Relativity Space: General Thread

[trimeta]

The thread title should be fixed. It is impossible to build and operate rockets on Earth without human labor. Even if all parts are printed and then assembled by robots, there will be some work left for humans, including programming and servicing the printers and robots. And then humans will sell launches, integrate payloads and operate the rockets.

"zero human labor" is just empty PR speak.

And I don't think it's even PR speak they use these days. I assume it's an artifact of their earlier claims.

[su27k]

Thirdly: copper >> Inconel. No, really? This company is what, 5 years old? And making statements like this? Kindergarden.

Copper based alloys are head and shoulders above nickel based alloys for conducting heat away. No alloy is a Swiss Army knife.

Yup.

Ringside, I have my share of skepticism for the idea of printing rocket tanks (etc), but you don’t know what you’re talking about for rocket engine alloys.

When I read that sentence, I assumed ringsider meant using copper in engines is common knowledge, not something you brag about in a press release. I mean I know close to nothing about rocket engines, even I know copper is used in engines due to the whole "green flame of death" thing.

[JEF_300]

Ringside, I have my share of skepticism for the idea of printing rocket tanks (etc), but you don’t know what you’re talking about for rocket engine alloys.

Attack the post, not the poster.

[ringsider]

Ringside, I have my share of skepticism for the idea of printing rocket tanks (etc), but you don’t know what you’re talking about for rocket engine alloys.

When I read that sentence, I assumed ringsider meant using copper in engines is common knowledge, not something you brag about in a press release.

Precisely.

[edzieba]

In an interview, Relativity CEO Tim Ellis said the company recently printed the second stage that will be used on the inaugural flight of the Terran 1 rocket, which is presently scheduled to take place before the end of 2021. The stage was printed at a rate of about 1 linear foot per day, so it took about three weeks in total to print the 20-foot tall second stage.

3 weeks for a Stage 2 tank? So that is what, 12-15 weeks for a Stage 1 tank? How is this a revolution in manufacturing? SpaceX are making a new Starship a week by manually welding steel plate!

Don't confuse end-to-end manufacturing time for parallelised manufacturing rate. You can spend a year to make a part and have parts rolling off the line once a day if you have 365 parts in production in parallel. Your comparison is apples to oranges, so of no value in comparing production time.
For example: the SN15 common dome was spotted being sleeved mid-November 2020. Even if we take that part as the first ever part of SN15 to be made and that the dome and barrel section were fabricated that same day, that's still a minimum build time of 5 months (assuming rollout and engine fitting later this month) at a build site with something close to 1000 staff now.

The lack of touch labour is the interesting part. The time-lapse for the S2 build showed one mid-build touch to flip the forward dome from dome-up to dome-down. The rest is hands-off manufacture. That means one team can monitor a large fleet of printers working in parallel, rather than one or more teams per rocket body. In addition, a lot of the manual fitout (brackets & mount-points, thrust structure, plumbing, etc) is built in as part of the initial fabrication, so finishing steps are reduced compared to sheet-stock fabrication. No need to fabricate brackets, fabricate piping, weld brackets to body (and inspect), weld piping to length, inspect piping, install piping, QC assembly, when your piping run was built as part of the body and inspect-as-you-build (you literally have a cross-sectional view of the entire vehicle) is in effect.

[Robotbeat]

Ringside, I have my share of skepticism for the idea of printing rocket tanks (etc), but you don’t know what you’re talking about for rocket engine alloys.

When I read that sentence, I assumed ringsider meant using copper in engines is common knowledge, not something you brag about in a press release.

Precisely.

Well in their defense, 3D printing copper is kind of a challenge, although they’re not the first to do it by far. The high thermal conductivity and high reflectivity to IR light means its pretty tough to laser sinter it. May need quite a bit of tweaking to get it to work.

