Additive manufacturing is spreading to every industry and is evolving way past the stringy plastic prints of the past. Learn how GE is utilizing this technology to print some things you may not expect such as ceramic components for airplane engines. Learn how they cut down on failures for the prints that take months by leveraging new technologies and some old ones too.
This episode is sponsored by GE, a leading company in a multitude of fields. Check out their research homepage here. Take a look at jobs here
This episode is sponsored by Materials Today, an Elsevier community dedicated to the creation and sharing of materials science knowledge and experience through their peer-reviewed journals, academic conferences, educational webinars, and more.
Thanks to Kolobyte and Alphabot for letting us use their music in the show!
If you have questions or feedback please send us emails at [email protected] or connect with us on social media: Instagram, Twitter.
Materialism Team: Taylor Sparks (co-creator,co-host), Andrew Falkowski (co-creator,co-host,editing assistance), Jared Duffy (production, marketing, and editing).
Keywords: Materials Science Additive Manufacturing 3D Printing General Electric
I would like to describe a field in which little has been done, but in which an enormous
amount can be done.
This field is not quite the same as the others in that it will tell us little of fundamental
physics, but it will tell us much about the strange phenomena that occur just below our
perception.
In contrast to the natural philosophers of the past, the scientists of this field delve
into the recesses of nature and show how she works in her hiding places.
Their quest is to understand and create the imperceptible.
After all, there is plenty of room at the bottom.
Hello and welcome to the Materialism Podcast, an exploration of the past, present, and future
of material science.
My name is Taylor Sparks and I'm joining you from Liverpool, England.
And I'm here with my co-host, Andrew Falkowski.
Andrew, how are you?
Doing well.
I didn't realize how early we had scheduled this, so I'm kind of regretting my past decisions.
Well, I'm glad that we woke you up here in like early afternoon.
For me, this time zone has been tricky.
Thankfully, we have people in the middle, right, in the middle of time zones between
you and I in Eastern time.
We're joined by Kathleen Hoyle and Justin Gambon from General Electric.
So guys, want to introduce yourselves?
Sure, I will start.
I'm Kathleen Hoyle.
I've been at GE Research 11 years, focusing on ceramics and ceramic processing and trying
to make very complex ceramic parts the whole time.
And I'm Justin Gambon, also from GE Research.
My main focus has been on metal added manufacturing for the past decade or so.
I'm the technology leader for it, where I help manage both our advancement of the technologies
and kind of the roadmap of where we want added manufacturing to go as we move through the
next several years.
So as you heard from the introductions today, we're talking about added manufacturing again.
Now, if you haven't caught our previous episodes, in our previous episode 37, we did a deep
dive already into the fundamentals of added manufacturing, which we think will be really
helpful to sort of put in context what we're going to talk about today.
And then in 41, we talked about some interesting case studies.
Today's episode is going to extend that with an emphasis on what's happening with added
manufacturing in industry and one of the big powerhouses, General Electric.
So we're curious to learn from you a little more about the history of added manufacturing.
We're going to talk about some of the challenges with industrializing it and eventually get
to what you think is happening going forward.
So with that said, tell us what GE is doing in the added manufacturing space.
Sure.
So GE has been working at manufacturing quite heavily since the early 2010s.
We began work really with a singular focus of the LEAP fuel nozzle and pretty heavily
discussed in the industry since it was really introduced around 2015.
But we have been using added manufacturing to advance both our aerospace engine components
and some of our GE power components to introduce additional complexity, increase efficiencies,
lightweight components, all to kind of advance our capabilities in the industry.
So tell us more about this fuel nozzle component.
Why that?
Like, why is this the starting point for you guys to look at?
So the fuel nozzle is, and for this one, we were talking about the LEAP fuel nozzle.
So for it was for an up and coming engine in the early 2010s, the fuel nozzle itself
is a few inches in size.
And it's a very complex component.
It essentially mixes the air with the fuel before combustion.
With advancements on engine, on engines, we're trying to get to leaner ferns, lighter
weight engines entirely.
We have ongoing challenge of increasing the complexity of the parts and with the fuel
nozzle, the small component, it was looking to be assembled through 20 or so different
individual components brazed and welded together.
And each of these are pretty small.
So manufacturing is becoming a challenge and then going any more complex was becoming more
impossible with traditional manufacturing methods.
So small part, complex shape, traditional routes are failing you.
This does seem like an ideal spot to bring in additive manufacturing.
How's it been?
It's been a wild ride over the past several years.
We've been expanding dramatically since that.
It was a large success.
We started off with the fuel nozzle being just a, an additive manufactured prototype.
But once we started seeing that capability at GE, we kind of dove headfirst into using
additive manufacturing as the main manufacturing method.
The LEAP fuel nozzle got stood up at a production site around 2015 on Auburn, Alabama.
And since then we've been expanding to new parts, new components for pretty much all
of our engines that are entering the market.
Most people, when we think of additive manufacturing, there is still a lot of people still see it
as more of a prototype technology.
And what is it about maybe the actual additive manufacturing technology or the material science
that is allowing it now to be used in production or in application?
Yeah.
So some of that is around just the maturation of the machines.
Early 2000s, the machines didn't necessarily have the reliability that they have today.
So ours was just robustness and reliability of the process, just with any other manufacturing
method.
So when you have a single machine, you can use it for prototyping, but there's a whole
new capability towards it and technology needs when you want to go from one printer
to dozens of printers printing the same thing years on end.
So that's been kind of the major advancement push in the past five or 10 years.
What does the difference look like between a prototype?
It just needs to maybe look correct, or maybe it needs to fit the fittings correctly.
But what sort of characterization and testing goes into qualifying a part that's actually
going to be used?
That's a good question.
So like you said, when you have just a prototype component, you just need it to roughly fit,
roughly be functional for short duration or for a very specific task.
When we're talking about actually having airworthy parts to fly on engines, it comes with a larger
amount of understanding required of the process.
