By Jer Faludi

3D printing is poised to revolutionize manufacturing; will it also revolutionize the ecological impacts of making things?  Will it really eliminate waste?  Will it eliminate shipping?  Will it create more problems than it solves?  Some colleagues and I in the UC Berkeley mechanical engineering department set out to find the answers, and what we found was surprising.

Electricity Use Dominates Environmental Impacts

In a nutshell, 3D printers are not necessarily less wasteful; their waste is not necessarily recyclable; their waste isn't even that important compared to their electricity use.

Even if 3D printers did eliminate transportation of goods, it wouldn't matter much--transport is a tiny part of most products' environmental impacts.  

At mass-manufacturing scale, 3D printers have far higher impacts per part than traditional injection molding.  But that also doesn't matter much, because that's not what they're replacing.  They are replacing small custom runs of parts that are machined out of blocks of material. Sometimes 3D printers are more eco-friendly than these machining processes they’re replacing – sometimes not.

Whether you’re milling or doing 3D printing, how you use the tool is the most important factor in its environmental impact. And there are many opportunities for 3D printers to improve, making huge leaps toward greener manufacturing.

How Impacts Were Measured

To see if 3D printing is beneficial or not, we compared it to a computer-controlled ("CNC") mill.  

There are many types of 3D printers; we measured two kinds: 

1) An "FDM" machine  Like a RepRap or Makerbot, rather like a hot glue gun with motors to move it in X,Y, and Z
2) An inkjet 3D printer  Lays down polymeric ink - layer by layer - and hardens it by curing with UV light. An example is the Objet PolyJet technology.

We only looked at making things out of plastic, since FDM and inkjet machines can't do other materials like metals.  Thus these results aren't universally applicable, though I believe they're broadly relevant to most additive manufacturing.

We used life-cycle assessment ("LCA") to compare the two 3D printers and the mill: including the materials and manufacturing of the machines themselves, transportation, energy use, material in the final parts, material wasted, and the end-of-life disposal of the machines.

We tested 22 different scenarios for how these machines are used. To make an apples-to-apples comparison, we made two parts in all three machines and calculated the ecological impacts per part per year.  

Comparing apples to apples: making two different parts on all three machines for a fair test.

Surprisingly, no one has published a comprehensive quantitative comparison like this before.  Most LCAs of these machines only measure energy use, which is not enough.  

If we only measured energy use, the mill would score the best.  But if we only measured material use and waste, the mill would score the worst.  So how do we balance greenhouse emissions, toxicity, air pollution, water pollution, etc.?  

We used a methodology called "ReCiPe Endpoint H" that normalizes and weighs the different impacts (kg of CO₂ and NO_x, ppm of particulate matter, etc.) into generic units called "points".  Single-score methodologies like this are controversial, as they are far from perfect. However, they have been developed over the past 20 years by environmental scientists through peer-reviewed processes and are far more credible than the uneducated guesswork of non-experts. 

The decision-makers who are using and purchasing tools like this (designers, engineers, business executives, etc.) need this simplification into single LCA scores, because none of them are environmental scientists.  To check against bias in the scoring method, we also ran the same analysis with a different method (IMPACT 2002+); the results were nearly identical, strengthening the credibility of the single-score systems. 

Which is Greener? 

The zero-waste myth was both confirmed and busted.  

The FDM machine actually can have negligible waste – but only if your model doesn't need any support material to shore it up while printing.  Hobbyist printers can't even print support material, so they qualify.  

However, the inkjet 3D printer wastes 40% of its ink, not even counting support material (which could be more mass than your final part, depending on geometry and orientation).  What's more, this waste can't be recycled (today).  Other researchers who have studied other kinds of 3D printers have found significant waste in some of them as well.

More surprising than the waste measured was whether the waste even matters.  For traditional machining, material use & waste is in fact the largest impact, but for 3D printing, it is energy use.  

Material does still matter, but it's not dominant.  Reducing the amount of material printed is definitely beneficial. One way to do this is to print hollow parts - and an FDM machine can print parts 90% hollow or more (an impressive feat). However, even in this scenario, the biggest benefit of hollow parts is the associated reduction in energy use, not the reduction of material use.

