Manufacturing guy-at-large.

Desk fab

Added on by Spencer Wright.

This weekend, in addition to setting up an Abaqus beam sizing optimization on my lattice bike stem, I got about 80% of the remaining work on my new desk done. Completing it (which I should get done in the next week) will be a long awaited upgrade to my home office, and I've put a lot of research and care into selecting my materials and designing a system that's functional, lightweight, aesthetically pleasing, and modular. The structure is based on unidirectionally would carbon fiber tubing - the same stuff that I use for my titanium-carbon road bike seatpost

I cut the tubing to length with a 10" diamond grit blade, and use a piece of aluminum oxide wet/dry cloth to knock down any slag (stray, partially cut fibers). I then wiped the tubes down with tack cloth and used a two-part high gloss clear coat to seal and protect the tubes. The result looks great - it's hard, smooth, and allows the fiber to shimmer when it catches the light. 

The last pieces to prep before assembly are the frame's nodes. These were printed on a Form 2 in their "Tough" resin, which most people would use for functional prototyping but which I'm planning to use indefinitely. I considered leaving the nodes in their natural blue-green (you'll note that I even changed the color on the parts in my model a few months ago), but ended up deciding to paint them black. 

I was a bit hesitant to take on a finish paint job - I've been painting more and more recently, but my spray skills are mostly untested on small, intricate parts - but Formlabs has a good two part (one, two) guide and anyway I didn't have any other options. I used the Tamiya primer and spray paint they suggested (a note on this: Tamiya's bottles are pretty small. I ended up buying three bottles of primer and five bottles each of black topcoat and matte clear.), which goes on soft and easy - much nicer than the big rattle cans that I'm used to.

I've still got to sand, prime and paint three more of the nodes, but once that's done I'll be able to assemble the entire desk and start attaching accessories - which, because I'm particular about my workspace, will be more complicated than maybe most people's setups :) 

More soon!

Snapshots of a day

Added on by Spencer Wright.

Some semi-random screenshots from a day's worth of lattice design in nTopology Element Pro:

There have been a bunch of big updates to Element recently, and this workflow takes advantage of a few of them. In particular, the new Warp to Shape tool is very helpful; I also used the Extract tool and the Remesher to make some nice selective surface lattices.

The last, and really the biggest, thing here is the conversion into Abaqus for beam analysis & sizing optimization - see the last photo above. I'll be working on & posting more about that in the next few days - stay tuned :)

An ode to VHB tape

Added on by Spencer Wright.

I mentioned VHB tape in a post a few weeks ago, and last week a friend, who had read it, looked at me quizzically and asked "what's up with VHB tape?"

The thing is, VHB tape is awesome. It's an industrial strength, double-sided foam adhesive, capable of creating *really* strong bonds between basically anything that you'd want to stick together. I've used it in a few contexts, and encourage anyone who takes hardware projects seriously to keep VHB tape in mind as an alternative to welding, bolting, or gluing parts together.

I first became aware of VHB tape when building bikes, and used it (as part of 3M's "mushroom head" tape, which is the same stuff that EZPass uses to stick sensors to your windshield) to create a modular front rack system that didn't require bungee cords. Later, when I was working on robot doors, I seriously considered using VHB tape to bond mahogany cladding to the doors' aluminum frames; we eventually decided on structural adhesives, and dealt with a slew of issues resulting from differences in thermal expansion rates.

Most recently, I'm using a small amount of VHB tape to adhere the 8020 subframe to the phenolic resin top of my desk. I chose a thin tape - 1/2" wide and .020" thick - and am using it on the short ends of the desktop as a secondary fastening method (the primary connections are two M5 bolts). I'm using it here for a few reasons, which highlight VHB's advantages:

  • It's really strong. Not as strong as a structural adhesive like DP420 (with its overlap shear strength of about 4500 PSI), but for applications like the ones above, VHB's ~100 PSI dynamic tensile and 1000 PSI static shear strengths are *well* within my design objectives. 
  • It's really easy to apply. In most cases, isopropyl alcohol is all I've needed to prep the surface, and it's kind of nice to not have to manually mix up epoxy or get out a mixing gun. 
  • It stretches to accommodate differences in CTE. In this way, VHB is *much* better than rivets or nuts/bolts, and has big advantages over structural adhesives as well. According to 3M, VHB tapes can be stretched up to 50% of their thickness, making them ideal for applications where dissimilar materials need to be bonded, or where some shock absorption & vibration dampening is desired.
  • It has a predictable thickness. Glue is great, but planning for a consistent glueline can be tricky. VHB is available in a variety of thicknesses, allowing you to design the assembly just the way you want it.
  • It's relatively easy to remove. Disassembling a glue joint can be a major pain in the ass, but VHB is just acrylic foam; if you can fit a knife into the joint, you can usually cut the tape in half and pull the assembly apart. 
  • It's easy to keep in stock. Glue is great, but inevitably I end up with a hardened tube laying around, or run out of mixing nozzles right when I need them. VHB doesn't need any supporting equipment, and is shelf stable for two years - whether or not you've opened the package. 