And so we typically look at it in two ways.
You first need to have geometry conformance, you need to make sure that the part has the
appropriate size in all the at least in the important areas that you care about to whatever
you design towards and parts do warp and change during the additive process.
So there's some science and technology has to go into ensuring that kind of fit.
And the second is material quality.
So you have to ensure that you have consistent material quality for whatever application
you're doing.
Maybe you care more about pencil properties, maybe it's more about LCF, but you have to
prove those types of capabilities for whatever application you're doing.
And prove that when you scale it to multiple printers, you have it over long durations
of time for production, that you maintain that quality.
That's got to be a pain in the butt.
I've messed around with a lot of 3d printers before and I feel like everyone like you have
to learn how to whisper it to get it just working right and there's so much like custom
calibration.
Now you're telling me that you're doing this across multiple parts because of course you're
not going to manufacture with one single printer.
How much of a pain is that or have you figured that out?
A lot of our work over the past 10 years has been figuring that out going from it being
more of an art where you have to have an expert user just working with a single machine
and like you said, just knowing the ins and outs of that process and being able to eyeball
a lot of things to really getting a production scalable understanding of the process.
And that's where the machine reliability comes in and improving its capabilities, better
process and procedures, more robust build parameters that you get the material quality
that you want, making sure it's reliable.
We did have to scale us because we've shipped over 100,000 fuel nozzles to date.
So it's a massive amount of production occurring over the past several years with across a
large fleet of printers.
So Kathleen, this has been metal so far.
You work in ceramics.
I remember when 3d printing first showed up and I remember thinking like, yeah, they'll
probably do some like here or there example of ceramics.
There's no way this is actually going to be a technology we use.
Has that changed?
It has changed quite a bit.
Ceramics are still behind metals and polymers by quite a bit.
However, there's been a lot of growth, especially in the last three or four years.
So one of the biggest issues that Justin already talked about is the machine maturation.
So there wasn't a machine that you could even buy to print a ceramic part until about 12
years ago.
So if you don't even have a machine, what are you supposed to do?
Other than simple extrusion based printers.
And when you're talking about something where bill layers can be a big impact on your properties,
those they're appropriate where they're appropriate, but there's a lot of spaces where they're
not.
So early, early days, were you making your own devices?
Or what were you doing?
No, we were not.
So we essentially just, we waited for those machines to be available.
We finally had a machine where we could go out and demonstrate what additive could do
for ceramics and start to find customers that were interested in using additive for ceramics.
What sort of machine is this?
What I'm familiar with when it comes to 3d printing ceramics is binder jet.
Is that what you were using?
Or was it something different?
The modality I work on is that polymerization.
And so that's similar to so for ceramics, there's two different ways you can approach
it.
One way is you do what the polymer guys are doing, but you now throw some ceramic particles
into your polymer, and you print up a polymer based part with the strength particles in
it.
And you throw in a furnace, you burn out all that polymer and you consolidate your ceramic.
And so it's like a very fancy way of making a clay pot.
And so then the other way to do it is to pretend you're metal.
And so we have metal powder, instead of that you have your metal, you have your ceramic
powder.
And so binder jet is the most adaptable one to that because you're just putting down your
little bit of glue and then you're burning it out.
In terms of laser fusion, ceramics are not really friendly with that because they're
so brittle, you need high temperatures.
People are still working on it, but that's the slowest modality to see quality parts.
I had no idea you could do that printing of ceramics, how you described, I guess it makes
sense.
Do you not end up with like huge porosity?
How does it compare to binder jet?
Is binder jet more dense?
Binder jet is actually less dense.
And that's because when you do binder jet, you need larger particles in order to spread
the powder.
And so for binder jet, you're typically at 40 microns.
I think the smallest I've seen is five.
And when you're doing ceramics and you need to densify, you need small particles.
So you really want to be more on the one micron size if you want high density.
So it's actually easier when you mix it with some type of polymer fluid and then you build
it out that way.
And so then it's just a matter of burning out the polymer so that you don't get a bunch
of cracks in it.
And that's a huge challenge.
That's the challenge, in my opinion, for ceramics.
It's not printing the part because polymer printing has come a long way.
It's after you've printed it, you got to get rid of that polymer without having a bunch
of cracks in your part and then densify it down without it distorting.
I have a million questions.
So, I mean, I don't know how much you guys can say, but how do you prevent that cracking?
Then are you doing things like, do you like hip it afterwards?
Do you just accept that you're not going to get it like fully dense, just get it to not
crack because you're always going to follow it up with like a hip step afterwards?
Or what do you do?
And for those aren't familiar with the hipping, that's hot isostatic pressing.
It's essentially you squeeze on it with a gas in a chamber, essentially you have to
seal the outside.
Then you let the gas do the centering pressure from isostatic conditions.
Yeah, we don't hip it, it's a lot of new chemistry development.
So for ceramics or traditional processing, you use simple polymers that burn out really
well.
And for certainly the vat polymerization, that's not the case.
And so it's actually that's been a big part of the work that we've done the last few years
is understanding what polymer chemistries we can use and how to fire these in order
to not have the cracks.
And I can't go into the big details on that, but there are companies out there that have
slurries that they sell you.
So when you buy a machine, typically the company that sells machine also has material to go
with it.
And a lot of it is these just very long, slow burnout cycles that just take a long time.
So you remove the polymer as gently as possible in order to avoid all that cracking, but it's
still a challenge and yeah, yeah.
Traditional ceramics approach.
You just got to do it really carefully.
Yeah, okay.
Pretty much.
You know, between both of these processes, there's lots of advantages and disadvantages,
but I think there's a lot of myths surrounding why someone would use additive manufacturing
as well.
Right?
Like you're talking about ceramics, 3D printing ceramics being a slow process, making ceramics
traditionally is a slow process, but one of the purported advantages of additive manufacturing
is that you can have much more complex parts and that it's easier to make those more complex
parts.
