Starship heat shield

[sevenperforce]

Very early ('Big Falcon Rocket' era) Starship designs were essentially scaled up Dragons utilising supersonic retropropulsion to expand the bow shock (works with canted-out engines, direct firing ones disrupt the shock) as Red Dragon was intended to demonstrate. Part of what killed Red Dragon was that the Starship design changed to sideways entry with aerodynamic lifting, so that demo mission became internally redundant.

Was there any proposal for getting around cosine losses or were they just going to grit their teeth and bear it?

[Action]

- Found the Phoenix papers. Gumdrop/capsule approach doesn't scale well for large, heavy, low ballistic coefficient vehicles. In order to handle the heating with known TPS you will need Shuttle, or lower, ballistic coefficient. To manage g's and atmospheric variations you will need the ability to generate an L/D of 0.5-1.0. A capsule shaped Starship design would need a diameter of ~26m to match Starship's sideways approach. Not very good for ascent aerodynamics.

- SSTO is a non-starter with current technology. You have to shave design margins pretty slim and payload fractions will be around 1% of gross, if all goes well. Any hiccups in development and you have no payload at all.

John

To the contrary, I think the big capsule shape is quite scalable; I'm not sure why you'd think otherwise.  You have the picture right - it would look like a regular rocket with a hammerhead fairing, maybe a little more conical than cylindrical.  Certainly entities like Boeing, Grumman, and Chrysler thought the big capsule made sense and proposed tremendously large vehicles with immense payloads.  They should have no problem achieving 3-g reentries, usually proposing cg-offsets or small aerodynamic features.  Maybe you think you need better than that, but I'm not sure why you would.  Plus, with lower total heat reentries they're amenable to ablatives, convection cooling, transpiration, or even heat sink heat shields.

The SSTO part of the Phoenix paper is not what I meant to point to - it's not terribly relevant to the current discussion.  Suffice it to say, if it was feasible to perform the proposed mission with a capsule-style SSTO, it would surely be feasible to perform it with a capsule-style second stage of a TSTO which would have much bigger margins.  Otherwise, I think the paper is kind of neat.  It reads like a first draft of the Starship architecture, with an RLV refueled in LEO and then sent to Mars and back.  It predates a lot of the thinking on Mars in situ resource utlization, so it suggests some scheme with the Martian moons, but I can't really fault the guy for not inventing everything.  Clearly he was born thirty years too soon.

[john smith 19]

- SSTO is a non-starter with current technology. You have to shave design margins pretty slim and payload fractions will be around 1% of gross, if all goes well. Any hiccups in development and you have no payload at all.

John

With the understanding you are talking about rocket based VTOVL SSTO.

Excellent find about the table of ballistic coefficients BTW.

[livingjw]

- SSTO is a non-starter with current technology. You have to shave design margins pretty slim and payload fractions will be around 1% of gross, if all goes well. Any hiccups in development and you have no payload at all.

John

With the understanding you are talking about rocket based VTOVL SSTO.

Excellent find about the table of ballistic coefficients BTW.

- SSTO by any known means is pretty iffy. I have modeled all the various approaches over the years for the USAF.

- Best (largest payload to empty wt. & lowest growth factor) SSTO approaches my team modeled over the years were:
    - VTOHL (or VTOVL) tri-propellant (LOx, LCH4, LH2) rocket
    - VTOHL (or VTOVL) tri-propellant (LOx, LCH4, LH2) rocket-scramjet

- All HTO concepts always faired poorly due to their large heavy wings, propulsion systems and takeoff gear. They have great EIsp but horrible ln(Mi/Mf).   DV = g EIsp x ln(Mi/Mf).   

- All SSTO concepts require the best available technology and have relatively low TRLs. They are risky propositions.

John

[livingjw]

- Found the Phoenix papers. Gumdrop/capsule approach doesn't scale well for large, heavy, low ballistic coefficient vehicles. In order to handle the heating with known TPS you will need Shuttle, or lower, ballistic coefficient. To manage g's and atmospheric variations you will need the ability to generate an L/D of 0.5-1.0. A capsule shaped Starship design would need a diameter of ~26m to match Starship's sideways approach. Not very good for ascent aerodynamics.

- SSTO is a non-starter with current technology. You have to shave design margins pretty slim and payload fractions will be around 1% of gross, if all goes well. Any hiccups in development and you have no payload at all.

John

To the contrary, I think the big capsule shape is quite scalable; I'm not sure why you'd think otherwise.  You have the picture right - it would look like a regular rocket with a hammerhead fairing, maybe a little more conical than cylindrical.  Certainly entities like Boeing, Grumman, and Chrysler thought the big capsule made sense and proposed tremendously large vehicles with immense payloads.  They should have no problem achieving 3-g reentries, usually proposing cg-offsets or small aerodynamic features.  Maybe you think you need better than that, but I'm not sure why you would.  Plus, with lower total heat reentries they're amenable to ablatives, convection cooling, transpiration, or even heat sink heat shields.

