Reuse business case

[Redclaws]

The economics of this industry is governed by fixed costs over launch rate. The reusability method that enables the highest launch rates will have a considerable advantage that can even overcome higher cost structures -- especially in an era of LEO constellations and hopefully orbital propellant.

The business case for reusable engine pod concept like SMART is really about the efficiency of the the turnaround flow. How often would they be able to launch each engine pod? What's the sustainable production rate for the booster tanks? How well does this system scale to 30, 40, 50 missions per year?

If one system can operate several times as frequently as another system of similar capacity, that's going to be the more economical of the two systems. Period.

I'm excited for the Kuiper launch campaign (whenever it eventually occurs) as an opportunity for ULA to show what kind of cadence they can achieve with the extremely mature Atlas V. The turnaround record for LC-41 is 42 days. ULA has suggested a 30-day turnaround capability. This is disconcerting in the context of contemporary developments in the industry, but it would be nice to see ULA reel off a string of launches at something close to a monthly rate.

I agree that with an RL-10 powered upper stage, a reusable booster really isn't an option. Maybe SMART is a good way to maximize Vulcan flight rate which might otherwise be limited by BE-4 production rate (which is a mystery). Vulcan tanks are not cheap, but Boeing decided way back when to build the production capacity for 60 Delta IV cores per year at Decatur, so it's not completely unreasonable.

SMART could make Vulcan significantly less uncompetitive with Falcon and either a bridge to New Glenn or a complement to New Glenn. A lot depends on how late New Glenn becomes. The farther away New Glenn really is, the more it makes sense for ULA to invest in SMART. If New Glenn is not going to be a high-rate launch system, then ULA should invest in SMART.

We could end up having a real-life demonstration of this thread's premise: reusable engine pod + low-thrust hydrolox upper vs. reusable booster stage + high-thrust hydrolox upper. Both based on the same booster engine. Which will be better? Well, if they both launch once a month, they'll probably cost about the same.

Not to be too much of a downer, but is there any indication whatsoever that ULA is actually working on SMART reuse?

[Veedrac]

The economics of this industry is governed by fixed costs over launch rate. The reusability method that enables the highest launch rates will have a considerable advantage that can even overcome higher cost structures -- especially in an era of LEO constellations and hopefully orbital propellant.

The business case for reusable engine pod concept like SMART is really about the efficiency of the the turnaround flow. How often would they be able to launch each engine pod? What's the sustainable production rate for the booster tanks? How well does this system scale to 30, 40, 50 missions per year?

SMART doesn't make any sense at 50 missions per year. All these analyses are under the assumption of totally inelastic demand, which, with the one exception of Starlink, seems to have held true to date. If there's demand for 50 Vulcans per year in a competitive market, you go with propulsive landings and don't look back.

[butters]

Not to be too much of a downer, but is there any indication whatsoever that ULA is actually working on SMART reuse?

Not to be too much of a downer, but I'm not ruling out the possibility that SMART will be developed because Blue Origin will pay ULA to do it as they become desperate for data on flight-proven BE-4 engines to help with New Glenn. Like I said: the worse things really are for New Glenn, the more likely it is that SMART will make sense, or enough sense that somebody will pay for it.

[soyuzu]

How has reentry wear do with refurbishment costs? You neither rebuild the Octaweb nor reapply heat shields for reentry wear. Simply retire it when the wear is deemed too much. Reentry wear should not be recounted for both stage life and refurbishment cost.

Inspections, repair, engine swaps, cleaning, etc. Note that this only represents a portion of the 10% cost, and note again that I also think 10% seems pessimistic.

These also apply for a rocket that flied up but didn’t experience reentry. Note I am not arguing with the (<)10% figure, but it is obvious  wear and tear is not a good excuse to justify all the extra (expendable) equipment and process incurred by SMART reuse.

I don't understand what point you're making here. Each useful orbit would have its Centaur V, so there'd typically already be one close by regardless of inclination.

There is no discrete “useful” orbits but a continuous spectrum of orbital parameters, and most of these orbits are undesirable as a midpoint to another higher energy orbit. The only useable cases are for interplanetary or lunar missions, but as you agree, these need fuel transfer between stages, which is very hard to coordinate. So unless you carefully planned, most Centaur V will be unusable.

Not to mention most exceed hydrogen will boil off before the opportunity to reuse it, which reduced the effective payload of expendable launches. Of course you can bring a solar powered refrigerator, but that also means mass penalty.

I agree but I do account for expended payload on the last flight. See the Reddit thread for details.

