Retiring Falcon 9 first stages vs expending?

[Jim]

Total remaining launches: 100.

There are roughly ten boosters in the active fleet. They must launch an average of ten more times each. No new boosters are needed.

Both are wrong. 

there are more than 60 launches scheduled in the next three year

ten boosters is not enough for 100 missions.  You are forgetting about Heavy missions and expended missions.

JIm, I may very well be wrong. However, My timeline showed 40+20+30 = 90 F9's in the next three years, not just 60. I did not neglect the FH or the expendables. The FH sides (2 per FH launch) are counted in that count of 90, which is a booster count, not a launch count. The expendables are counted in the average remaining launches per booster. They will need an average of 10 additional flights each. The fleet seems to have a current average flight count of about 6, so with none expended the average would need to go to 16, or if you expend half of them on the first remaining flight the average would need to go to about 25. Note also that SpaceX has not actually quit building F9 boosters as far as I know. They will need to continue to build FH cores.

My 60 count was just launches

There are planned expended missions that will use new or low cycle boosters

[whitelancer64]

*snip*
2026: 2 crew dragon. CRS has move to Starship, NASA has not qualified Starship for crew to ISS. 2, not 1 because NASA is avoiding Starliner due to cost.
2027: 2 crew dragon. As above.
2028: 2 crew dragon As above.
*snip*

That's not going to happen. Starliner will fly whatever number of missions are contracted, regardless of cost. NASA wants the dissimilar redundancy.

[DanClemmensen]

Total remaining launches: 100.

There are roughly ten boosters in the active fleet. They must launch an average of ten more times each. No new boosters are needed.

Both are wrong. 

there are more than 60 launches scheduled in the next three year

ten boosters is not enough for 100 missions.  You are forgetting about Heavy missions and expended missions.

JIm, I may very well be wrong. However, My timeline showed 40+20+30 = 90 F9's in the next three years, not just 60. I did not neglect the FH or the expendables. The FH sides (2 per FH launch) are counted in that count of 90, which is a booster count, not a launch count. The expendables are counted in the average remaining launches per booster. They will need an average of 10 additional flights each. The fleet seems to have a current average flight count of about 6, so with none expended the average would need to go to 16, or if you expend half of them on the first remaining flight the average would need to go to about 25. Note also that SpaceX has not actually quit building F9 boosters as far as I know. They will need to continue to build FH cores.

My 60 count was just launches

There are planned expended missions that will use new or low cycle boosters

OK. For new or low-cycle expended, SpaceX will need to build new boosters to keep the active fleet at 10, we can then remove those flights from my very speculative count of 100. This reduces the average number that the fleet of 10 must cover as it is slowly diminished by "ordinary" expendables and accidental loses.

Do you have an approximate count of these new-or-nearly-new expendables? It looks like time to create a spread sheet.

[DanClemmensen]

*snip*
2026: 2 crew dragon. CRS has move to Starship, NASA has not qualified Starship for crew to ISS. 2, not 1 because NASA is avoiding Starliner due to cost.
2027: 2 crew dragon. As above.
2028: 2 crew dragon As above.
*snip*

That's not going to happen. Starliner will fly whatever number of missions are contracted, regardless of cost. NASA wants the dissimilar redundancy.

Agreed, unless Boeing decides to walk away. As of now, that's a total of six Starliner missions. My number two-per-year is sloppy. If Starliner flies one operational mission per year starting in 2023, then they fly the last mission in 2028. I was being conservative by accommodating the 2 Crew Dragon "just in case". These might also be Axiom missions to ISS, which presumably need NASA crew certification. It might be better to say that there are 16 CCP missions in the next eight years, and Starliner will fly six of them.

[AmigaClone]

Personally, I can see only three reasons SpaceX would choose to retire F9 boosters.

First would be if a major museum specializing in space or air and space makes a request for a launch vehicle - possibly with  specific booster.

Second would be a life leader after SpaceX has determined when they would be retiring the F9 (easier/cheaper/faster to refurbish newer boosters), and they have enough younger boosters to complete the planned launches.