But is that actually true?
No, I wish it was.
So I like to say complexity is accessible.
It's not free.
It's accessible.
You can get it, but it does cost something.
Just I guess one example was a project that I was working on with the Army Research Lab.
We were 3D printing a little prototype, a ceramic cooler.
So a little fluid would go through this box that was centimeter by centimeter, two inches
long.
Sorry to mix my units.
How my brain works sometimes.
And it had this internal helix.
So the fluid would swirl as it goes through it.
And so it was a box with a tube inside of it.
And you would think, and that is a little difficult to extrude.
It was a little asymmetric too.
So it would have been difficult to extrude and you certainly couldn't do the helix inside.
And so looking at it, I was like, oh yeah, no big deal.
We can print that.
No problem.
And we printed it, but when we fired it, we got, say, we got these cracks and they're
all in the bill layers.
Bill layers love to crack and ended up said, well, if we didn't have a square, if it was
like a symmetric tube, then it didn't crack at all.
It wasn't that much more complexity, but it made a huge difference and just being able
to process this part successfully.
How much of those sort of innovations do you feel like are trial and error or like who
would have thought sort of moments as opposed to like, oh, we had a really clear hypothesis
why this would be a change in parameter that would lead to optimization?
Cause I feel like there's still like a lot of just like, we just tried it cause it was
a parameter we could try.
There's some of that.
I think as you do it, you get some intuition for what parameters make sense.
I know in the early days we're messing around with cure time a lot and has an impact, but
not as much as you want it to.
So design is a huge design is a huge player in that.
And so I think what it really comes down to is being really well partnered with whoever's
doing your design to ice that.
What about on the metal side?
Is complexity more or less accessible there?
I'd say it's very similar to the ceramic side.
These machines are capable of printing very complex shapes.
And one of the challenges we've always been finding with using Adam A factory on the metal
side is when you give designers the freedom to make really complex shapes, they capitalize
on that.
So as complex as they can think of for their application, whatever they're trying to do,
fuel flow or aerodynamic quality, whatever it is, they push to the limits and it absolutely
can be a limiting factor for additive.
When you're working with metal additive, just part development itself can sometimes take
years.
And if you have a simple part, simple design, very quick, but to actually get the right
material quality, to get the right shape conformity, you can go through iteration after iteration.
When you have a very complex part, lots of thin walls and nested features to make sure
that you get the right end result, that you can get all the powder out, that you can inspect
it appropriately.
It just takes a long time to actually get all that nailed down.
And like I said before, it's, it, there's a lot of art to add manufacturing.
That's where a lot of this comes in.
Designers just know what builds well, what won't build well.
And that's been the push of the industry as a whole for the past decade or so is to get
more and more of that understood.
And then in different ways, trying to figure out how to preempt some of those decision
makings, design rules and simulation tools and anything else you can give to your designers
and your manufacturing engineers so that they can get the right part out of the printer
sooner so you can get that stuff into actual production and out to the market itself.
But definitely not free.
It comes with a lot of development needs, but try to make some strides in making that
smooth process.
Simulation tools have been very increasing, very popular for good reason, too.
I think they're going to be a critical part of helping the industrialization.
Yeah, it's a, it's an interesting area with the simulation tools available for added
manufacturing, at least for the metal side, we typically work more heavily with laser
based metal added manufacturing.
And it's amazing to see how much the parts can actually deflect and warp during the build
process or as it comes out of the machine, because you have such a local amount of energy
being put into that system.
Lasers, maybe 100 microns in diameter when it's hitting the metal and creating a very
small melt pool and it moves miles and miles of welds as you're building up that part.
And so stress formations can just be very complex and very shape dependent as you're
printing that component.
And so we've had to not just figure out how to simulate that, the stress generation in
that part, but then also figure out how to predistort the part so that it warps into
its correct shape.
For real.
So in the design, you're actually, you're accounting for the warping.
I had no idea this is something they did.
Absolutely.
So if you know a wall is going to shift maybe 200 microns in a certain direction, you just
build it a little bit the opposite direction.
So it comes into its final shape as you want.
Wow.
That's wild.
Yeah.
I've seen some of the simulation on these, like for example, a guy at Utah, he was doing
like, he could actually look at where the laser would hit it.
He could actually simulate the cross section, like where the melt pool is, where the
volatilization is, all the heat flow, like this incredibly complicated process, all as a
function of laser energy and the raster rate and the underlying materials.
Like it is wild how powerful that is.
I could see why that would be a useful tool for you guys in the design process, but yeah,
that's cool that you can actually take it and then infer what the design part should be
accounting for distortions.
That's new to me.
So actually maybe on that point.
So when people think about laser-based printing, Oh, great.
You avoid the problem of sintering.
You don't have to consolidate afterwards.
Is this always the case or are you still consolidating?
No, we get it to as dense as we can.
Essentially, there is still, especially for aerospace parts, there's still a HIP process
that we go through to ensure full density, but we work towards having fully dense or
near fully dense parts from the outsets.
So you've described to us some small components, is GE doing anything big components
and how does that even work with a big printer or do you put things together or what?
Yeah, so in the past, like the fuel nozzle, we have some other parts that have been
printed in the past as well.
There are a few inches in size, so they would be printed in what we call a medium
class printer, about a 250 by 250 millimeter size, put a bunch of them on the build plate
and then they can be printed in the course of maybe a few days, one, two, three days.
We have a new class of parts coming out for the catalyst engine and
these are large format parts.
So these parts are closer to half a meter in size and they have to go in large format
machines, which are half a meter in size.
So it does get more complex.
The process is larger over larger volume.
The parts themselves are just bigger and that runs into additional challenges, not
just the print, but how do you simulate parts that are a half meter in size?
It's a little bit different than simulating a part that's a few inches in size.
So, I mean, yeah, like all the problems you've described before just get amplified.
Yeah.
