Reuse business case

[edzieba]

This is probably because Falcon 9 was designed with Starlink as its main customer.

Kind of, but only ain a roundabout came-at-it-from-the-back kind of way. Falcon was very well placed for launching large masses to LEO but not to great at smaller masses to GEO. That mean large LEO constellations The assumption was that once the capability for cheap(er) mass launches to LEO was available, satellite designers would switch from designing filigree GEO birds to simpler LEO birds and launching more of them. This didn't actually happen in practice - whether just inertia, long lead times, payload customers not demanding that fro msatellite builders, or everyone being burnt by the last time mass constellations were proposed and resulted in several rounds of billion dollar bankruptcies so not wanting to be the first mover in case it happened again - so SpaceX went for "well, we'll do it ourselves them".
tl;dr Starlink was built to fit Falcon 9, but Falcon 9 was built for a lots-of-cheap-sats market that never materialised.

[DanClemmensen]

...Falcon 9 was designed with Starlink as its main customer...

How could that be possible?

It's not. I messed it up. Their stated design vision was "cheap access to space", starting in 2002, and this led to the decision to design and implement booster reuse. The Starlink plan, starting in about 2014, took advantage of cheap access to space. So reuse conceptually came first.

But this means that we must contrast ULA's design philosophy of full service to all NSSL missions with SpaceX's design philosophy of cheap access to space. The two timelines don't align, so comparisons are not perfect, but F9 design started before 2010 and Vulcan in about 2014 and Vulcan design started before any F9 booster had ever landed.

[rpapo]

...Falcon 9 was designed with Starlink as its main customer...

How could that be possible?

It's not. I messed it up. Their stated design vision was "cheap access to space", starting in 2002, and this led to the decision to design and implement booster reuse. The Starlink plan, starting in about 2014, took advantage of cheap access to space. So reuse conceptually came first.

But this means that we must contrast ULA's design philosophy of full service to all NSSL missions with SpaceX's design philosophy of cheap access to space. The two timelines don't align, so comparisons are not perfect, but F9 design started before 2010 and Vulcan in about 2014 and Vulcan design started before any F9 booster had ever landed.

F9 v1.0 design started in 2007 or 2008, with the first launch in June 2010.  It was an exercise in getting the minimum viable product out the door as soon and as cheaply as possible (2-3 years and $400M).  F9 v1.1 design probably started as soon as v1.0 flew, or perhaps a few months before, and that design formed the basis of the rocket we have today.  All the changes since then have been incremental improvements.  Even the legs and gridfins.

[DanClemmensen]

...Falcon 9 was designed with Starlink as its main customer...

How could that be possible?

It's not. I messed it up. Their stated design vision was "cheap access to space", starting in 2002, and this led to the decision to design and implement booster reuse. The Starlink plan, starting in about 2014, took advantage of cheap access to space. So reuse conceptually came first.

But this means that we must contrast ULA's design philosophy of full service to all NSSL missions with SpaceX's design philosophy of cheap access to space. The two timelines don't align, so comparisons are not perfect, but F9 design started before 2010 and Vulcan in about 2014 and Vulcan design started before any F9 booster had ever landed.

F9 v1.0 design started in 2007 or 2008, with the first launch in June 2010.  It was an exercise in getting the minimum viable product out the door as soon and as cheaply as possible (2-3 years and $400M).  F9 v1.1 design probably started as soon as v1.0 flew, or perhaps a few months before, and that design formed the basis of the rocket we have today.  All the changes since then have been incremental improvements.  Even the legs and gridfins.

Sure, but I was trying to figure out why a bunch of really good engineers at a really good company chose in 2014 to design a rocket that did not have a reusable booster, since lots of folks here seem to think they should have. I now think that ULA's decision at the time was correct (not that anyone should care about my opinion). It only looks wrong in retrospect. Had Vulcan flown in 2019 as originally planned, it would look even better. ULA and Tory Bruno are putting the best face they can on the situation they are in.

[rpapo]

ULA and Tory Bruno are putting the best face they can on the situation they are in.

Absolutely.  Not to slight Tory Bruno in the least.  Good man, tough spot.

[meekGee]

SMH.
Price is not cost.
Cost is reduced via reusability.
Price can be anything SpaceX wants, subject to market constraints.

Any increase in price reflects on the inaptitude of the competitors who to this day haven't been able to cause SpaceX to reduce pricing.

It does NOT in any way indicate anything about their costs, as you're trying to claim.

How is this still lost on you?

The difference between SpaceX's price and cost, their margin, is how they're able to afford all the nice things. (And why investors think they are a real company)

Do you have their costs?

Of course you don't, and if you did, it would be proprietary and you wouldn't be allowed to release it.