The SSTO part of the Phoenix paper is not what I meant to point to - it's not terribly relevant to the current discussion.  Suffice it to say, if it was feasible to perform the proposed mission with a capsule-style SSTO, it would surely be feasible to perform it with a capsule-style second stage of a TSTO which would have much bigger margins.  Otherwise, I think the paper is kind of neat.  It reads like a first draft of the Starship architecture, with an RLV refueled in LEO and then sent to Mars and back.  It predates a lot of the thinking on Mars in situ resource utilization, so it suggests some scheme with the Martian moons, but I can't really fault the guy for not inventing everything.  Clearly he was born thirty years too soon.

- I should have been clearer. As the vehicle mass and volume grows (assuming constant vehicle density), the capsule shape will get flatter to maintain ballistic coefficient. This is the direct result of the cube-square relationship between volume and area. Fine for reentry but, it makes ascent through an atmosphere harder and harder.

- My primary concern is ascent. The supersonic drag of a capsule shaped vehicle with a 26 m diameter will be ~10x larger on ascent than Starship (see attached diagram). Some of this can be offset by lowering thrust to weight of the booster, but at the expense of gravity losses. It will have higher losses and thus will require higher propellant mass ratio.

- The diagram shows a capsule design with the same volume and reentry area as the Starship. Each square is 3 meters. I superimposed the capsule's circular reentry area on top of the Starship sideview. I sketched in a 520m^2 rectangle to show that it more or less matches the Starship dimensions. I hope this helps.

John

[HMXHMX]

This thread is not the place to discuss Phoenix; one of these days I may start a thread somewhere at NSF to talk about my designs/ideas from decades past...but not today.  I will note a few quick points to get them on the record.

Several Phoenix vehicles (including the one in the Mars paper) were dual-fuel, mixed-mode.  Some were single-fuel, mixed-mode (variable LH2 mix ratios).  Most would have used active cooling base TPS; one of the benefits of active cooling is it can be used during ascent to alleviate base heating and also can be dialed up when conducting an entry from hyperbolic (or GTO) orbits.  I also proposed running the engines in gas generator mode during entry – essentially a "supersonic retro-propuslion" approach to reduce base heating on entry by blowing off the shock.  This approach also has the potential to increase L/D by reducing drag.

Most all concepts also would have used nose-mounted "aerospikes" (not aerospike engines, but drag reduction devices similar to that used on Trident).  Such spikes can reduce supersonic drag by as much as 75%. 

And the mass fraction argument won't be settled until someone flies one.

Back on topic!

[edzieba]

Very early ('Big Falcon Rocket' era) Starship designs were essentially scaled up Dragons utilising supersonic retropropulsion to expand the bow shock (works with canted-out engines, direct firing ones disrupt the shock) as Red Dragon was intended to demonstrate. Part of what killed Red Dragon was that the Starship design changed to sideways entry with aerodynamic lifting, so that demo mission became internally redundant.

Was there any proposal for getting around cosine losses or were they just going to grit their teeth and bear it?

Deceleration was intended to be more from the 'inflated' shock rather than the retropropulsion, so the angle was a feature rather than a bug.

Though remember that this design was dropped because it traded poorly against an unpowered aerodynamic lifting entry (propellant mass vs. structural mass), at least at Starship's scale. Might trade better at a different scale, and there is a gulf between vehicles the size of Starship, and the largest viable pure drag capsule for Mars entry (MSL is close to that limit) where retropropulsion, inflatable decelerators, negative lift trajectories, etc, can play in the trade space.

[Action]

- I should have been clearer. As the vehicle mass and volume grows (assuming constant vehicle density), the capsule shape will get flatter to maintain ballistic coefficient. This is the direct result of the cube-square relationship between volume and area. Fine for reentry but, it makes ascent through an atmosphere harder and harder.

- My primary concern is ascent. The supersonic drag of a capsule shaped vehicle with a 26 m diameter will be ~10x larger on ascent than Starship (see attached diagram). Some of this can be offset by lowering thrust to weight of the booster, but at the expense of gravity losses. It will have higher losses and thus will require higher propellant mass ratio.

- The diagram shows a capsule design with the same volume and reentry area as the Starship. Each square is 3 meters. I superimposed the capsule's circular reentry area on top of the Starship sideview. I sketched in a 520m^2 rectangle to show that it more or less matches the Starship dimensions. I hope this helps.

