should be included in the analysis. m s accounts for both powder losses and material losses attributed to support structures. As far as powder recyclability is concerned, there is still a need for systematic studies to demonstrate the stability of the powders for the use over extended periods of time. Some articles dealing with Ti-6Al-4V powder recyclability in EBM processes have recently been published (Petrovic and Niñerola 2015;Nandwana et al. 2016). These research activities have proved that EBM-processed Ti-6Al-4V powder properties are conserved in consecutive builds with recycled powder. A significant amount of raw material can be saved by means of powder recycling given that, despite numerous reuses, the chemical contents of the powder material are maintained, and the aeronautical standards are thus satisfied. As for the support structures, the material losses are generally related to the geometrical complexity of the components that have to be additively manufactured. In the present article, m s has been hypothesized to vary for AM as a percentage of the mass of the component plus the machining allowance (m p + m a ).
The energy necessary for material production was computed according to equation (1). The titanium ingots were assumed to be subjected to an atomization process in order to obtain suitable powders for the EBM process. Therefore, the environmental impact attributed to the atomization process has to be added to those of material extraction, refinement, and preprocessing. Only a few articles provide useful information on how to model the impact of titanium powder production (Serres et al. 2011;Baumers et al. 2017). The material for the gas atomization process is melted and then atomized by means of high-pressure jets of gas (argon or nitrogen). The energy necessary for melting as well as for gas production should be included in the LCI step. Energy and resource demands for titanium powder production were reported in Serres and colleagues (2011): It is possible to compute a specific electric energy consumption of 1.95 kilowatt-hours per kilogram (kg) from the researchers' data and an argon flow rate of 0.21 liters (L) per kg. According to Weir and Muneer (1998), the embodied energy for argon can be assumed to be 0.672 kilojoules per L. On the basis of such information, the energy necessary for gas atomization should be around 20 MJ/kg. On the other hand, other researchers (Baumers et al. 2017) estimated 31.7 MJ/kg for the atomization process. In the present article, the latter value was conservatively assumed to consider the gas atomization of Ti-6Al-4V.

Machining
Several energy consumption assessment models have been proposed for machining (such as the ones by Kara and Li [2011]; Mori et al. [2011]; Mativenga and Rajemi [2011]; Li et al. [2013]). Black-box approaches empirically correlate the input machining parameters to the energy demand, whereas bottomup approaches divide the total energy consumption into the contributions attributed to machine states and/or components (Guo et al. 2015). In the present research, the specific energy consumption (SEC) was related to the material removal rate (MRR) by applying the model shown in equation (6), proposed by Li and Kara (2011). The specific coefficients C 0 and C 1 of the machine were experimentally assessed for when Ti-6Al-4V is machined using a Graziano 101 SAG CNC lathe.
Only a few studies (such as those of Rajemi et al. [2010]; Mativenga and Rajemi [2011]; Liu et al. [2016]) account for indirect energy consumption attributed to cutting tool usage. The tooling footprint could be modeled according to equation (7) (Mativenga and Rajemi 2011;Ingarao et al. 2016), where E ct is the energy necessary for tooling (including the embodied energy of tungsten carbide and tool manufacturing), n e is the number of cutting edges of the turning insert, t c is the cutting time necessary to produce a part, and T L is the tool life of each cutting edge.   Figure 4 plots the direct and indirect contributions to the energy demand when a component with an ID 1 geometry is produced. According to the SEC model, the energy necessary for the turning process (which was computed directly by referring to the primary energy source consumption) decreases when the material removal rate is increased. Vice versa, the increase in the material removal rate (obtained by increasing the cutting speed in the present case) results in an increase in the tooling footprint, attributed to an increase in tool wear. The cutting parameters that can minimize the direct and indirect energy demand should be identified for each specific case study. However, in the present research, because low MRRs were chosen experimentally, only the direct electric energy demand for machining has been considered. ID 1 was assumed to be produced by longitudinal external turning, whereas minimum internal diameters of 20 and 25 mm were obtained for ID 2 and ID 3 by considering drilling with twist drills. The cutting speed for turning was 90 meters per minute (m/min), the feed was 0.15 millimeters per revolution (mm/rev), and the depth of cut was 0.5 mm. The cutting speed for drilling was 35 m/min and the feed was 0.05 mm/rev. All the cutting operations were performed under conventional flood cooling, which is a widely applied industrial standard.

Additive Manufacturing
As far as the inventory analysis of the EBM process is concerned, the electric energy consumed by the used platform has to be monitored and, subsequently, ascribed to the functional unit. An SEC value for EBM-sintering Ti-6Al-4V was proposed by colleagues (2011, 2017). They quantified the SEC for each deposited kilogram of material as 59.96 MJ/kg, which includes the energy consumption due to both productive and non-productive modes. The authors proved there is no direct link between energy consumption and shape complexity for EBM (Baumers et al. 2017). As a consequence, such an SEC value might be used to quantify the production energy of any component made of Ti-6Al-4V and manufactured via EBM. However, it is worth mentioning that the utilization of the build volume affects the process energy, and therefore the SEC

S198
Journal of Industrial Ecology value itself (Baumers et al. 2011). In the present research, the electric energy consumption for EBM has been assumed to be 59.96 MJ/kg, and operation at full machine capacity has been considered. The energy for the post-processing finish machining operations was computed, for the three considered case studies, by applying the cutting conditions for turning detailed in Machining.