Don't get me wrong - I keep a big selection of bolts on hand, and usually have a active tubes of cyanoacrylate, Titebond, and DP420 on hand. But my shop is much better for having VHB in stock, and my desk is much better off for having used it :)

Recent lessons, pt. 2

Added on by Spencer Wright.

More lessons from the past few years and beyond - these more personal than my previous post. 

  1. Few things feel better than getting consistent exercise, but few things feel worse than being told that you should be getting more exercise.  
  2.  There's nothing more fulfilling than setting someone else up to do what they love. Corollary: few things hurt more than to be denied the chance to do something that you're great at. 
  3. People will hate you for saying that having a baby is like having a dog - and for sure, there are significant differences - but that doesn't change the fact that it's true.
  4. Unless you have access to an actual soda fountain, you should add the root beer to the ice cream and NOT the other way around. 
  5. The ability to give advice well is a direct outgrowth of understanding and judgement; the ability to receive advice well is a direct outgrowth of humility and a sense of politics.

Desktop fabrication/assembly

Added on by Spencer Wright.

In between bouncing Nora around the past few weeks, I've been able to sneak in a little work on my new desk. I now have both the desktop (a 30" x 60" x 1" phenolic resin surface from JHC Lab Resin) and its subframe (a 20 mm x 40 mm extrusion frame from 8020), and have partially assembled them to each other:

The subframe only attaches mechanically to the desktop in two places - one bolt in the middle of each of the long frame legs. I *think* I'm also going to use VHB tape between the frame and the desktop, but I'm holding off on that until I get a few more of the parts in & prepped.

I also received the shelf (which is 12" x 48" x 3/4") and all of the carbon fiber tubing. I'll be cutting the tubing to length with a diamond grit saw blade, and need to drill & tap a bunch of holes in the shelf still as well. I'm also planning on mounting my old Wilton Bullet vise to the desktop using some nut inserts, but I'll probably hold off on drilling more holes in the desktop until I get the whole thing together and use it for a few days.

I should be getting the printed parts back from Formlabs in the next week or so, and because of the way it assembles I should be able to get the desk put together really quickly once they arrive. Fun :)

Recent lessons

Added on by Spencer Wright.

I've been on paternity the past few weeks, and have spent some time reflecting. Below are some lessons - thoughts, really - that might have been useful had they occurred to me earlier. Note: The list below is decidedly incomplete; your mileage may, also, vary. 

  1. It's really hard to overcome a bad say/do ratio. Some people prefer to use ambitious time estimates as an incentive to work harder, but I've never seen that approach work. Do what you say you're going to do, when you say you're going to do it.
  2. If you're not being listened to, stop talking. I learned this first while training my dog. Being ignored doesn't help anyone.
  3. Strong argumentative skills beat weak argumentative skills, but it's only a debate if the other person cares.
  4.  "No" is a vastly underrated  answer. One would expect it to be right about half of the time, yet most people (citation needed) are shocked when someone actually says it. 
  5. Long games are really hard to win if you don't pace yourself. Know what kind of game you're playing, and calibrate your patience level accordingly.
  6. Jobs only add up to careers if you're capable of sacrifice. If not, you'll probably need a business of your own. 
  7. The satisfaction of watching Michael Clayton again never really translates into material effects on the rest of your life. Escapism has diminishing marginal returns. 

Distributed Manufacturing

Added on by Spencer Wright.

Note: This draft was started about a year and a half ago. In the name of valuing what's shipped more than what's (theoretically) perfect, I publish it now with considerably less preciousness than originally planned. 

When I took these pictures - on the street in Dongguan, PRC, in the summer of 2015 - I was thinking about the emphasis that American startup culture has placed on distributed manufacturing over the past few years. According to the narrative, distributed manufacturing is being enabled by a combination of 3D printing, streamlined digital documentation standards, and web/mobile outsourcing marketplaces. Through these, we're ostensibly moving towards a paradigm that offers unparalleled improvements in efficiency, variety, and speed-to-market.