Third would involve a booster being recovered, but heavily damaged. After the post-mission checkup SpaceX decides it to not be viable to refurbish it.

[alugobi]

Yes, and it was over-the-top rah-rah stupid.  The very thing that I advocate must stop. 

WRONG.  After an absence of an American crew launcher for more than 10 years, it was a milestone worth celebrating.  Additionally, it was the first commercial crew vehicle which was another milestone that could stand on its own.ib

It was worth celebrating.  It's NASA's style, delivery, and overcompensation that should be revisited.  Their broadcasts make me wonder if their target audience is anyone outside of NASA at all.

Edit:  missed a word

[Jim]

Yes, and it was over-the-top rah-rah stupid.  The very thing that I advocate must stop. 

WRONG.  After an absence of an American crew launcher for more than 10 years, it was a milestone worth celebrating.  Additionally, it was the first commercial crew vehicle which was another milestone that could stand on its own.ib

It was worth celebrating.  It's NASA's style, delivery, and overcompensation that should be revisited.  Their broadcasts make me wonder if their target audience is anyone outside of NASA at all.

The target audience is certainly not NASA.  And it is for the the unwashed masses outside of NASA.  The public likes them.

[Barley]

4 - It's hard to imagine SpaceX abandoning a capability that is still leagues better than any other provider can currently provide in many ways. At least anytime soon. SpaceX is likely a decade ahead of competitors, but who knows?

Why is this hard to imagine? SpaceX believes that they will make a larger profit by replacing an F9 launch with a Starship launch, for any F9 launch, and as the F9 launch rate goes down the F9 fixed operations go up.

I can imagine it going either way.

SpaceX beliefs are not the only thing that matter.  If NASA or DoD is willing to pay enough for what they think is a unique capability or a dissimilar redundancy I imagine SpaceX will take the money and keep building F9.

Where I worked when we transitioned from one generation of product to the next, we would increase the price of the old version while pointing out the availability of the cheaper, more capable replacement.  Usually demand for the obsolescent version vanished and we halted production.  But if customers were willing to pay, we would keep selling the old version, treating it as a cash cow.

In one case the DoD preferred mechanical controls, so a 1960's design remained in limited production for over 40 years, outlasting about 6 generations of replacements.  Last I knew in 2005 the DoD was paying about five times more than for the modern COTS equivalent.  But as long as the DoD wanted something that is EMP proof and was willing to pay for a corner of the factory and training technicians in obsolete technology we were willing to keep supplying.

There were elements of patriotism and keeping a customer happy.  We didn't price gouge, but we were fully compensated for the inconvenience.

[DanClemmensen]

4 - It's hard to imagine SpaceX abandoning a capability that is still leagues better than any other provider can currently provide in many ways. At least anytime soon. SpaceX is likely a decade ahead of competitors, but who knows?

Why is this hard to imagine? SpaceX believes that they will make a larger profit by replacing an F9 launch with a Starship launch, for any F9 launch, and as the F9 launch rate goes down the F9 fixed operations go up.

I can imagine it going either way.

SpaceX beliefs are not the only thing that matter.  If NASA or DoD is willing to pay enough for what they think is a unique capability or a dissimilar redundancy I imagine SpaceX will take the money and keep building F9.

Where I worked when we transitioned from one generation of product to the next, we would increase the price of the old version while pointing out the availability of the cheaper, more capable replacement.  Usually demand for the obsolescent version vanished and we halted production.  But if customers were willing to pay, we would keep selling the old version, treating it as a cash cow.

In one case the DoD preferred mechanical controls, so a 1960's design remained in limited production for over 40 years, outlasting about 6 generations of replacements.  Last I knew in 2005 the DoD was paying about five times more than for the modern COTS equivalent.  But as long as the DoD wanted something that is EMP proof and was willing to pay for a corner of the factory and training technicians in obsolete technology we were willing to keep supplying.

There were elements of patriotism and keeping a customer happy.  We didn't price gouge, but we were fully compensated for the inconvenience.