For example, like you can't just rely on a mesh of a couple inches.
Now your computing time explodes.
So how do you extrapolate maybe small scale simulations to large parts?
It's like a whole nother length scale, which means it's a whole nother model
on top of your previous model.
Same thing with like distortions, right?
If something comes at a small scale, what happens now at a larger scale,
is it going to be continuous?
Will it scale evenly with volume?
Or is there still going to be non, non-anticipated scaling relation there?
We've had to work on scaling for modeling almost from the outset.
You've brought up a really interesting type of simulation where it's a
laser powder interaction model.
You can understand the melt pool physics in relation to how the energy is being
delivered by the laser, gas flow, powder flow, really complex simulation.
We have an internally developed capability along those same lines,
as long as several national labs have them out in the market too.
And that provides really detailed physics of the process.
It also takes a really long time to run those simulations.
One laser strike, maybe a few millimeters long, you're looking at least an hour,
maybe a day to run that type of simulation.
So at most, you typically look at just running a few millimeters of that type
of simulation over the course of days or weeks.
So from the outset, we had to figure out ways to do things faster,
even just for the small parts.
We had to look at simplifying the models.
How do we work on doing things from more of a layer based approach,
generalizing the energy inputs and letting us understand that interaction
to the complex geometry.
Even then it was difficult to get the simulations up and running
because it still took a while.
When we first started the simulations up around the first time,
it would take us potentially days to run simulations, even on the small parts.
It's hard to get somebody to run a simulation when they can print the part
in almost the same amount of time.
So our focus for the past, I don't know, eight years has been just making
the simulations more accurate and faster.
And so now we do have the capability.
We're taking that same type of idea of layer based understanding of the process.
So some generalization to it.
We're able to run it far faster so they can be run well under
the normal build time of a part.
And that's been important when we go up to these large format parts,
because these half meter sized parts can sometimes take weeks
to build for some of them.
So anywhere from seven days, I've seen 28 day prints.
I've seen for some extra large components, 90 day prints.
So these can be, sometimes these very large parts can take a very long time to print.
So it's important.
We don't have 90 day simulations.
We want to get the simulations to be a fraction of that.
So we can iterate, we can learn from the simulation.
Where do you want to put supports to help hold it down better?
Where's it going to deflect more or less?
Figure out those problems before you start committing to these multi-week prints.
Kathleen, how's the simulation software in the ceramics space?
Is it there?
Is it like way behind?
It's not there.
Yeah, it's.
And I think that's just because the whole ceramics in general is behind.
Metalhead have had a little bit of a leg up on this one as well,
because when we, we started seeing these deformation issues on parts,
I think like back in 2012, I was involved in some projects back then
when we were first starting to see it occur.
And what we started showing this to like the simulation groups at the research center.
So the research center has different groups working in different technology areas all
across GE.
We started getting this across to other people.
The first reaction was we have some casting and welding simulation tools.
Let's grab those and we can just start tweaking them for laser based additive.
Now, a lot of extra development went into it, but we had some groundwork already laid.
There's already a lot of technology that was developed that we could just pull into
additive and then fine tune it and develop it further for the specific application.
So we had something already going even from the very outset.
That's awesome.
On these larger parts, one of the, another one of the sort of advantages or at
least perceived advantages of additive manufacturing is that you can use less
material and ostensibly with being able to simulate a lot of parts and identify errors
ahead of time, you can potentially reduce that further.
But does that, how does that actually shake out in practice?
Do you end up using less material or do you often have to use maybe more or about the
same amount as conventional casting methods?
Yeah.
And you certainly can use less material with additive manufacturing.
That's something we've been striving towards understanding long-term.
The main driver of that you have to understand is powder reuse.
So with a metal-based printer, you have to typically for most of the technologies, at
least, you have to fill up a large volume with metal powder and then you melt, you
solidify the actual part within that volume.
So you'll have excess powder after the fact.
Typically with additive, you don't really print big bulky parts either.
Additive lends itself towards having thinner walls and lighter weight components
by nature of having this capability for designers.
And so you might only end up solidifying a small fraction of the powder actually
use 5%, 10%, 15% of the total powder.
We have been working towards understanding what happens to that
powder after the additive process.
And so at the very early days of additive manufacturing, you would have to take just
all that powder that was used and potentially either set it aside or scrap it.
So you didn't always use brand new powder for your next build.
And at that point, when you're looking at just using one used
powder, it creates a lot of waste.
So a large push over the past several years, and as we scale up our production
capability, has been towards understanding powder reuse.
How does the powder behave after multiple prints of it being reused?
You take the powder out, you sieve it to get rid of any type of large conglomerates
in there, and you reuse the powder.
And to understand that you have to understand how the build process
will occur with slightly aged powder.
And we've had to go through and validate the material quality
really around that type of behavior.
How does it change with powder size distribution?
How does the chemistry change in the powder?
Or if it does, and that's been something that is looked at for every alloy
you introduce for the manufacturing.
I would imagine you're getting things like oxidation, obviously
these metals volatilize.
And so there's this sublimation and then you're changing the
maybe chemistry iteratively.
So what have you learned?
Is it possible that you can reuse these things?
It just requires, I guess it would depend on the alloy, how many times and what's
happening to it specifically, but in general, it can be done pretty easily.
Yeah, so it can be done.
And you're right.
There is an alloy by alloy case.
You have to look at one for every material that you're really scaling up.
But part of this is just better on better machine control and capabilities.
So having inner environments, when you're processing to reduce the amount
of oxidation of your powder, and then also improving some of your processing
parameters, you introduce less conglomerates into the excess powder.
So you don't have as many of the spatter particles occurring in the
powder and having a methodology to sieve them out, but really it just goes
down to the validation of the capability and also understanding how
many reuses are actually allowable.
So can you get away with two times reuse?
Can you get rid of, get away with 100 times reuse?
