Chapter 5. 3D printing and its environmental implications

Printing processes

Many 3D printing processes exist – far more than can be adequately described in this chapter. Four of the most popular processes are described here: thermoplastic extrusion, inkjet, light-polymerised, and laser sintering, based on their ubiquity in 3D printing and potential for environmental sustainability.

Thermoplastic extrusion melts a solid filament of material through a heated nozzle onto a platform bed or stage to deposit material in the X, Y and Z dimensions. It includes fused deposition modelling (FDM), also called Fused Filament Fabrication (FFF), which is one of the earliest additive technologies and was the first to be open-sourced. FDM is generally lower-resolution than other technologies, but has the particular advantages of being simple, reliable, and low cost.

Common plastics used with this technology are acrylonitrile butadiene styrene (ABS), the same plastic used to produce Lego bricks; polylactic acid (PLA), a plant-derived and biodegradable plastic often used in food packaging; and polyethylene terephthalate (PET), the same plastic used to create most drink bottles and other common packaging. FDM can also be used with modelling clay, plasticine, rubber-like polymers, and eutectic metals.

FDM can be very low waste, when part geometry does not require additional support material. In principle it can extrude any thermoplastic, even with additives. Extrusion printers can only print one part at a time, and print time directly correlates to the amount of material printed.

Inkjet prints like a 2D inkjet printer, but instead of printing liquid ink on paper, it prints liquid binder onto a bed of powder, and then a mechanical wiper spreads another layer of powder over the bed to print the next layer.

It can print many different materials, from plaster to sugar (Molitch-Hou, 2015) to ceramics (fired after printing) to metal powders (sintered after printing). Experimenters at the University of California at Berkeley have used inkjets to print in sawdust, concrete, salt, starch, and more.

This method of printing can offer very high quality – both high resolution and full colour models. In theory, the process can be almost zero waste, because the powder can be reused and no support material is required. Of the few printing methods which have undergone full LCA, inkjet printing of green materials has shown some of the greatest potential for sustainability (Faludi et al., 2015b). However, experimental materials often do not yet meet quality standards for consumer products. Inkjet printers can print more than one part at a time, and print time correlates more strongly to part height than to the amount of material printed.

Figure 5.1. Experimentally printed part via ink jet

Source: Faludi, J. et al. (2015b), “Does material choice drive sustainability of 3D printing?”, https://waset.org/Publication/does-material-choice-drive-sustainability-of-3d-printing-/10000327.

Light-polymerised printing uses liquid photopolymer that hardens when UV light strikes it. This includes stereolithography (SLA) using a UV laser, digital light processing (DLP) using a UV digital projector, and continuous liquid interface production (CLIP) using a UV projector with modified polymer chemistry. SLA, also known as optical fabrication or resin printing, is the oldest type of 3D printing. DLP generally prints faster, and CLIP is an order of magnitude faster.

A popular hybrid of photopolymer and inkjet methods is the PolyJet system, using inkjet print heads to print photopolymer onto a surface and then hardening it with a UV lamp. All these methods can make high resolution parts with excellent surface finishes, acceptable for commercial products.

SLA usually requires no support material; others vary by part geometry, and PolyJet requires more support material than other printers listed here. For SLA and DLP, unsolidified liquid polymer can be reused, though not infinitely. Polyjet ink waste is not reusable or recyclable, and current machines are high-waste, averaging 43% even before counting support material (Faludi et al., 2015a).

All 3D printing photopolymers today are somewhat toxic in their liquid form, frequently with a Hazardous Materials Identification System (HMIS) health score of two out of four (3D Systems, 2012), similar to many epoxies. However, they are usually considered non-toxic once solidified.

Photopolymer printers can print more than one part at a time, and print time correlates more strongly to part height than to the amount of material printed.

Figure 5.2. SLA (left) and PolyJet (right) printed parts

Source: Faludi, J. et al. (2015a), “Comparing environmental impacts of additive manufacturing vs. traditional machining via life-cycle assessment”, https://doi.org/10.1108/RPJ-07-2013-0067.

Laser sintering or melting binds a powder together in certain places by heating it with a laser. Most machines such as selective laser sintering (SLS), selective laser melting (SLM), and direct metal laser sintering (DMLS), use a bed filled with powder; the laser strikes the top surface to melt the powder together only in specified locations, then a mechanical wiper spreads another layer of powder over the bed to print the next layer. Other laser melting machines spray the powder from a nozzle while heat-fusing it, such as direct metal deposition (DMD) and construction laser additive directe (CLAD).

Laser melting machines can print in thermoplastics, metals, ceramics, or glass. The most common plastic is nylon and the most common metals are steel, aluminium, and titanium; exotic metal-ceramic alloys are sometimes used for aerospace parts.

Unmelted metal or plastic powder can be reused for later prints, perhaps five to ten times (Slotwinski et al., 2014; Dotchev and Yusoff, 2009), though this is often prevented by quality risk. Support material is not needed to fight gravity, but is often needed to prevent warping. When removed, support plastic is usually landfilled, but support metal is often recycled as machining scraps would be.

Laser sintering printers can print more than one part at a time, and print time correlates more strongly to part height than to the amount of material printed. Laser melting machines can have long warm up and cool down times, as well as significant time required for part removal, so there is an incentive to batch operations.