John

Understood, thank you.  You won't need to maintain the same ballistic coefficient exactly, because you have a flatter bottom with a higher coefficient of drag.  You also get a better mass fraction to work with from a more structurally efficient shape, so it's lighter per unit volume.  And nothing stops you from using extra aerodynamic features to increase that drag as Starship does, so generally I'd envision a steeper cone for similar descent aerodynamic performance.  Besides, ascent drag losses are relatively small on such large vehicles. 

Even if you are skeptical about all the above, if Starship's engines and mass fraction work as advertised it has performance to spare.  Performance that could reasonably be allocated to safety and reliability, and perhaps more importantly, development time.  If the second stage was VTVL, it would be flying by now, and we wouldn't have this heatshield thread.

I've derailed the thread enough.  I'll let this go.  Thanks all.

[edzieba]

If the second stage was VTVL, it would be flying by now

The second stage IS VTVL, and it HAS been flying.

[sevenperforce]

Very early ('Big Falcon Rocket' era) Starship designs were essentially scaled up Dragons utilising supersonic retropropulsion to expand the bow shock (works with canted-out engines, direct firing ones disrupt the shock) as Red Dragon was intended to demonstrate. Part of what killed Red Dragon was that the Starship design changed to sideways entry with aerodynamic lifting, so that demo mission became internally redundant.

Was there any proposal for getting around cosine losses or were they just going to grit their teeth and bear it?

Deceleration was intended to be more from the 'inflated' shock rather than the retropropulsion, so the angle was a feature rather than a bug.

Though remember that this design was dropped because it traded poorly against an unpowered aerodynamic lifting entry (propellant mass vs. structural mass), at least at Starship's scale. Might trade better at a different scale, and there is a gulf between vehicles the size of Starship, and the largest viable pure drag capsule for Mars entry (MSL is close to that limit) where retropropulsion, inflatable decelerators, negative lift trajectories, etc, can play in the trade space.

I meant cosine losses on ascent. Or was it expected to have separate engines for orbital insertion? And if so, where would they have been placed?

[Action]

If the second stage was VTVL, it would be flying by now, and we wouldn't have this heatshield thread.

If the second stage was [conventional] VTVL [as described on the past few pages], it would be flying [regularly, more than once] by now, and we wouldn't have this heatshield thread.

[inaccurate_reality]

Sorry if it's come up before, but has there been discussion of larger, slab-like TPS tiles to replace hundreds of individual tiles?

I suppose the obvious problem is thermal expansion, which is way more manageable with a small hexagonal tile than a single rectangular tile that stretches across the entire ship. But the upside is, assuming the tiles follow the shuttle recipe exactly and need to be replaced with waterproofed ones every flight, the replacement process for large portions of the ship - anything with simple geometry over a large area, is arguably faster

[john smith 19]

- SSTO by any known means is pretty iffy. I have modeled all the various approaches over the years for the USAF.

I've always found that any modelling's realism is only as good as the realism of the assumptions used to set it up. GIGO is eternal

- Best (largest payload to empty wt. & lowest growth factor) SSTO approaches my team modeled over the years were:
    - VTOHL (or VTOVL) tri-propellant (LOx, LCH4, LH2) rocket
    - VTOHL (or VTOVL) tri-propellant (LOx, LCH4, LH2) rocket-scramjet

The SCramjet result is truly astonishing given AFAIK the best T/W of one of these is 4:1, and they need a big rocket to get they up to operating speed.
  

- All HTO concepts always faired poorly due to their large heavy wings, propulsion systems and takeoff gear. They have great EIsp bu

t horrible ln(Mi/Mf).   DV = g EIsp x ln(Mi/Mf).   

With the understanding that you are talking about air breathing HTO. AFAIK there is no reason an HTO would have any better Isp than a VTO rocket.
But when the Isp is 6x the best rocket Isp available you can now afford those wings 

- All SSTO concepts require the best available technology

True.

and have relatively low TRLs. They are risky propositions.
John

Although there has been significant work by some companies to raise those TRL's.

But then there's actual risk in development and there's perceived risk.

SS's perceived risk is viewed as low.  Time will tell wheather that perception is accurate.

[john smith 19]

Sorry if it's come up before, but has there been discussion of larger, slab-like TPS tiles to replace hundreds of individual tiles?

Some earlier in the thread.  There was a NASA Tech Brief from the 80's or 90's that described a method of slip casting the tiles over a porous mold that was under a slight vacuum to draw the water through it. This allowed near net shape slabs to match contours without machining. The test pieces were IIRC 18"x18" IE 9 standard tiles in 1. Apparently there not worried about thermal effects at this scale.

I suppose the obvious problem is thermal expansion, which is way more manageable with a small hexagonal tile than a single rectangular tile that stretches across the entire ship. But the upside is, assuming the tiles follow the shuttle recipe exactly and need to be replaced with waterproofed ones every flight, the replacement process for large portions of the ship - anything with simple geometry over a large area, is arguably faster

Waterproofing something so large is likely to be tricky. The small tiles each had a specific injection hole through the thin, brittle glass coating for injection of the waterproofing compound. I'd find it very doubtful that one hole would allow the fluid to penetrate a significantly bigger area. My instinct would be you'd need multiple holes.