I agree in principle: sell the customer what the customer wants, and charge enough to make a nice profit. But  DoD (specifically, the NSSL program) has already had to accommodate the upcoming retirement of their favored rockets (Atlas and Delta)  by adding new rockets: F9, FH, and (ahem, real soon now) Vulcan Centaur, so if SpaceX announces a plan to retire Falcon, NSSL will likely begin a plan to qualify Starship.

SImilarly, when/if SpaceX annouces a plan to retire F9 and Dragon 2 (crew and cargo), NASA might choose to qualify Starship for crew and for ISS docking.

If you look at my totally speculative and made-up timeline, I show F9 being used to support Crew Dragon and NSSL through 2029, pretty much based on your reasoning.

[Barley]

If you look at my totally speculative and made-up timeline, I show F9 being used to support Crew Dragon and NSSL through 2029, pretty much based on your reasoning.

It's not unreasonable, perhaps it's even a maximum likelihood estimator, but there are a range of reasonable timelines. 

I do think your timeline is a little short.  You have F9 mostly phased out in 2024, possibly with the last NSSL F9 in 2025.

Given the lacuna between Shuttle and Crew Dragon I'd expect NASA, NSSL and some commercial customers to wait for things to show up before committing to a transition.  Committing to retire F9 before Starship, Vulcan or Starliner are in service seems unnecessary.  That could push the start of any transition into 2023 or later; Given the lead times substantially completing by 2025 seems rushed.

Then there are special situations that might show up.
Maybe SS is too big to safely dock at the ISS.
Ideally the SS launch pad is rather busy. Perhaps NSSL prefers to keep a private pad in a Space Force base.  Conversely SpaceX might prefer not to interrupt SS launches for a week to satisfy security requirements.  It might be easier to keep launching F9 from Canaveral and Vandenburg than build additional SS launch pads.
Or SpaceX might not want to freeze the design of SS.  F9 wasn't frozen until Block 5, 8 years after the first success.  Keepping F9 available for the fuddy-duddies more conservative customers might be the easier option than partial premature freezes of SS.
Do we even know how SS will deploy payloads?  Some redesign may be needed to throw them out an airlock. (Ok, it's not an airlock, but it is different.)

[AmigaClone]

2025.
Then there are special situations that might show up.
Maybe SS is too big to safely dock at the ISS.

I can see that being a potential issue - both in terms of mass of the SS and its payload and possibly by the SS preventing docking or undocking from both the zenith and nadir docking ports on the Harmony module if it docks on the forward docking port.

2025.
Ideally the SS launch pad is rather busy. Perhaps NSSL prefers to keep a private pad in a Space Force base.  Conversely SpaceX might prefer not to interrupt SS launches for a week to satisfy security requirements.  It might be easier to keep launching F9 from Canaveral and Vandenburg than build additional SS launch pads.
Or SpaceX might not want to freeze the design of SS.  F9 wasn't frozen until Block 5, 8 years after the first success.  Keepping F9 available for the fuddy-duddies more conservative customers might be the easier option than partial premature freezes of SS.

I would not be surprised if the first versions of SS to be finalized be the 'tanker SS', the Depot SS, and the HLS. The Depot would be a variant of the tanker optimized for that purpose. A Moon/Mars cargo variant might be another early version to be finalized.

2025.
Do we even know how SS will deploy payloads?  Some redesign may be needed to throw them out an airlock. (Ok, it's not an airlock, but it is different.)

I am not certain SpaceX has finalized the design that will be used to deploy payloads from SS. It's not impossible that there be several designs that make it to orbit before the final one is chosen.

[DanClemmensen]

If you look at my totally speculative and made-up timeline, I show F9 being used to support Crew Dragon and NSSL through 2029, pretty much based on your reasoning.

It's not unreasonable, perhaps it's even a maximum likelihood estimator, but there are a range of reasonable timelines. 

I do think your timeline is a little short.  You have F9 mostly phased out in 2024, possibly with the last NSSL F9 in 2025.

