Relativity Space: General Thread

[Fmedici]

Yeah it's kinda humorous that with everything that has happened all 3 companies in this category finish shipping stages for orbital attempts in a period of <1 month. Firefly in July, ABL are targeting August according to the Kodiak range schedule and Relativity probably no sooner than the ABL launch.

Looking at the schedules I think that it could also be possible to have SpaceX, ULA (with both Atlas V and Delta IV Heavy), NASA (with SLS), Rocket Lab, Virgin Orbit, Astra, Northrop Grumman, Firefly, Abl and Relativity all launching or attempting a launch of their rockets in the span of one and a half month (let's say from mid-July to late August).

[DanClemmensen]

Yeah it's kinda humorous that with everything that has happened all 3 companies in this category finish shipping stages for orbital attempts in a period of <1 month. Firefly in July, ABL are targeting August according to the Kodiak range schedule and Relativity probably no sooner than the ABL launch.

Looking at the schedules I think that it could also be possible to have SpaceX, ULA (with both Atlas V and Delta IV Heavy), NASA (with SLS), Rocket Lab, Virgin Orbit, Astra, Northrop Grumman, Firefly, Abl and Relativity all launching or attempting a launch of their rockets in the span of one and a half month (let's say from mid-July to late August).

You might even get three different SpaceX LVs: F9, FH, and Starship.

[Robotbeat]

I think 3D printing is still a bad tech for making rocket tanks, but it is a good tech for making smaller rocket engines, and the 3D printing tech for that is slowly getting even better, so in some ways it’s not a worse approach than SpaceX took with Falcon.

Early clustering with Terran 1 helps them prepare for Terran-R a little bit. The Aeon-R is probably the largest engine you can reasonably 3D print, although that scale keeps increasing as the tech improves and there are ways to improve the scalability of 3D printing to larger engine thrusts. SpaceX also 3D prints parts of their engines, including Raptor, I think (you can see characteristic design features of 3D printing on some parts of Raptor). And as we know, SpaceX is extremely aggressive with cost on Raptor, so they must think 3D printing is appropriate from a cost perspective for those parts.

But I do think Relativity would be better off with some other method of making rocket tanks.

Printing tanks gives them lot flexibility in early stages while perfecting RLV design. After that printing is ideal for low production rate needed.

They can scale up production for initial ELVs then later use surplus machine capacity for contract work.

This the approach Firefly want to use. Instead of scaling back production facilities when RLVs are on line they just which to contract work.



Sent from my SM-A528B using Tapatalk

Look at how poor the surface quality is on their 3D printed tanks. That translates to lower useful strength and greater weight. And the toollessness is a bit exaggerated. You can’t just make any shape. 3D printing itself has a whole bunch of design constraints, and any change in shape means a bunch of test parts to validate it.

The industry-standard powder bed metal 3D printers they use for engines and stuff, though, COULD be used for contract work. The large-scale 3D printers? Probably not as the design and the process necessarily are highly coupled and it’s not an industry-standard process.

[Robotbeat]

Yeah it's kinda humorous that with everything that has happened all 3 companies in this category finish shipping stages for orbital attempts in a period of <1 month. Firefly in July, ABL are targeting August according to the Kodiak range schedule and Relativity probably no sooner than the ABL launch.

Looking at the schedules I think that it could also be possible to have SpaceX, ULA (with both Atlas V and Delta IV Heavy), NASA (with SLS), Rocket Lab, Virgin Orbit, Astra, Northrop Grumman, Firefly, Abl and Relativity all launching or attempting a launch of their rockets in the span of one and a half month (let's say from mid-July to late August).

We truly are in a Golden Age of new launch vehicle development. Unprecedented except for the earliest decade or two of the space race (actually we might have even more vehicles now) and probably will peter out in the next few years as the market pulls back from investing in new space launch.

[edzieba]

I think 3D printing is still a bad tech for making rocket tanks, but it is a good tech for making smaller rocket engines, and the 3D printing tech for that is slowly getting even better, so in some ways it’s not a worse approach than SpaceX took with Falcon.

Early clustering with Terran 1 helps them prepare for Terran-R a little bit. The Aeon-R is probably the largest engine you can reasonably 3D print, although that scale keeps increasing as the tech improves and there are ways to improve the scalability of 3D printing to larger engine thrusts. SpaceX also 3D prints parts of their engines, including Raptor, I think (you can see characteristic design features of 3D printing on some parts of Raptor). And as we know, SpaceX is extremely aggressive with cost on Raptor, so they must think 3D printing is appropriate from a cost perspective for those parts.

But I do think Relativity would be better off with some other method of making rocket tanks.

Printing tanks gives them lot flexibility in early stages while perfecting RLV design. After that printing is ideal for low production rate needed.