[schuttle89]

If the second stage was VTVL, it would be flying by now, and we wouldn't have this heatshield thread.

If the second stage was [conventional] VTVL [as described on the past few pages], it would be flying [regularly, more than once] by now, and we wouldn't have this heatshield thread.

So how would this craft scrub off velocity? A capsule? Which earlier in the thread was shown to have to be huge and cause other problems. Otherwise you're back to this whole heat shield issue. Also you're assuming that the reason starship isn't flying right now is the heat shield (it could be gse, ffa approval, raptor 2 problems or anything else). There are thousands of pages on this part of the website talking about design tradeoffs. People argue about one part and ignore everything else that has to work in sync to make the whole system work.

[livingjw]

- This is really not the place for this alternative design discussion, so let me end with one of our reports that covers quite a few alternative designs, unfortunately it does not include a SSTO VTHL or VTVL rocket. Wish it did.

John

[john smith 19]

- This is really not the place for this alternative design discussion, so let me end with one of our reports that covers quite a few alternative designs, unfortunately it does not include a SSTO VTHL or VTVL rocket. Wish it did.

John

Thanks for this. What a very interesting collection of authors in one place

[octavo]

- This is really not the place for this alternative design discussion, so let me end with one of our reports that covers quite a few alternative designs, unfortunately it does not include a SSTO VTHL or VTVL rocket. Wish it did.

John

Thanks for this. What a very interesting collection of authors in one place

Thank you both for all of your informative posts over the years - I mostly lurk, but wanted to say I enjoy reading both of your posts, wherever I find them. The armchair engineers who are convinced their pet idea is a better solution than the team at SpaceX have been able to come up with never fail to stagger me with their hubris.

[edzieba]

Sorry if it's come up before, but has there been discussion of larger, slab-like TPS tiles to replace hundreds of individual tiles?

I suppose the obvious problem is thermal expansion, which is way more manageable with a small hexagonal tile than a single rectangular tile that stretches across the entire ship. But the upside is, assuming the tiles follow the shuttle recipe exactly and need to be replaced with waterproofed ones every flight, the replacement process for large portions of the ship - anything with simple geometry over a large area, is arguably faster

The big problem is not breaking the thing. A large shell made of sintered ceramic is incredibly fragile. I'd be willing to bet it would not even come vaguely close to being self-supporting. If you embed a structural support into it, you are now adding non-functional dry mass. Even if you try and get away with only external handling equipment (e.g. similar to how the STS orbiter payload bay doors required external frames if opened in a 1g environment) then you need to figure out how to attach it to the tank wall in a way that supports it across the entire surface (so a lot of blind attachment points. And ones that need to be unlatched remotely if you ever want to remove it intact) without restraining it when it expands and contracts between cryogenic and re-entry conditions. Along with all the dewatering and waterproofing issues mentioned. And the difficulty in fabricating such an item - first, construct the world's largest monolithic 2200°C furnace...

You earn yourself a lot of extra headaches, for no real benefit.

[RamsesBic]

Sorry if it's come up before, but has there been discussion of larger, slab-like TPS tiles to replace hundreds of individual tiles?

I suppose the obvious problem is thermal expansion, which is way more manageable with a small hexagonal tile than a single rectangular tile that stretches across the entire ship. But the upside is, assuming the tiles follow the shuttle recipe exactly and need to be replaced with waterproofed ones every flight, the replacement process for large portions of the ship - anything with simple geometry over a large area, is arguably faster

The big problem is not breaking the thing. A large shell made of sintered ceramic is incredibly fragile. I'd be willing to bet it would not even come vaguely close to being self-supporting. If you embed a structural support into it, you are now adding non-functional dry mass. Even if you try and get away with only external handling equipment (e.g. similar to how the STS orbiter payload bay doors required external frames if opened in a 1g environment) then you need to figure out how to attach it to the tank wall in a way that supports it across the entire surface (so a lot of blind attachment points. And ones that need to be unlatched remotely if you ever want to remove it intact) without restraining it when it expands and contracts between cryogenic and re-entry conditions. Along with all the dewatering and waterproofing issues mentioned. And the difficulty in fabricating such an item - first, construct the world's largest monolithic 2200°C furnace...

You earn yourself a lot of extra headaches, for no real benefit.

I could add to that. If you were to make the worlds largest furnace and make that shield, how would you fit it to the ship without cracking it? You can't bake it with the ship - the steel would melt.
If any section of that shield breaks off during re-entry the size of the gap would very likely be orders of magnitude larger than any tile.