3D printing rocket engines

[Stormbringer]

http://www.spacedaily.com/reports/Australia_researchers_create_world_first_3D-printed_jet_engines_999.html

printed jet engines.

[Asteroza]

http://www.spacedaily.com/reports/Australia_researchers_create_world_first_3D-printed_jet_engines_999.html

printed jet engines.

By appearances, it looks like they took apart an old jet turbine APU that Safran didn't mind being shown disassembled, 3D scanned the components, then 3D printed copies of the components in roughly appropriate materials. There is no mention of a full fit check or actual operation of the printed components fully assembled and running as an APU. Spinning turbine parts would likely need some final balance machining at the least, unless one has grossly high resolution 3D metal printers.

[john smith 19]

Note that although "3D printing" is fairly new you should keep in mind that additive techniques are at least half a century old.

Quite a lot of Aerojet designs used a combination of photoetched foils diffusion bonded into stacks.

The technology is also used by Velocisys and Meggit to build "printed circuit" heat exchangers and chemical reactors to deliver so called "process intensification."

Personally I always thought Aerojet could have pushed it much harder. They tended to do the stuff flat and then press (or use high pressure gas) to get it to shape.

Obvious extensions that came to mind were :-

Constructing parts as blocks but with either the final part inside the block, or internal cavities, defined by "perforations" around the outline. The little segments left holding parts inside the block would be quick to etch away, freeing the component.

Stretching or bending the unbonded foils should be much easier than doing it to the finished product, provided layer alignment can be preserved. It would mean that once the layers were bonded together they would need to have their edges trimmed to give the right size.

A technique in MEMS mfg is the use of "sacrificial" layers that can be preferentially etched to release objects.

Making structures that are curved as you go down the layers smoothly is probably too difficult. However by using a smaller number of masks could give a more viable "stepped" structure. Those steps should be preferentially etched, giving a (relatively) smooth result.

OTOH curves in the plane are relatively simple. Generally curves give smoother fluid flow.

It should be possible to fabricate in situ sensors based on fluids effects on the resonance frequencies of various structures, being driven and read by various acoustic transducers. Embedded electrical sensors are likely to more difficult due to the need to create insulating and encapsulating layers inside the structures.

By combining sub units split along different planes it would be possible to make more complex structures. this is relevant because of the difficulty of putting curves through layers.

Layer thickness can also be varied. Historically they have been foils the same thickness, but they could be substantially thicker, from a few 0.002" up to say 1 or 2 mm thick.

It should be possible to dispense with a photo resistant and go with a "direct write" exposure of the foils in a liquid, with the laser activating the liquid to etch the foil.

While these methods don't have the total flexibility of metal deposition of 3D printing they are likely to be much faster to produce a large unit quickly (or many small units as a block).

Just some possibilities which are also additive but not 3D printing.

[HMXHMX]

Note that although "3D printing" is fairly new you should keep in mind that additive techniques are at least half a century old.

Quite a lot of Aerojet designs used a combination of photoetched foils diffusion bonded into stacks.

The technology is also used by Velocisys and Meggit to build "printed circuit" heat exchangers and chemical reactors to deliver so called "process intensification."

Personally I always thought Aerojet could have pushed it much harder. They tended to do the stuff flat and then press (or use high pressure gas) to get it to shape.

Obvious extensions that came to mind were :-

Constructing parts as blocks but with either the final part inside the block, or internal cavities, defined by "perforations" around the outline. The little segments left holding parts inside the block would be quick to etch away, freeing the component.

Stretching or bending the unbonded foils should be much easier than doing it to the finished product, provided layer alignment can be preserved. It would mean that once the layers were bonded together they would need to have their edges trimmed to give the right size.

A technique in MEMS mfg is the use of "sacrificial" layers that can be preferentially etched to release objects.

Making structures that are curved as you go down the layers smoothly is probably too difficult. However by using a smaller number of masks could give a more viable "stepped" structure. Those steps should be preferentially etched, giving a (relatively) smooth result.

OTOH curves in the plane are relatively simple. Generally curves give smoother fluid flow.

It should be possible to fabricate in situ sensors based on fluids effects on the resonance frequencies of various structures, being driven and read by various acoustic transducers. Embedded electrical sensors are likely to more difficult due to the need to create insulating and encapsulating layers inside the structures.

By combining sub units split along different planes it would be possible to make more complex structures. this is relevant because of the difficulty of putting curves through layers.

Layer thickness can also be varied. Historically they have been foils the same thickness, but they could be substantially thicker, from a few 0.002" up to say 1 or 2 mm thick.

