SpaceX progress towards a 100 launch year

[ZachF]

 Also note, Starship launching 40x/year lifting say ~60x/150t of Starlink v2 each is going to be alone surpassing the cumulative combined 50y+ totals of the Chinese and European launch industries on annual basis... 

[Vahe231991]

So far, almost 60 SpaceX launches have been conducted, and if the second Starship launch is conducted, SpaceX could reach a total of more than 100 launches this year for the first time in its history.

[spacenuance]

So far, almost 60 SpaceX launches have been conducted, and if the second Starship launch is conducted, SpaceX could reach a total of more than 100 launches this year for the first time in its history.

/s LLaMA math

60 + 1 > 100

math checks out /s

Honestly I am amazed every time I read that bot - sorry, I mean that guy's messages : )

[FutureSpaceTourist]

I think hitting 100 launches this year is pretty unlikely now, but we shouldn’t lose sight of just how capable SpaceX is.

For example, this all happened within 48 hours:

https://twitter.com/michaelbaylor_/status/1695844121590771810

😎

[redneck]

Maybe unlikely with almost 2/3 of the year gone and 59% of the 100 launched. OTOH, it might hit 10 launches for August bringing it to 60-61. If 10 per month can be sustained the rest of the year, then the 100 is attainable assuming no weather delays or technical problems.  My guess is low 90s. Either way a fine record. And starting about next week, each launch will set a new class record.

[meekGee]

Maybe unlikely with almost 2/3 of the year gone and 59% of the 100 launched. OTOH, it might hit 10 launches for August bringing it to 60-61. If 10 per month can be sustained the rest of the year, then the 100 is attainable assuming no weather delays or technical problems.  My guess is low 90s. Either way a fine record. And starting about next week, each launch will set a new class record.

They needed to average 8/mo.  But we should always assume an increasing cadence, we're just not sure about the growth coefficient.

Since the first 6 months were 7/mo, they need to average 9/mo in the last 6 months.
If they get those two launches on the 31st, it'd be 8 and 10, and they can then settle on a 9/mo average.

If improvement continues, I don't see any reason why they won't surpass 9/mo on this second half of the year, except for the always-possible LoV.

[meekGee]

Maybe unlikely with almost 2/3 of the year gone and 59% of the 100 launched. OTOH, it might hit 10 launches for August bringing it to 60-61. If 10 per month can be sustained the rest of the year, then the 100 is attainable assuming no weather delays or technical problems.  My guess is low 90s. Either way a fine record. And starting about next week, each launch will set a new class record.

They needed to average 8/mo.  But we should always assume an increasing cadence, we're just not sure about the growth coefficient.

Since the first 6 months were 7/mo, they need to average 9/mo in the last 6 months.
If they get those two launches on the 31st, it'd be 8 and 10, and they can then settle on a 9/mo average.

If improvement continues, I don't see any reason why they won't surpass 9/mo on this second half of the year, except for the always-possible LoV.

[Robotbeat]

We do know, actually. Growth coefficient needed is 60-65% per year. Which they are on track for.

[meekGee]

We do know, actually. Growth coefficient needed is 60-65% per year. Which they are on track for.

We don't.  We have a couple of data points, and they are not consistent.  They're always positive though.

As you've noted yourself, it's silly to look only at end-of-years.  If instead you look at "instantaneous" (let's say last 60 days) launch count for current date vs. 12 months earlier, you won't see a consistent year-over-year improvement.  There's way too much variance.

If you try to look at sub-year growth, it's even worse. For example, the first 6 months of 2023 were entirely flat.

And, without adding pads and/or barges, there's an upper limit somewhere (above 100, no doubt), but as you get closer to it, growth becomes more difficult.

So while I'm confident the rates will keeping increasing (I was very confident the 7 launches/month flat curve was only temporary) I can't tell by how much.

I think 9/month for the next 6 is easily within reality. I think 10/month is too.

[Robotbeat]

*shrug*, the required growth rate is basic arithmetic. 100/61 = 164%. my fit to the compound-improvement curve given the 61 launches last year and the 61 launches they’ve had so far this year puts them at ~102 launches by the end of the year.

Which is pretty consistent with what I’ve gotten every other time I’ve done the curve fit in this thread.

This is all automated. I just put in the number of launches so far and the number of days so far. There’s nowhere to fudge the numbers.

[abaddon]

39a has been dragging them down this year; last year 39a saw 18 launches (30% of all launches) whereas this year there have been only nine to date (15% as of now).  SLC40 has already exceeded last year's totals by three (+10%) and Vandenberg by four (+30%).  It doesn't seem likely that the factors that have kept 39a launches under the 2022 pace are going to be lifted this year, so that's probably going to be what keeps them under the 100 goal.  Which is pretty remarkable when you think about it, regarding the increase the other two pads have seen.