[playadelmars]

In an interview, Relativity CEO Tim Ellis said the company recently printed the second stage that will be used on the inaugural flight of the Terran 1 rocket, which is presently scheduled to take place before the end of 2021. The stage was printed at a rate of about 1 linear foot per day, so it took about three weeks in total to print the 20-foot tall second stage.

3 weeks for a Stage 2 tank? So that is what, 12-15 weeks for a Stage 1 tank? How is this a revolution in manufacturing? SpaceX are making a new Starship a week by manually welding steel plate!

Don't confuse end-to-end manufacturing time for parallelised manufacturing rate. You can spend a year to make a part and have parts rolling off the line once a day if you have 365 parts in production in parallel. Your comparison is apples to oranges, so of no value in comparing production time.
For example: the SN15 common dome was spotted being sleeved mid-November 2020. Even if we take that part as the first ever part of SN15 to be made and that the dome and barrel section were fabricated that same day, that's still a minimum build time of 5 months (assuming rollout and engine fitting later this month) at a build site with something close to 1000 staff now.

The lack of touch labour is the interesting part. The time-lapse for the S2 build showed one mid-build touch to flip the forward dome from dome-up to dome-down. The rest is hands-off manufacture. That means one team can monitor a large fleet of printers working in parallel, rather than one or more teams per rocket body. In addition, a lot of the manual fitout (brackets & mount-points, thrust structure, plumbing, etc) is built in as part of the initial fabrication, so finishing steps are reduced compared to sheet-stock fabrication. No need to fabricate brackets, fabricate piping, weld brackets to body (and inspect), weld piping to length, inspect piping, install piping, QC assembly, when your piping run was built as part of the body and inspect-as-you-build (you literally have a cross-sectional view of the entire vehicle) is in effect.

Alright, long post but having never seen anyone try to put objective numbers to Relativity’s costs for printing the rocket primary structure (domes, stiffened tanks, thrust structure, baffles, etc) I wanted to make a go at it. I’m definitely a fan of their long term mission for Mars, but wanted to see for myself how good near term competitiveness could be.

Starting with Terran 1. Taking vehicle dimensions of 7 ft by 100 ft tall, and their print rate of about 1 foot of rocket per day, that is about 22 square feet of rocket shell printed per printer, per day. There are at least 5 printers shown in recent videos, so that’s 110 square feet per day – or a total vehicle print time of 20 days spread across all the printers. But that is sans any internal features like domes, etc, so we can add in a factor of say 1.25x for those features for a total print time of 25 days. Let’s assume it’s not perfectly 24/7 operations, so the 5 printers build about 12 rockets per year. Looking at the printer design, it’s mostly 6-axis robotic arms, a welder of some kind, cameras, and a structure. That maybe costs $1 million unless some special “sauce” I’m missing. A 5 printer factory costs $5M, and produces 60 rocket fuselages over 5 years, or about $83,000 per fuselage amortized cost. Raw aluminum wire looks to be about $6 per pound searching online, theirs is a special custom alloy with unknown cost, but let’s say its closer to $10 per pound due to customization. The rocket thickness I really don’t know, maybe 0.25”? So for the whole height and surface area, that’s about 10,000 pounds give or take, or $100,000 in raw material. Labor rates it appears are just some people watching the printers going automatically, and is perhaps a team of 4-6 people over a year at rate production, or another $40,000 ish per rocket produced.

Added all up, we get $223,000 per Terran 1 rocket fuselage, including labor. Is that cheap? I don’t know at this scale, but it seems so and depends on just how many features are integrated into one print.

Now let’s scale that to Terran R, their new Falcon 9 sized rocket. We don’t know exact dimensions, but let’s scale off of the same cost per square foot printed as Terran 1 – that’s about how this printing approach will scale, is linearly with print surface area and multiplied by number of printers. Cost per square foot printed is $81/sq ft given Terran 1 cost and dimensions above, and includes all the material/labor/printer costs, etc. This assumes the tank thickness for Terran R is the same 0.25”, maybe it is, maybe it’s not. If Terran R is same size as Falcon, it would be 12 foot wide and 230 foot tall, plus a factor for all the extra features. Let’s say about 11,000 sq ft of print area. That would translate to $891,000 cost for a Falcon 9 sized fuselage, and depending on how many printers, could still be built in weeks.