It should be possible to dispense with a photo resistant and go with a "direct write" exposure of the foils in a liquid, with the laser activating the liquid to etch the foil.

While these methods don't have the total flexibility of metal deposition of 3D printing they are likely to be much faster to produce a large unit quickly (or many small units as a block).

Just some possibilities which are also additive but not 3D printing.

I got the chance to try "platelet" fabrication technology in 1998-99 when we built this engine, which was LOX cooled, 2400 psia Pc design pressure, 6.6K-lbf.  As can be seen from the photos, individual copper foils were assembled in a stack and then diffusion bonded together.  It wasn't cheap at the time costing about $80K, but we fired it 40 times and it worked well.

Edit: spelling

[john smith 19]

I got the chance to try "platelet" fabrication technology in 1998-99 when we built this engine, which was LOX cooled, 2400 psia Pc design pressure, 6.6K-lbf.  As can be seen from the photos, individual copper foils where assembled in a stack and then diffusion bonded together.  It wasn't cheap at the time costing about $80K, but we fired it 40 times and it worked well.

Wow. 

That is a really nice piece of hardware. Beautiful surface finish.  Nice flame (are those shock diamonds?) . What was the fuel?

I'd seen references to a LOX cooled engine (was not sure it was pressure fed) but did not realize it was also a platelet and at such a high pressure (given the troubles of the SSME I'd guessed people would have preferred to keep the chamber pressure below say 1500 psi).

Just to be clear this is a LOX cooled engine in Copper. The common belief is such a thing would burn out at the slightest imperfection as the hot pure O2  hits the equally hot Copper.

Except it didn't.   

Am I right in thinking the stack was not inside a heated press but the pressure was applied by bolts? Putting the stack in a furnace and using the expansion differential between the bolts and the stack to generate the pressure?

The other question would be did this stack include the nozzle section or did this run nozzleless?

[HMXHMX]

I got the chance to try "platelet" fabrication technology in 1998-99 when we built this engine, which was LOX cooled, 2400 psia Pc design pressure, 6.6K-lbf.  As can be seen from the photos, individual copper foils where assembled in a stack and then diffusion bonded together.  It wasn't cheap at the time costing about $80K, but we fired it 40 times and it worked well.

Wow. 

That is a really nice piece of hardware. Beautiful surface finish.  Nice flame (are those shock diamonds?) . What was the fuel?

I'd seen references to a LOX cooled engine (was not sure it was pressure fed) but did not realize it was also a platelet and at such a high pressure (given the troubles of the SSME I'd guessed people would have preferred to keep the chamber pressure below say 1500 psi).

Just to be clear this is a LOX cooled engine in Copper. The common belief is such a thing would burn out at the slightest imperfection as the hot pure O2  hits the equally hot Copper.

Except it didn't.   

Am I right in thinking the stack was not inside a heated press but the pressure was applied by bolts? Putting the stack in a furnace and using the expansion differential between the bolts and the stack to generate the pressure?

The other question would be did this stack include the nozzle section or did this run nozzleless?

I don't want to take us too far off-topic, but the quick answers are:

The chamber was designed for 2400 psia at a 2.7:1 o/f mix ratio using LOX-Jet A, but I believe we never tested above about half pressure, though there was no indication it wouldn't have operated at full design pressure w/o incident.  I can't recall all the details.

The throat that was used wasn't flight-weight, and was transpiration-water-cooled.  Yes, those are shock diamonds.

I'm not conversant with all the manufacturing details, beyond knowing it was diffusion bonded, but I do believe the bolts provided the pressure under the expansion induced loading, as you suggest.

One side note about LOX cooling.  Much nonsense has been written about it.  Of course, GOX and hot metal don't mix.  But we inadvertently ran the "LOX-leak-into-the-chamber" experiment with this TCA.  The part was delivered with a hairline flaw – there was a microscopic failure to bond between two foils, about 1/3 of the way down the chamber and running maybe 5% of the circumference.  So LOX leaked out of that tiny gap and into the chamber.

Contrary to popular belief, not only didn't it catch on fire, there was absolutely no difference in the coloration or surface finish after dozens of firings.  The reason is obvious: the local o/f ratio goes significantly LOX "rich" at the crack, and thus the surface cools rather than heats.  Several years before NASA Lewis engineers saw the same thing when they deliberately induced flaws into a LOX-regen test article.

[john smith 19]

<chamber description removed for brevity>

Thank you, that was very interesting to me. Pressure feeding a 2400psi chamber for any significant length of time is going to need a very substantial test set up. Was the nozzle added after chamber fabrication? It's getting a smooth interior contour I'm having trouble with working it out. 