Parts of this narrative may well be true. I'm certain, however, that neither additive, nor the model-based enerpeise, nor any digital matchmaking service is a prerequisite for distributed manufacturing. Really, all you need is real estate and some demand for (in this case) overnight EDM and machined parts. 

I tell you, seeing this was really breathtaking.  

Desk update

Added on by Spencer Wright.

I was able to spend a little more time on my desk this past weekend, and got it nearly done:

A quick changelog:

  • Added a frame to the desktop. I'm planning on using 1" phenolic resin (or *maybe* epoxy resin, if I can afford it) for the surface, and it needs to span almost 5'. At that length it seemed like a good idea to support the span, and I'm using an 8020 frame to do so. It's not the cheapest option, but means that attaching the surface will be easy. It also has the advantage of allowing me to attach other accessories (a power strip, my monitor stand, an architect's lamp or two) really securely as well.
  • Added adjustable feet. These are nylon with a rubber pad, and they thread into nut inserts that'll be installed in the leg caps. 
  • Added shelf clips to hold the lower shelf onto its supports. I think that the shelf will just rest there - the clips should snap more or less securely onto the supports, and I don't think I need to attach them any more firmly than that.
  • Other small changes to the leg brackets. I had an early prototype printed by Form Labs in their "Tough" resin, and was pretty impressed with the results; the part came out true to size and clamps onto my carbon fiber tubing really well. I'm having a few more parts printed now, and will rig up a bigger test assembly to confirm - but with the combination of my leg geometry and the Form Tough's high tensile strength and elongation at break, I'm confident that it'll work well.

Time allowing, I'll order the rest of the stock and cut-to-order parts in the next week or so. Pretty excited :)

DMLS lattice sample prints

Added on by Spencer Wright.

I'm *very* excited about these parts from C&A:

These parts were printed in titanium 6/4 by C&A Tool in Churubusco, Indiana; they were designed in nTopology Element. 

This is a pure lattice structure - the entire geometry is designed as beams and nodes, with no explicitly defined solid regions. The beam lengths are on the order of 2-3 mm; their thicknesses range between .45 mm and 1.1 mm. In some areas (for instance, the bolt holes) this results in a fully solid part, but the transition from lattice to solid is continuous rather than discrete. The result is a structure that's solid where it needs to be and sparse elsewhere, with no stress risers where solid and lattice meet.

The parts are, of course, sample regions of the bike stem that I've been working on for some time now. The intent of the samples was to prove the printability of the structure and identify any potential difficulties. The results were overwhelmingly positive: With the exception of a few small flaws, the parts printed very well, and I believe the problematic areas can be addressed in the design pretty easily.

Given the good quality of the sample prints, I'm planning on printing a full version of the part soon. I'm also experimenting with a few other design variations (intended for a variety of different metal AM machines), and am running them through a beam sizing optimization process with Abaqus and Tosca in order to reduce mass and decrease strain energy. More on these soon :)


Thanks to Rich Stephenson for his ongoing help on this project - and for continuing to educate me on the metal AM industry.

Build process simulation

Added on by Spencer Wright.

Early last year, Andrés Bellés Meseguer reached out to me with a proposition. He had read my piece in Metal AM magazine, and wanted to use my printed parts from DRT Medical Morris to verify a build process simulation workflow that he created using Abaqus. I agreed, and with Dave Bartosik's help I got him the build files necessary to simulate the print.

Andrés' full results were published in a paper titled Prediction of Distortion of a Titanium Bike Part Built by DMLS, which he presented at a NAFEMS conference in November. The simulation used a fine hexahedral mesh at the part itself, and a coarser mesh for the surrounding powder bed and the build platform. At each timestep, heat (representing energy applied by the laser) is applied to nodes throughout the model; it then dissipates throughout the structure. Below, see a thermal map of the part about 70 minutes into the build:

Image courtesy Andrés Bellés Meseguer and Prime Aerostructures

You can also use this simulation to model distortion in the part - seen here at the end of the build:

Image courtesy Andrés Bellés Meseguer and Prime Aerostructures

The distorted areas in the simulation correspond well to the as printed part, but Andrés notes that the magnitude values don't match perfectly; it's likely that some of the discrepancy can be narrowed by adjusting thermal coefficients.

This field - simulating additive processes to predict and compensate for built in stress and distortion - is one that I've been excited about since I began working with AM. Thanks to Andrés for sharing his work - I'm looking forward to more progress on this soon.

Desk update

Added on by Spencer Wright.