And neither do you, but it was you who was making the claim that costs have not been going down (and actually that reuse is not reducing cost), as evidenced (per you) by rising prices.  And I'm telling you that even if prices go up, it doesn't mean costs do too. Pricing is determined by what the company can charge, and if all the competitors are expensive, than SpaceX can be too.

OTOH I pointed to the fact that boosters are being turned around in 3 weeks now, as evidence that turn-around pricing is much lower than fabrication and lower than it used to be.  Since the booster spends over a week on barge-back and on pre-launch, my claim is that there's just so much you spend in 1.5 weeks...  What do you think they're doing in there that's costing so much money, as per your claim?

[niwax]

SMH.
Price is not cost.
Cost is reduced via reusability.
Price can be anything SpaceX wants, subject to market constraints.

Any increase in price reflects on the inaptitude of the competitors who to this day haven't been able to cause SpaceX to reduce pricing.

It does NOT in any way indicate anything about their costs, as you're trying to claim.

How is this still lost on you?

The difference between SpaceX's price and cost, their margin, is how they're able to afford all the nice things. (And why investors think they are a real company)

Do you have their costs?

Of course you don't, and if you did, it would be proprietary and you wouldn't be allowed to release it.

And neither do you, but it was you who was making the claim that costs have not been going down (and actually that reuse is not reducing cost), as evidenced (per you) by rising prices.  And I'm telling you that even if prices go up, it doesn't mean costs do to. Pricing is determined by what the company can charge, and if all the competitors are expensive, than SpaceX can be too.

OTOH I pointed to the fact that boosters are being turned around in 3 weeks now, as evidence that turn-around pricing is much lower than fabrication and lower than it used to be.  Since the booster spends over a week on barge-back and on pre-launch, my claim is that there's just so much you spend in 1.5 weeks...  what do you think they're doing in there that's costing so much money, as per your claim?

Do we really have to go through all this again? We have it on the record from SpaceX that even the very first reuse including all the additional checkouts was 50% cheaper than a new one.

[JayWee]

Do you have their costs?
Of course you don't, and if you did, it would be proprietary and you wouldn't be allowed to release it.

Well, there was that quickly deleted video of a meeting with SpaceX higher-up saying launch costs were about $30M two years ago or so.

But answer a simple question: If reuse wasn't really cheaper - would SpaceX launch Starlink 52 out of 53 Starlink launches so far on a reused booster, some of them reused 12 times?
 

[Lee Jay]

SMH.
Price is not cost.
Cost is reduced via reusability.
Price can be anything SpaceX wants, subject to market constraints.

Any increase in price reflects on the inaptitude of the competitors who to this day haven't been able to cause SpaceX to reduce pricing.

It does NOT in any way indicate anything about their costs, as you're trying to claim.

How is this still lost on you?

The difference between SpaceX's price and cost, their margin, is how they're able to afford all the nice things. (And why investors think they are a real company)

Do you have their costs?

Of course you don't, and if you did, it would be proprietary and you wouldn't be allowed to release it.

And neither do you, but it was you who was making the claim that costs have not been going down (and actually that reuse is not reducing cost), as evidenced (per you) by rising prices.  And I'm telling you that even if prices go up, it doesn't mean costs do to. Pricing is determined by what the company can charge, and if all the competitors are expensive, than SpaceX can be too.

OTOH I pointed to the fact that boosters are being turned around in 3 weeks now, as evidence that turn-around pricing is much lower than fabrication and lower than it used to be.  Since the booster spends over a week on barge-back and on pre-launch, my claim is that there's just so much you spend in 1.5 weeks...  what do you think they're doing in there that's costing so much money, as per your claim?

Do we really have to go through all this again? We have it on the record from SpaceX that even the very first reuse including all the additional checkouts was 50% cheaper than a new one.

Do you have their costs?
Of course you don't, and if you did, it would be proprietary and you wouldn't be allowed to release it.

Well, there was that quickly deleted video of a meeting with SpaceX higher-up saying launch costs were about $30M two years ago or so.

But answer a simple question: If reuse wasn't really cheaper - would SpaceX launch Starlink 52 out of 53 Starlink launches so far on a reused booster, some of them reused 12 times?
 

Marginal or full-wrap?

Elon is ON RECORD saying that developing a flyback reusable booster was going to cost at least 1 billion dollars.  Let's assume some sort of Elon multiplier and guess it actually cost 1.5-2 billion.

Does that $30M include amortized development costs?  I'd bet it doesn't.

The business case probably doesn't close until you're launching a lot of rockets.  Maybe SpaceX is at that point now, but they had to become their own customer to get there.

[AmigaClone]

Sure, but I was trying to figure out why a bunch of really good engineers at a really good company chose in 2014 to design a rocket that did not have a reusable booster, since lots of folks here seem to think they should have. I now think that ULA's decision at the time was correct (not that anyone should care about my opinion). It only looks wrong in retrospect. Had Vulcan flown in 2019 as originally planned, it would look even better. ULA and Tory Bruno are putting the best face they can on the situation they are in.

