Starship Expendable Upper Stage and other FrankendRocket Concepts

[Vultur]

If both vehicles are able to manoeuvre during docking, it makes the risk from, for eg, a stuck thruster on the approaching vehicle lower, since the passive vehicle can get out / stay out of the way while the problem is being resolved, reducing the risk of loss of vehicle(s). Likewise, the vehicles can switch roles, with the faulty / less-controllable vehicle becoming the passive partner, reducing the risk of LoM.

That's much harder if one of the vehicles is a large space-station.

I'm not worried about the docking--other than the time it takes.  But I'm very worried about two vehicles doing 2km/s+ burns simultaneously, close to each other. 

If the vehicles are in real time communication with each other and know their distance, doesn't that help a lot?

I agree two large burns with independent errors (even if they're small errors) would be bad.

But if the burns are constantly being adjusted by real time feedback from the other vehicle, then they're no longer independent errors. Or am I missing something?

[JaimeZX]

It makes intuitive sense that if they can coordinate 27 Merlins across three sub-vehicles, they should be able to coordinate 12-18 Raptors across two vehicles. Naturally the vehicle separation imposes more time delay in the signals, but I *feel like* that should be smaller than the time it takes to adjust throttle settings or gimbal angles.
Inter-vehicle positive comms with latency below some threshold would presumably be a go/no-go criteria.
 
I say the above with the caveat that IANARS nor systems engineer and there are almost certainly layers of complexity I'm overlooking.

[xvel]

One has nothing to do with the other, there is load-bearing structure betwen falcon heavy boosters

[JaimeZX]

Sure, and no doubt that compensates for small-scale asymmetries, but there still has to be an active system evaluating and tweaking individual motor thrust / vector and vehicle acceleration.

In military formation take-offs, pilots do this manually by having the lead jet use slightly less than maximum throttle so that the other jet(s) in formation can maintain relative position.  Surely this can be done more effectively when you delete atmosphere (turbulence) and have high-power computers with low-latency communications providing feedback. Heck, I could see SpaceX trying it with a (radiation-hardened) Tesla FSD computer trained on Starship images, since it's a known shape and size and simpler visual background.

[TheRadicalModerate]

If the vehicles are in real time communication with each other and know their distance, doesn't that help a lot?

I agree two large burns with independent errors (even if they're small errors) would be bad.

But if the burns are constantly being adjusted by real time feedback from the other vehicle, then they're no longer independent errors. Or am I missing something?

You have to think about failure modes.  What if they're suddenly not in communication?  What if the lidars fail, or are unexpectedly masked by exhaust plasma?  What if a gimbal gets stuck?  What if you have a multi-engine shutdown on one of the Starships?

You mitigate all of those problems by opening the distance between the two Ships, which means that one of them will have to chase the other (i.e., make an orbital change that forces it to nearly intersect with the target, in such a manner than a short burn can give them essentially the same orbital parameters).  You can mitigate further by not having the Ships start with the same altitude.  But again:  what does the post-burn chase and rendezvous look like when you do that?

If the distance between the two Ships is only a km or two, I suspect this isn't a big deal.  But the trailing ship can cover a few km awfully fast if the leading ship has a multi-engine failure or complete shutdown.  If the gap is, say, 50km, then I'd think that the chance of a collision during the burn is negligible.  But now we're in the realm of a non-trivial post-burn chase.

As Twark will no doubt point out, I'm doing all of this without the benefit of a full-up numerical analysis, because I don't know much about rendezvous techniques, and certainly not techniques for eccentric orbits.  I admit I could be wringing my hands for nothing.  But if there's a pusher, there's no need to wring my hands at all, because the problem doesn't exist, and a couple of other ones go away as well.

There's only one extra trip through the Van Allen belts, easily handled by the crew staying in the solar storm protected location one should have for long interplanetary flights anyways.

Nope, it's a pair for each HEEO orbit.

Normal TLI/TEI:  Two, one just after TLI, another before Earth perigee on return.

TLI/TEI with HEEO refueling:  Four, one up to HEEO apogee, one down to perigee (where TLI is performed), one just after the TLI, and one at Earth perigee on return.

Furthermore, the one around HEEO apogee lasts a long time, and it's during an RPOD where the crew would need complete access to the cockpit.  If you can radiation-shield the whole cockpit, there's no problem.  But that sounds kinda heavy to me.

[Vultur]

If the vehicles are in real time communication with each other and know their distance, doesn't that help a lot?

I agree two large burns with independent errors (even if they're small errors) would be bad.

But if the burns are constantly being adjusted by real time feedback from the other vehicle, then they're no longer independent errors. Or am I missing something?

