SpaceX progress towards a 100 launch year

[meekGee]

On your tablet/phone you can hit the full site button, like, then click the mobil link at the bottom to get back to mobil view.

I have done that on occasion.

OT but I've tried that - for some reason the button doesn't show up on my phone, even in chrome and in non-mobile view.

Probably karma for using TnT.

[Robotbeat]

End of March

Okay, now you're just teasing me lol.

Using my previous method from here: https://forum.nasaspaceflight.com/index.php?topic=58240.msg2466409#msg2466409

((b*61/(e^(b) -1))/b)*(e^(b*(1+(31+28+31)/365)) - e^b) = 21., solve for b, gets us to b = ~0.554712 (1/year)
A =0.554712*61/(e^(0.554712) -1) = 45.6374897312 launches/year (that being the instantaneous annualized launch rate at Jan 1st, 2022), so, they're on track for:
integral(45.6374897312*e^(0.554712*t),t,1,2) = 106.228 launches in 2023. (Note this is better than last time I estimated, where it was hovering around 100 launches in 2023... so they've more than caught up to the rate-of-compounding-improvement needed to meet 100 launches in 2023.)

https://www.wolframalpha.com/input?i=integral%2845.6374897312*e%5E%280.554712*t%29%2Ct%2C1%2C2%29


This basically just uses the past number of launches in 2022 (61 launches, 365 days), the number of launches (21) and days (31, 28, 31) there as been since the start of 2023, and the idea that the overall launch rate is improving at some compounding rate which fits these data points (which seems reasonable to me, especially compared to just assuming it's a constant value that arbitrarily changes on January 1st). A full least-squares type fit would be better, but this is easier for me as I'm sick of messing with weirdly formatted spreadsheets.

(Note in this graph, "1.0" corresponds with the start of 2023, 2.0 with the end of 2023, and the y axis is the instantaneous annual launch rate.)

Okay, updating this so it's just in python (with numpy and scipy libraries), not using WolframAlpha, which has been failing on some of these requests:

import numpy as np
from scipy.optimize import fsolve

from scipy.integrate import quad

daysinto2023 = 31+28+31+2
timenow = 1+daysinto2023/365 #time in years since Jan 1, 2022
print("How many days has it been since the start of 2023:", daysinto2023)

launchesnow = 22 #launches that there has been from Jan 1, 2023 until now
print("How many launches has there been since the start of 2023:", launchesnow)
def equation(b,timenow,launchesnow):
    return (61/(np.exp(b) - 1))*(np.exp(b*(timenow)) - np.exp(b)) - launchesnow

b = fsolve(equation, 0.5,args=(timenow,launchesnow))
print("b:",b)

A = b*61/(np.exp(b) -1)
print("A:",A)

def integrand(t,A,b):
    return A*np.exp(b*t)

result, error = quad(integrand, 1, 2,args=(A,b))

print("The estimated number of launches in 2023:", result)

...gets me:

>python fitexp.py
How many days has it been since the start of 2023: 92
How many launches has there been since the start of 2023: 22
b: [0.59424445]
A: [44.66012711]
The estimated number of launches in 2023: 110.51135940155463

...with the latest launch today.

I promise I won't update this every single launch LOL, but I just wanted to get beyond relying on WolframAlpha.

(Note that what I think will happen, now that they're getting ahead of the exponential rate, is that the curve or launch rate will be sort of sigmoid-esque. If they continue increasing in the mid-term, they'll have headroom to level out a bit toward the end of the year. Not that they'll intentionally be slowing down or anything, but I think the transition to Starship will likely entail a deceleration of the launch rate increase for Falcon 9.)

[meekGee]

End of March

Okay, now you're just teasing me lol.