Some changes to my desk design:

In no particular order:

  • I rebuilt about 90% of the model so that it's driven off of an Excel spreadsheet. It took a little while, but makes updating tubing diameters *so* much faster.
  • The tabletop and shelf are designed to be phenolic resin; the shelf doubles as a footrest. I'm a bit worried that the long span on the tabletop will result in bounce, but that'll be relatively easy to fix if it happens.
  • The frame tubes are filament wound carbon fiber throughout.
  • The lugs/brackets are all slotted; they'll be printed. The band clamps are a little bulky for my taste, but the practicality of making the frame modular is just too appealing to me. Plus, I'm pretty sure they'll distribute the clamp force evenly and be plenty strong to hold the frame together.
  • My current plan is to drill and countersink the tabletop and then use flat head screws with washers & nuts to fasten the leg brackets. 
  • The legs are angled, and I'm planning to use swivel mount feet with rubber pads on them. I'll install nuts in the bottom of the legs (method TBD) to hold the feet in place.
  • Having dummy models to serve as Utah teapots is *really* nice. I'm still missing my Wilton vise, Gerstner tool chest, and monitor/laptop stand, but none of those is worth the time it'll take to model them.

I'm pretty happy with this so far - hoping to have it together soon :)

EBM and chemical surface finishing

Added on by Spencer Wright.

As I've written here before, the field of high performance surface finishes is fascinating - and complex. Surface finish plays a big role in the mechanical and aerodynamic properties of a part, and (in the case of consumer products) it can have a huge effect on marketability too. And so, as I've gone through the process of developing my titanium 3D printed bicycle seatpost, I've been conscious to evaluate many different surface finishing options to find a manufacturing process chain that's both effective and economical.

And so, this past spring, I reached out to Dr. Agustin Diaz to see how REM Surface Engineering could help my parts.

For a bit of context, allow me to quote myself (from EBM surface finishes and MMP):

The part's nomenclature

The part in question is the head of a seatpost assembly for high end road bikes. The part itself is small - about 70mm tall and with a 35mm square footprint. As built, it's just 32g of titanium 6/4. Add in a piece of carbon fiber tubing (88g for a 300mm length) and some rail clamp hardware (50g), and the entire seatpost assembly should be in the 175g range - on par with the lightest seatposts on the market today.

As a product manager who's ultimately optimizing for commercial viability, I had three questions going into this process:

  1. How do the costs of the different manufacturing process chains compare? 
  2. How do the resulting parts compare functionally, i.e. in destructive testing?
  3. Functionality being equal, how do the aesthetics (and hence desirability) of the parts compare?

Towards these ends, Dr. Diaz and REM finished three parts for me:

The parts were printed in titanium 6/4 on an Arcam A2X by Addaero Manufacturing. They were then HIP'd (hot isostatic pressing is a whole other area of interest - more on it soon, I hope) before being treated by REM's isotropic superfinishing process.

The results are very interesting, and contrast in many ways with MicroTek's MMP process. REM ISF is a chemical accelerated vibratory finishing process. In it, parts are placed in vibratory finishers with a nonabrasive media and a chemical activating agent. The chemical agents are selected by the part's composition: Different metal alloys will react to different chemicals. The media, on the other hand, is selected depending on the part's geometry: Parts with small features will require smaller media, etc.

As with any vibratory finishing process, then, REM tunes the frequency and amplitude of the machine to adjust the aggressiveness of material removal. The media chambers in these machines are shaped like toruses, and parts take a rotating helical path around them as they vibrate. Adjusting the frequency and amplitude of the vibration affects that helical path, and REM tunes the rolling angle to produce the result that's needed. "It's an art and a science," Dr. Dia zaddtold me.

REM finished three parts for me, in two slightly different design variations. The first part went through their Extreme ISF process - a quick treatment that removes excess powder and some higher-order roughness. The second two parts went through Extreme ISF *and* an additional ISF treatment, removing .020" (about .5 mm) and producing a much smoother surface. In order to compensate for the material removal, one of the Extreme ISF + ISF parts was printed with .020" of  extra stock on all of its surfaces (if you look closely, you can see the additional stock as a stair-stepping effect on the inside of the skirt edge). The results are below - click on the photos to enlarge.

Incidentally, I found REM's process nomenclature a bit confusing at first. As described above, all of these processes include some chemical agent and some media. The difference in the different processes has to do with the balance of those two factors: Extreme ISF uses aggressive chemistry but relatively little media interaction, whereas ISF is a longer process with less aggressive chemistry and more media interaction. REM also offers a Rapid ISF process, which is to ISF much as a lathe is to a milling machine. In it, the parts are fixtured and then moved through the chemical/media mixture. It's a much faster process, but one which requires more tooling and setup and hence is reserved for high volume parts.