Lastly, scaling to Starship, which is 390 feet by 30 feet. More uncertainty here because it uses stainless steel not aluminum, is a lot larger, and who knows the print speed of something like this. But using similar logic above, if it holds true for stainless, would yield a cost of about $3.7 million for something Starship sized. It would probably start to take longer to print, but those are roughly the estimated costs.

As to whether any of this is competitive, it sure sounds promising to me, and I imagine Relativity investors, to take a shot at building a launch vehicle in this way.

[Pueo]

Regarding the tanks, do we have any idea what, if anything, they're doing with the tank wall structure?  One of the big criticisms of printing tanks is that it's cheaper and faster to just use sheet metal.  The only way 3D printing begins to make sense is if the design needs something like isogrid for a better dry mass fraction.  However in all the images of Relativity's tanks the inside wall is smooth except for the baffles, so they're clearly not using isogrid, nor a ring and stringer design.

They could be doing some sort of sandwich structure, but the old timlapse  of the tank being printed certainly looks like a monocoque:

And the January tweet of the test firing with significant frosting on the tank.
https://twitter.com/relativityspace/status/1352342386060505088

On the other hand the time lapse of the stage 2 flight print is set at an angle that makes it impossible to see a wall cross-section and the Dev 2 Mission Duty Cycle Test shows a tank completely free of frost.
https://twitter.com/relativityspace/status/1362129115898212354

So do they have a new tank wall design that they are keeping hidden for proprietary reasons?  Are they using a graduated monocoque like SpaceX so we only see the monocoque portion of the tank during print videos because they print upside down?  Did they determine that an entirely monocoque structure was sufficient for the second stage or weren't able to lighten it further because they are limited by minimum line width?  If so did they still choose to 3D print the second stage out of pride, or because they determined it was simplest given the relatively small size and ability to include mounting and plumbing in the print?

[FutureSpaceTourist]

https://twitter.com/relativityspace/status/1380267027206602752

We are making exciting progress as we work towards first launch of Terran 1! Check out this timelapse to see how our Stage 1 fuel barrel tank is being 3D-printed at Stargate, our #factoryofthefuture. #RelativitySpace

[matthewkantar]

https://twitter.com/relativityspace/status/1380267027206602752

We are making exciting progress as we work towards first launch of Terran 1! Check out this timelapse to see how our Stage 1 fuel barrel tank is being 3D-printed at Stargate, our #factoryofthefuture. #RelativitySpace

Very impressive that all of those ladders and lifts move around with zero human labor.

[ncb1397]

Very impressive that all of those ladders and lifts move around with zero human labor.

Maybe they aren't doing anything? Just watching...

[russianhalo117]

https://twitter.com/relativityspace/status/1380267027206602752

We are making exciting progress as we work towards first launch of Terran 1! Check out this timelapse to see how our Stage 1 fuel barrel tank is being 3D-printed at Stargate, our #factoryofthefuture. #RelativitySpace

Very impressive that all of those ladders and lifts move around with zero human labor.

Probably chroma keyed the people out.

[JEF_300]

https://twitter.com/relativityspace/status/1380267027206602752

We are making exciting progress as we work towards first launch of Terran 1! Check out this timelapse to see how our Stage 1 fuel barrel tank is being 3D-printed at Stargate, our #factoryofthefuture. #RelativitySpace

Very impressive that all of those ladders and lifts move around with zero human labor.

Probably chroma keyed the people out.

What a colossal and pointless waste of time that would be. This video is going really fast, it may just be that we can't see them. Or the camera is going at like 1 frame every 5 minutes or something. But they definitely did not edit people out of a timelapse just to comply with a claim that they don't even make anymore, and that was pure PR nonsense to begin with.