You have to think about failure modes.  What if they're suddenly not in communication?  What if the lidars fail, or are unexpectedly masked by exhaust plasma?  What if a gimbal gets stuck?  What if you have a multi-engine shutdown on one of the Starships?

well, sure, but couldn't those be dealt with by redundancy (in LIDARs, by doing the burn at less than full throttle so you can compensate for engines shutting down by increasing the throttle on the remaining engines, etc.)?

If this is long-term (not first few Artemis missions) were probably looking at hardware that is much better tested and understood than any space hardware today except maybe F9.

ISS safety rules were developed with the assumption of multiple countries' hardware, all built in very small numbers and not all reusable at all, using computer systems which were by 2020s standards hyper-primitive.

In this scenario we're talking about same-manufacturer, industrially produced and highly reusable, systems using mid-2030s+ computers and talking to/docking with only same-manufacturer systems.

 Not only will the same rules not apply to Starship refueling, the whole philosophy may very well be different.

For example...

Furthermore, the one around HEEO apogee lasts a long time, and it's during an RPOD where the crew would need complete access to the cockpit.

Is it actually necessary that crew do anything during a RPOD? Starship will do totally uncrewed docking and orbital refueling a number of times before any humans ever get onboard, and even once it does fly humans uncrewed will be the majority (tanker -> Ship). So is there actually a need for the crew of a Starship to be at all involved in the process?

NASA might require it, but since we're not discussing the first few Artemis missions, current rules and contracts may not be in place then.

[Paul451]

by doing the burn at less than full throttle so you can compensate for engines shutting down by increasing the throttle on the remaining engines, etc.)?

Probably the other way around, just cut an engine on the other vehicle. The prop load is the same, hence same total delta-v. And it's orbit-to-orbit, you're not dealing with gravity losses as in launch, thrust matters less; as long as the burn is sufficiently "instant" for Mr Oberth. If you blow out a ten minute burn to fifteen minutes, does it matter?

[Speaking of which:

TRM,
What's the Oberth gain from doing the TLI burn from a HEEO perigee rather than LEO? Does it gain you enough advantage (since the TLI ship is effectively "stealing" energy from the depot/tanker's burn to HEEO) to just circularise both ships in HEO and do a much more mundane RPOD there? Then de-circ back into HEEO for the perigee burn. I can't recall if you've mathed this previously in the refuelling thread. Apologies if so. If it works, the burns into HEEO could be done days apart. No coordination required until POD.]

[InterestedEngineer]

by doing the burn at less than full throttle so you can compensate for engines shutting down by increasing the throttle on the remaining engines, etc.)?

Probably the other way around, just cut an engine on the other vehicle. The prop load is the same, hence same total delta-v. And it's orbit-to-orbit, you're not dealing with gravity losses as in launch, thrust matters less; as long as the burn is sufficiently "instant" for Mr Oberth. If you blow out a ten minute burn to fifteen minutes, does it matter?

[Speaking of which:

TRM,
What's the Oberth gain from doing the TLI burn from a HEEO perigee rather than LEO? Does it gain you enough advantage (since the TLI ship is effectively "stealing" energy from the depot/tanker's burn to HEEO) to just circularise both ships in HEO and do a much more mundane RPOD there? Then de-circ back into HEEO for the perigee burn. I can't recall if you've mathed this previously in the refuelling thread. Apologies if so. If it works, the burns into HEEO could be done days apart. No coordination required until POD.]

circularizing a 42,000km elliptical orbit (GTO) takes 1.5km/sec, definitely not worth it.  You'd burn 3km/sec, or half the prop load, doing and undoing it. 

[meekGee]

You have to think about failure modes.  What if they're suddenly not in communication?  What if the lidars fail, or are unexpectedly masked by exhaust plasma?  What if a gimbal gets stuck?  What if you have a multi-engine shutdown on one of the Starships?

You mitigate all of those problems by opening the distance between the two Ships,....

When two jetliners want to use both runways at SFO (which are closely spaced) they have to fly formation, with one jet designated as the lead.

If they get too close to each other, the follower jet breaks off and goes around.

Interestingly though, the same applies when they get too far from each other.

The formation is deemed safe when not too close but also not too far.

[TheRadicalModerate]

TRM,
What's the Oberth gain from doing the TLI burn from a HEEO perigee rather than LEO? Does it gain you enough advantage (since the TLI ship is effectively "stealing" energy from the depot/tanker's burn to HEEO) to just circularise both ships in HEO and do a much more mundane RPOD there? Then de-circ back into HEEO for the perigee burn. I can't recall if you've mathed this previously in the refuelling thread. Apologies if so. If it works, the burns into HEEO could be done days apart. No coordination required until POD.