Given the lacuna between Shuttle and Crew Dragon I'd expect NASA, NSSL and some commercial customers to wait for things to show up before committing to a transition.  Committing to retire F9 before Starship, Vulcan or Starliner are in service seems unnecessary.  That could push the start of any transition into 2023 or later; Given the lead times substantially completing by 2025 seems rushed.

Then there are special situations that might show up.
Maybe SS is too big to safely dock at the ISS.
Ideally the SS launch pad is rather busy. Perhaps NSSL prefers to keep a private pad in a Space Force base.  Conversely SpaceX might prefer not to interrupt SS launches for a week to satisfy security requirements.  It might be easier to keep launching F9 from Canaveral and Vandenburg than build additional SS launch pads.
Or SpaceX might not want to freeze the design of SS.  F9 wasn't frozen until Block 5, 8 years after the first success.  Keepping F9 available for the fuddy-duddies more conservative customers might be the easier option than partial premature freezes of SS.
Do we even know how SS will deploy payloads?  Some redesign may be needed to throw them out an airlock. (Ok, it's not an airlock, but it is different.)

These are all good points. Note that my F9 booster count for 2022 is 40, which exceeds the actual count in 2021 of 31. I think that without any Starship in 2022 the F9 booster count would be higher, say 50.  So I'm even more optimistic than you thought.

"Rather busy": maybe. but a single Mechazilla is supposed to be able to launch multiple flights per day, so the entire commercial load can be handled from one launch site. I think your point about NSSL security is a good one and I had not considered it. It basically means that the USSF will need its own separate launch site that is not shared with commercial launches. This will stretch out the "tail" of the F9 curve by a year. But there are not a lot of NSSL lauches. The question then is how expensive is a USSF-only launch site versus the cost difference between F9/FH and Starship.

Starship-to-ISS: I've been thinking(?!) about this. If NASA won't let Starship dock to ISS, then Starship can deliver a cargo pod. Old pod undocks from ISS and moves our about 1000 meters. Starship delivers new pod to about 1000 meters from ISS and kicks it out. New pod moseys on over to ISS and docks. Old pod "docks" to a bracket inside starship, and starship returns to Earth. Much cheaper than a Cargo Dragon mission.  For crew (after 2029 on my timeline), I assume a crew-rated Starship, but if it cannot dock to ISS, I envision a crew taxi pod. Details depend on exact capabilities of the crewed Starship design. One simple possibility is that the cargo pod acts as the crew taxi. The crew will only use the taxi for the 1000 meter flight between Starship and ISS with max thrust well below 0.1 g.

[spacenut]

I would think SpaceX might continue to make F9 boosters at least until 2028 due to NASA launches, but not much longer after Starship is operational. 

[DanClemmensen]

I would think SpaceX might continue to make F9 boosters at least until 2028 due to NASA launches, but not much longer after Starship is operational.

SpaceX will (or might) continue to fly F9 boosters. I was trying to quantify (crudely) how many they will need to manufacture. Clearly, they will need to manufacture second stages, but my speculative estimate puts the total number of remaining booster flights at about 100. The current fleet is 10 boosters with an average "age" of six flights.  If SpaceX can squeeze out an average of ten more flights per booster, they need no new boosters. Attrition (deliberately expended, retired, otherwise lost) will require the remaining fleet to fly even more to maintain the average.

BUT: Jim points out that some customers will insist on and pay more for new or nearly-new boosters. This has some effects: first, SpaceX will need to continue to build these boosters so the production facility must continue to operate or they must stockpile some new boosters. Next, these flights will reduce the stress on the active fleet. third, since the production line or stockpile is there anyway, the incremental cost of a few "extra" boosters are small.

So now the question is: to meet this (wild-guess) demand for 100 remaining flights, How many new boosters must SpaceX build? Answer: one for each "new or nearly-new" customer, plus enough more to match the attrition rate for the existing fleet. Let's speculate that there will be ten "new or nearly-new" boosters. That's ten "new" flights and ten "nearly-new" flights.  We would need to also speculate on their distribution, but just guessing, I think SpaceX will need to build about twelve new boosters before EOL in 2030. It probably makes sense to build and stockpile them.