Figure 5.3. A laser sintered metal sculpture

Source: Grossman, B. (2003), “Oldest news”, www.bathsheba.com/artist/news_old_0.html (accessed 1 May 2015).

Table 5.1 presents a snapshot of the different 3D technologies covered in this chapter in addition to details on how they function on their cost curve.

Choosing between print quality and time

3D printing is largely used for prototyping today because high-quality prints remain expensive. However, print quality has improved, and some printer types already print parts acceptable for finished products at a lower cost than other manufacturing methods (e.g. SLA-printed dental crowns) (Bammani, Birajdar and Metan, 2012). Print quality is determined by resolution, tolerance, structural robustness, smoothness of finish, dimensional stability, and other factors. It can also include special abilities, such as the capacity to print multiple colours or multiple materials in the same object. Comparing Figures 5.2and 5.3 to Figure 5.1 shows how SLA, PolyJet, and SLS generally make higher resolution and smoother surface-finish parts than FDM or the experimental inkjet printing in salt.

Improving resolution often increases production time, thus also cost and energy use. For example, doubling FDM resolution means the print head must traverse twice as many layers, roughly doubling print time. Doubling SLA resolution means using a polymer that needs much longer exposure by the UV laser to harden, thus increasing print time. In both cases, the longer print times mean more energy use because the machine runs at similar power consumption for a much longer time. However, this may not be inevitable. A DLP printer with twice the resolution in the projector might not require noticeably more energy than a lower-resolution projector, because that component of the machine is a small fraction of total power use.

Print cost versus time

The largest barrier to high-quality, low-cost 3D printing at scale is long print time, though this barrier is beginning to diminish. Today, a hollow-shell part just 5 cm x 5 cm x 2.5 cm, such as in Figures 5.1 to 5.3, somewhat similar in mass to that of a mobile telephone case, can take hours to print as opposed to seconds in injection-moulding. While the industry has worked to speed up 3D printing, its main market so far has been prototypers, who appreciate but do not require high speed. The shift in 3D printing towards manufacturing has now made faster print speeds a higher priority for printer manufacturers.

Faster speeds are often simply achieved though larger versions of existing printers, so more parts can be printed at once, but some designs are revolutionary. The CLIP printer by Silicon Valley start-up Carbon3D claims to be 25 to 100 times faster than traditional 3D printers, printing the part described above in minutes rather than hours. This increase in speed can be attributed to the printer’s chemistry, which allows a liquid photopolymer to harden continuously as the stage moves, rather than pausing for each layer as all other 3D printer types require (Rolland and Desimone, 2016). Another company, 3D Systems, was developing a printer claiming to be 50 times faster than current speeds in 2014 (McKenna, 2014), though at the time of going to press it had not been released. Such speed increases should greatly reduce the cost per part of 3D printing. Still, it is unclear how much or how fast prices will decrease for customers, because the razor and blade business model is common in the industry. For example, Carbon3D’s photopolymer material may be expensive enough to only improve cost per part moderately, despite its large speed increase.

Considering print size

3D printing is expanding print sizes both to very large objects and very small objects. For very large objects, the architecture industry includes KamerMaker’s canal house in the Netherlands (Wainwright, 2014) as well as WinSun’s concrete panels for a 5-storey apartment building in China (Starr, 2015) and the University of California Los Angeles/Contour Crafting’s partnership with NASA to build a moon base (Khoshnevis et al., 2012). For cars, Local Motors has prototyped an electric car with a 3D printed body, the Strati (Davis, 2014). On the small scale, most micro or nano 3D printing has actually been macro-scale printing of materials with micro- or nanoscale properties (Campbell and Ivanova, 2013). However, there has been actual printing at these scales as well. Hybrid microelectromechanical systems (MEMS) techniques have allowed the production of a robotic bee weighing just 90 milligrammes (Sreetharan et al., 2012), quantum-dot light-emitting diodes have been 3D printed (Kong et al., 2014), and lithium-ion batteries have been printed at the millimetre scale, with near-record high energy density (Sun et al., 2013).

In the coming years there will be more progress in the range of sizes 3D printers can produce at low cost. For economic and technical reasons, this evolution will probably come from the large scale rather than the small scale. Small-scale 3D printing is economically disadvantaged because all nanoscale manufacturing must be mass manufacturing if it is to produce useful amounts of a product. Computer chips often contain billions of transistors (Meindl, 2003), and millions of carbon nanotubes only fill a cubic centimetre (Andrews et al., 1999). 3D printers must become much faster to compete economically in this realm (Li et al., 2011). Technology limits play a part as well: lasers used for small-scale 3D printing are physically incapable of the higher resolution of some other nano-fabrication methods (Li et al., 2011). Hybrids of traditional and 3D printed steps in the electronics manufacturing process can be economic and environmental improvements (Miettinen et al., 2008), but decades will probably pass before 3D printing can fully replace the semiconductor industry for chip fabrication.

Figure 5.4. Local Motors 3D printed car body

Photo © Local Motors, Inc.