No, in effect you didn’t. Firstly the payload loss is less than 32% you mentioned, Starlink are launched to near ISS inclination, higher than 200km. Judging by ULA’s number for Vulcan https://www.ulalaunch.com/rockets/vulcan-centaur , it likely incurs a near 10% payload loss compared with a minimum LEO corresponds to 22.8t figure, and there are rods released before Starlink separation. So I think the actual recoverable payload is 75% of 22.8t, which you set for the last expendable launch.

For GTO launch it is closer to two thirds, but expendable GTO launches can easily use up all capabilities to boost the payload to a higher SSTO.

[Robotbeat]

Not to be too much of a downer, but is there any indication whatsoever that ULA is actually working on SMART reuse?

Not to be too much of a downer, but I'm not ruling out the possibility that SMART will be developed because Blue Origin will pay ULA to do it as they become desperate for data on flight-proven BE-4 engines to help with New Glenn. Like I said: the worse things really are for New Glenn, the more likely it is that SMART will make sense, or enough sense that somebody will pay for it.

But the conditions would be totally different.

Crazier things have happened, but that makes no sense to me.

[TrevorMonty]

SMART is still being worked on, waiting on Atlas mission next year that will test HIAD. Flying as secondary payload on NASA mission.

SMART will help keep Vulcan launch costs down until they come up with replacement. Any replacement won't be available till late 2020s, assuming they are planning on one.

Sent from my SM-T810 using Tapatalk

[Veedrac]

I don't understand what point you're making here. Each useful orbit would have its Centaur V, so there'd typically already be one close by regardless of inclination.

There is no discrete “useful” orbits but a continuous spectrum of orbital parameters, and most of these orbits are undesirable as a midpoint to another higher energy orbit. The only useable cases are for interplanetary or lunar missions, but as you agree, these need fuel transfer between stages, which is very hard to coordinate. So unless you carefully planned, most Centaur V will be unusable.

Not to mention most exceed hydrogen will boil off before the opportunity to reuse it, which reduced the effective payload of expendable launches. Of course you can bring a solar powered refrigerator, but that also means mass penalty.

Your Centaur can scoot over a few degrees if need be. The stage is designed for these long loitering lifetimes. You'll have to ask Tory Bruno about precisely how the cooling works, but I think it is mostly insulation-based, using boil-off to provide power. Anyhow it's not that relevant whether you believe it'll work; what matters is that ULA intends to do it.

I agree but I do account for expended payload on the last flight. See the Reddit thread for details.

No, in effect you didn’t. Firstly the payload loss is less than 32% you mentioned, Starlink are launched to near ISS inclination, higher than 200km. Judging by ULA’s number for Vulcan https://www.ulalaunch.com/rockets/vulcan-centaur , it likely incurs a near 10% payload loss compared with a minimum LEO corresponds to 22.8t figure, and there are rods released before Starlink separation. So I think the actual recoverable payload is 75% of 22.8t, which you set for the last expendable launch.

For GTO launch it is closer to two thirds, but expendable GTO launches can easily use up all capabilities to boost the payload to a higher SSTO.

If you want to argue reusable payload is more than I gave, aight, but this is a completely different argument.

I don't buy your numbers though. We have two official numbers that both give about 68%, we have a Shotwell saying it's about 30%, we have Musk saying it's <40%, and we have ULA using a 30% hit in their model. And on the other side we have nothing but your unsourced guesswork.

[soyuzu]

Your Centaur can scoot over a few degrees if need be. The stage is designed for these long loitering lifetimes You'll have to ask Tory Bruno about precisely how the cooling works, but I think it is mostly insulation-based, using boil-off to provide power. Anyhow it's not that relevant if you believe it'll work; what matters is that ULA intends to do it.

Change inclination also cost propellant, and a lot more than you think.

And please have a look at ULA’s own paper, passive thermal insulationLH2 boil off can only get down to 2.5%/day, with IVF (use propellant to drive refrigerator) the ideal goal is still as large as 0.1%/day

https://www.ulalaunch.com/docs/default-source/extended-duration/centaur-extensibility-for-long-duration-2006-7270.pdf

I don't buy your numbers though. We have two official numbers that both give about 68%, we have a Shotwell saying it's about 30%, we have Musk saying it's <40%, and we have ULA using a 30% hit in their model. And on the other side we have nothing but your unsourced guesswork.

30% is likely the average for LEO and high energy orbits like GTO. The “official number” you claim are for different orbits, and my“guesswork” are based on the actual Delta-v difference, official M1DVac Isp and widely accepted S2 dry mass.

[Robotbeat]

SMART is still being worked on, waiting on Atlas mission next year that will test HIAD. Flying as secondary payload on NASA mission.