Where do you start finding a detriment to the quality that you have to kind
of pull that, pull that powder out of circulation?
Is it the same with ceramics, Kathleen?
Can you just wash the glue off?
Is there like a powder processing step that you can do to get it back
to pristine or is that not possible?
Well, for ceramics.
So in terms of a binder jet type modality, I think it would be pretty
comparable in a sense that you have to watch your particle size distribution
shifts, that would be the biggest thing.
You generally don't have to worry so much about oxidation in that case.
For the, say the vat polymerization, where you have a slurry, you have your
resin polymer resin mixed with your ceramic particles, then now you have
some different challenges such that ceramics don't like to stay in a
nice suspension all the time.
They tend to settle out.
And so a question as well, I finished this print yesterday.
Is the slurry still good in my printer or did it settle out
and now I need to throw it away.
And then even if you can recover it, let's say it looked okay.
You recover it.
There's still questions about it's light sensitive.
And so your curing will change over time just because you have different
stabilizers and such in there that evolve over time, react with oxygen.
And so that long-term use of it also comes into question of I've had some slurries
that I bought and from the vendor and they just sat around the lab for a few
months, we said, Hey, we want to use this now it was gelled, I could not use it.
And then they're like, Oh, you should put it in the fridge.
Oh, okay.
Whoops.
Which goes a long way.
So, you know, just storage and such, but yeah, in terms of reuse, it really, it's
all about, do you have the same material that you started with or did it shift
just through all the processing?
So it's to me anyway, sounding really like a marvel that this actually is
off the ground that we're actually making parts given the incredible
complexity that you face, but even at the end of the day, let's say we make
a device, right, you still then have to think about this from a industry
standpoint where you're now bringing in new parts, tell us about that process.
Right.
How did these things stack up against legacy components?
What goes into bringing in a part that's been manufactured in such
a completely different modality?
Yeah, that's an interesting area for additive is a lot of the times the
area that I work in is new part designs.
So you have designers coming in.
They're making something from scratch and you can have some guidance on how
you will add certain features to the geometry that makes it well printable
for additive, right, makes it robust and reliable for the process, but there's
also a whole other push of legacy parts, legacy components, where you have
something that maybe was manufactured 50 years ago and now new parts are needed.
And the traditional manufacturing methodology is not around anymore.
Or like you don't have the casting capability to do or take a long
amount of time to make the dyes for it.
And so now you start looking towards using additive manufacturing
to print those components.
And there, you don't have as much freedom to optimize a geometry to make it more
printable for the additive process, but at the same time, additive is well suited
to work with those components to do small lot production of those
areas at the time of need.
And so there's a lot of work going into understanding how we can take
in old parts, take things like CT scans of them, recreate the geometry to CAD,
work it through the entire additive pipeline so we can start producing those
parts in a manufacturing environment and help fill some of the gaps that
you might find for old components.
This is a very typical area of interest for the DoD, save a lot of legacy
components from the Navy, Army, Air Force that they would like to see replaced in
modern day.
But how awesome if you don't have to have the dye, like the tooling, if your
tooling is flexible, that's the whole advantage of additive, right?
If something does break 30 years later, as long as you still know how to make
this thing, which I think we can figure out, then you don't have to have
the custom tooling to go with it.
I didn't think about that, but that is a huge advantage.
Absolutely.
That's where the large amount of interest is.
It's the flexibility.
It's the fast responsiveness that you can get with additive or you typically
can't get with the least aging areas of the industry.
When you have the company, maybe the manufacturer of that part, maybe that
company has been acquired three or four times over, and there's a lot of
information and understanding lost.
So let's say you figured it out.
You're making things, you still have different sort of concerns with
an additive manufacturing part.
You talked about it, right?
You're doing miles of welding, for example, right?
We know that cracking at the build layers is something that you might
have a distinct concern there relative to a part that's cast, for example.
So you must have different sort of characterization when
you qualify these materials.
Can you talk about that a little bit?
Sure.
There are certainly different things that can happen in the additive process.
You're right.
You're doing a lot of small welds.
And so you need to understand that you're getting consistent, good
quality across this complex geometry.
And we look at it in two paths.
There's in process, looking at the build as it's being assembled, as it's being
printed, and then post-process inspection.
Within a process, it could be some sensors, leverage like cameras that you
take a snapshot every layer to look at the quality of your powder recode.
And that can tell you a lot because if your part's starting to warp, maybe
it's bending upwards a little bit because it was too hot in certain
sections, you'll see an inconsistency in the powder, you'll see a part
poking through the powder layer.
That's a big red flag that something's going wrong there, that you may need
to improve your part design or something's going wrong with the
additive process itself.
There's also new sensors coming out that are on machines that let you look
at the melt pool itself, so it's a really dynamic process, difficult area to measure.
What are they doing there?
Is that like a reflectance thing?
What are they doing?
You're typically looking at say the emission, light emission of the melt
pool, so you can look at it with say a photodiode, a single pixel camera,
essentially very fast acquisition.
Try melt B40 is a 60 kilohertz data acquisition that just tells
you how bright is the melt pool.
But there's also a slew of other ones.
You can look at cameras that can measure around the melt pool region.
It's a lot of data being generated over long periods of time.
So you have to have ways to analyze and interpolate that data.
And it's a big push by the industry is to understand how we can process that
information, how we can interpret that information to get better and better
quality metrics within the print.
And then you have after the print, CT, CMM, all these different inspection
methodologies, cutups when you're going through and understanding the process
behavior of the part and sure you have good material quality.
All of those are important steps to understand you have consistent quality
when you're scaling up that process.
And all that kind of leads into several post-processing steps that you might
be going through after a part has been printed.
Oh, this is cool.
I I'm just like imagining ideas.
Have they done any sensors from where the tip itself as a part from depositing
material is also making the measurement.
So for example, you could do like a conductivity measurement to the base or
something and see how you'd expect it to change as opposed to what it's actually
doing, or I don't know, are you familiar with other sensors embedded near the tip?