Source: Davis, J. (2014), “This is The world’s first 3D printed electric car”, www.businessinsider.com/this-is-the-worlds-first-3d-printed-electric-car-2014-12#ixzz3QB6iqz5x (accessed 15 January 2015); Local Motors, Inc.

Material choice and multiple materials

Expanding material choices allow greater market reach, and the 3D printing industry’s material palette is expanding rapidly. The range of materials available for use with 3D printers include: plastics, metals, ceramics, paper and food. Even living human cells can be 3D printed today. They may also come in different states such as powder, resin, paste, or filament. Low-cost desktop FDMs can print flexible or rigid plastics, plastics that change colour with temperature, and plastics infused with wood fibre or carbon nanotubes (Krassenstein, 2014).

Most 3D printer manufacturers use a razor and blade revenue model, selling relatively high-cost proprietary materials. However, open-source communities and hackers frequently make and share new materials. For example, 3D inkjet users have online communities to share recipes for powder and binder mixes (such as vodka and starch, or sawdust).

Perhaps the most important material expansion in 3D printing in coming years will be the capability to print multiple materials in the same part or alter a material’s properties from one zone in the object to another. Today, it is impossible to 3D print a mobile telephone – one can only print simple components such as the plastic case, then assemble the electronics into the case. In fact, the only whole products that can be 3D printed are materially simple things like plastic toys or metal jewellery. Everything else is printing components. Printing in multiple materials, however, lets a single part perform the role of whole assemblies of other parts. Researchers have printed pain-reducing splints, sound-absorbing chaises longues, and other multi-material products (Oxman, 2010), and some commercial printers can already print 14 materials at once (Stratasys, 2015). Still, industry is probably a decade or more from printing a working telephone, both because of material limits and the print size limits discussed above. Multi-material machines printing wiring within plastic parts have been experimentally proven (Silverbrook, 2004), and may soon emerge commercially. If so, it would greatly improve industry’s ability to replace assemblies with printed parts.

Printing dissimilar materials together severely reduces their recyclability (Dahmus and Gutowski, 2007), which is problematic for sustainability. However, tuning the properties of a single material within the same part could provide an object that is recyclable, and which replaces an assembly of different materials not recyclable together. Similarly, if different printed materials are all compostable, such as some bioplastics or composites, then there is no need for separation (though injection-moulding could equally well co-mould two different compostable bioplastics). Even wiring could be compostable in the future, as conductive carbon inks can already be made biodegradable (Kilner, 1993). The same is true for incineration, where energy can be recovered from mixed plastics.

Industry expansion

3D printing will expand into many more industries in coming years, but it will be limited by a hyperbolic growth curve: easily entering markets with fewer units produced at high or moderate costs, but unable to penetrate markets with many units produced at very low costs. Thus its growth will affect certain market sectors most; for other sectors it will only affect start-ups and small business groups for some time. Its economics will probably have significant social impacts in both developed and developing countries.

Prioritising costs or volume

3D printing has a higher cost per unit than traditional manufacturing methods, but no set-up costs between batches of different product lines; this determines the economics of manufacturers switching to 3D printing. Equipment costs vary widely, but both injection-moulding machines and commercial 3D printers range from tens of thousands to hundreds of thousands of dollars. Assuming that equipment costs amortise away to be roughly equal means that switching from injection-moulding to 3D printing becomes economical when higher costs per part are outweighed by avoiding batch set-up costs. This occurs when:

where N = number of units per batch, C = cost of tooling/set-up for a new batch, C_3D = cost per part of 3D printing N units, and C_IM = cost per part of injection-moulding N units (Askin and Goldberg, 2007). Supply chain costs such as transportation and logistics, which are also saved by 3D printing, can be considered part of the set-up cost for this simplistic model. Obviously the lower the 3D printing cost, the larger the market that should switch, but it is a hyperbolic, not linear, curve, as shown in Figure 5.5.

Figure 5.5. Generalised cost curve of 3D printing versus injection-moulding

Source: Authors’ calculations.

For example, if tooling and set-up for the plastic apple shape in Figures 5.1and 5.2 cost USD 10 000 and the price per part of injection-moulding is USD 0.50, then 3D printing costing USD 5.50 per part (over ten times the price per part for injection-moulding) is still economical for fewer than 2 000 units. But at 100 000 units, the price of 3D printing must fall to USD 0.60 per part for it to be as economical as injection-moulding. At 1 000 000 units, the price of 3D printing must fall to USD 0.51 per part (within one penny of the injection-moulding cost) for it to be economical.

In industry, these cost crossovers can occur at a range of values. While some studies have found cost crossovers in the tens of parts for metal casting (White and Lynskey, 2013; Atzeni and Salmi, 2012), or hundreds of parts for plastic injection-moulding (Bhasin and Bodla, 2014; Sculpteo, 2014), Conner found that for some parts, plastic SLS can already be less expensive than injection-moulding for runs of fewer than 10 000 parts (Conner et al., 2014). Thus 3D printing will continue to expand rapidly from the prototype market to small-run production, as it is currently doing, but will have much more difficulty expanding from small-run production to mass manufacturing, until 3D printing costs become radically lower.