SMART will help keep Vulcan launch costs down until they come up with replacement. Any replacement won't be available till late 2020s, assuming they are planning on one.

Sent from my SM-T810 using Tapatalk

That’s LOFTID, and it’s being paid for by NASA. It’s a test of inflatable heatshields, but other than that it has little to do with SMART. The velocity will be much higher than SMART. It’s testing orbital recovery, not suborbital SMART.

It’s still completely consistent with SMART being all but abandoned by ULA/parents.

[Veedrac]

Change inclination also cost propellant, and a lot more than you think.

A 3 degree inclination change costs ~500 m/s delta v.

And please have a look at ULA’s own paper, passive thermal insulationLH2 boil off can only get down to 2.5%/day, with IVF (use propellant to drive refrigerator) the ideal goal is still as large as 0.1%/day

https://www.ulalaunch.com/docs/default-source/extended-duration/centaur-extensibility-for-long-duration-2006-7270.pdf

That's from 2006. Bruno is going around saying today that they intend Centaur V Mk2 to last 5 to 6 months, and Mk3 to last longer.

https://twitter.com/torybruno/status/1366820061860618241

He has also talked about spacecraft loitering for years in storage orbits.

https://www.reddit.com/r/ula/comments/mngsmk/i_cant_find_a_real_use_for_the_improved/gu9ht2r/

Unfortunately my memory seems to have conflated the “5” in “5 months” with the “years” from the second post, so that part is wrong, but still, there's clearly a goal for these to last a long time.

I don't buy your numbers though. We have two official numbers that both give about 68%, we have a Shotwell saying it's about 30%, we have Musk saying it's <40%, and we have ULA using a 30% hit in their model. And on the other side we have nothing but your unsourced guesswork.

30% is likely the average for LEO and high energy orbits like GTO. The “official number” you claim are for different orbits, and my“guesswork” are based on the actual Delta-v difference, official M1DVac Isp and widely accepted S2 dry mass.

There are two official numbers; the LEO comparisons and the GTO comparisons. Both agree.

If your comments involve math, show your workings so it can be evaluated on its merits. I'm not going to take your word over Musk's and Shotwell's.

[rpapo]

Come on guys.  This argument has been had several times.  The bottom line is: SpaceX has been reusing rockets for several years now.  There have been 19 launches since the last time a fresh Falcon first stage was last used.  The last four launches were all of boosters with more than five launches under their (virtual) belt already.

In the meantime, AFAIK, SMART is still vapor.  A concept without any known implementation and no timetable.  And even if it is implemented, it will save half of the Vulcan booster cost at most.  I won't argue about the time required, because if they actually implement SMART correctly, it should be possible to pre-position new stage bodies for quick mating to recovered engine compartments.

User "Veedrac" lost me the instant he claimed that the fleet average usage for Falcon 9 was less than two flights per rocket.  While that is true if you count every launch since 2010, recovery didn't start until late 2015, and reuse only hit its stride with the Block 5 variant of the rocket, which took advantage of the lessons learned from recovery.  The Block 5 boosters B1046 through B1063 are 18 first stages.  Between them they have launched 68 times.  Now that is an average of less than four flights apiece, but this is skewed low by the fact that three boosters were expended, two were stored or scrapped and six were lost at sea for various reasons.  The seven surviving boosters average more than five launches apiece.

[Veedrac]

User "Veedrac" lost me the instant he claimed that the fleet average usage for Falcon 9 was less than two flights per rocket.  While that is true if you count every launch since 2010, recovery didn't start until late 2015, and reuse only hit its stride with the Block 5 variant of the rocket, which took advantage of the lessons learned from recovery.  The Block 5 boosters B1046 through B1063 are 18 first stages.  Between them they have launched 68 times.  Now that is an average of less than four flights apiece, but this is skewed low by the fact that three boosters were expended, two were stored or scrapped and six were lost at sea for various reasons.  The seven surviving boosters average more than five launches apiece.

Excluding v1.0, which didn't use propulsive landings, doesn't meaningfully change the picture, because there were only 5 of those.

You cannot reasonably amortize only over Block 5 if you want to look at the economics of developing reuse, because previous generation boosters were necessary steps in learning how to build Block 5. They weren't built for fun, and they weren't free.

Similarly, you can't just ignore scrapped or expended boosters and you definitely cannot ignore boosters lost at sea. Those also weren't free. SpaceX didn't get a refund on the reuse work they did. God doesn't say, “whups, I guess your booster fell off the barge, here's your money back.”

