Why isn’t 3D metal printing more popular yet?

With the recent news that GE offered a total of $1.4 billion to buy two industrial 3D metal printing companies, 3D metal printing has been getting a lot of discussion. Spencer Wright has a great writeup.

Spencer’s analysis is that GE is making moves toward offering a complete array of industrial 3D metal printing solutions, including the various software packages that are part of an industrial 3D printing workflow.

GE is betting that metal printing will replace more traditional processes like machining, in which one starts with a block of material and removes the undesired stuff, or fabrication, in which one makes or buys, say, tubes, and welds them together. And, come to think of it, those traditional processes  sound so…inelegant compared to the efficiency of just printing whatever design one’s heart desires. So if metal printing is mature enough for GE (not known for high-risk maneuvers) to invest in it, why isn’t everything metal made by metal printing today?

1. Metal printing is different from plastic printing 
Unlike plastic printing, metal printing remains a largely industrial affair. The machines cost roughly $250,000 and up. There’s been no shortage of attempts to develop a consumer-grade, desktop metal printer, but none have had success even remotely approaching that of Formlabs and MakerBot. Meanwhile, plastic printers can now cost under $500.

The optimist in me says this is simply because the market is young and new. As one metal printing CEO put it,  “We’re in the early, early days of 3-D printing. If you think of aviation, we’re at the World War I stage, when planes were made of canvas and wood. This is an industry with a very long arc.” The pessimist wonders if the technical differences between metal printing and plastic printing will keep the price point too high for widespread consumer market penetration. As MakerBot co-founder Bre Pettis put it, There’s things that happen when metal melts. Aluminum you [melt] around 700 degrees. Does your oven go up to 700 degrees? You probably don’t want that in your house. (He must’ve meant Celsius; pure aluminum melts at 1,221° F, which is 660° C.)

Pouring the metal

2. Metal printing cost-benefits
Metal printing is neat, and will only become more commonplace and important with time. That said, it may be like autonomous vehicles in that it’s been a little overhyped; while the potential for disruption is huge, the “3D printed everything!!!1!!” future is a little further away than one might be led to believe based on the media coverage.

That’s because, for the moment, for most manufactured metal parts, 3D printing doesn’t offer enough benefit to justify rapidly displacing more traditional metal manufacturing methods. The chief benefits appear to be:
  • It’s possible to 3D print geometries that can’t be produced economically by other methods (machining, casting, forging, stamping, fabricating, etc.)
  • Material savings, because the printed part is closer to net shape and therefore uses only the minimum necessary material, where traditional methods produce more scrap (e.g. chips in machining and flash in forging).

(I’ve heard some chatter about an advantage in mechanical properties, i.e. that it’s possible to produce 3D printed metal parts with better hardness/strength/toughness/etc. than would be possible with other methods, which would be interesting for some specialized applications, but not for most mainstream applications that currently use cheap materials like mild steel or 6061 aluminum.)

At the moment, for most metal parts, material costs are insignificant. The cost of raw materials is such a small percentage of the cost of goods sold (COGS) that even relatively large changes in material costs have a relatively tiny impact on overall part COGS. There are a few reasons for this:

  • Metals are amazingly cheap. My own rule of thumb when buying steel is to budget $1/lb, and that’s for relatively tiny quantities; manufacturers buying thousands of pounds are probably paying far less. When you buy a block of, say, aluminum, you get that exact amount of usable aluminum, unlike, say, wood, where it’s not unusual to throw away 30% or 40% of a board. And metals are easy to store; they aren’t prone to distortion and they last forever (steel rusts eventually, but the oxidation rate is pretty slow, especially indoors).
  • Labor is so expensive that every other cost is small by comparison. In the welding industry, for example, labor generally makes up ~70% of costs. Welding filler metals typically represent only a small portion of overall welding costs, despite being relatively pricey as metals go.
  • Scrap metal isn’t necessarily waste. Industry magazines are full of ads for devices to convey, compress, and clean scrap chips, all to improve their sale price. I have a buddy who used to work for GE in supplier qualification. He qualified one supplier that did a lot of high-end machining of massive, intricate, expensive nickel castings used in GE jet engines. They told him they only broke even on the machining services; the profit came from selling their chips.