[AmigaClone]

Looking at a list of future F9/FH flights, it appears that SpaceX will be attempting 5 Falcon Heavy flights this year. Three are for the USSF (USSF-44, USSF-52, and USSF-67). At this point it appears that both USSF-44 and USSF-67 call for the center core to be expended and at least USSF-44 calls for new side boosters.

NASA's Pysche mission calls for the attempted recovery of all three boosters (the side ones at LZ-11 and LZ-2 and the core on a drone ship.) There also is a commercial FH flight (ViaSat-3 Americas) - although I have not seen if it requires the same launch profile as at least two of the USSF missions.

Three FH Cores (B1066, B1068, and B1070) have been spotted in FL, along with two FH side boosters (B1064 and B1065). Another side booster (B1072) has been spotted being tested at McGregor, TX.

Depending on customer requirements, just those 5 missions could account for 4 boosters being intentionally expended this year at the end of their first flight.

[DanClemmensen]

Looking at a list of future F9/FH flights, it appears that SpaceX will be attempting 5 Falcon Heavy flights this year. Three are for the USSF (USSF-44, USSF-52, and USSF-67). At this point it appears that both USSF-44 and USSF-67 call for the center core to be expended and at least USSF-44 calls for new side boosters.

NASA's Pysche mission calls for the attempted recovery of all three boosters (the side ones at LZ-11 and LZ-2 and the core on a drone ship.) There also is a commercial FH flight (ViaSat-3 Americas) - although I have not seen if it requires the same launch profile as at least two of the USSF missions.

Three FH Cores (B1066, B1068, and B1070) have been spotted in FL, along with two FH side boosters (B1064 and B1065). Another side booster (B1072) has been spotted being tested at McGregor, TX.

Depending on customer requirements, just those 5 missions could account for 4 boosters being intentionally expended this year at the end of their first flight.

My speculative estimates did not include FH cores, as they are  not part of the F9 booster fleet. I did include FH sides since they can be converted to/from F9 boosters. Thus for my speculative accounting I assume new and non-recoverable FH cores, but I failed to state that explicitly.

Speculative extrapolation: They will need to keep building FH cores, almost one per launch, for the life of the FH. If the booster production facility is building both types of booster on the same line, they may choose to keep the line open instead of stockpiling. NSSL will account for up to 34 launches in 2022-2027 and SpaceX will get 40% of them, or up to 14 launches, some on FH and some on F9. Looking at past years, probably about half on F9.  Let's assume the F9s and FH sides are already accounted for on my silly timeline. let's further assume USSF/NSSL will qualify Starship by 2027 and will have built a dedicated Mechazilla by the end of 2027. If so, SpaceX will need to build about 7 FH cores for NSSL launches. Also assume that commercial customers will have migrated from FH to Starship by about 2025, so let's further assume that non-NSSL FH launches will use recovered FH cores.  All up, this adds these seven FH cores to the twelve new F9 boosters I calculated earlier, for a total production run of 19 new boosters by the EOL of the F9/FH.  You are of course free to make your own guesses and they will probably be better than mine.

[Barley]

Speculative extrapolation: They will need to keep building FH cores, almost one per launch, for the life of the FH. If the booster production facility is building both types of booster on the same line, they may choose to keep the line open instead of stockpiling.

Stockpile v. keep the line open probably depends more on the details of how the line works, and the nature of the differences between stage 2, stage 1 and FH core than anything else.  If you have to stockpile it may come down to individual parts rather than final assembly.

If you're using CNC or 3D printing it can be easy to reliable slip different parts into the assembly line.  My colleagues in design engineering sometimes slipped singleton prototypes through the factory without much difficulty.  It's different if you're using jigs and molds.  Keeping old jigs, and the knowledge of their quirks, is a lot harder than keeping old cad files.