Bhasin estimated that 3D printing costs would fall by roughly one-third by 2020 (Bhasin and Bodla, 2014), which is not enough to approach cost crossover for millions of units. Bhasin also showed that although 3D printing could radically reduce costs of product transportation, logistics, warehousing, and other holding costs, these are only a small percentage of most products’ total cost, meaning that they will not significantly speed up industry takeover. Thus, 3D printing is not likely to replace injection-moulding at the largest scales for at least a decade or two, even after it has become the default production technology for small-run parts. If technical breakthroughs occur this timeline would be shortened, but it will still be limited by simple industry inertia. In the 2D paper printing industry, digital printing became commercially available in the early 1980s, but did not become widespread until the mid-1990s, with many companies taking five years or more from initial use to full integration in production systems (Parnell, 2007).

This conservative assessment should not be taken as an underestimation of 3D printing’s growth curve; many industries produce parts in the thousands or fewer. High-technology industries such as aerospace, defence and certain medical devices often produce parts in the mere hundreds. Some of these parts are already being made by 3D printers more economically than with other methods, as mentioned in the introduction. Mass-customised products such as individually fitted hearing aids and tooth fillings are already cheaper to 3D print because each unit is different even though they are produced in the millions. The potential exists for medical prosthetics and elite sporting equipment to be the next frontier in low-volume printing.

Other industries that have not yet been penetrated by 3D printing, but where this is likely in the next several years include those manufacturing toys, precision machinery, optics and miscellaneous luxury goods. Objects produced in a five- to ten-year timeframe may encompass designer housewares, furniture and clothing. These industries all have what McKinsey calls a relatively high “value density” and “labour density” (McKinsey Global Institute, 2012), meaning they are relatively high value per mass of material, and a relatively high percentage of the cost of manufacturing is from labour, which 3D printing reduces or eliminates. Toys may be an especially likely sector, because they are a highly volatile seasonal market with unpredictable demand for “must-have” items.

3D printing can also increase production instantly to arbitrarily high levels, without the need to manufacture and store goods in warehouses beforehand. Another advantage is the elimination of tooling costs on several production lines on short notice. The production of furniture and clothing are more difficult to predict because most 3D printer research and development is not focused on those markets. One example from the textile industry is Nike’s Flyknit shoe line which uses 3D computer-controlled knitting, and been a great commercial success (Townsend, 2012).

3D printing compared to traditional machining

The next five to ten years will probably see 3D printing supplant much (perhaps even most) machining of parts, but this is unlikely to create large shifts in the environmental impact of the manufacturing sector on a global scale, because machining’s market niche is small. Machining is the main method for prototyping and producing limited amounts of complex-geometry custom parts. 3D printing is already significantly altering these markets for both plastics and metals. For plastics, prototypes and small-run production parts are 3D printed by FDM, SLA, and SLS; for metals, they are printed by e.g. DMLS, SLM, DMD or CLAD. Boeing alone has already replaced machining with 3D printing for over 20 000 units of 300 kinds of parts (Davidson, 2012). However, machining is a small niche – statistics from the US Bureau of Economic Analysis show machining categories to comprise less than 1% of total manufacturing dollars (BEA, 2014).1 Thus, even dramatic environmental improvements from replacing machining with 3D printing will not noticeably reduce manufacturing sector impacts on a global scale.

3D printing generally produces parts with lower environmental impacts per part than machining, but there are many exceptions – some part geometries are more efficient to machine than to print, different printers have very different impacts per part, and even a given printer can have very different impacts per part depending on several factors. For plastic parts, two studies (Faludi et al., 2015a; Faludi et al., 2014) showed that changing printer utilisation causes more variation than the difference between 3D printing versus machining. Machines which predominantly sit idle have far higher environmental impacts per part than machines producing parts 24 hours per day, seven days per week, whether they are 3D printers or milling machines. Thus, maximising utilisation is a top priority for sustainability in 3D printing. These studies also showed that some 3D printers have higher impacts per part than others due to higher energy use, waste, and material toxicity. Finally, comparing the two studies shows that producing hollow-shell part geometry rather than solid-block geometry reduces environmental impacts per part for 3D printing, but increases them for machining. Thus, some solid-block geometries can be machined for less impact than they can be if 3D printed, even if 3D printing is lower-impact for other parts.

For example, Figures 5.6and 5.7, derived from the aforementioned studies, show LCA comparisons of machining via a CNC mill versus several 3D printers, all producing parts 24 hours per day, seven days per week. The LCAs use the ReCiPe Endpoint H method (Goedkoop et al., 2009), which creates a single overall environmental impact score from 17 different types of impact, including climate change, acidification, eutrophication, atmospheric particulates, fossil-fuel depletion, mineral depletion, human toxicity, and others. Colours show what part of the life cycle causes these impacts: shades of blue are manufacturing, transport, and disposal of the printer or CNC mill itself (not the parts created), yellow is electricity use during part creation and/or idling, dark brown is the material used in the final parts produced, light brown is any material consumed that does not end up in the final part (be it model material or support material), and grey is the mill’s cutting fluid and lubricating oil.