[rpapo]

Similarly, you can't just ignore scrapped or expended boosters and you definitely cannot ignore boosters lost at sea. Those also weren't free. SpaceX didn't get a refund on the reuse work they did. God doesn't say, “whups, I guess your booster fell off the barge, here's your money back.”

That's why I gave the "slightly less than four" uses number.  As for blocks 1, 2, 3 and 4, you cannot count them against the average reuse count because they were not a finished product.  Saying that Falcon 9 averages fewer than two uses per rocket is disingenuous, since the first 50-60 launches either could not attempt recovery, or recovery was still an experiment, or reuse was still an experiment.  It only became real with Block 5.

And don't try to claim that SMART will have a perfect recovery record.  Now if and when it gets done, it will be perfectly valid to compare the results once they have stabilized (apples to apples).  But that presumes the project ever gets implemented.  You cannot compare projected results with real results.

To be fair, I also will wait to judge Starship.  Time will tell whether it succeeds or not.  I am not of the camp that says "Starship has made everything else obsolete already!"

[Redclaws]

Not to be too much of a downer, but is there any indication whatsoever that ULA is actually working on SMART reuse?

Not to be too much of a downer, but I'm not ruling out the possibility that SMART will be developed because Blue Origin will pay ULA to do it as they become desperate for data on flight-proven BE-4 engines to help with New Glenn. Like I said: the worse things really are for New Glenn, the more likely it is that SMART will make sense, or enough sense that somebody will pay for it.

This seems an enormous stretch.  Blue Origin desperate enough they’d pay another company to develop a technology that makes their - not BOs - launch vehicle cheaper?  So they can get more data on flown engines?? Uh

[Redclaws]

User "Veedrac" lost me the instant he claimed that the fleet average usage for Falcon 9 was less than two flights per rocket.  While that is true if you count every launch since 2010, recovery didn't start until late 2015, and reuse only hit its stride with the Block 5 variant of the rocket, which took advantage of the lessons learned from recovery.  The Block 5 boosters B1046 through B1063 are 18 first stages.  Between them they have launched 68 times.  Now that is an average of less than four flights apiece, but this is skewed low by the fact that three boosters were expended, two were stored or scrapped and six were lost at sea for various reasons.  The seven surviving boosters average more than five launches apiece.

Excluding v1.0, which didn't use propulsive landings, doesn't meaningfully change the picture, because there were only 5 of those.

You cannot reasonably amortize only over Block 5 if you want to look at the economics of developing reuse, because previous generation boosters were necessary steps in learning how to build Block 5. They weren't built for fun, and they weren't free.

Similarly, you can't just ignore scrapped or expended boosters and you definitely cannot ignore boosters lost at sea. Those also weren't free. SpaceX didn't get a refund on the reuse work they did. God doesn't say, “whups, I guess your booster fell off the barge, here's your money back.”

So while I agree with you about this, reuse development cost can be amortized over all the launches over *quite* a long time.  I mean really when you get down to it, it can be amortized over *all future reused launches*.  Even if there isn’t a lot of market elasticity and some boosters are lost, this has to count for a lot.

A capability has to be really low profit and high cost for it to exist in perpetuity and not pay down its development cost...

[yoram]

User "Veedrac" lost me the instant he claimed that the fleet average usage for Falcon 9 was less than two flights per rocket.  While that is true if you count every launch since 2010, recovery didn't start until late 2015, and reuse only hit its stride with the Block 5 variant of the rocket, which took advantage of the lessons learned from recovery.  The Block 5 boosters B1046 through B1063 are 18 first stages.  Between them they have launched 68 times.  Now that is an average of less than four flights apiece, but this is skewed low by the fact that three boosters were expended, two were stored or scrapped and six were lost at sea for various reasons.  The seven surviving boosters average more than five launches apiece.

Excluding v1.0, which didn't use propulsive landings, doesn't meaningfully change the picture, because there were only 5 of those.

You cannot reasonably amortize only over Block 5 if you want to look at the economics of developing reuse, because previous generation boosters were necessary steps in learning how to build Block 5. They weren't built for fun, and they weren't free.

Similarly, you can't just ignore scrapped or expended boosters and you definitely cannot ignore boosters lost at sea. Those also weren't free. SpaceX didn't get a refund on the reuse work they did. God doesn't say, “whups, I guess your booster fell off the barge, here's your money back.”

So while I agree with you about this, reuse development cost can be amortized over all the launches over *quite* a long time.  I mean really when you get down to it, it can be amortized over *all future reused launches*.  Even if there isn’t a lot of market elasticity and some boosters are lost, this has to count for a lot.

A capability has to be really low profit and high cost for it to exist in perpetuity and not pay down its development cost...