They can scale up production for initial ELVs then later use surplus machine capacity for contract work.

This the approach Firefly want to use. Instead of scaling back production facilities when RLVs are on line they just which to contract work.



Sent from my SM-A528B using Tapatalk

Look at how poor the surface quality is on their 3D printed tanks. That translates to lower useful strength and greater weight. And the toollessness is a bit exaggerated. You can’t just make any shape. 3D printing itself has a whole bunch of design constraints, and any change in shape means a bunch of test parts to validate it.

The industry-standard powder bed metal 3D printers they use for engines and stuff, though, COULD be used for contract work. The large-scale 3D printers? Probably not as the design and the process necessarily are highly coupled and it’s not an industry-standard process.

The limitations of a additive manufactured part depend on the manufacturing mechanism. the constraints on an FDM part are not the same as an SLS/SLM part, and the constraints on a 3-axis plastic FDM part are not the same as those on a 6-axis metal FDM part.
Surface finish also does not directly correlate to part quality (or cast parts would all be junk). As long as the part itself is structurally sound - and here Relativity's print mechanism has the advantage of continuous inspection during deposition, giving you the benefits of a destructive cross-section test, but throughout the entire part - whether you utilise a post-machining op depends on whether surface quality is a necessity or not. Bearing surfaces and mating planes would need it, exterior surfaces not so much.

Anyone who has seen a robotic palletised light-out CNC shop (stock comes in, parts come out, nobody in the building) can see there is no barrier to integrating additive manufacture into unattended manufacture if there is a demand for it. For initial rapid iteration for the very first product trying to add production optimisation prematurely is just adding a rod for your own back, but none of their manufacturing steps preclude it.

[Robotbeat]

[…
Surface finish also does not directly correlate to part quality (or cast parts would all be junk). …

It absolutely does! And yes, cast parts usually have much worse qualities than machined or forged parts, but even polishing the surface alone can improve the strength of cast parts.

The big fat paths used for large scale 3D printing cause stress concentrations between each layer. This is one big reason why FDM parts are stronger in the x-y directions than z direction. Even more than just the reduction in effective cross section.

You can directly see part strength reduce as surface roughness increases.

This sort of thing is literally part of my day job and I’ve studied this sort of thing for years.

[JayWee]

.... even polishing the surface alone can improve the strength of cast parts.

You can directly see part strength reduce as surface roughness increases.

What is the reason for that?

[trimeta]

[snip]

The industry-standard powder bed metal 3D printers they use for engines and stuff, though, COULD be used for contract work. The large-scale 3D printers? Probably not as the design and the process necessarily are highly coupled and it’s not an industry-standard process.

For their large-scale 3D printers, I assume the "contract work"  would include Relativity working closely with the customer on design, not just "give us the .stl" or whatever format is used in industry. It would be a whole service.

[Robotbeat]

[snip]

The industry-standard powder bed metal 3D printers they use for engines and stuff, though, COULD be used for contract work. The large-scale 3D printers? Probably not as the design and the process necessarily are highly coupled and it’s not an industry-standard process.

For their large-scale 3D printers, I assume the "contract work"  would include Relativity working closely with the customer on design, not just "give us the .stl" or whatever format is used in industry. It would be a whole service.

Right, so it’s unlikely they’d find many willing customers for that, as it’s all Relativity-proprietary stuff.

[Robotbeat]

.... even polishing the surface alone can improve the strength of cast parts.

You can directly see part strength reduce as surface roughness increases.

What is the reason for that?

Stress concentrations.

[niwax]

As long as the part itself is structurally sound - and here Relativity's print mechanism has the advantage of continuous inspection during deposition, giving you the benefits of a destructive cross-section test, but throughout the entire part - whether you utilise a post-machining op depends on whether surface quality is a necessity or not.

Not much to inspect when you shred the crystal structures and everything a material scientist cares about into a fine powder. I guess it doesn't come up much anyway when you have to use an allow that can easily be melted and will be suspended in all kinds of goo so it can be extruded. The general reply among all manufacturers, quality sensing companies and engineers I've worked with so far has been "Yeah, it would be nice if someone looked into material properties at some point"

[TrevorMonty]

[…
Surface finish also does not directly correlate to part quality (or cast parts would all be junk). …

It absolutely does! And yes, cast parts usually have much worse qualities than machined or forged parts, but even polishing the surface alone can improve the strength of cast parts.

The big fat paths used for large scale 3D printing cause stress concentrations between each layer. This is one big reason why FDM parts are stronger in the x-y directions than z direction. Even more than just the reduction in effective cross section.

You can directly see part strength reduce as surface roughness increases.

This sort of thing is literally part of my day job and I’ve studied this sort of thing for years.