Figure 5.6. Environmental impacts of one CNC mill versus two 3D printers, all running at maximal temporal utilisation, making solid parts

Source: Faludi, J. et al. (2015a), “Comparing environmental impacts of additive manufacturing vs. traditional machining via life-cycle assessment”, https://doi.org/10.1108/RPJ-07-2013-0067.

 https://doi.org/10.1787/888933473851

Figure 5.6 shows one CNC mill making solid-block plastic parts, compared to two different scenarios of a commercial FDM printer and three different scenarios of a PolyJet printer making the same solid-block parts. The CNC mill performs better than two scenarios of the PolyJet, and is within error bars of the commercial FDM. However, Figure 5.7 shows two CNC mills making hollow-shell parts, compared to eight 3D printers in 11 scenarios making the same hollow-shell parts. There, the CNC mills always have higher environmental impacts than all the 3D printers (though sometimes within error bars). Figure 5.7 more fairly represents most plastic consumer product parts, such as a telephone case, but a few product categories are solid plastic.

Figures 5.6and 5.7 also show the wide variation among printers and scenarios. Some printers have far lower impacts per part than others, and the impacts change based on part type and machine utilisation. Policy makers should keep in mind that this is not a monolithic field. For example, the inkjet printing in salt has the lowest impact of all machines shown, reducing impact by over 90% compared to machining, but PolyJet printing sometimes scores worse than machining. Unfortunately, today some of the lowest impact machines, such as the desktop FDM, have the poorest print quality. Conversely, some of the highest-impact machines, such as PolyJet, have some of the best print quality.

Figure 5.7. Environmental impacts of two CNC mills versus eight 3D printers, all running at maximal temporal utilisation, making hollow parts

Source: Faludi, J. et al. (2015a), “Comparing environmental impacts of additive manufacturing vs. traditional machining via life-cycle assessment”, https://doi.org/10.1108/RPJ-07-2013-0067; Faludi, J et al. (2015b), “Does material choice drive sustainability of 3D printing?”, https://waset.org/Publication/does-material-choice-drive-sustainability-of-3d-printing-/10000327.

 https://doi.org/10.1787/888933473864

In both graphs above, the environmental impacts of 3D printing are largely dominated by energy use during printing. For CNC mills, however, material waste is a significant, sometimes dominant, impact. Thus, switching from machining to printing will reduce material waste impacts, but may not reduce energy impacts. It is possible to increase energy impacts beyond the savings in waste impacts. The embodied impacts of producing the machines are insignificant for any printer or CNC mill when producing parts 24 hours per day, seven days per week, though this is no longer true at low utilisation (not shown in these graphs, but in the cited studies) (Faludi et al., 2014; Faludi et al., 2015b).

Other studies have confirmed a wide variation in energy use per part across different printer types and scenarios, including metal printers such as SLS, DMLS, SLM, CLAD; they also confirm that while 3D printing is often lower energy use per part than machining, some circumstances can cause it to be higher (Yoon et al., 2014; Serres et al., 2011; Morrow et al., 2007). High utilisation can also be extremely important for metal printers. Baumers showed that energy use during laser sintering is “job-dependent, time-dependent, geometry-dependent, and Z-height-dependent” (Baumers et al., 2011a) and that high utilisation reduced energy use per part from as little as a few per cent to as much as 98%, depending on the printer type (Baumers et al., 2011b).

3D Printing versus injection-moulding

Injection-moulding is the most common manufacturing method for plastic consumer product parts. The potential for major changes in environmental impact due to replacement by 3D printing could be significant if they happen on a large scale. For example, Figure 5.8 compares the environmental impact of injection-moulding (based on standard LCA data) to that of different 3D printing technologies (based on empirically measured data from Figures 5.6and 5.7). Rather than dividing each bar into components of e.g. energy or waste, the graph displays one bar per printer and material type, faded at the end to show uncertainty more intuitively than an error line.

Figure 5.8. Environmental impacts per part of injection-moulding versus several kinds of 3D printing at maximal utilisation

Source: Authors’ calculations.

 https://doi.org/10.1787/888933473878

In Figure 5.8 injection-moulding has lower environmental impacts per part than any 3D printing technology in widespread use today, for large print runs. Injection-moulding’s fixed impacts, such as creating tooling and set-up time, are amortised across hundreds of thousands or millions of parts produced. Desktop FDM of PLA plastic causes 20% higher impacts than injection-moulding of ABS (although this is nearly within uncertainty margins), while one commercial FDM of ABS causes roughly ten times higher impacts. This does not apply to small production runs, such as hundreds of parts (Telenko and Seepersad, 2012); both economics and environmental impacts can favour 3D printing for small-scale production. These results do not include increased exposure to toxic particulates of personnel in offices or homes with 3D printers rather than industrial environments with safety measures (Stephens et al., 2013). Some researchers have shown that 3D printers powered by solar panels or printing in PLA can have lower impacts than injection-moulding ABS (Kreiger and Pearce, 2013), but these arguments may not capture the full picture, because injection-moulding machines can be solar-powered or use PLA plastic with equivalent ease to 3D printers. If they were operated in such a way their impacts would be similarly lower.

The only technology in the graph above that scores better than injection-moulding is 3D inkjet printing in salt. This is a low-energy process using chemistry instead of melting to bind material together. This is an experimental material that currently does not match the resolution or surface finish of injection-moulded parts. However, its score shows an almost 70% reduction in impact per part. A compromise between this method and more conventional technologies could be an environmental improvement over injection-moulding.