[meekGee]

*shrug*, the required growth rate is basic arithmetic. 100/61 = 164%. my fit to the compound-improvement curve given the 61 launches last year and the 61 launches they’ve had so far this year puts them at ~102 launches by the end of the year.

Which is pretty consistent with what I’ve gotten every other time I’ve done the curve fit in this thread.

This is all automated. I just put in the number of launches so far and the number of days so far. There’s nowhere to fudge the numbers.

I know they're improving.  I was the one saying that the flat performance for 6 months was NOT indicative of having reached a maximum.

But we don't know the rate, because there isn't one.  Your model did not predict 6 months of no improvement, right?  It's very uneven.

Anyway - I'm was also saying they're more or less on track for 100, +/- a few, if only they can keep up 9/month and maybe inch up a little.  I don't know what this argument is about.

[Robotbeat]

There has been improvement tho? It was like 7 launches in May and June, 8 in July, and 8 or 9 (or 10) in August.

We know the REQUIRED rate of annualized improvement. That’s not a mystery, and that was my point. It also is the same rate of improvement as /demonstrated/ from 2021 to 2022.

And from the current data, they are on track.

There’s no point in getting granular with the data. You only get single digit percentage statistical power if you average over like 4-12 months like I’m doing in the model. And the model has shown consistently about 100-105 projected launches in 2023 since I started doing it in like April or May.

[meekGee]

There has been improvement tho? It was like 7 launches in May and June, 8 in July, and 8 or 9 (or 10) in August.
.....

That's cherry-picking.  There were 7/month in Dec, Jan, Feb, Mar, April, May, June.  That's 7 months of no growth.
It's not shameful to admit that momentarily, the launch graph ran flat.

During that time I argued that it's not indicative of "the final plateau", and indeed we're seeing an uptick now. Hopefully they're able to sustain it and even grow from it.  That's where my money's at.

The overall pattern is that of growth. You can always ask excel to curve fit any formula (e.g. exponential) to it, but then you need to look at the quality of the fit...

In fact, look back and you'll see that the launch rate has always grown in spurts, never a smooth exponential.  Launches per year were 29,31,60, and hopefully 100.  That's not a constant ratio year-to-year, which is what an exponent looks like (not to mention the actual drop in launch rate in the year before that.)

But uneven growth is natural.  There's also a lot of noise introduced by factors such as weather, high workload launches which may cluster up, etc.

All we can do is tell people that using a constant launch rate is a terrible predictor and they should expect growth, month after month, with the growth no necessarily happening on a year boundary (e.g. December 2022 was already at 7/month)

[Robotbeat]

There has been improvement tho? It was like 7 launches in May and June, 8 in July, and 8 or 9 (or 10) in August.
.....

That's cherry-picking. ..

It’s not; those are just the last few months. But generally: That’s why I just use the entire year to date plus last year instead of looking month to month. month to month data will be noisy, with some months we’ll above trend and others well below.

[meekGee]

There has been improvement tho? It was like 7 launches in May and June, 8 in July, and 8 or 9 (or 10) in August.
.....

That's cherry-picking. ..

It’s not; those are just the last few months. But generally: That’s why I just use the entire year to date plus last year instead of looking month to month. month to month data will be noisy, with some months we’ll above trend and others well below.

Well, if you choose a small section of the dataset to support your claim, ignoring the data that's immediately before it - that's already cherry picking.

But you're doing something worse. Earlier  this year we were both arguing with people that said this year's best predicted total was 12x7=84 and you were showing that in fact growth happened month to month, September-december 2022.

So now you've used two carefully chosen short subsections of the timeline, each of which shows growth, while ignoring the intervening 6 month.

And for those few data points, of course you can fit an exponential function.

And then you ignore the overall year-to-year trend which is increasing, but not exponentially.

--

Also, go back to fundamentals.  Why in the world would it be exponential?  Growth always slows down when it becomes resource constrained.  And without new pads or barges, they clearly are.  The limit is likely above 100/yr, but it's not like growth will be exponential and then hit a ceiling.

[steveleach]

There has been improvement tho? It was like 7 launches in May and June, 8 in July, and 8 or 9 (or 10) in August.
.....

That's cherry-picking. ..

It’s not; those are just the last few months. But generally: That’s why I just use the entire year to date plus last year instead of looking month to month. month to month data will be noisy, with some months we’ll above trend and others well below.

Well, if you choose a small section of the dataset to support your claim, ignoring the data that's immediately before it - that's already cherry picking.

