SpaceX progress towards a 100 launch year

[eeergo]

Steve, the growth meekgee is referring to is the *change* in slope (i.e. the rate's acceleration with time, which corresponds to the exponential nature of the trend). If there's zero growth, it means the rate is constant. And indeed, throughout the last few months there have been slightly under 8 launches/month.

[meekGee]

For 7 months, the monthly total was exactly fixed.  Zero growth.

What do you mean by that exactly?

Here's the data from Jan 21 with the first 7 months of 2023 highlighted.

Go back amd count launches.

Dec 2022 was 7 launches
Jan 2023 was 7 launches
Feb 2023 was 7 launches
Mar 2023 was 7 launches
Apr 2023 was 7 launches
May 2023 was 7 launches
Jun 2023 was 7 launches

... And all this time, when people said the best predictor for the year was 7x12=84, I (and others) pointed out that we mustn't assume a fixed rate per year, and likely it'll be something like 7-7-7-7-8-8-8-8-9-9-9-9...

That didn't happen, so after June concluded with 7 launches, we said it'd have to be a more aggressive ramp up if we have to meet 96/yr falcon launches , something like 7-7-7-7-7-7-8-8-9-9-10-10.

Still possible.

But:

What you guys are doing is pre-supposing he rate is exponential.

You then take the ratio of two end points, n months apart, take the n-th root, and find your "monthly growth rate".

Then, you plot the graph along with the noisy data and rejoice that it "looks good".  And since you have the two end points pre-pegged, of course it'll predict the few next ones rather well, especially with all the noise...  any increasing model would do that!

And in all the excitement you completely ignore the drastic reduction in growth which is staring you straight in the face.

So:

Forget the graphs for a second.

+100%, +50-60%, +20-33%.  Those are hard numbers.  Explain how that's exponential, and then let's get back to curve fitting methodologies.

[edzieba]

Plotted a day-to-day 1-year rolling average of launch totals since 2020, rather than just looking at calendar years periods. Orange is the rolling average, green is smoothed by week.
This makes it clearer that the annual launch rate already achieved is around 90 launches per year period, and continuing to rise.

[Barley]

[2021] 31
[2022] 60
[2023] 90-100  estimated
[2023] 120  projected by Spacex

That's +100%, +50-60%, +20-33%

Emphasis added.

+100%, +50-60%, +20-33%.  Those are hard numbers.

one hard number, one projection and one estimate.  Not hard numbers.

Not enough data, but that hasn't stop anybody from trying to torture it.  It's not like the model even matters.

[gsa]

Plotted a day-to-day 1-year rolling average of launch totals since 2020, rather than just looking at calendar years periods. Orange is the rolling average, green is smoothed by week.
This makes it clearer that the annual launch rate already achieved is around 90 launches per year period, and continuing to rise.

How do you get 90? I only got 85, and that with IFT-1.

[ZachF]

For what it's worth, I found a "show R²" button on the charts.

For the exponential trend line it is 0.663 and for the linear trend it is 0.673 which (I think) means that the linear trend is a slightly better fit (?)

I also tried going (yearly not monthly) back to 2012, and can get an exponential trend line with an annual growth of 34% out of it (significantly less than the 55% that best matches the current year) and an R² of 0.949; the linear trend gives negative launch rates in 2012 & 2013, so I'm not sure it is valuable.

That's it: the annual launch rate is compatible with a linear trend, within the noise, and has been so in recent years. An exponential trend is also compatible but slightly worse and with a very small exponent (i.e. approaching linear).

Moreover, the fine-tuning necessary to avoid over/under-shooting of an exponential function is much stronger, while multi-year-fitting exponentials that cover a few years back don't predict close to 100 launches this year, plus require piecemeal treatment of certain periods or tampering with the slope in approximately-yearly periods to fit the data well enough and not explode toward unfeasible values (like the linear trend does too, but subject to more error in case of inaccuracies in the fit).

The overall historical launch trend, while still noisy, is much more clearly exponential, as a simple visual inspection clearly tells you - although in reality it will taper off at some "saturation" value achieving something like a sigmoid, at least in the medium term.