The inkjet printing method is well-established and commercially successful, warranting further research and development to improve print quality without compromising environmental impacts. This technology presents low operating costs, because rather than buying an expensive proprietary powder and liquid from the printer manufacturer, the powder and liquid are an open-source recipe of common inexpensive materials such as fine-ground salt, maltodextrin, isopropyl alcohol, and water. While at an individual level expertise is required to prepare the ingredients, at scale such materials would be inexpensive and comparable to the cost of plastic granulate for injection-moulding.

Impacts on greenhouse gas emissions

Widespread replacement of injection-moulding with 3D printing could notably increase or decrease greenhouse gas emissions on a global scale. These changes would probably be gradual. In terms of global resource depletion, 3D printing is unlikely to make significant changes even in the long term.

For greenhouse gas emissions, all industry worldwide accounts for roughly 29% of emissions, with other major sources being buildings, transportation, and agriculture (Ecofys, 2013). Statistics on what percentage derives from injection-moulded plastics are difficult to find, but comparing LCA data from two sources (Hendrickson et al., 1998; Bjorn and MacLean, 2003) suggests plastic injection-moulded products may account for 0.5% to 2% of global greenhouse gas emissions. Shrinking all of these emissions by 70% would be a noticeable improvement, and increasing them ten-fold would pose a severe problem. However, even with these large differences in technology impacts, the global-scale consequences of shifting from injection-moulding to 3D printing will probably occur slowly. As previously described, 3D printing’s cost curve implies that printing will only replace injection-moulding for small-run production in the next several years, taking over moderate production volumes as costs fall and quality improves, with great difficulty (and therefore many years) replacing high-volume low-cost production.

As regards resource depletion, worldwide plastic use for injection-moulding totals approximately 39 million tonnes per year (Thiriez and Gutowski, 2006). The percentage of total world material extraction this represents is uncertain, but for the United States, less than 5% by mass of total material extraction is petroleum products (Matos, 2012), and of that less than 3% is used to make plastics (US EIA, 2014); therefore, all plastic use (including injection-moulded plastic) comprises perhaps 0.1% of total material consumption by mass. As the difference between 3D printing material use and injection-moulding material use are within a factor of ten of each other, switching to printing is unlikely to make significant differences on the global scale.

Green 3D printing misconceptions

One popular fallacy is that 3D printing will drastically reduce externalities related to transportation of goods, because manufacturing will shift away from centralised factories to regional factories or even consumers’ own homes. This is not quite true, and even if it were true, this effect would not have a large aggregate environmental impact. It is not quite true because 3D printers today can only print parts, not whole products, unless those products are extremely simple. Most products must still be assembled in factories, out of a combination of printable and non-printable parts, and shipped to customers. Even for printing 100% on site, feedstock materials for 3D printers still need to be transported, even for multi-material printers that can print whole products rather than just components. 3D printing could indeed reduce transport for businesses selling many different products all made from one material. For example, an automotive repair shop might stock 500 different components made from the same steel alloy, and could radically reduce their transportation impacts by only shipping and warehousing that steel alloy powder to print different parts on demand. Furniture shops might indeed reduce their costs linked to shipping bulky items that are mostly air, instead shipping compact spools of wood-plastic composite.

Even when transport reductions are large, their impacts are usually small compared to manufacturing impacts. Multiple studies (Hanssen, 1998; Hunter, 2013; Apple, 2014) have shown that for most consumer products (be they electronics, furniture, car parts, housewares, clothing, etc.), transportation generates a small percentage of the product’s total cradle-to-grave environmental impacts, with occasional exceptions. Even if 3D printing did entirely eliminate transportation related externalities via molecular-scale local material sourcing (Stephenson, 2003), it would barely matter for most products. Aside from a few exceptions, the environmental impacts of manufacturing and energy use during product life are far more important, making these areas a priority for policy makers.

Another persistent fallacy is that 3D printing will automatically be more sustainable because it “eliminates waste” whether of final products or input materials. Some types of 3D printers can indeed produce parts with almost no waste. For example, when FDM machines print parts with no support material, waste can be a fraction of a per cent of part mass. SLA and inkjet machines can also be run with negligible waste. However, long part geometries with steep overhangs can require more support material than actual model material (i.e. greater than 50% waste). Most 3D prints fall in between these extremes, requiring a small to moderate amount of support material.

However, support material is not the only source of waste; other sources vary by printer type. A PolyJet printer wastes 43% of all its liquid polymer in both model and support material, so a print requiring as much support material as model material would actually cause 65% waste (nearly twice as much waste as the final part). Moreover, liquid PolyJet model material is more toxic per kilogramme than standard injection-moulding plastics like PET or ABS. For plastic sintering, Telenko and Seepersad calculated that “up to 44 per cent of the material that enters the SLS process might be wasted” (Telenko and Seepersad, 2012). For metal sintering, Kellens reported unused powder losses of 20% of part mass (Kellens et al., 2011), with the rest reused. Even some FDM printers generate excessive waste through poor design choices. For example, the 3D Systems Cube desktop FDM printer uses proprietary cartridges of printing filament. The cartridges are produced from two types of plastic over-moulded, which renders the cartridge unrecyclable (Grenchus et al., 1998). Since the mass of plastic comprising the cartridge is greater than the mass of printing filament within, this causes over 50% waste even if the printer is operated at perfect efficiency. Such situations are rare, and some manufacturing methods produce more waste. Machining a hollow part from a block of plastic typically creates large amounts of waste, often over 80%, therefore 3D printing can be generally assumed to cause less waste compared to such methods. However, mass-manufactured plastics are typically moulded, not machined.