The long time might not be that long. After all SpaceX has already announced they're planning to phase out Falcon over a few years, so there might not be that many (even with a short term high launch rate for Star Link). But their intended successor wouldn't be feasible without the reusability development on F9, so even then it wouldn't be a loss. Such a complex relationship is really though to capture in a simple economic model, you would need to model the successor too. Reality is unfortunately often messy.

[Veedrac]

So while I agree with you about this, reuse development cost can be amortized over all the launches over *quite* a long time.  I mean really when you get down to it, it can be amortized over *all future reused launches*.  Even if there isn’t a lot of market elasticity and some boosters are lost, this has to count for a lot.

A capability has to be really low profit and high cost for it to exist in perpetuity and not pay down its development cost...

Yes, exactly, you need to look at the total number of launches over the timeline you expect a return on investment. Since this analysis is anchored around what's relevant for ULA, 10 launches per year for a decade (100 total) seemed reasonable. For Falcon 9, probably it'll be more like 200 launches before Starship retires it—though that dev cost also pays down for Starship, which isn't a modelled effect—and if not for Starship it might well have reached multiples of that.

[joek]

Excluding v1.0, which didn't use propulsive landings, doesn't meaningfully change the picture, because there were only 5 of those.
...

Better way to look at this type of historical data is ratio of reflights to total flights over trailing N months-quarters-years-whatever. A ratio of 1.0 means 100% reuse; a ratio of 0.0 means no reuse. That provides better visibility into trend-trajectory and avoids the "you cherry-picked period X or v.Y", the "lie of the average" and a lot of other noise. The data is available in Wikipedia, but I'm too lazy to dig it out and to the visualization. (Would not be surprised if some has not already done that.)

Obviously does not apply to forward-looking models such as yours.

[soyuzu]

m
A 3 degree inclination change costs ~500 m/s delta v.

Which is going to cost 15% of total propellant for a Centaur with 10t propellant left. And you omit change of other orbital parameters.

That's from 2006. Bruno is going around saying today that they intend Centaur V Mk2 to last 5 to 6 months, and Mk3 to last longer.

https://twitter.com/torybruno/status/1366820061860618241

He has also talked about spacecraft loitering for years in storage orbits.

Life of stage ≠ Extent stage can last will ZERO boil off. The 2006 paper also targeted 1 year life, albeit with propellant loss. It’s physically impossible to keep Liquid hydrogen from boil off in space without external power input.

Unfortunately my memory seems to have conflated the “5” in “5 months” with the “years” from the second post, so that part is wrong, but still, there's clearly a goal for these to last a long time.

Then good luck with your reuse scheme, ULA only have a couple mission before stage life runs out.

There are two official numbers; the LEO comparisons and the GTO comparisons. Both agree.

If your comments involve math, show your workings so it can be evaluated on its merits. I'm not going to take your word over Musk's and Shotwell's.

Like I said before, you don’t understand orbital dynamics, there is no single “LEO” but an infinite spectrum of orbits regarded as LEO.
 
Launch to 53deg from Cape requires a 43deg launch azimuth, compare with zero azimuth from 28.5deg, which means the earth rotation has about 110m/s less contribution to the final velocity vector due to cosine loss. By vis-viva equation, the 255x286km orbit Starlink separate requires ~60m/s more Delta-v than a 185x185km orbit. So starlink need 170m/s more Delta-v to reach orbit. S2 Isp is 348s, which corresponds to 5% less burn out mass for S2+Starlink.

Stage 2 dry mass is 4500kg, assume 500kg residual fuels, 200kg for separable rods, then total stack mass is 20.8t. So F9 can have 1100kg more payload if launched to 185x185 28.5degree orbit from the Cape, or 16.9t in total, 74% of expendable payload, more if de-orbit is not required (which is very likely for the 22.8t figure)

Frankly speaking the lack of common sense from you (like it’s more costly to launch to ISS, LH2 is not space storageable) make me hardly to believe you are not biased.

[soyuzu]

You cannot reasonably amortize only over Block 5 if you want to look at the economics of developing reuse, because previous generation boosters were necessary steps in learning how to build Block 5. They weren't built for fun, and they weren't free.

Similarly, you can't just ignore scrapped or expended boosters... Those also weren't free. SpaceX didn't get a refund on the reuse work they did.

Why? You can consider pre Block5 and those expended Block5 a totally different vehicle with a reusable Block5.

In this case the first expendable vehicle have launched about 50 times, more than the total of many launch vehicles, which should already justify all development costs. Then the second reusable vehicle only need a fraction of development cost compared with a clean sheet RLV, and has already obtained >5 fleet average life.