The thing is, it looks like the growth is accelerating, not tapering off. 

From 2012 to 2022 it was trending at 35% a year.
From 2021 to 2023 it has been trending at 55% a year.

Is it common to see exponential growth turn into linear growth at a higher rate than it was while exponential?

Data can be noisy year to year.


Here’s what SpaceX dV adjusted payload growth looks like:

2008: 179
2009: 217
2010: 9,734
2011: 0
2012: 14,896
2013: 16,938
2014: 47,000
2015: 42,024
2016: 69,380
2017: 174,965
2018: 184,835
2019: 151,674
2020: 333,141
2021: 409,922
2022: 778,787
2023ytd: 887,136

SpaceX is on pace for 1,300t+ adjusted payload mass this year, already ahead of last year with ~four months to go. To provide more context with this, the cumulative total for the entire European space industry since its first launch is 3,224t, and the Chinese total is 2,241t (SpaceX is now 3,121t). The previous record tonnage in a year was 608t by the USSR in 1988.

Tonnage will be a better barometer than launches as the switch to Starship happens… as impressive as projected SpaceX’s massive 1,300t total for 2023 is, that’s only like a dozen Starship launches. We may be soon coming to a point where SpaceX is lifting as much quarterly as the entire combined launch histories of Chinese or European space industries… 

[eeergo]

[Tonnage discussion]

Fine, and certainly jaw-dropping, but please read the title of this thread. Other metrics can be illuminating, but the discussion is about the number of launches, its trend and the company's aims regarding that.

[Robotbeat]

...
"Exponential" a wrong model. If the rate of increase is decreasing, it's not an exponent.

And the rate of increase is decreasing by a LOT.
...

Based on what? I've gotten about the same estimated number of launches since I started doing an exponential fit like 5-6 months ago. 100-105.

I don't expect the current exponential increase to hold past this year. They were targeting an increased rate, and because you can't flip a switch and just start launching at a higher launch rate starting January 1st, you have to model the increase in launch rate somehow, and a gradual compounding improvement is basically the simplest of models for that. It's a simple assumption that won't hold forever, but it sure as heck beats piecewise linear.

You need to decide if you're looking long or short term.

For 7 months, the monthly total was exactly fixed.  Zero growth. And since this was a temporary condition, any analysis of a period comprising these 7 month plus the more recent increase is just nonsensical.

Looking back over a longer time period, 2021->2022  showed a huge increase (100%), 2022->2023 is looking like a little more than half that (50-60%) and 2023->2024 is forecast by Musk to be only a 20% increase.

If your curve fitting methodology tells you there's an exponent there, question the methodology...

i tried explaining why it's misleading but this gets lost in the back and forth so I gave that up.

linear curve fitting is no better and in fact is much worse. “If your curve fitting methodology tells you there’s a constant slope there, question the methodology.”

“Linear means a CONSTANT and NOT-increasing launch rate throughout the year.”

[Robotbeat]

[Tonnage discussion]

Fine, and certainly jaw-dropping, but please read the title of this thread. Other metrics can be illuminating, but the discussion is about the number of launches, its trend and the company's aims regarding that.

That’s why even if folks think the compounding tonnage trend will last, I don’t think the compounding growth trend in pure launch numbers will last past this year because they’re switching emphasis to a vehicle with 10 times the mass to orbit. In fact, logically it WOULDNT last if the switch is quick because you’d otherwise have a super-exponential increase in launch mass rate.

(In case it’s not clear, it means I agree with eeergo on this.)

[RedLineTrain]

Once starship is launching starlink, what is F9 going to launch 12 times a month? Besides starlink is there any other payload on the manifest for this many launches? Will they continue to use F9 for starlink?

Starlink v2 minis.  As I read it, SpaceX has been guarding against the possibility that Starship takes longer than wished to introduce and ramp.  The introduction of the Starlink v2 minis smooths over any timing mismatch between the constellation and launch.