Injection-moulding is an efficient process: waste or scrap rates are reported ranging from 10% (Thiriez and Gutowski, 2006) to 5% (Olmsted and Davis, 2001) to 1% (Frischknecht et al., 2005). Thus, while some 3D printing wastes less than injection-moulding, some wastes significantly more. Even if waste were always reduced, it would not always be important. As shown in Figures 5.6and 5.7, energy use is by far the largest environmental impact of 3D printing. Increasing energy use per part produced can overwhelm material savings to increase total environmental impacts. When printers do reduce total environmental impacts, they do so by combining low waste with low-impact material choice and low-energy processing.

Aligning economic incentives

The clearest way in which 3D printing is likely to improve manufacturing sustainability is by aligning economic incentives with environmental impacts. In traditional production, both prototyping and mass manufacturing, complexity of design is more expensive than materials or energy. This has led to material and energy inefficiency when producing parts. In 3D printing, by contrast, materials are expensive, but design complexity is free. In the coming years this should lead to parts made complex to save material. Software already exists to optimise part geometry for minimum material mass while fulfilling specified physical requirements such as stresses of given directions and magnitudes (Within, 2011). Energy is less expensive, but for printer operators, energy use is proportional to print time, effectively causing energy to be expensive and incentivising printer operators to save energy through the complexity of print setups. However, there is no such incentive for printer manufacturers to design energy-efficient printers. At the time of going to print, there was not a strong correlation between power use and print time within a given printer type (Yoon et al., 2014). Thus, the incentive to save energy is weaker than the incentive to save material mass.

Legacy product repair

The most important way that 3D printing could encourage material reuse is not through recycling, but through repair. Many legacy products are discarded because a single component breaks and there is no longer a supply chain to replace it. These components are often single-material plastic or metal parts, which 3D printing is already well suited to replace. The primary barrier to ubiquitous printing of replacement parts is intellectual property (IP): both the accessibility to end users of 3D models for printing, and the legal right to print models that are available. Some websites such as Thingiverse are already open repositories for free downloads of 3D models to print replacement parts for dishwashers, refrigerators, bicycle accessories, and more. The 3D models are legally ambiguous, arguably not violating IP laws because they are reverse-engineered, but challenges came as early as 2011 (Coetzee, 2011). Businesses may learn to monetise such sites for IP rights, just as Napster’s illegal music sharing evolved into paid subscription services such as Pandora or Spotify.

Policy could easily encourage 3D printing for repair, as several companies are already in a position to provide this service, including billing and royalty payments. It is merely a question of legal rights and contracts. Thingiverse could charge a fee for users downloading the file to print at home, and send royalties to the original product manufacturer. Companies such as Shapeways, which provide a web-based interface to mail-order 3D printed parts, do not even require the end user to have a printer in their home. It is unclear whether such markets will be large enough to avoid significant amounts of discarded products, as repairs require not only parts, but labour – often skilled labour. Still, it is a worthwhile experiment, as it requires no new technology or investment, only legal clarity for existing entities to engage in such practices.

Saving use-phase energy

It may be that 3D printing’s most important contribution to sustainability will come not from manufacturing impacts, but from saving energy in the use-phase of products’ lives. This is speculation, as such savings may only accrue in aerospace, automotive, or other industries where use-phase energy impacts dominate the entire product life cycle, and where such energy use depends heavily on the mass of material in the product.

The higher complexity of manufacturing enabled by 3D printing can allow for lower-mass parts, which can save fuel in transportation products. For example, General Electric’s 3D printed parts for the jet engine described earlier reduce weight, improving fuel efficiency by 15% (Beyer, 2014). Even if the manufacturing of the parts themselves has a greater environmental impact than previous methods such as machining or welding, these large use-phase energy savings can easily overcome increases in the manufacturing phase. Such weight savings would also reduce lifetime impacts of trucking and cars, but would not reduce impacts for other product categories, such as consumer electronics or buildings. However, there are other potential routes for use-phase energy reduction.

The higher complexity enabled by 3D printing can also improve the efficiency of fluid flow and heat transfer for some systems. For example, the SpaceX SuperDraco engine achieves more efficient fuel combustion by having its “regenerative cooling” fluid channels within the walls of the combustion chamber, rather than welded to the outside, as was required for previous manufacturing methods (Post, 2014; Dankhoff, 1963). Use-phase energy efficiency may even be enabled for other sectors, such as architecture. Roughly 90% of an average building’s lifetime environmental impact is from its energy use (Faludi, Lepech and Loisos, 2012), and air leaks through gaps in construction are a major culprit (Liaukus, 2014), especially at corners and joints. Though it has not yet been demonstrated, in theory printing all of a building’s walls as one continuous piece could help avoid such leaks.