I can think of two good reasons why engineers at ULA decided to design a non-reusable launch vehicle as opposed to one at least partially reusable.

1) The decision to not design a launch vehicle that was at least partially reusable was made early on using recovery costs based on the costs of recovering the Shuttle boosters and refurbishing them as well as refurbishing the orbiters.

2) The number of failed attempts to introduce a reusable launch system in the two decades prior to 2014. Part of ULA's engineering team in 2014 might have even included some who had worked under engineers involved with the McDonnell Douglas Delta Clipper project or the Lockheed Martin VentureStar.

[meekGee]

...Falcon 9 was designed with Starlink as its main customer...

How could that be possible?

It's not. I messed it up. Their stated design vision was "cheap access to space", starting in 2002, and this led to the decision to design and implement booster reuse. The Starlink plan, starting in about 2014, took advantage of cheap access to space. So reuse conceptually came first.

But this means that we must contrast ULA's design philosophy of full service to all NSSL missions with SpaceX's design philosophy of cheap access to space. The two timelines don't align, so comparisons are not perfect, but F9 design started before 2010 and Vulcan in about 2014 and Vulcan design started before any F9 booster had ever landed.

F9 v1.0 design started in 2007 or 2008, with the first launch in June 2010.  It was an exercise in getting the minimum viable product out the door as soon and as cheaply as possible (2-3 years and $400M).  F9 v1.1 design probably started as soon as v1.0 flew, or perhaps a few months before, and that design formed the basis of the rocket we have today.  All the changes since then have been incremental improvements.  Even the legs and gridfins.

Sure, but I was trying to figure out why a bunch of really good engineers at a really good company chose in 2014 to design a rocket that did not have a reusable booster, since lots of folks here seem to think they should have. I now think that ULA's decision at the time was correct (not that anyone should care about my opinion). It only looks wrong in retrospect. Had Vulcan flown in 2019 as originally planned, it would look even better. ULA and Tory Bruno are putting the best face they can on the situation they are in.

Even at the time, SpaceX's reuse intentions and plan were public, and if you weren't turning a blind eye to their ability and appetite, you'd understand that they were going to succeed in following through.

I said at the time that SpaceX lies at the intersection of Vision, Balls, and Competency.  If ULA's board/mgmt had understood that, they'd have reacted differently.

In essence ULA's first reaction was the block buys, which are a procurement-centric weapon.  Basically try to crowd the new guy out, limit his prospect.  All's fair in love and business, but ULA should have realized that the block buy was only a temporary stop gap measure.  It bought them time, but it didn't buy them the future.

So in that time, they had to do something real, and here they fucked up. When they announced Vulcan, (partly as a response to the engine ban) it was clear they were taking the minimalist way forward, essentially betting on SpaceX not being able to make reuse economical or survive in general.

That was a mistake, even at the time. I think some observers saw it even then, but if you look at output from ULA management (whether it's technical like the Sowers spreadsheet, or PR like "their idea of making noise") you'd understand that it was an echo chamber.  They were delusional.

Reuse is not magic.  ULA could have done it. It would have been a lot harder than Vulcan, but as others have pointed out, there's no new physics involved, or even fundamentally new tech.  Just Vision, Balls, and Competence.

In my view of things, ULA had exactly one out of these three.

[freddo411]

I worked up my own  simple model of costs and profit associated with launching expendable vs reusable rockets.   Note that the exact numbers are hypothetical, but meant to be reasonable estimates

You can play with the numbers directly:
https://docs.google.com/spreadsheets/d/1bCKWc9SjZI9TqTFvhGmbp_CaL4mWEGxXLv5eGZDQTiE/edit?usp=sharing

As this is a simple spreadsheet, it is not attempting to scale to multiple years, or to very, very high flight rates.

NOTE, this spreadsheet DOES focus on fixed and recurring costs.  So marginal cost is a very sensitive function of flight rate.

Key takeaways ...

* Fixed costs, particularly labor, dominate both expendables and reusables.   

* Reusables have a relatively small advantage at low or medium cadence. Flying in the region of 5 to 10 flights per year the costs are roughly similar

* Expendables probably have a sharp upper limit of flights per year (20 perhaps) while Reusables can probably scale many, many times this upper limit.   Big win here for reusables -- IF there is demand for high cadence

* A Factory of equal cost doesn't drag reusable profit AT ALL, regardless of how much the factory is or is not used compared to expendables.  I think this is a huge canard when discussing expendables vs. reusables.
 

(Numbers in millions of dollars)

[Robotbeat]

SMH.
Price is not cost.
Cost is reduced via reusability.
Price can be anything SpaceX wants, subject to market constraints.