The surface character of the REM parts differs significantly from the parts that I had MMP'd. ISF interacts with the full surface of the part, with the result being that both peaks and valleys are rounded out. Note that the color scale in the images below are not constant; click on the images to see the color scale key.

The roughness values for the two methods are also quite different. The key metrics are below; full roughness profiles & filter data are here for Untreated, Extreme ISF, and Extreme ISF + ISF parts (thanks to REM).

Ra - Roughness average

All values in µm. Evaluation length = .5"

Rq - Roughness, root mean square

All values in µm. Evaluation length = .5"

Rsk - Roughness skewness

All values in µm. Evaluation length = .5"

Rt - Roughness total

All values in µm. Evaluation length = .5"

As with most things, the numbers above both a) capture interesting differences between these five finishing methods, and b) are abstractions which ultimately fail to capture the entirety of the physical parts. Such is the nature of data; in and of itself, it's not particularly insightful.

As I've described previously (and above), my interests are functional, aesthetic, and economical. The latter two of these sit more or less in balance, but the former is bound by the composition & arrangement of matter. To that point, Agustin referred me to a paper by Kwai Chan called "Characterization and analysis of surface notches on Ti-alloy plates fabricated by additive manufacturing techniques," which shows a correlation between notch depth and a shortened fatigue life in EBM parts. To quote:

The presence of surface notches is likely to promote crack initiation and reduce the fatigue performance of LBM and EBM materials. Since the depths of the surface notches correspond to the maximum valley depths on the surface, fatigue life of the various Ti–6Al-4V materials is expected to decrease with increasing maximum Rvm values...
To improve fatigue performance, the surface notches on the EBM and LBM materials must be removed by machining.

Of course, those last two words - "by machining" kind of begs the core question that I'm asking here. To wit, see the Rvm numbers from REM:

Rvm - Maximum valley depth

All values in μm. Evaluation length = .5".

The goal, here, is to improve fatigue life - and avoid the fate that my last parts met during testing. My hope - and one that's balanced by my lingering concerns about my glue joint design - is that the much reduced Rvm numbers here will help significantly.

More soon.

Intellectual influence

Added on by Spencer Wright.

From a 2011 review of Daniel Kahneman's Thinking, Fast and Slow - which, as I've written before, is very good:

But intellectual influence is tricky to define. Is it a matter of citations? Awards? Prestigious professorships? Book sales? A seat at Charlie Rose's table? West suggests something else, something more compelling: "Kahneman's career shows that intellectual influence is the ability to dissolve disciplinary boundaries."

That's a pretty compelling definition to me.

My favorite tools of 2016

Added on by Spencer Wright.

I like tools a *lot*; my favorite gifts (both given and received) have been tools. So I thought: Wouldn't it be fun to put together a guide for folks like myself who want/need gifts for loved ones & themselves?

Aside from five items here (each marked with an asterisk), I own and have used all of the stuff below extensively. My filter: It's gotta be a tool; it's gotta be a pleasure to use and/or be exceptionally useful; and it's gotta be something that you might not think about or stumble across otherwise.

And so, without further ado: My favorite tools of 2016!

Planning & Strategy.

  • A serious particulate respirator. You know those paper dust masks? They're BS, and you'll never look back once you use a real respirator.
  • Eye protection! I have a pair of these. I've had foreign objects (metal, sand) removed my my corneas on more than one occasion - it *sucks.*
  • Workflowy. The free version is everything you'd ever want out of a list app - iOS or web access, supports hashtags and sharing, etc. It's a *great* tool for task & information management.

Making & Manufacturing.

Maintenance, Repair & Operations.

Distribution & Logistics.

Inspection & Testing.

Tangents.

And.

A Lego model of the Panama Canal (which you can only buy in Panama).


 

 

Good at it

Added on by Spencer Wright.

This week I experienced a few moments - fleeting, but palpable - of being good at it. 

It was a nice feeling: A combination of knowledge, experience, and at least a bit of maturity. It was the feeling of having all of your tools laid out in just the right place. The feeling of finding a rhythm. The feeling of knowing exactly how much energy you've got, and pacing your sprint so that you *just* exhaust all of it.

Of course, I continue to make mistakes. Things rarely move as quickly as I hope, and rarely am I satisfied with the results. But when I look back, I can see my own work working - and I'm pretty sure it's not accidental.

I know the feeling isn't permanent; even writing this now, it seems a far way off. But I've been building for a while now, and even a fleeting reprieve is pretty great :)

New desk

Added on by Spencer Wright.