So we have done a few things in that area.
It's a really interesting area to explore.
And one program that was on, we wanted to understand the thermal cycling of the
part was actually seeing during the build process, it was partially just to
understand what was happening in the machine, but it's also to help calibrate
some of our thermal models to get some information for our thermal models.
And so we would, we designed a part where we actually inserted a bunch of small
holes in it and into the actual CAD geometry, we printed up the part of
about halfway through, we got through those hole prints, then we stopped the
build, opened up the chamber, removed all the powder from the chamber, and we
added in thermocouples all into those sections of the part, put all the powder
back in place and resumed the build.
And we did it within 10 minutes or so so that the whole system didn't
necessarily cool down or anything.
And we resumed the build so we could start getting information around the
thermal cycling of the process.
When we finished the second half of the print, we can understand the
different locations in the geometry.
What temperatures did it see?
How hot did it get?
And try to use that as a way to start getting some fundamental understanding
of the thermal processing of the geometry.
So we're always looking for different ways to get that type of information.
So cool.
I think there's just going to be an explosion of additional
research happening in that area.
Kathleen, I'm kind of curious from a ceramic standpoint, where flaws are
just like killers because of your brittle components, right?
It's not like you can just fudge it.
Oh, the little deformable, you get plastic deformation.
That'll just rammed out those.
That doesn't happen in ceramics.
So is it, what's your take on that?
What does inspection have to do differently given the propensity for
flaws to be deadly for ceramics?
So for ceramics, traditional ceramics, you already have to deal with flaws.
And so you need to understand what your process yields in terms of your flaw.
And so one way of doing that is looking at your Weibull strength.
And that's just where you measure the strength of a whole
bunch of different bars.
And you look at the statistical distribution of that and you say, well,
10% of them fail at this strength, 50% at this strength, 90% the strength.
And that gives you an idea of what type of, of how you can put your ceramic in
the design such that you have high reliability and understand that.
And it's the same way for additives.
Ultimately additive, you'd be doing the same thing of saying, here's my
process, here's how I build it.
The interesting twist with additive is that you have now
build orientation effects.
So when we look at this, we say, well, I built my bar such that my strength bar,
such as the length of it, where all the load is going to go is by the build layers.
And so all of the, all the loads on the built layers and, or you could say, well,
I built my bar horizontally.
So now the load is perpendicular to the build layers and you get very, it's, it
can be a two times difference between those strength values.
And so big aspect is understanding really more about that orientation effect.
And so that coupled with then some of this inspection in order to see, do I
have a big pore in the middle?
That could be a huge problem.
Yeah.
Just saying, if I have a huge pore in the middle, it could be a big
problem or some other flaw like that.
So do you do a lot of non-destructive testing then?
Or how do you find a pore like that?
Other than the CT, like what do you do?
It would have to be CT.
Yes.
It would have to be CT.
Would be looking for large enough pores.
What about ultrasound?
I guess it's similar, but.
On such a complex geometry though.
Like how would you even.
Yeah.
I guess if you have internal features, how do you differentiate, but.
Yeah.
Right.
Yeah.
I haven't, we haven't explored that.
I guess if you have one part that you know is good, cause you CT tested it and
you measure like an ultrasound response, you can just look for deviations from that.
So maybe it would work.
I'm not an expert there.
Yeah.
That's one of those things that, so when you're talking about R and D versus.
Industrialization, your methods are going to be very different.
So in R and D we use the most killer methods you can to look at as much as we
can, when you've settled down on what it's one thing is supposed to look like,
you can now get a very narrow response.
And so now you can bring in these cheaper, faster, easier inspection
methods, and you can just say, I'm doing my one measurement, is it in or out?
And so for, you could say, yeah, I got to do CT now to look for my core, some
flaw, but perhaps you could get some, ultimately just some simple ultrasound
or something and say, Hey, I did some response and is it appropriate or not?
And if not, let's take a closer look at it.
So through this whole conversation, we keep talking about, this is like
a lot of different moving pieces.
You've got the simulation people, you have testing, you've
got the actual fabrication folks.
Is this all happening at GE up in, I assume you guys are in
Schenectady, New York, or are you somewhere else?
Is this a much broader coalition of people working on this?
So yeah, the GE research campus does encapsulate all
these different technologies.
We work though very closely with the different business segments.
So it's usually a team effort between ourselves, GE aerospace and other
groups, but we do have, well, when we pull together essentially a project
to either work on part development or work on type of new machine modality,
those projects tend to be pulling on designers, material experts, software
experts, control experts, simulation experts, kind of all coming together
to figure out some of these challenges, develop new tools, develop new processes.
And so it's usually a very collaborative kind of team that we generate here
at the site and work in coordination with similar contemporaries
at the businesses as well.
In one of our early episodes, we talked about Gore-Tex, right?
And so one of the philosophies, which I think is kind of interesting
behind Gore-Tex is that you can't get too big or you can't run a business
efficiently and GE is sort of on, I don't know, my impression is that it
might be on the other end of that, that you have a massive corporation.
Is it hard to be, I don't know, is it, can you really do this
across such a large company?
Or do you break it down into small teams?
Like how's it actually work?
I want to say it, it feels like a double-edged sword to me being in a big
company because it is difficult to manage.
And how do you align people and teams in a way that you have enough people
involved that are in the loop, but you're not in constant meetings,
keeping everyone in the loop.
And so there is a lot of balance and I feel like we're constantly trying
new things in order to manage all of that and manage the teams to work.
But still stay connected enough.
The, but what's such a big advantage of being a big company is just the
amount of resources available.
The fact that there are experts, the fact that we have equipment and sometimes
you have equipment that you bought 10 years ago, you forgot about and said,
Hey, let's just dig this up again and use it again.
So just the sheer amount of resources and network within the company, I
think is just such a massive advantage to really maturing this technology.