Any increase in price reflects on the inaptitude of the competitors who to this day haven't been able to cause SpaceX to reduce pricing.

It does NOT in any way indicate anything about their costs, as you're trying to claim.

How is this still lost on you?

The difference between SpaceX's price and cost, their margin, is how they're able to afford all the nice things. (And why investors think they are a real company)

Do you have their costs?

Of course you don't, and if you did, it would be proprietary and you wouldn't be allowed to release it.

And neither do you, but it was you who was making the claim that costs have not been going down (and actually that reuse is not reducing cost), as evidenced (per you) by rising prices.  And I'm telling you that even if prices go up, it doesn't mean costs do to. Pricing is determined by what the company can charge, and if all the competitors are expensive, than SpaceX can be too.

OTOH I pointed to the fact that boosters are being turned around in 3 weeks now, as evidence that turn-around pricing is much lower than fabrication and lower than it used to be.  Since the booster spends over a week on barge-back and on pre-launch, my claim is that there's just so much you spend in 1.5 weeks...  what do you think they're doing in there that's costing so much money, as per your claim?

Do we really have to go through all this again? We have it on the record from SpaceX that even the very first reuse including all the additional checkouts was 50% cheaper than a new one.

Do you have their costs?
Of course you don't, and if you did, it would be proprietary and you wouldn't be allowed to release it.

Well, there was that quickly deleted video of a meeting with SpaceX higher-up saying launch costs were about $30M two years ago or so.

But answer a simple question: If reuse wasn't really cheaper - would SpaceX launch Starlink 52 out of 53 Starlink launches so far on a reused booster, some of them reused 12 times?
 

Marginal or full-wrap?

Elon is ON RECORD saying that developing a flyback reusable booster was going to cost at least 1 billion dollars.  Let's assume some sort of Elon multiplier and guess it actually cost 1.5-2 billion.

Does that $30M include amortized development costs?  I'd bet it doesn't.

The business case probably doesn't close until you're launching a lot of rockets.  Maybe SpaceX is at that point now, but they had to become their own customer to get there.

$1 billion divided by the 100 reused flights so for is around $10 million apiece. They’re flying around 50-60 times per year, so in a few years (unless they go to another reusable launcher), the amortization cost will fall to around $5 million.

$5 million is pretty low.

[edzieba]

Sure, but I was trying to figure out why a bunch of really good engineers at a really good company chose in 2014 to design a rocket that did not have a reusable booster, since lots of folks here seem to think they should have. I now think that ULA's decision at the time was correct (not that anyone should care about my opinion). It only looks wrong in retrospect. Had Vulcan flown in 2019 as originally planned, it would look even better. ULA and Tory Bruno are putting the best face they can on the situation they are in.

I can think of two good reasons why engineers at ULA decided to design a non-reusable launch vehicle as opposed to one at least partially reusable.

1) The decision to not design a launch vehicle that was at least partially reusable was made early on using recovery costs based on the costs of recovering the Shuttle boosters and refurbishing them as well as refurbishing the orbiters.

2) The number of failed attempts to introduce a reusable launch system in the two decades prior to 2014. Part of ULA's engineering team in 2014 might have even included some who had worked under engineers involved with the McDonnell Douglas Delta Clipper project or the Lockheed Martin VentureStar.

3) Vulcan was designed to meet an existing set of requirements (NSSL orbits). Vulcan was also designed around the engines available on the market at the time or already in development for availability in the near future. None of these engines were in the gap between high thrust engines (the RD18x/RDF19x series, BE-4, AR-1, etc, all in the 2-4 MN class) and low thrust upper stage engines (the next step down in non-hydrolox US engines is the NewtonThree at 0.27MN, but that did not exist back then so even further down is the RL-10 on an alternate propellant, or a small hypergol motor like the ~0.1MN AJ10). For re-use to makes sense, the vehicle would need to have been sized to cover for the largest NSSL payload, so would have need many many more engines per core. More engines means higher cost per core, unless each individual engine managed to be a fraction of the cost of a pair of BE-4s (i.e. if a newly developed engine had a 0.5MN thrust, you would need 10 of these new engines for the same or lower cost, or 1/5 the cost per engine). Comparing with the GEM 63 XL costs added in means you now need to add an extra 12MN of thrust, and to add the cost of those new engines to every launch, rather than strap-on motors that can be omitted. That makes the up-front per core manufacture cost even higher still, and a higher financial risk for each launch in the event of a lost core.

4) Vulcan was designed with the painful lesson of Atlas V and Delta IV in mind: that if you expect a large boom in commercial launch to appear and require that boom to occur to make your launch vehicle development financially viable, you will be left holding the bag when that demand does not materialise. Government support (in the guise of "Buy III" and alter Assured Access) is not something ULA could rely upon receiving a second time around, so Vulcan would need to be profitable for NSSL launches in and of itself without additional commercial sales.