About a year ago, when I refurbished an old Wilton vise for my home office, I noted that I intended to build myself a new desk as well. Like many projects, I ended up moving a bit more slowly on that than I expected - which in this case was convenient, as it allowed me to settle into a new house and think about what my work will look like over the next few years. And so last weekend I opened up Inventor and started putzing with my desk design again:

Like so many of the things I've designed in the past few years, the idea here is to use modern materials & assembly methods, and make something that is highly functional and also aesthetically pleasing. The tabletop will be lab-grade phenolic resin - a material that is strong, seamless, and durable to impact, scratching, and liquids. The legs will be monofilament wound carbon fiber tubing, which I'll probably apply a clear sealer to. And the whole structure will be held together with - you guessed it - 3D printed node connectors.

The exact dimensions and assembly methods are still TBD; I'm also considering a few details for attaching/mounting things (my vise, my monitor stand, lighting, the power strip that I like). I'm also still considering integrating the desk with the 7-drawer tool cabinet that I have, although at this point it's more likely that they just sit side-by-side.

Timeline on this moving forward is... medium? But hoping to show some progress soon :)

Intentions & Modes of Communication

Added on by Spencer Wright.

This is a bit of a blast from the past: A draft blog post that I wrote, but never published, back in July of 2014. I was working at Undercurrent at the time, and thinking a lot about the way that we communicated to our clients (mainly through slide presentations), and about the degree of intellectual seriousness and honesty of that presentation. 

What I wrote below is a bit out of context, so I'll put it bluntly here: I think that slide decks are fine and good as supplemental info during presentations, but do not generally encourage the kind of thought that real strategy and/or evaluation require. 


From an excellent 2001 report by Smithsonian Institution's Office of Policy and Analysis titled "Art Museums and the Public":

Over the last 25 years, many museums have engaged in studying the impact of their exhibitions on their visitors...The research has been largely evaluative, comparing outcomes with intentions, and has been directed towards improving the mechanisms of presentation so that desired outcomes are more likely...
One of the most striking results of this generation-worth of museum audience studies is that the explicit aims of exhibition planners are rarely achieved to any significant degree. In study after study at the Smithsonian, in all types of settings, researchers found that the central goals of the exhibition team (which are usually learning goals) were rarely met for more than half of the visitors, except in those cases where most visitors entered the museum already possessing the knowledge that the museum wanted to communicate. Rather than questioning their aims, most museums, at the Smithsonian and elsewhere, reacted to such results by attempting to improve their exhibition designs and information delivery systems, and by downplaying the importance of such outcome measures.

Here I see a striking disconnect between the stated goals of museums and those of public museum goers. Curators feel a mandate to *educate* their audience, and tailor their exhibition plans towards that end. But visitors remain uneducated, and when curators are presented with that fact, they react defensively.

Later in the report:

Art museums serious about enhancing their public role may also need to reconsider their internal structures to better express their priorities. How are decisions about public programs to be made? Must exhibition subjects be determined solely by the interests of the museum's research staff? Who will be responsible for maintaining the dialogue with present and prospective visitors? Will the dialogue function be called marketing? Program research? Audience research? What role will those specialists play in directing the museum's program plans? What role can be shared with the public directly?

Ultimately, it might even be necessary to review the subject matter distinctions that currently separate museums. If the aim of a museum is to serve a public that is often less interested in the authorship and style of an art object than in the culture that gave rise to it, or the meaning that is currently found in it, there may be little practical reason to maintain the subject matter boundaries that museums have inherited from the departmental structure of academic institutions...

If a museum wants to seriously address its public role, it needs to find a way to engage in an extensive, prolonged, multi-faceted dialogue with that public. There needs to be a way for the museum to listen, especially to those who do not believe that the museum has anything to offer them. And there needs to be a way for the museum to respond to what it hears.

And lastly, in a section titled Rethinking "quality":

The operations of some Smithsonian art museums are deeply affected by a concept of quality that discourages innovation, experimentation, and flexibility. If museums are going to find ways to connect with new audiences, they will have to experiment. Many of those experiments will fail and many will have to look very different from what is currently being done. Unless the museums are willing to take such chances, they will not change.

In my opinion, a fundamental rethinking of purpose is appropriate here: away from pseudo-objective quality and towards popularity. At minimum, museums should be honest with the fact that their curatorial decisions are largely based on an elitist form of popularity. I will leave it to them whether they want to broaden the range of aesthetic and cultural perspectives that they aim to serve, but I find the focus on quality intellectually dishonest and ultimately counterproductive.