And, but I do have to say, we do quite a bit of collaboration with external
companies and universities and government agencies as well, because
this on its own is impossible.
It's just building up a whole new supply chain and manufacturing process.
No, no one company can do, but certainly there's a lot we can do internally to
get things really kicked off the ground.
And on that one, communication is a major priority for us at all times,
cause it is a large group and we are spread out.
So like we, we have our team at GE Research in upstate New York, but we
are working with a variety of people across GE on every one of these projects.
Just last week, I was helped host a working session for somebody's
large part additive capabilities.
And we have some ongoing projects around the industrialization of large
format and printers, and we pulled, we had people flying in from Bangalore,
India from our sister research site that's located there, we have people
flying in from Italy, from Avio Aero.
So Northern Italy area, people from GE additive from Lichtenfels, Germany's
well North of Munich, and a large group from Cincinnati, Ohio coming
in from the aerospace division.
And so all of us coming together just to talk about the singular focus.
And we spent hours upon hours just working through different technical
challenges, having discussions, planning, and that's required when you have such
a large team and a global team to kind of stay aligned, keep communication channels
open, if you don't have that, then you start running into issues.
Jeez.
Yeah, I can imagine.
I naively think that because you can prototype so quickly with this, then it
must be just a fast fabrication technique.
Is that true or not?
And additive is well known as being kind of a fast process.
And that's really from its prototyping dates, where you have a design that you
come up with, you put in a printer and it comes out, plastic printers, even
metal printers, you'll see this.
The issue of, you do save time on post manufacturing, right?
You're not like CNCing stuff away.
So it kind of makes sense where you might think it'd be faster.
Absolutely.
And it can be much faster than several other manufacturing methods.
The misnomer is typically people think that you print the part and you're done.
So I've had many times where I'm given a design for a geometry to maybe help
with a rig test or something along those lines where we need to have good material
quality to it and when it's being supplied back.
So we'll get the geometry, we put it into a printer.
A day later, I have a picture of the part.
It's been printed.
It's out of the system.
And then that correspondent will ask us, you know, when can it be on their desk?
Well, next we have to bring it through stress relief, heat treatment.
Then we'll go through wire EDM to remove from the build plate.
Then we'll have to probably ship it out for the HIP process.
Then we'll do some solution, heat treatments, and then we'll get it over to you.
So give us about a month.
So it's that different.
So that's the area we have to constantly educate people on.
It can be a very short period of time, but if you want that good material quality, if
we need to get it to be kind of robust and reliable part, it takes a long time to go
through all those extra steps, a lot of handoffs, a lot of different procedures
you have to go through.
Yeah.
You think looking into your, oh, go ahead, Kathleen.
Let's say that's what I tell her.
One month.
I need a month.
It takes one day to print the part, but a month for me to give you the part.
And a big part of that is the firing takes such a long time and yeah.
And some post-processing to get rid of some stuff.
And so looking into your crystal ball in the future, however far out will we bring
that time down or is that always just can be inherent to this process?
That's going to be relatively slow.
For the metal side, there might be some possibility to streamline some of the
capability, but depends on the industry as well.
Not every industry needs to do a HIP process.
So for some areas, some users, you don't necessarily need to worry about that.
Potentially there's everybody's looking towards ways that you can get to
hipless processes, even for fully dense materials, but that's probably
a long ways off for development.
So it's typically going to be.
Always requiring post-processing now, whether you lean out the process or
anything like that's a whole other factor when you try to scale production
manufacturing, but for a while, we're going to have to live with
most of those post-processing steps.
And in the end, you just build that into your manufacturing pipeline.
And like you said before, other processes take just as long or longer.
Is this more so educating people that the print itself is one of many steps that
you have to go through for the manufacturing process?
Yeah.
And same thing for the ceramics.
I think that the firing will always take longer than traditional ceramics, but I
do think that there is ability to get it more efficient just in terms of
getting the right piece of equipment.
Ceramics is traditionally batch processing.
And so it's all about throughput.
And so just being able to have the right equipment to maximize your throughput.
Oh yeah.
So the other big thing is automating because sometimes it's not
really about how long it takes.
It's how much labor it takes to do it.
And so having one process where you have an operator do something for a full
day could actually be more expensive than having a automated process where a
machine is doing something for a week.
And that's just because you could just set it up and say, just go and you can
have your operators go do more important things.
Yeah.
That makes a lot of sense given the number of processes that you have and a
lot of the just checking that you have to do, right?
If you're going to do all these checks on dimensionality, automation makes a lot
of sense there because for a single person to go and take all these measurements,
even with a tool, even with a digital tool, it's very time consuming.
And so that certainly makes a lot of sense is maybe just reducing the downtime
into the human components, right?
It's kind of like a systems engineering approach.
Like you're just looking for all of your delays.
So what else do you see in terms of advancements in this field?
So maybe it's not the speed at which these are produced, but maybe it concerns
the quality or the consistency, or maybe there's new opportunities and new
options for complexity that are introduced.
Absolutely.
So typically our standpoint is we try to maintain extremely high quality for
the additive process and just find new ways to accelerate the process to go
faster or to add some new feature capability.
So a larger push in recent years has been heat exchangers for
metal-based additive manufacturing.
So before when we were printing, like some of the components are now in
production, they're complex, they have thin walls, but heat exchangers
takes it to a whole other level.
Where now you're building walls that might be on the order of half
millimeter thick, or sometimes even smaller, depending on the different
application areas in the industry.
And these are just very complex thin walls as well, lattice structures and
the like, where now you're dealing with something where you're only getting
maybe a few laser strikes to actually generate that wall thickness.
So how do you get good quality in that area and consistency in the thickness?
And then kind of getting the final shape that you want while also
driving new complexities and simulation.
Now you have these very tiny features, just the voxel size of simulations grow.
So definitely a new challenging area that we're really pushing into is going to
these really fine feature type of print capabilities.