One side note about LOX cooling.  Much nonsense has been written about it.  Of course, GOX and hot metal don't mix.  But we inadvertently ran the "LOX-leak-into-the-chamber" experiment with this TCA.  The part was delivered with a hairline flaw – there was a microscopic failure to bond between two foils, about 1/3 of the way down the chamber and running maybe 5% of the circumference.  So LOX leaked out of that tiny gap and into the chamber.

Contrary to popular belief, not only didn't it catch on fire, there was absolutely no difference in the coloration or surface finish after dozens of firings.  The reason is obvious: the local o/f ratio goes significantly LOX "rich" at the crack, and thus the surface cools rather than heats.  Several years before NASA Lewis engineers saw the same thing when they deliberately induced flaws into a LOX-regen test article.

This really needs to be more widely known.

TBH I'd expected some signs of a leak but none at is even better. I'd also note that the results with NASA (which ran LH2/LO2) are even better, given the very wide combustion range of H₂

Something I've never understood about SpaceX is that if you're interested in engine reuse and avoiding coking issues logically you need to run on LOX for the coolant, but they don't, which seems odd.

[Prober]

I got the chance to try "platelet" fabrication technology in 1998-99 when we built this engine, which was LOX cooled, 2400 psia Pc design pressure, 6.6K-lbf.  As can be seen from the photos, individual copper foils where assembled in a stack and then diffusion bonded together.  It wasn't cheap at the time costing about $80K, but we fired it 40 times and it worked well.

Got to love it

Excellent examples of additive thinking.  This engine would translate perfect in todays environment. 

As is the design could be duplicated by laser cutting or plasma jet cutting.  Know many 17-21 year olds who have built lasers that could do the job.  One guy I know built a plasma cutter he has installed in his dad's garage.

Might take some effort but the design would also translate very well into additive printing methods, just about all methods.

HMX if sometime in the future you wish to have some real fun.  Put the design out there on a blog and let one of the colleges pick it up as a project.  Would be fascinating to see it further developed.

[Prober]

When I first started at Masten in 2004, we were working on a GOX/GH2 catalytic igniter, and were looking at doing a metal 3d printed part as a way to get the intimate mixing you need to make that type of system work. While I agree wholeheartedly that our change to just doing spark torch igniters was the right call, I almost wish we had gone through with it, because we probably could've claimed to be the first company using 3d printing for rocket engine parts... Oh well. :-)

~Jon

you might find this interesting

3D printed hot gas, film cooled thrusters.

 

[HMXHMX]

<chamber description removed for brevity>

Thank you, that was very interesting to me. Pressure feeding a 2400psi chamber for any significant length of time is going to need a very substantial test set up. Was the nozzle added after chamber fabrication? It's getting a smooth interior contour I'm having trouble with working it out. 

One side note about LOX cooling.  Much nonsense has been written about it.  Of course, GOX and hot metal don't mix.  But we inadvertently ran the "LOX-leak-into-the-chamber" experiment with this TCA.  The part was delivered with a hairline flaw – there was a microscopic failure to bond between two foils, about 1/3 of the way down the chamber and running maybe 5% of the circumference.  So LOX leaked out of that tiny gap and into the chamber.

Contrary to popular belief, not only didn't it catch on fire, there was absolutely no difference in the coloration or surface finish after dozens of firings.  The reason is obvious: the local o/f ratio goes significantly LOX "rich" at the crack, and thus the surface cools rather than heats.  Several years before NASA Lewis engineers saw the same thing when they deliberately induced flaws into a LOX-regen test article.

This really needs to be more widely known.

TBH I'd expected some signs of a leak but none at is even better. I'd also note that the results with NASA (which ran LH2/LO2) are even better, given the very wide combustion range of H₂

Something I've never understood about SpaceX is that if you're interested in engine reuse and avoiding coking issues logically you need to run on LOX for the coolant, but they don't, which seems odd.

Yes, the nozzle was added to the flight-weight chamber.  There was no particular problem with tolerances or seams.

[john smith 19]

Yes, the nozzle was added to the flight-weight chamber.  There was no particular problem with tolerances or seams.

Yes I suspected the nozzle would be added after the chamber was built. I didn't think tolerances would a problem as long as the foils did not slip relative to each other.

Thank you for that detailed description. While 3D printing (especially of metals) is quite new it's important to remember that "additive" mfg has a long history and includes both these platelet methods and "electroforming" (which has also been used to make TCA's in the past).