[Pueo]

Very impressive that all of those ladders and lifts move around with zero human labor.

Also impressive how the dark oxidized patches of the recently printed sections magically become shiny every now and then, almost as if a pixie comes by with an angle grinder to polish it.  Of course, pixies aren't human so it isn't human labor. 

[Alberto-Girardi]

Regarding the tanks, do we have any idea what, if anything, they're doing with the tank wall structure?  One of the big criticisms of printing tanks is that it's cheaper and faster to just use sheet metal.  The only way 3D printing begins to make sense is if the design needs something like isogrid for a better dry mass fraction.  However in all the images of Relativity's tanks the inside wall is smooth except for the baffles, so they're clearly not using isogrid, nor a ring and stringer design.

They could be doing some sort of sandwich structure, but the old timlapse  of the tank being printed certainly looks like a monocoque:

And the January tweet of the test firing with significant frosting on the tank.
https://twitter.com/relativityspace/status/1352342386060505088

On the other hand the time lapse of the stage 2 flight print is set at an angle that makes it impossible to see a wall cross-section and the Dev 2 Mission Duty Cycle Test shows a tank completely free of frost.
https://twitter.com/relativityspace/status/1362129115898212354

So do they have a new tank wall design that they are keeping hidden for proprietary reasons?  Are they using a graduated monocoque like SpaceX so we only see the monocoque portion of the tank during print videos because they print upside down?  Did they determine that an entirely monocoque structure was sufficient for the second stage or weren't able to lighten it further because they are limited by minimum line width?  If so did they still choose to 3D print the second stage out of pride, or because they determined it was simplest given the relatively small size and ability to include mounting and plumbing in the print?

Maybe they 3d printed the 2nd stage for pride, but I think it isn't the inly reason. They could be testing an refining the technology. Maybe starting a second different production line isn't worth it.

[harrystranger]

This photo taken from the International Space Station shows Relativity's pad SLC-16, with what looks like the hangar up at the south end of the pad!

Credit: https://eol.jsc.nasa.gov/

[Phil Stooke]

PM3: "The thread title should be fixed. It is impossible to build and operate rockets on Earth without human labor. Even if all parts are printed and then assembled by robots, there will be some work left for humans, including programming and servicing the printers and robots. And then humans will sell launches, integrate payloads and operate the rockets.

"zero human labor" is just empty PR speak."


The title is unfortunate too because, though we know lots of people are in fact working on it, it seems to be saying 'orbital rockets with massive layoffs in the aerospace sector'.  Not true, but hardly a confidence builder for someone thinking of getting into the industry. 

[Alberto-Girardi]

PM3: "The thread title should be fixed. It is impossible to build and operate rockets on Earth without human labor. Even if all parts are printed and then assembled by robots, there will be some work left for humans, including programming and servicing the printers and robots. And then humans will sell launches, integrate payloads and operate the rockets.

"zero human labor" is just empty PR speak."


The title is unfortunate too because, though we know lots of people are in fact working on it, it seems to be saying 'orbital rockets with massive layoffs in the aerospace sector'.  Not true, but hardly a confidence builder for someone thinking of getting into the industry.

I agree that it is pr speak, but I also question if not having human labor is necessary better. And, aren't current rocket building methods low human labor? For example spacex in boca chica uses many automatixation, but human labor doesn't seem to be the limiting factor in current rocketry.

[trimeta]

Relativity Space is showing off a newly-built structural test stand. I've seen some on Twitter compare this to SpaceX's "nosecone jail" in terms of role and purpose.

https://twitter.com/relativityspace/status/1393248848185679877

[edzieba]

More akin to the blue frameworks at McGregor for S1 and S2 testing than the 'nose jail'.
It's odd that there are two identical (or near-identical) structural stands though. Unless they're expecting to pump out extreme numbers of stages in short enough order to form a queue, one stand would be sufficient.