How much of this is driven by hardware changes?
I know one thing that's happening is that the lasers used, right?
During the solidification process, they themselves are changing.
So there there's certainly a lot of advancements around the
machine architectures themselves.
It is possible to go to say smaller diameter lasers, you can get smaller
melt pools that also though runs contrary to another big push of the
eye of industry to go to lower cost components, because when you get smaller
melt pools, it takes longer to build apart.
And if your parts already taking a week, now you're talking
about maybe multiple weeks.
So you actually have two conflicting things where you're trying to go to
the say finer features, but also lower costs.
So major answer from the industry has been let's add more lasers to the system.
So let's do instead of one laser printing apart, let's do four
lasers printing apart or more.
And it allows you to have parallel printing, like parallel processing on
the computer, same idea for the printer.
We can have multiple running at the same time to print your part and just then
optimizing the process to be able to print some of those finer features.
So if you have a hundred micron beam diameter, you can print features
down to pretty much that same size.
You can go down to a hundred microns, 150 microns, it's just making
sure you have consistent quality.
That's the real key attribute we have to really develop towards and have new ways
to assure ourselves of that quality.
So one of the things about the ceramics that I'm focused on and very interested
in why I think it's going to be ultimately important in the future is that the
additive lets us make ceramics that can be as complex as say polymers and metals
can be.
So ceramics have typically been difficult to make very complex parts, but with
the additive, since we have more options for printing things, we can now do things
for high temperature insulation, for example.
And so polymers is a huge limiting factor for how hot you can get an electronic
stack, for example, because they can only go at about 200 C at best for a short
period of time.
And their thermal conductivity is very low where ceramics, you can go really
hot and they're also better at pulling heat out as well.
And same thing for metals, harder to replace a metal than a polymer from a
temperature standpoint, because you start to get into things such as high
temperature materials.
But again, that's another area we're really focused as to how can we get to
very high temperatures past what metals can handle.
And that's where ceramics have a lot of interest.
What do both of you think are some of the challenges that just haven't been
addressed yet and that there hasn't been a lot of movement towards solving?
So I'd say major area for the metal additive is towards fully leveraging in
machine sensors to say, certify components.
That's an area that people have been talking about for several years now.
There's a lot of interest in that.
There's a large push by the industry, by the government to really make that
happen, but it has some significant challenges that are, that still have to
be overcome.
And part of it's just really understanding better the fundamental physics of the
process and how we can best view and understand what's occurring during the
build.
We said before, you're laying down miles of welds.
And so getting a clear understanding of the melt pool dynamics and how you can
measure that in process, we're getting more sensors on the machine, but to be
able to fully understand when it's deviating from an expected type of size
or shape or temperature is a challenge because the additive process itself has
some variations to it.
And we build that into how we design our parameters, such as our laser power and
our scan speeds.
But when we're building these complex geometries, a bulky center section of a
part might have a different type of melt pool shape and size than when you're
building closer to the edge of a part, an overhanging region where you're above
metal powder.
We design our parameters to actually be robust around that variation, right?
So we quantify that and we test it and we validate it, but then also being able to
understand that variation in process physics with our sensors to really fully
understand the dynamics of the process towards certifying that a part is good.
And maybe you can bypass downstream inspection steps is a significant
challenge of the industry.
And it's something we're working towards.
And even at the research center, it leads us to having very interesting new
technology areas to kind of investigate.
And we get to work on instrumenting the machines with more complex sensors, just
better understand the physics of the process.
So one example is with one of our systems, we're able to leverage a fast
acquisition, high resolution IR camera that a different lab has.
So one of our other labs for combustion technologies has a very high end FLIR
camera that we'll borrow at times.
We'll put it onto our system to be able to measure and see the actual melt pool
striking the metal powder with an IR camera.
So you can get some understanding of the thermal effects of the process, capturing
it, maybe 120,000 frames a second, getting snapshot images.
And that's the kind of area we are starting to use to build towards
understanding the melt pool dynamics better.
And furthermore, we're partnering with different groups.
We work with Argonne National Lab at the Advanced Photon Source.
They do kind of in situ x-rays of the process to see during a laser strike,
actually getting x-ray images to create a video of what the melt pool looks like.
Not just the width and the length, you're also getting the depth.
You can actually see the profile into the part because you have that
x-ray running throughout the scan.
And it's like getting all of that fundamental understanding is still
critical to continue improving our sensors, to help us better understand
the process and certify good parts using that data.
This was a fantastic conversation.
We thank you guys so much for joining us.
General Electric continues to amaze me at the breadth of things they do, at
the depth at how willing they are to interface with folks like us.
So bringing the community into understanding how this emerging
technology is actually being commercialized is so cool that we love that you do it.
Thanks for joining us.
In addition to GE, the Materialism podcast is also sponsored by Materials Today.
You can visit materialstoday.com to stay up to date on the latest happenings
in the material science field and read some of the fantastic
articles that they've published.
You can also head over to elsevier.com to find out more about their journals,
books, conferences, and related programs.
As always, thank you for listening to this episode of the Materialism podcast.
If you have questions or feedback, please send us emails
at materialism.podcast at gmail.com.
Make sure you subscribe to the show on iTunes, Spotify, Google podcasts, or
wherever you find your podcasts.
And if you like the show and want to help us reach more people,
consider leaving a review.
It helps us improve and it exposes the show to new people.
Finally, check out our Instagram page at materialism.podcast and connect
with us to tell us what new material you'd like to hear about next.
We'd also like to give a shout out to Alphabot and Colabyte for
making the music for the podcast.
They both make a ton of really cool synthwave music, and you can go check
them out on Spotify and YouTube.
Catch you next time.
The inventors of fire, electricity, magnetism, iron, lead, glass, silk,
cotton, the makers of tools, the captors of lightning, the architect,
the engineer, the musician are all beneficiaries of the materials of this
world and are bound only by their imaginations in manipulating those materials.