Social benefits and the need for green 3D printing to scale

One of 3D printing’s sustainability impacts may be social, not environmental. Economic empowerment is a significant part of sustainability. 3D printing empowers small businesses and others with limited access to capital by radically reducing economic barriers to entry into manufacturing. Anyone with a moderate-quality computer and CAD software can create 3D models for parts, then have the parts printed in high quality without the expense of tooling a mass-manufacturing line. Printing is often even less expensive than traditional forms of prototyping, such as machining (Beyer, 2014). Such empowerment of small businesses could help to reduce income inequality in developed countries, and help developing countries to industrialise.

The widespread replacement of injection-moulding for mass manufacturing would probably be an environmental problem with current 3D printers, but the replacement of machining for hollow-shell parts would probably bring about environmental benefits if two conditions are met: greatly reducing energy use per part printed, and reducing the embodied impacts of printing materials. These goals can be achieved through many strategies, e.g. maximising utilisation, eliminating toxic chemistry and reducing material waste. Table 5.2 lists some specific strategies for these goals by printer type.

Table 5.2 shows that because 3D printers vary so much in their design and operation, few sustainability strategies apply to all printer types equally. Reducing energy use per part made is the top priority for sustainability in 3D printing, but it can require different strategies for different printer types. One energy-reduction strategy common to all printers is minimising idle energy. Idle energy can be reduced both by having printers automatically enter low-power sleep states when they finish printing, and/or by sharing fewer printers among more users. Also common to all printers is simply designing them for energy efficiency. The relative low cost of electricity today means that many printers lack simple energy-saving features such as insulation around heated components. The final energy-reduction strategy common to all printers is reducing print time. Many 3D printers use the majority of their electricity to simply keep support systems running, the additional power for printing is often small (Faludi et al., 2015b). Thus more parts could probably be printed with less energy per part if print times were reduced, even if more print heads, lasers, etc. were required.

A material impact reduction strategy common to all printers is tuning material properties, as (in theory) all types of 3D printers should enable this in the future, albeit by different means. Much of the table lists technology-specific strategies. For extrusion printers, energy savings correlate to material savings, so hollowing parts and minimising support material save both material and energy. Extrusion printers can also significantly reduce energy use by chemical solidification rather than melting plastics. Photopolymer, inkjet, and laser sintering machines can reduce energy per part by maximising the number of parts per print, because they can print several parts using almost the same energy as printing a single part, regardless of material use. For material impact reduction, most printers can minimise support material, but inkjet printers do not generally need support material, so it is not a priority for them. Reducing toxicity is a higher priority for photopolymer systems than others because non-toxic materials already exist for other printer types. Compostable biopolymers are one option for photopolymer printers because photopolymers are inherently unrecyclable, and incineration with energy recovery is another. Compostability is already possible for some inkjet and extrusion printer materials, though today options are limited. Development is required to expand commercially available choices and improve physical performance of such materials, if they can be produced at competitive cost and ways can be found to deal with their potential to disrupt existing recycling systems. Metals are not generally compostable, so metal sinterers should improve material recyclability, both in terms of powder reuse and producing powder from recycled metal.

An example of 3D printing’s potential is the Solar Sinter by Markus Kayser. It functions similarly to SLS, but instead of sintering plastic or metal powder with a laser, it uses a large Fresnel lens to sinter sand into glass with the heat of the sun. Its motors and electronics are powered by solar panels and the sand used is non-toxic, abundant, local, and requires no energy-intensive processing before use in the printer. The Solar Sinter was merely a demonstration project, probably not practical for industry, but it can serve as a beacon for what is possible. Even without such extreme measures, some commercial 3D printers today are capable of printing non-toxic, abundant, renewable, compostable materials with low-energy printing processes – close to the ideal for sustainable manufacturing. For example, inkjet printing in salt or sawdust, or MIT’s WBDF.

Policy mechanisms

This section outlines the priority areas to improve the environmental implications of 3D printing. Table 5.3 highlights some of these areas. The table lists high, medium and low-priority goals for printer design, materials, printer operation, IP and use-phase energy efficiency. These priorities serve the goals of minimising energy use and waste, as well as transitioning to materials which are non-toxic, abundant, renewable, and low-embodied-energy, with useful end-of-life.

The 3D printing industry is at an inflection point, meaning that interventions implemented now rather than a few years from now would have greater impact. Some possible interventions include banning specific practices, taxes, subsidies, certification systems (eco-labels), and preferential purchasing programmes. Interventions in other industries could also fix impacts in the 3D printing industry: since its largest impact is from the electricity to print parts, converting national electrical grids to 100% renewables (as is already being done in some countries) could cut printing impacts by perhaps 75%. Of course, such measures would also improve other electricity-intensive manufacturing methods. 3D printing industry interventions should primarily target printer design, materials, and operation. As mentioned above, interventions encouraging 3D printing of parts in industries such as aerospace in order to reduce use-phase energy demand should target those industries, not the printing industry, because 3D printing is just one possible means to improved fuel efficiency. The following sections consider several policy mechanisms for directly influencing the 3D printing industry.