---

On a personal note, I am struck with how this report contrasts with my output at Undercurrent. Here there is a remarkably legible and compelling analysis. It is both academic and personal. It describes the extant goals and performance metrics of both the organization at hand and its broader marketplace, and asks serious questions about how the reader should interpret them. It then suggests both specific action steps *and* general frameworks to consider - all while allowing for some ambiguity in what an optimal outcome will look like. And it does so without sounding jargony or academic, something that I am often bothered by in the world that Undercurrent inhabits.

To be fair, some of the work I've done at Undercurrent has shared these features. I've also not been tasked with such a high-level analysis of an organization's objectives (most of the product strategy we do tends to be more visual). But for the most part, we produce decks, and I can't compellingly argue that they are as thoughtful or intellectually honest as this Smithsonian document.

Now, I should note that there are benefits to being short-winded, and throwing a couple of pretty pictures in with your pitch isn't necessarily a bad thing. But I wonder: In a business that communicates with busy and varyingly interested stakeholders, what is the place of producing text-only reports? Are there specific traits of organizations that are well suited to integrating and acting on such output? And what are the constraints which should be applied to it, such that it can have as great an impact as possible?

Notes on Arcam and SLM

Added on by Spencer Wright.

Yesterday, GE announced that they had put in bids to acquire both Arcam and SLM for a combined total of $1.4B. This move poses some interesting questions about the next few years in industrial AM, and will no doubt have a big impact on both the companies involved and their customers and competitors. I don't have any privileged insight into any of these companies' decision making process, but I have a longstanding interest in the industry and what they're working on. Here are a few observations & questions that occurred to me about the deals and their impact.

Background

In 2012, GE Aviation made three large acquisitions in industrial AM. The first was the combined purchase of Morris Technologies and Rapid Quality Manufacturing, two sister companies based in Cincinnati who had already been a big supplier to GE Aviation (terms of the deal were not disclosed). Later that year, they bought Avio Aero, and Italian parts supplier, for $4.3B. These two purchases showed an interesting balance in technologies. While Morris and Avio had very similar business models (both were job shops that produced parts for GE Aviation and other business units; Avio also produces parts by traditional manufacturing methods), they focused on different additive technologies: Morris on laser, and Avio on EBM.

I've written about the difference between laser and EBM in the past, but a few points here:

  • The fuel nozzle that GE is so famous for printing is made by laser in Auburn, Alabama on EOS machines. I believe that their (lesser known) T25 temperature sensor is made on the same machines.
  • The laser (Note: I'm using "laser" here to refer to processes that are variously called DMLM, SLM, DMLS, lasercusing, and the generic "laser metal powder bed fusion." Note also that SLM can be used to refer both to the printing process AND to the machine manufacturer that GE just acquired.) machine market has a number of providers: Aside from EOS and SLM (the two machine manufacturers that GE is most known for using) there's Renishaw, Concept Laser, Additive Industries, 3D Systems, and a variety of Chinese entrants.
  • While GE Aviation has tended towards EOS machines (see the video above), GE Power & Water uses machines made by SLM in their Greenville, SC plant (Note: Here I'm drawing from an AMUG 2015 and other industry sources; sorry for the lack of a hyperlink reference).
  • Arcam sits apart from these: it's currently the only company making machines for EBM (electron beam melting, or "electron beam metal powder bed fusion" if you're picky).
  • Avio Aero has done some really interesting things with EBM since the GE acquisition. Perhaps most notably, last year they printed low pressure turbine blades out of titanium aluminide, an intermetallic alloy. TiAl has excellent mechanical properties at high temperatures (an important feature of any part that's in the hot stage of a jet engine), and is traditionally cast by companies like Precision Castparts Corp (PCC). Printing in TiAl brings advantages but is extremely difficult due to the material's tendency to fracture. Printing TiAl could be a big deal as GE ramps up production of TiAl blades for the GEnx engine, and it was very interesting to note that after the successful prints, Avio bought an additional ten Arcam systems - the largest purchase that Arcam had ever accepted.

So: GE already had a strong portfolio in additive. What are the implications of the Arcam and SLM acquisitions, and how will this impact the industry?

A full stack, in-house

The most interesting part of the acquisition to me is the fact that GE will now be able to in-house the entire industrial AM supply chain (minus software; more on that below). Previously, they were focused primarily on basic research and applications development (Morris, Avio, CATA, and the Niskayuna Global Research Center) and serial part production (Auburn, Greenville, and Avio). Now, they'll own not one but two machine manufacturers - allowing them to push upstream and make a more direct impact on the development of the additive industry.