If SpaceX can continue to ramp Falcon 9 launches with modest incremental investment, it seems like a wise thing to do.  Eventually, Starship will come on line in a big way and even the modest incremental investment on Falcon 9 launch capacity will make less sense.

[DanClemmensen]

[Tonnage discussion]

Fine, and certainly jaw-dropping, but please read the title of this thread. Other metrics can be illuminating, but the discussion is about the number of launches, its trend and the company's aims regarding that.

That’s why even if folks think the compounding tonnage trend will last, I don’t think the compounding growth trend in pure launch numbers will last past this year because they’re switching emphasis to a vehicle with 10 times the mass to orbit. In fact, logically it WOULDNT last if the switch is quick because you’d otherwise have a super-exponential increase in launch mass rate.

(In case it’s not clear, it means I agree with eeergo on this.)

This effect will not affect 2024, since there will be only a few Starship launches in 2024 and they will carry either no payload or unique new payloads that were never on the F9 manifests. My personal guess is that Starship will launch a total of ten times before 2028 and most of those will carry such unique payloads.

[Robotbeat]

But even if it’s just 5 Starship launches, that’s the equivalent of 50 Falcon 9s. That would already be a substantial effect.

[abaddon]

[Tonnage discussion]

Fine, and certainly jaw-dropping, but please read the title of this thread. Other metrics can be illuminating, but the discussion is about the number of launches, its trend and the company's aims regarding that.

It's a nice break from the constant fruitless arguing over methodology.

And, it's at least related.  I don't think it makes sense to create a new thread that's the same thing but "tonnage" so why not let it live here?

[meekGee]

...
"Exponential" a wrong model. If the rate of increase is decreasing, it's not an exponent.

And the rate of increase is decreasing by a LOT.
...

Based on what? I've gotten about the same estimated number of launches since I started doing an exponential fit like 5-6 months ago. 100-105.

I don't expect the current exponential increase to hold past this year. They were targeting an increased rate, and because you can't flip a switch and just start launching at a higher launch rate starting January 1st, you have to model the increase in launch rate somehow, and a gradual compounding improvement is basically the simplest of models for that. It's a simple assumption that won't hold forever, but it sure as heck beats piecewise linear.

You need to decide if you're looking long or short term.

For 7 months, the monthly total was exactly fixed.  Zero growth. And since this was a temporary condition, any analysis of a period comprising these 7 month plus the more recent increase is just nonsensical.

Looking back over a longer time period, 2021->2022  showed a huge increase (100%), 2022->2023 is looking like a little more than half that (50-60%) and 2023->2024 is forecast by Musk to be only a 20% increase.

If your curve fitting methodology tells you there's an exponent there, question the methodology...

i tried explaining why it's misleading but this gets lost in the back and forth so I gave that up.

linear curve fitting is no better and in fact is much worse. “If your curve fitting methodology tells you there’s a constant slope there, question the methodology.”

“Linear means a CONSTANT and NOT-increasing launch rate throughout the year.”

That's a new (and wrong) definition of linear.

And even if you decide to label the integrated quantity rather the the actual rate, then you're just one derivative away..

So exponential still means exponential (and equally poorly apply)

Linear is a great fit till now - they increased by +30,+30 to +40, and projected to increase by, +20 to +30.

Again:
If you increase by a fixed RATIO then it's exponential
If you increase by a fixed AMOUNT then it's linear
If you don't increase, then you're not increasing

You can also increase by less and less, following something like log or a (square) root

You can also grow with an upper asymptote.

--

Current and projected growth are actually more or less constant over time, when smoothed out, which is amazing, given that usually it'll become bounded by something, like demand, or infrastructure, or other resources..

But they are not exhibiting constant ratio between successive time periods. The ratio is decreasing.  Even if you're plotting the total instead of the rate (which was NOT the metric being discussed)

[meekGee]

[2021] 31
[2022] 60
[2023] 90-100  estimated
[2023] 120  projected by Spacex

That's +100%, +50-60%, +20-33%

Emphasis added.