If you're looking to have an apogee of 500,000km (a little less than your typical BLT, a little more than your typical approach speed for a direct insertion into LLO), going from 200 x 25,000 direct to 200 x 500,000 costs 928m/s.  Going from 200 x 25,000 to 25,000 x 25,000 costs 406m/s, then going on to 25,000 x 500,000 costs 1327m/s.  Total = 1734m/s.  Difference between the two conops is 806m/s.  Non-trivial.  Even if you only want to look at the final TLI, the eccentric version is still 399m/s cheaper than the circular version.

PS:  If you put your HEEO apogee at 12,500, which is in the gap between VAB #1 and #2, the difference is still 853m/s.

[Paul451]

What's the Oberth gain from doing the TLI burn from a HEEO perigee rather than LEO?

If you're looking to have an apogee of 500,000km [...] going from 200 x 25,000 direct to 200 x 500,000 costs 928m/s. Going from 200 x 25,000 to 25,000 x 25,000 costs 406m/s, then going on to 25,000 x 500,000 costs 1327m/s.

No, it would still deorbit from 25,000 x 25,000 back to 200 x 25,000 after refuelling, then burn 200 x 500,000. Circ, then PROD + refuel, then decirc.

What I was trying to compare was a pusher going from LEO direct to TLI (ie, 200 x 200 -> 200 x 500,000) versus a docking-refuelling doing low-middle-low-high, with a circularisation in the middle (ie, 200 x 200 -> 200 x 25,000 -> 25,000 x 25,000 -> (RPOD, refuel, separate) -> 200 x 25,000 -> 200 x 500,000). The only difference I was suggesting was the circularisation, to simplify RPOD. But the 406 + 406 m/s (circ + decirc) cost would be too high to justify it.

[TheRadicalModerate]

What's the Oberth gain from doing the TLI burn from a HEEO perigee rather than LEO?

If you're looking to have an apogee of 500,000km [...] going from 200 x 25,000 direct to 200 x 500,000 costs 928m/s. Going from 200 x 25,000 to 25,000 x 25,000 costs 406m/s, then going on to 25,000 x 500,000 costs 1327m/s.

No, it would still deorbit from 25,000 x 25,000 back to 200 x 25,000 after refuelling, then burn 200 x 500,000. Circ, then PROD + refuel, then decirc.

What I was trying to compare was a pusher going from LEO direct to TLI (ie, 200 x 200 -> 200 x 500,000) versus a docking-refuelling doing low-middle-low-high, with a circularisation in the middle (ie, 200 x 200 -> 200 x 25,000 -> 25,000 x 25,000 -> (RPOD, refuel, separate) -> 200 x 25,000 -> 200 x 500,000). The only difference I was suggesting was the circularisation, to simplify RPOD. But the 406 + 406 m/s (circ + decirc) cost would be too high to justify it.

Circ + decirc from 200 x 25,000 would be 2 * 2227 1467 = 4454m/s 2934m/s.  So it's even costlier.

[Twark_Main]

  If the gap is, say, 50km, then I'd think that the chance of a collision during the burn is negligible.  But now we're in the realm of a non-trivial post-burn chase.

As Twark will no doubt point out, I'm doing all of this without the benefit of a full-up numerical analysis, because I don't know much about rendezvous techniques, and certainly not techniques for eccentric orbits.  I admit I could be wringing my hands for nothing.  But if there's a pusher, there's no need to wring my hands at all

Always better to spend huge amounts of (someone else's) money — pusher hardware R&D, evaluation, and the subsequent extensive flight test campaign — just to save a bit of math.

Anyway, anyone able to run a porkchop plotter is capable of doing the math. Set muSun = muEarth, set the origin "planet" as the chaser vehicle (ie initialized to a circular orbit starting ~100 km behind the perigee intersection), and the destination "planet" is the vehicle already in elliptical orbit. The total delta-v plot will then directly give the total maneuver delta-v (start+end), with no need for the usual v_hyperbolic gravity well correction.

[InterestedEngineer]

the math isn't fancy.

A control loop that keeps two ships within a km of each other during a burn is trivial control loop math compared to say landing a booster or a ship on a pair of chopsticks.

[TheRadicalModerate]

the math isn't fancy.

A control loop that keeps two ships within a km of each other during a burn is trivial control loop math compared to say landing a booster or a ship on a pair of chopsticks.

Boosters aren't crewed.  It remains to be seen if crewed Starships will land on chopsticks--or if they will land crews at all.  A lot depends on how fault-tolerant the system can be made.  The same would be true for close-by, simultaneous, high delta-v maneuvers.