But perhaps more importantly, GE gains both powder production (through AP&C, the Canadian powder supplier that Arcam acquired for CAD 35MM in 2014) and final parts manufacturing (through DiSanto, the medical implants manufacturer that Arcam acquired for $18.5M later the same year). When Morris was acquired, they shut down their sales organization and focused on printing parts for internal GE customers. DiSanto is a very different business, though, and I wonder whether it might make sense as part of GE's healthcare unit - with its traditional focus on medical imaging and healthcare IT.

AP&C is a bit of a different beast. They manufacture the raw materials for not only powder bed fusion but also MIM, HIP, and other powder metallurgy applications. Their website advertises commercially pure titanium and ti64, but Arcam also markets cobalt-chrome - which both the fuel nozzle and the T25 sensor housing are made of. I'll be very curious to see whether AP&C continues selling powders to the public, or if they focus on internal use.

Improving - and selling - manufacturing machines

Separate and apart from the implications to GE's internal operations, I'm particularly interested in the way that GE's involvement at these new levels of the tech stack will affect how the industry matures. Try though they may, it's difficult for a company whose bottom line depends on selling machines (as opposed to, say, selling machines AND printing parts, or selling machines AND developing manufacturing software) to truly impact the end-to-end engineering process much. And while GE has been at the forefront of additive research (and, no doubt, collaborates very closely with both their hardware and software providers), I'm hopeful that having them at the helm will push both EBM and laser powder bed fusion forward in a cohesive way. 

Some might suggest that keeping expertise in house would be the strategic choice here, but I disagree. Powder bed fusion today suffers from both a lack of talent (easiest to change by expanding the number of use cases for the process, many of which GE ultimately is not going to compete on) and a lack of process predictability and reliability. GE has more knowledge about improving AM part yield than just about anyone else in the world, but that doesn't mean that there aren't other approaches that they're not trying. Inasmuch as sharing whatever process improvements they come up with will encourage others to try their own approaches, I hope that GE does just that. And they have in the past: through their involvement with standards and industry organizations like America Makes, 3MF, and ASTM F42; through their participation in sessions at AMUG and RAPID, and through their open innovation work (some of which I worked on while at Undercurrent) with GrabCAD and NineSigma.

My hope would be that GE continues to market and improve both SLM and Arcam machines. The latter is particularly dear to me, as EBM equipment isn't currently made by anyone else (and because I've had many parts printed on Arcam machines). SLM is a bit different, as the market for laser powder bed fusion is already so rich. But by that same rationale, the potential impact that any updates to SLM's machines would have could be huge, as they would force other players to respond in kind.

Software

I take GE at their word: They want to be a contemporary engineering company, and they believe that contemporary engineering companies need formidable software capabilities. And if they're going to truly make their mark with a software solution, it would be wise to do so in a realm where they know the problems well.

Even before these acquisitions, GE knew the pain points in additive as well as anyone else. Adding a few machine manufacturers, plus a raw materials supplier and a finished parts manufacturer, will only help that along. So my question is this: Why wouldn't GE make a play in additive manufacturing software? This is, after all, the whole subtext behind the Brilliant Factories initiative: GE knows how hard it is to make things, and they can help you make them better.

As you'll know from my previous writing, I'm excited for advances in build processing (see netfabb and Magics), build simulation (see 3DSim and Pan Computing, and research by Wayne King at LLNL), and in-process monitoring & control (see Sigma Labs, plus the product spec sheets for a *lot* of the current class of laser printers). Each of these is extremely hard in itself, and recreating the entire stack would be extraordinarily complex; I don't expect any one company to solve (or even attempt) them all. But whether they build their own solutions or work with external providers to build them, GE will be a huge stakeholder in the next generation of additive manufacturing software. And if they're serious about being a formidable software company, then why wouldn't they take a shot at building it themselves?

Regardless of how these acquisitions shake out, the next year should be interesting. I'm looking forward to it :)

The first 14mm

Added on by Spencer Wright.

This week I got some good news: Researchers at The MTC had begun printing one of my latticed bike stems.

The first 14mm of my latticed bike stem, printed in titanium on an Arcam A2x. The part is upside down (relative to the build orientation) in this photo.

This part was printed in titanium 6/4 on an Arcam A2X. Unfortunately the build failed at 14mm high; on the upside, it appears that the failure was *not* caused by my part. It's a bit early to make any judgments about its feasibility, but I'm pleased to see that these beam diameters (which are between .8mm and 1.8mm) seem to print without support structures. As you can see below, many of them (almost all, in fact) had very low angles relative to the XY plane.

The build orientation of my latticed bike stem.

I'm hoping to have more progress on this soon. Thanks to my friends at The MTC for their help with printing - and with debugging the design!