Best case keeping the line open means reserving a few thousand square meters of the factory*, a single jig for S1 and reassigning techs from the S2 line to use the skills they use on S2.  Worse case there are thousands of distinct parts with jigs, some made of unique materials, and S1 needs a standing army to do things you don't do on S2.

* This is not always as easy as you think.  Any parts of a factory not in active use tend to be misappropriated as warehouse space.

[LouScheffer]

If NASA or DoD is willing to pay enough for what they think is a unique capability or a dissimilar redundancy I imagine SpaceX will take the money and keep building F9.

Where I worked when we transitioned from one generation of product to the next, we would increase the price of the old version while pointing out the availability of the cheaper, more capable replacement.  Usually demand for the obsolescent version vanished and we halted production.  But if customers were willing to pay, we would keep selling the old version, treating it as a cash cow.

In one case the DoD preferred mechanical controls, so a 1960's design remained in limited production for over 40 years, outlasting about 6 generations of replacements.  Last I knew in 2005 the DoD was paying about five times more than for the modern COTS equivalent.  But as long as the DoD wanted something that is EMP proof and was willing to pay for a corner of the factory and training technicians in obsolete technology we were willing to keep supplying.

As this story points out, don't underestimate how hard it is to kill an old product without enraging your customers.  I worked at Hewlett-Packard back when it was an instrument company.  We kept making an old, vacuum tube signal generator long after it was fully surpassed by cheaper, better, more reliable, solid-state products.  But the military found it simpler to keep buying the old one.  It was written into the procedures, they did not need the increased performance, and the cost was not a significant part of their budget.  Re-writing the procedures would have been hard work for them, to little benefit, as the people who wrote the originals were long gone.  So we kept making the original until we could no longer get the vacuum tubes they required.  When we told the military we were phasing it out, they sighed and ordered 99 more, hopefully a lifetime supply (ordering 100 would have required another level of signatures).

Overall, a forced phase-out may enrage your customers, who then need to develop new procedures and authorizations, on a time scale not of their choosing, to little benefit. It's often better to keep making your old product, charge a little more for it, and have a happy customer.

[markbike528cbx]

If NASA or DoD is willing to pay enough for what they think is a unique capability or a dissimilar redundancy I imagine SpaceX will take the money and keep building F9.

Where I worked when we transitioned from one generation of product to the next, we would increase the price of the old version while pointing out the availability of the cheaper, more capable replacement.  Usually demand for the obsolescent version vanished and we halted production.  But if customers were willing to pay, we would keep selling the old version, treating it as a cash cow.

In one case the DoD preferred mechanical controls, so a 1960's design remained in limited production for over 40 years, outlasting about 6 generations of replacements.  Last I knew in 2005 the DoD was paying about five times more than for the modern COTS equivalent.  But as long as the DoD wanted something that is EMP proof and was willing to pay for a corner of the factory and training technicians in obsolete technology we were willing to keep supplying.

As this story points out, don't underestimate how hard it is to kill an old product without enraging your customers.  I worked at Hewlett-Packard back when it was an instrument company.  We kept making an old, vacuum tube signal generator long after it was fully surpassed by cheaper, better, more reliable, solid-state products.  But the military found it simpler to keep buying the old one.  It was written into the procedures, they did not need the increased performance, and the cost was not a significant part of their budget.  Re-writing the procedures would have been hard work for them, to little benefit, as the people who wrote the originals were long gone.  So we kept making the original until we could no longer get the vacuum tubes they required.  When we told the military we were phasing it out, they sighed and ordered 99 more, hopefully a lifetime supply (ordering 100 would have required another level of signatures).

Overall, a forced phase-out may enrage your customers, who then need to develop new procedures and authorizations, on a time scale not of their choosing, to little benefit. It's often better to keep making your old product, charge a little more for it, and have a happy customer.

Can anyone remember the New Coke furor?
I was in a business where we had old processes, that were obsolete to the point of near-dangerous-ness, but a customer wanted them because they were invented in the customer's country.   
We would have had to pay a royalty on using them.  We had not used those processes for 20 or more years, so our institutional knowledge and procedures had decayed.