But you're doing something worse. Earlier  this year we were both arguing with people that said this year's best predicted total was 12x7=84 and you were showing that in fact growth happened month to month, September-december 2022.

So now you've used two carefully chosen short subsections of the timeline, each of which shows growth, while ignoring the intervening 6 month.

And for those few data points, of course you can fit an exponential function.

And then you ignore the overall year-to-year trend which is increasing, but not exponentially.

--

Also, go back to fundamentals.  Why in the world would it be exponential?  Growth always slows down when it becomes resource constrained.  And without new pads or barges, they clearly are.  The limit is likely above 100/yr, but it's not like growth will be exponential and then hit a ceiling.

I have to say, I do find Robotbeat's arguments more convincing than yours on this topic.

[spacenut]

Florida is in hurricane season.  So that can slow them down.  Still, even if they reach 90, that is more than anyone else.  Another hold back, might be the age of some of the boosters.  I know they are concentrating on Starship at Boca Chica, but wish they would hurry along the crew access at pad 40 so 39a can get more launches as well as 40.  They have a lot on their plate.  Starlink production, Starship/Raptor production, new F9 booster and second stage production, launch pad improvements, production facility construction.  They are actually doing far more than NASA alone. 

[Robotbeat]

There has been improvement tho? It was like 7 launches in May and June, 8 in July, and 8 or 9 (or 10) in August.
.....

That's cherry-picking. ..

It’s not; those are just the last few months. But generally: That’s why I just use the entire year to date plus last year instead of looking month to month. month to month data will be noisy, with some months we’ll above trend and others well below.

Well, if you choose a small section of the dataset to support your claim, ignoring the data that's immediately before it - that's already cherry picking.

But you're doing something worse. Earlier  this year we were both arguing with people that said this year's best predicted total was 12x7=84 and you were showing that in fact growth happened month to month, September-december 2022.

So now you've used two carefully chosen short subsections of the timeline, each of which shows growth, while ignoring the intervening 6 month.

And for those few data points, of course you can fit an exponential function.

And then you ignore the overall year-to-year trend which is increasing, but not exponentially.

--

Also, go back to fundamentals.  Why in the world would it be exponential?  Growth always slows down when it becomes resource constrained.  And without new pads or barges, they clearly are.  The limit is likely above 100/yr, but it's not like growth will be exponential and then hit a ceiling.

I don’t think they’re near the ceiling. Even with existing barges and pads, the limit is like 200-300. Technically about 1000 if they do a lot of RTLS with 24 hour pad turnaround.

Exponential just means continuous, compounding improvement. More important than pads and barges is the fact that their constellation demands higher and higher flightrates. Starlink is an economic engine for turning launches into more money, unlike any other ever developed.

[meekGee]

There has been improvement tho? It was like 7 launches in May and June, 8 in July, and 8 or 9 (or 10) in August.
.....

That's cherry-picking. ..

It’s not; those are just the last few months. But generally: That’s why I just use the entire year to date plus last year instead of looking month to month. month to month data will be noisy, with some months we’ll above trend and others well below.

Well, if you choose a small section of the dataset to support your claim, ignoring the data that's immediately before it - that's already cherry picking.

But you're doing something worse. Earlier  this year we were both arguing with people that said this year's best predicted total was 12x7=84 and you were showing that in fact growth happened month to month, September-december 2022.

So now you've used two carefully chosen short subsections of the timeline, each of which shows growth, while ignoring the intervening 6 month.

And for those few data points, of course you can fit an exponential function.

And then you ignore the overall year-to-year trend which is increasing, but not exponentially.

--

Also, go back to fundamentals.  Why in the world would it be exponential?  Growth always slows down when it becomes resource constrained.  And without new pads or barges, they clearly are.  The limit is likely above 100/yr, but it's not like growth will be exponential and then hit a ceiling.

I don’t think they’re near the ceiling. Even with existing barges and pads, the limit is like 200-300. Technically about 1000 if they do a lot of RTLS with 24 hour pad turnaround.

Exponential just means continuous, compounding improvement. More important than pads and barges is the fact that their constellation demands higher and higher flightrates. Starlink is an economic engine for turning launches into more money, unlike any other ever developed.

Exponential or Compounding means there's a fixed ratio between any two points separated by an equal amount of time.

That's a very poor fit to what we're seeing.

We see growth spurts, then flat regions.  When zooming out to see if we can "even out" the perturbations, it doesn't look any more exponential.

Any assertion about 1000 flights a year, well, I doubt F9 will last that long. Also why would they do more RTLS? If they wanted 200/yr, they could add another barge and not take the payload hit.

At best we'll see another record year for F9, and then Starship will take over most of Starlink.