Extra tank mass means they trade a bit of performance for lower build cost.  Flexibility of design means less bolted parts and human welds, so they should save mass else where on vehicle construction. Still comes down to $kg to orbit.



Sent from my SM-T733 using Tapatalk

[Robotbeat]

[…
Surface finish also does not directly correlate to part quality (or cast parts would all be junk). …

It absolutely does! And yes, cast parts usually have much worse qualities than machined or forged parts, but even polishing the surface alone can improve the strength of cast parts.

The big fat paths used for large scale 3D printing cause stress concentrations between each layer. This is one big reason why FDM parts are stronger in the x-y directions than z direction. Even more than just the reduction in effective cross section.

You can directly see part strength reduce as surface roughness increases.

This sort of thing is literally part of my day job and I’ve studied this sort of thing for years.

Extra tank mass means they trade a bit of performance for lower build cost.  Flexibility of design means less bolted parts and human welds, so they should save mass else where on vehicle construction. Still comes down to $kg to orbit.



Sent from my SM-T733 using Tapatalk

Except I’m pretty confident they have HIGHER build cost by building the tanks entirely out of welds. They’re also not making it all in one piece, but in multiple sections they have to bolt or weld together. Otherwise it’d take too long to print.

[JayWee]

Any guesstimate how much is the material they are using per kg?

[Robotbeat]

Any guesstimate how much is the material they are using per kg?

Dunno what they’re using for the large scale stuff, but metal powder they’re using for engines can be very expensive for the good stuff, like $500-1000/kg, plus the unused powder can only be recycled so many times and needs sifting, etc.

They may be using metal wire for the large scale printing, which should be MUCH cheaper than that, not really any different than sheet metal. But the process itself is pretty expensive. It requires, ironically, a lot of work to ensure good, strong parts. Again, they’re essentially making it entirely out of welds.

[playadelmars]

Doesn’t this only matter if fatigue strength limits are reached? AFAIK, rockets aren’t the same as aircraft and other than turbopumps don’t really see as much LCF and HCF issues vs raw yield and ultimate strength which should be fine pending good elongation.

[Bananas_on_Mars]

The big fat paths used for large scale 3D printing cause stress concentrations between each layer. This is one big reason why FDM parts are stronger in the x-y directions than z direction. Even more than just the reduction in effective cross section.

Good thing that in a thinwalled pressure vessel under internal pressure, hoop stress is a lot higher than axial stress which is favorable to how they print their tanks.

[TrevorMonty]

Any guesstimate how much is the material they are using per kg?

But the process itself is pretty expensive. It requires, ironically, a lot of work to ensure good, strong parts. Again, they’re essentially making it entirely out of welds.

They've spent lot of their R&amp;D money on automated quality control systems. Especially making sure physical properties of finished product are within specifications.

All the launch companies specialize in some fancy technology to make a point of difference then use proven technology elsewhere ie pick your battles.
With RL it was carbon Fibre and electric pump engines. For the the Neutron carbon Fibre with low tech engines and reuseability.
SpaceX F9 low cost build with initially low performance. Then reuseability and higher performing engines. Space Ship its low cost SS construction with ultra high performance Raptor engines.

Astra low tech LVs with low cost mass production being their secret sauce.

ULA speciality is high performance Centuar US. Rest LV is well proven construction methods with high performance engines from externally suppliers.





Sent from my SM-A528B using Tapatalk

[edzieba]

As long as the part itself is structurally sound - and here Relativity's print mechanism has the advantage of continuous inspection during deposition, giving you the benefits of a destructive cross-section test, but throughout the entire part - whether you utilise a post-machining op depends on whether surface quality is a necessity or not.

Not much to inspect when you shred the crystal structures and everything a material scientist cares about into a fine powder. I guess it doesn't come up much anyway when you have to use an allow that can easily be melted and will be suspended in all kinds of goo so it can be extruded. The general reply among all manufacturers, quality sensing companies and engineers I've worked with so far has been "Yeah, it would be nice if someone looked into material properties at some point"

The in-fab-inspection system RS have developed is basically inspecting the weld-pool in real-time as it cools and solidifies, to provide a volumetric record of conditions throughout the entire part volume. Whilst not implemented yet (AFAIK, this was a while ago) they have mentioned in the past use of laser post-heating to control local cooling rates and tune the microstructure of the deposited material.

[Hog]

Doesn’t this only matter if fatigue strength limits are reached? AFAIK, rockets aren’t the same as aircraft and other than turbopumps don’t really see as much LCF and HCF issues vs raw yield and ultimate strength which should be fine pending good elongation.

Bold emphasis mine.

LCF=Low Cycle Fatigue
HCF=High Cycle Fatigue