The manufacturing, transport, and end of life of all the machines (both 3D printers and mill) were a small portion of impacts, amortized away by high utilization.  However, if you only make one part per week, those embodied impacts can be significant for the FDM and the mill.

Overall, we found the inkjet printer had significantly worse ecological impacts than traditional machining, but the FDM was significantly better.  

The variation and uncertainties are also significant, however.  The graph below shows the bars fading out at the tops to show the degree of variation among 22 scenarios. Each tool may score as well as the bottom of the fading, or may score as badly as the top of the bar, depending on the usage scenario and fundamental data uncertainty.  

LCA results for different manufacturing processes  (more points = more environmental impacts) A well-run milling machine could perform better than a badly-run FDM machine, and an extremely well-run inkjet might score better than a moderately-run mill.  Injection-molding is also shown, for scale.

This is not the final verdict, though. This graph is for all the machines producing parts all the time (nearly 24 hrs/day, 7 days/week).  As you can see, the impact variations are between 30% (for FDM) and roughly double (for inkjet and CNC).  The truly big differences in environmental impact are between using the machines a little and using them a lot.  

Any of these tools making a part just once a week, but left on the rest of the time, had roughly ten times the impact of the same machine at maximum utilization.  This 10x difference is obviously far larger than the differences between tools.  

So even more important than what machine you choose is to have the fewest tools run the most jobs (by sharing tools, for example). This not only amortizes the impacts of manufacturing the machines themselves, but eliminates idle-time energy use.  (Note, though, that an FDM machine turned off between runs performs fairly close to the best-case scenarios shown in the graph above.)

How to Print 3D Parts Better

ENERGY

Because energy use dominates 3D printers' impacts, the best way to reduce impacts is to reduce run-time.  Here are three simple strategies for that:

1. Print hollow parts rather than solid.  The hollow version may require support material, but this is often still ok, because support material may print faster, and/or may have the side benefit of being less toxic than model material.  You can test your setup to see if this is true for your printer and materials.

2. Orient parts for the fastest printing. For example, laying a tall part on its side may print faster, or choosing a certain orientation might eliminate the need for much support material, reducing both energy use and waste.

3. Fill the printer bed with multiple parts.  FDM machines get no benefit from this, but the inkjet we tested had roughly the same print time whether it was printing a single part or several parts, cutting its eco-impacts per part nearly in half.  The same may be true for other kinds of printers.

MATERIALS

Choosing good materials can also be important.  Not only do better materials reduce resource use and waste, they reduce toxicity and even reduce energy use.  

Metals obviously require extreme heat to melt together, using more energy than plastics.  On the other hand, some printers can use wood pulp with an adhesive binder, using radically less energy than melting plastics.  While most 3D printers use plastic, literally hundreds of materials are available--glass, starch, plaster, ceramic, etc.

Toxicity may not be obvious, but remember that a 3D printer's melting plastic fumes get breathed in by whoever is nearby, and these fumes are not good for you.  Some plastics are less bad than others.  Find a material safety data sheet ("MSDS") and look for the "NFPA" or "HMIS" numbers.  These are standardized scales of flammability, toxicity, and reactivity.  For the lowest toxicity, you want a zero score; most plastics have scores of one, but many support materials score zero.  If those numbers are tied, you can look for "LD50" or "LC50" numbers (measurements of toxicity in mice), and pick the lowest.

One promising example is PLA bio-plastic: It requires less energy to print (and less energy to manufacture) than ABS plastic; it's also less toxic, and even has better print quality.  Because of all this, it is rapidly becoming a standard 3D printing material for hobbyists.  

The most impressive example of green 3D printing is Markus Kayser's "solar sinter." It is a self-contained cart that fuses sand into glass with the rays of the sun.  He rolls it into the desert, pours some sand in, and the giant lens on top focuses sunlight enough to fuse the sand into glass.  A small solar panel runs the motors and electronics to move the lens and level the sand in the bed.  It uses 100% renewable energy, with non-toxic, local, plentiful materials.

Such examples may be only the beginning--it will be exciting to see how much greener 3D printing can get.

Note from the author: Many thanks to Lawrence Berkeley National Lab's Joint Center for Artificial Photosynthesis and UC Berkeley's mechanical engineering machine shop for the use of their tools.