3. Beachead applications
Given the insignificance of material costs, the most important benefit of 3D printing in most applications is likely part geometry flexibility. The applications where this matters are being converted to printing.
Typically these beachead applications will be where:

  • costs to machine the part would be high, e.g. because
    • part geometry is complex or
    • the material is difficult/expensive to machine;
  • production quantities are low;
  • part volume is low (i.e. the part would fit in a relatively small box);
  • mix is high (i.e. there’s variation between similar parts, as in medical implants);
  • material costs are high (see above); and/or
  • weight is important (it can make more sense to print a part to be a lighter, rather than paying to machine away more material)

Aerospace is rife with these applications, and indeed SpaceX and Blue Origin reportedly use printed parts in their rocket engines. Another field that comes to mind is medical implants, especially those rendered in titanium, which can be a pain to machine..

I expect most of these low-hanging-fruit applications will be converting from machining, as it’s a process also generally well-suited to complex geometries, low production volumes, etc.

Automated turret punch press

4. Inertia
Even in applications where 3D printing demonstrably makes more sense than a traditional process, and especially outside specialized fields like aerospace and medicine, traditional processes will persist. Manufacturing is generally a change-averse, risk-averse industry. Traditional methods are traditional for reasons that matter, especially to the accountants:

  • Costs like maintenance and tooling are well understood.
  • Manufacturing machinery, like machining centers and turret punch presses, is expensive; it’s often either A. still being paid off, and therefore needs to be used frequently in order to justify its purchase, or B. been paid off already, in which case it generates (relatively) pure profit; either way, the accounts want the machinery used until it’s dead, which usually takes decades.

Perhaps more importantly, the large, well-resourced companies that make traditional machinery and tooling have long-established relationships with their customers, increasing the costs of entering these markets.

Training ≠ Manufacturing

A factory in Faritabad, India, I visited in January 2014. This is not a training environment.

Recently I was at a customer site in the Midwest. These folks do…a lot of welding. They’re consistently one of the biggest 3 or 4 buyers of steel plate in the country. They buy so much plate that they don’t have to buy it in standard thicknesses like 1/2 inch or 3/8 inch; rather, they buy it in whatever size their engineers decide is best. One employee told me they can buy custom thicknesses like .433 inch cheaper than most companies can buy normal sizes like 1/2 inch. They buy a lot of plate, and then they do a lot of welding on it.

This site (one of dozens) is massive–easily the size of a couple of football fields. I was taken at one point to the “training area” at this plant. The training area felt like they had picked a random corner of the factory to stick it in.

It was loud. Not from the welding or the training that was happening, but from ambient noise.

It was poorly-lit. That is, there was more than enough light to see things happening around you, but nowhere near enough light for detail work, and welding training is most certainly detail work.

It was filthy. Not just in the sense that they needed to sweep up. More like there was just a lot of visual clutter around–old welding machines, spools of wire, work in progress, none of which were part of the welding training process.

Here’s the thing. Welding training is not welding. It’s not even manufacturing. It requires a different approach to do well.

The goal of a training center is to mold minds, not make widgets. The goal in a production environment is, primarily, efficiency; the main concerns are things like deposition rates and defect rates. The goal of a training center is to take untrained humans and turn them into (better-) trained humans. The goals are different, so the environments need to be different.

In a production environment, there shouldn’t be a lot going on besides production. As Peter Drucker put it, “A well-managed factory is boring. Nothing exciting happens in it because the crises have been anticipated and have been converted into routine.” The factory environment, then, doesn’t really need to foster communication beautifully. Communication isn’t the main priority; welding is. Yes, welders need to communicate, but not every second of the day, or even most seconds.

In a training environment, on the other hand, communication is everything. Training is communication. Communication moves the training process forward, and a lack of communication will hold it back.

So it’s imperative that a training area be quiet.

Similarly, it’s important for trainers and trainees to be able to see what they’re looking at. Details often matter in welding, so one might argue that good lighting is always necessary. But experienced welders can produce good work even under sub-optimal lighting conditions; this is in fact one way to define a welder as experienced. In my experience, inexperienced welders require good lighting to even do mediocre work.

There’s a school of thought that suggests that trainees should have to endure hardships identical to those they’ll experience on the job. I agree. In fact, the ideal training environment might expose trainees to hardships worse than those they’ll ultimately experience on the job. But, those hardships must be introduced gradually. Training is inherently stressful; adding hardships from the start only increases the dropout rate and makes the training take longer.

To sum up, let me contradict myself: in a sense, training actually is manufacturing, if we define “manufacturing” broadly. The goal is to produce something, even if that something is very different from a typical manufactured widget. As Peter Drucker said, we still need to anticipate crises and convert them to routine…it’s just that training involves a vastly different set of crises and routines. To be effective, training areas must reflect that.