When I first started at Masten in 2004, we were working on a GOX/GH2 catalytic igniter, and were looking at doing a metal 3d printed part as a way to get the intimate mixing you need to make that type of system work. While I agree wholeheartedly that our change to just doing spark torch igniters was the right call, I almost wish we had gone through with it, because we probably could've claimed to be the first company using 3d printing for rocket engine parts... Oh well. :-)

Interestingly it was Aerojet (using platelets) that got this project to work in the first place, after an earlier attempt (IIRC without premixing before injection into the bed) produced poor results.

This was on the NTRS and was done sometime in the early 1970's.

[Prober]

NJ Engineer 3D Prints Entire Open Source Liquid Fueled Rocket Engine  http://3dprint.com/48179/3d-printed-rocket-engine/

"Sortino used a binary mixture of stainless steel and bronze to 3D print the engine components because of its hardness and high heat transfer. The total cost to have the parts 3D print was rather low. The 3D printed igniter ran Sortino $60, the injector $80, and the Engine $260, for a total of just $400 for the entire setup."

“While others (SpaceX/NASA) have 3D printed rocket engines recently, I’m pretty sure that I’m one of the first (or only) people to open source a rocket engine design,” explained Sortino. “A big reason for this is that there was traditionally a lot concern about releasing rocket engine information online due to ITAR requirements."

[john smith 19]

NJ Engineer 3D Prints Entire Open Source Liquid Fueled Rocket Engine  http://3dprint.com/48179/3d-printed-rocket-engine/

"Sortino used a binary mixture of stainless steel and bronze to 3D print the engine components because of its hardness and high heat transfer. The total cost to have the parts 3D print was rather low. The 3D printed igniter ran Sortino $60, the injector $80, and the Engine $260, for a total of just $400 for the entire setup."

“While others (SpaceX/NASA) have 3D printed rocket engines recently, I’m pretty sure that I’m one of the first (or only) people to open source a rocket engine design,” explained Sortino. “A big reason for this is that there was traditionally a lot concern about releasing rocket engine information online due to ITAR requirements."

keep in mind that (like HMX's package) this is also a pressure fed "Thrust chamber" Assembly"

Like HMX's TCA it's not clear that this unit had a nozzle (although in principle) that should be possible.

But for the price it's impressive.

This leaves the challenge of some kind of pump.

A couple of options exist for this. Doing a turbo pump is going to be very demanding. Probably easier to do individual blades and assemble afterward. SOP for gas turbines.

Option B is to do a positive displacement pump like John whiteheads team at Livermore, or the work at XCOR or flometrtrics. Flow metrics is (was, I'm not sure they are still around) the simplest, using a high pressure tank to pump low pressure fluids.

Whitehead believes a pump fed SSTO design small enough to put on the back of a pickup truck is possible. 

[Asteroza]

NJ Engineer 3D Prints Entire Open Source Liquid Fueled Rocket Engine  http://3dprint.com/48179/3d-printed-rocket-engine/

"Sortino used a binary mixture of stainless steel and bronze to 3D print the engine components because of its hardness and high heat transfer. The total cost to have the parts 3D print was rather low. The 3D printed igniter ran Sortino $60, the injector $80, and the Engine $260, for a total of just $400 for the entire setup."

“While others (SpaceX/NASA) have 3D printed rocket engines recently, I’m pretty sure that I’m one of the first (or only) people to open source a rocket engine design,” explained Sortino. “A big reason for this is that there was traditionally a lot concern about releasing rocket engine information online due to ITAR requirements."

I fail to see how opensourcing the design will absolve him of ITAR duties...

[Paul451]

Doing a turbo pump is going to be very demanding. Probably easier to do individual blades and assemble afterward. SOP for gas turbines

The trend for pumps and turbines is towards all-in-one printing. Specifically because 3d printing allows you to produce all the fiddly bits at the same time, drastically reducing time/labour in subsequent assembly.