+100%, +50-60%, +20-33%.  Those are hard numbers.

one hard number, one projection and one estimate.  Not hard numbers.

Not enough data, but that hasn't stop anybody from trying to torture it.  It's not like the model even matters.

Yeah but those are the numbers being discussed.

You can zoom in to monthly rates, but if the yearly ones are not exponential the monthly ones won't be either.

In fact the announced projection by SpaceX of 120 and 144 should have squashed any talk of exponential growth since it's now down to +20% annually..

[RedLineTrain]

Pushing Falcon 9 launch rate has a lot of benefits, so long as the incremental investments aren't very high.  In case they aren't obvious to all, here are a few of these benefits...

SpaceX is learning as an organization how to launch often, which eventually will help set the bar for Starship.

It is working on minimizing marginal cost of launch.  The cost of the 101st launch (or 301st...) is dirt cheap as far as these things go.

In the extreme, SpaceX could satisfy the FCC that it can launch its entire ~34,000 Starlink constellation comprising v2 minis.

The company's credibility as the premier launch provider in the history of spaceflight is cemented.

[Robotbeat]

No, I’m not using an incorrect definition of linear.

As far as your latter point: Right, and I think ratio is a far better fit to what they’re doing. Again, they’re not flipping a switch and magically getting better on Jan 1st, they’re building on past knowledge and capacity to do slightly better from one month to the next. The actual resulting launch rate will be noisy, and it is unlikely they’ll maintain it forever, but a compounding improvement just seems like a much better approximation to what they’re doing than a linear fit. That’s why company growth or a nation’s  GDP growth is also given in terms of an exponential.

[DanClemmensen]

But even if it’s just 5 Starship launches, that’s the equivalent of 50 Falcon 9s. That would already be a substantial effect.

OK, please make a prediction on the year in which there will be 5 F9-replacing Starship launches. (i.e., not depot, tanker, HLS,...)

Also note that I don't think we know enough about Starlink V.2 versus Starlink V.2 mini. It's possible that SpaceX want to launch the same number of satellites, not the same payload mass. In either case, Starship will replace all F9 Starlink launches as soon as it reaches high cadence. I still think it's 2028, but this is a mostly just an uneducated guess.

If they eventually complete the 42,000-satellite system, including replacing all earlier satellites with V.2, and if the SS delivers 72 V.2/launch, then they will need 584 SS launches. If they eventually get to an average satellite lifetime of ten years, this will stabilize as 59 Starlink lunches per year after some peak year of maybe 200 SS Starlink launches.

[Barley]

[2021] 31
[2022] 60
[2023] 90-100  estimated
[2023] 120  projected by Spacex

That's +100%, +50-60%, +20-33%

Emphasis added.

+100%, +50-60%, +20-33%.  Those are hard numbers.

one hard number, one projection and one estimate.  Not hard numbers.

Not enough data, but that hasn't stop anybody from trying to torture it.  It's not like the model even matters.

Yeah but those are the numbers being discussed.

You can zoom in to monthly rates, but if the yearly ones are not exponential the monthly ones won't be either.

In fact the announced projection by SpaceX of 120 and 144 should have squashed any talk of exponential growth since it's now down to +20% annually..

They are the numbers being discussed by you.  Other people are using the actual number of launches in prior years.

"SpaceX projects xxx launches in 202x" is a fine argument in support of xxx launches in 202x.  But it is not the same as historic data.  Curve fitting to projections will match the projections but the curve fitting adds nothing.

And so I'm not just picking on you.  If you are fitting an exponential you should use a semi-log plot.  The usual way of fitting an exponential is to perform a linear fit to the log of the data.  This changes the weighting of the errors in a way that is better matches a semi-log.

[steveleach]

Dec 2022 was 7 launches
Jan 2023 was 7 launches
Feb 2023 was 7 launches
Mar 2023 was 7 launches
Apr 2023 was 7 launches
May 2023 was 7 launches
Jun 2023 was 7 launches

Where do you get those numbers from?

The launch history on wikipedia that I've been using has 7, 7, 6, 8, 6, 9, 7.