I agree that this is easy if everything is working properly.

[TheRadicalModerate]

I only have the quick-and-dirty formula for only moving the apse line, which isn't what you want to do.  (You also want to do TLI in the same burn, and it's not as simple as applying the law of cosines and having done with it.)  Here's what I get:

ΔvRadial = -tan 6º *  μEarth * (eccentricity=0.73) / ((r = 6562E03) * (vTangential=10,245m/s)) = -454m/s

So it's pretty bad.

I finally got around to figuring out the combined apse line rotation and energy change for if you missed a TLI insertion opportunity and had to incur an extra number of HEEO orbits.  The spreadsheet is here.

With a 200 x 25,000 HEEO inserting into a 200 x 500,000 TLI, a single missed orbit only increases the delta-v by 79m/s.  Two missed orbits require 264m/s.  So it's not nearly as bad as just rotating the apse line all on its own.

[Twark_Main]

Kudos to TheRadicalModerate for doing (by hand!) the math on this.

I presume, at such a low LEO, there's not really an opportunity to "cheat" and have some negative apse line rotation on the first pass in order to make the second/third pass more favorable, correct? I'm pretty sure the trajectory would intersect the atmosphere in that case...


EDIT: Just to immediately contradict myself...

I suppose in theory, since it's a non-impulsive burn, you could start the burn fully prograde and then tilt negative toward the end of the burn. You pay for some tiny cosine loss, but the tradeoff is that (in case the burn ends prematurely) the instantaneous trajectory never dips too far into the atmosphere.

[TheRadicalModerate]

I suppose in theory, since it's a non-impulsive burn, you could start the burn fully prograde and then tilt negative toward the end of the burn. You pay for some tiny cosine loss, but the tradeoff is that (in case the burn ends prematurely) the instantaneous trajectory never dips too far into the atmosphere.

If you know it's gonna take more than one HEEO orbit, then you set the HEEO's apse line so it'll point at the right spot for lunar insertion during the orbit where the TLI is planned.  This is only an issue if some contingency makes the TLI late.  But if you're planning on 2-3 orbits in HEEO, then you have to have some way to dealing with the 2-3x radiation, and some way to mitigate the 2-3x MMOD risk.

You can also choose a higher-energy HEEO, with a higher apogee, so the orbit period is longer.  That'll lengthen your time of flight, but so will incurring more orbits.  However, that will also increase the delta-v penalty if you miss the TLI.  For example, if you double the HEEO apogee altitude to 50,000km, the Moon moves farther in its orbit (7.8º vs. 3.7º for the 25k apogee), and the delta-v penalty increases to 427m/s.

[Twark_Main]

I suppose in theory, since it's a non-impulsive burn, you could start the burn fully prograde and then tilt negative toward the end of the burn. You pay for some tiny cosine loss, but the tradeoff is that (in case the burn ends prematurely) the instantaneous trajectory never dips too far into the atmosphere.

If you know it's gonna take more than one HEEO orbit, then you set the HEEO's apse line so it'll point at the right spot for lunar insertion during the orbit where the TLI is planned.  This is only an issue if some contingency makes the TLI late.

Sure, I got it.

The concept behind "cheating" the apse line negative on the first perigee is to reduce the worst-case (ie limiting) delta-v out of all the available contingency trajectories.

But if you're planning on 2-3 orbits in HEEO, then you have to have some way to dealing with the 2-3x radiation, and some way to mitigate the 2-3x MMOD risk.

Indeed. I presume that (for real mission planning) you'd only be planning those trajectories if you already decided it's okay to accept the risk.

If you have (say) a 90% chance of it working on the first orbit, a 9% chance on the second orbit, and a 0.9% chance on the third orbit, then of course you're going to account for those probabilities when calculating the risk.

You can also choose a higher-energy HEEO, with a higher apogee, so the orbit period is longer.  That'll lengthen your time of flight, but so will incurring more orbits.  However, that will also increase the delta-v penalty if you miss the TLI.  For example, if you double the HEEO apogee altitude to 50,000km, the Moon moves farther in its orbit (7.8º vs. 3.7º for the 25k apogee), and the delta-v penalty increases to 427m/s.

Agreed, the orbital height is an important parameter when it comes to planning. It definitely wouldn't be chosen "willy nilly," it would be planned around exactly this sort of constraint, among others.

[Vultur]

, and some way to mitigate the 2-3x MMOD risk.

Wouldn't a HEEO spend the vast majority of its time at higher altitudes where there's less artificial debris, so the risk would be less than in lower orbits?