Either directly sintered:


Or using 3d-printed sand-casting moulds:

[docmordrid]

More on impellers

http://www.calraminc.com/newsletters/Impeller_Paper.pdf

Discussion

This limited study of the electron-beam, layer-build process produced three impellers with all required drawing details. It also demonstrated that surface finishing techniques presently available are capable of producing finishes sufficiently smooth for operational use. Work planned for the coming year will include a detailed dimensional capabilities analysis; however, preliminary findings are favorable.  The mechanical properties results indicate that tensile and yield strengths are comparable to wrought product, such as forgings, while the ductility and toughness at cryogenic temperatures are superior. The very good ductility and notched toughness obtained are due undoubtedly to the very fine grain size resulting from the rapid solidification pattern of this particular process. Even without an oxygen content meeting that specified for ELI grade, the elongation and reduction in area values obtained at liquid hydrogen temperatures are over twice those typical for wrought Titanium-6Al-4V ELI and the notched-to-unnotched ratio is nearly equivalent to the more ductile, but less producible Titanium-5Al-2.5Sn ELI alloy. (Note that the minimum -253C notched tensile ratio for AMS 4930 Ti-6Al-4V ELI was 0.75 before this requirement was dropped from the latest versions of the specification). Although more work needs to be done, it would appear that the electron beam, layer-build process is viable for the production of complex hardware. The only limitation is one of size, as the working bed of present machines is a 12” diameter x 8” high.

Also,

http://www.esmats.eu/amspapers/pastpapers/pdfs/2014/merriam.pdf

[Prober]

Aerojet Rocketdyne Hot-Fire Tests Additive Manufactured Components for the AR1 Engine to Maintain 2019 Delivery   http://finance.yahoo.com/news/aerojet-rocketdyne-hot-fire-tests-130000241.html


Aerojet Rocketdyne Successfully Completes Hot-Fire Tests of 3D Printed Rocket Engine Injectors
 http://3dprint.com/51525/3d-printed-rocket-engine-injectors

A lot more info in this 2nd article including a video (not working atm)


Edit: add 2nd article

[john smith 19]

More on impellers

http://www.calraminc.com/newsletters/Impeller_Paper.pdf

This was a very interesting paper, showing both the (current) capabilities of the technology and the limitations.

Note that while the core materials being laid down certainly seem up to the job you need "post processing" to produce a finished part. 350 micro inches is about 8.9 micro metres, whereas machined blades (IIRC) run about 1/2 to 1/4 of that surface roughness, hence the need for "liquid honing" to polish the part. This is a nice process as it's fairly close to room temperature and pressure, as opposed to the highly specialized equipment for HIP.

An interesting idea (which I think has been done in plastic) is to switch to a finer grade of metal powder to give the surface finish. The downside is a more complicated feed mechanism.

I was quite surprised at size limit however. I'm sure I recalled someone printing Titanium like this in a chamber 19 feet long. 

 

[docmordrid]

http://www.mmsonline.com/blog/post/metal-additive-manufacturing-for-parts-up-to-7-feet-tall

>
One of the constraints on additive manufacturing machines that make metal parts from powder has been the relatively small build envelope of these machines. Rapid City, South Dakota-based RPM Innovations is now prepared to challenge that constraint with laser deposition additive manufacturing machines that have a build envelope of 5 ×5 ×7 feet. An 83-inch-tall rocket-like part made from Inconel 625 that was grown in one of this company’s machines will be on display in the Advanced Manufacturing Center at IMTS.
>
Nearly 80 percent of its applications have been related to aerospace or defense, including aircraft engine components and aircraft structural components for “companies whose names you’d recognize,” he says. Inconel 625, Inconel 718 and titanium 6-4 are among the alloys that the machines apply routinely.

The rocket-like part took around 340 hours to build is approximately 7,000 layers, he says. And to the RPM staff, that is not all that long. “We have had big parts—not as tall as this, but broader and a lot more complex—that took us 1,800 hours to build,” Mr. Mudge says.
>

[john smith 19]

http://www.mmsonline.com/blog/post/metal-additive-manufacturing-for-parts-up-to-7-feet-tall

>
One of the constraints on additive manufacturing machines that make metal parts from powder has been the relatively small build envelope of these machines. Rapid City, South Dakota-based RPM Innovations is now prepared to challenge that constraint with laser deposition additive manufacturing machines that have a build envelope of 5 ×5 ×7 feet. An 83-inch-tall rocket-like part made from Inconel 625 that was grown in one of this company’s machines will be on display in the Advanced Manufacturing Center at IMTS.
>
Nearly 80 percent of its applications have been related to aerospace or defense, including aircraft engine components and aircraft structural components for “companies whose names you’d recognize,” he says. Inconel 625, Inconel 718 and titanium 6-4 are among the alloys that the machines apply routinely.

The rocket-like part took around 340 hours to build is approximately 7,000 layers, he says. And to the RPM staff, that is not all that long. “We have had big parts—not as tall as this, but broader and a lot more complex—that took us 1,800 hours to build,” Mr. Mudge says.
>

That's impressive given these are high temperature Nickel based alloys but the one I'm thinking of is another electron beam in a vacuum system outfit.

Can't for the life of me think who they are.