Transportation
Given that the two manufacturing approaches involve different amounts of material, different amounts of material have to be shipped, which, in turn, leads to different transportation-related environmental impacts. The impact attributed to material shipment from the titanium production site to the manufacturing plant and attributed to the scrap shipment to the recycling plant was modeled according to the guidelines provided in EcoInvent, that is, 300 kilometers (km) for the shipment to the manufacturing plant and 300 km for shipping the process scraps to the recycling plant. Moreover, impacts attributed to the transportations identical for the two manufacturing paths were not included. Energy and the emission factors (equal to 0.71 MJ/tonne·km and 0.05 kg/tonne·km, respectively) concerning a 55-tonne truck were used (Ashby 2013).

Results and Discussion
The life cycle results for the primary energy demand and CO 2 emissions are presented in figures 5 and 6. The contribution of each considered factor as well as the shares of those of most influence are highlighted. Overall, it is possible to note that the material-related impact always plays a significant role. In fact, it accounts for more than 80% in all the considered scenarios for CM, whereas it is lower for the AM approach. For the ID 1 part geometry, CM results in substantial energy and CO 2 emission savings, even though the process scraps for AM (m s ) are nullified, as shown in figure 5. Vice versa, for the ID 3 part geometry (which exemplifies a thin-walled component), AM appears to be the best solution for both the considered metrics, even when a certain amount of process scraps is included. Figure 6 plots the results as a function of m s (computed as a percentage value of m p +m a ) with reference to this claim. Specifically, the m s value has been varied from 0% to 40%. The material-related impact as well as the process-related impact have been estimated, together with the impact of transportation. When m s increases, the energy demand and the CO 2 emissions consequently increase. However, under the chosen hypothesis, the impacts of AM are always lower than those of CM. Higher m s values than 105% and 124% would be necessary for the energy demand and CO 2 emissions, respectively, to make the CM process preferable over the AM one.
The impact of transportation appears to be negligible ( Therefore, volumetric and handling differences related to workpiece and powder shipment were not analyzed in detail. Overall, it can be observed that the outcomes are mainly affected by the different amounts of used material and by the different SEC values of the two manufacturing approaches. When moving from ID 1 to ID 3, the machined-off material increases for the CM approach. As a result, the gap in material usage between the two approaches increases. On the other hand, the manufacturing step is energy intensive for AM, especially when a large amount of material has to be deposited layer by layer (ID 1). In fact, the energy demand and the CO 2 emissions for ID 1 attributed to AM largely surpass the savings attributed to the reduction in material usage. On the contrary, when the solidto-cavity ratio decreases, the performance of the CM approach is affected negatively by the larger amount of used material, and, as a result, the AM approach appears to be the best strategy.

Conclusion and Outlooks
A methodology for quantifying the factors of influence for both additive and subtractive manufacturing has been proposed. A CM approach has been compared with an EBM process (followed by finish turning operations) for the production of parts made of Ti-6Al-4V. The proposed methodology does not focus only on the direct energy intensity of the processes, but also includes all the impacts related to the energy and material requirements throughout an LCI analysis. Three component shapes, characterized by different solid-to-cavity ratios, have been considered to analyze the effects of material-related aspects on process performance. The energy demands and CO 2 emissions for AM and CM have been shown to be affected by material usage. When enabling significant material savings, the AM approach appears to be the best strategy from the environmental viewpoint. On the other hand, when a small amount of material has to be machined off, the high energy intensity of the EBM process has a bad effect on the AM performance. Therefore, it is not correct to a priori label the AM or CM approach as energy efficient, given that the component features should also be taken into account.
In the present study, simple part geometries have been assumed as case studies, given that the basic idea has been to analyze the impact related to material usage and, in particular (for CM), to the amount of the machined-off material with respect to the volume of the part. It is worth pointing out that the technological potential of AM has already been shown for the creation of complex geometries or parts redesigned for additive manufacturing (i.e., by means of topological optimization practices). The AM processes might also enable new product design scenarios that could result in a substantial weight reduction. This, in turn, could lead to a further reduction in material usage for AM approaches over CM ones, which could result in further energy and CO 2 emission savings. In addition, the manufacturing process can be considered to have a significant effect on the impacts of the use phase, because it establishes the quality and functionality of the finished part. From this perspective, the decision on which process to select comes down to which process can provide the right balance of quality per unit resources consumed. In this context, the identification of the main factors of influence when all the possible production scenarios (such as material ecological properties, shape complexity, and batch size) are varied becomes a milestone to stimulate the scientific debate on the environmental performance of processes. Such knowledge will enable new green manufacturing guidelines to be defined.