Using my previous method from here: https://forum.nasaspaceflight.com/index.php?topic=58240.msg2466409#msg2466409

((b*61/(e^(b) -1))/b)*(e^(b*(1+(31+28+31)/365)) - e^b) = 21., solve for b, gets us to b = ~0.554712 (1/year)
A =0.554712*61/(e^(0.554712) -1) = 45.6374897312 launches/year (that being the instantaneous annualized launch rate at Jan 1st, 2022), so, they're on track for:
integral(45.6374897312*e^(0.554712*t),t,1,2) = 106.228 launches in 2023. (Note this is better than last time I estimated, where it was hovering around 100 launches in 2023... so they've more than caught up to the rate-of-compounding-improvement needed to meet 100 launches in 2023.)

https://www.wolframalpha.com/input?i=integral%2845.6374897312*e%5E%280.554712*t%29%2Ct%2C1%2C2%29


This basically just uses the past number of launches in 2022 (61 launches, 365 days), the number of launches (21) and days (31, 28, 31) there as been since the start of 2023, and the idea that the overall launch rate is improving at some compounding rate which fits these data points (which seems reasonable to me, especially compared to just assuming it's a constant value that arbitrarily changes on January 1st). A full least-squares type fit would be better, but this is easier for me as I'm sick of messing with weirdly formatted spreadsheets.

(Note in this graph, "1.0" corresponds with the start of 2023, 2.0 with the end of 2023, and the y axis is the instantaneous annual launch rate.)

Okay, updating this so it's just in python (with numpy and scipy libraries), not using WolframAlpha, which has been failing on some of these requests:

import numpy as np
from scipy.optimize import fsolve

from scipy.integrate import quad

daysinto2023 = 31+28+31+2
timenow = 1+daysinto2023/365 #time in years since Jan 1, 2022
print("How many days has it been since the start of 2023:", daysinto2023)

launchesnow = 22 #launches that there has been from Jan 1, 2023 until now
print("How many launches has there been since the start of 2023:", launchesnow)
def equation(b,timenow,launchesnow):
    return (61/(np.exp(b) - 1))*(np.exp(b*(timenow)) - np.exp(b)) - launchesnow

b = fsolve(equation, 0.5,args=(timenow,launchesnow))
print("b:",b)

A = b*61/(np.exp(b) -1)
print("A:",A)

def integrand(t,A,b):
    return A*np.exp(b*t)

result, error = quad(integrand, 1, 2,args=(A,b))

print("The estimated number of launches in 2023:", result)

...gets me:

>python fitexp.py
How many days has it been since the start of 2023: 92
How many launches has there been since the start of 2023: 22
b: [0.59424445]
A: [44.66012711]
The estimated number of launches in 2023: 110.51135940155463

...with the latest launch today.

I promise I won't update this every single launch LOL, but I just wanted to get beyond relying on WolframAlpha.

(Note that what I think will happen, now that they're getting ahead of the exponential rate, is that the curve or launch rate will be sort of sigmoid-esque. If they continue increasing in the mid-term, they'll have headroom to level out a bit toward the end of the year. Not that they'll intentionally be slowing down or anything, but I think the transition to Starship will likely entail a deceleration of the launch rate increase for Falcon 9.)

I'm with you on all the data analysis, but to make forward looking statements, it's best to look beyond just the curve fit.

There are several potential bottlenecks:
- Barge round-trip times
- Payload availability
- Weather
- There may be some component of pad turn-around time (especially on pad 40) that will dictate a lower limit.

If we want 9/month later this year, that's 3/month per barge until ideal utilization. (and ignoring RTLSs).  How long is the round trip for Starlink launches?   

[kevin-rf]

It's more than that. 

SLC-40 is already maxed.

LC-39A is booked through at least June with Heavy missions, and let's not forget Dragon or Psyche.This essentially throttles the launch rate.We should see an uptick in August and again once Psyche clears the pad in October. LC-39A can match SLC-40's flight rate when it doesn't have payloads that require extra care.

SLC-4E has several RTLS mission,  they greatly help the flight rate.  This April we may even see three RTLS missions in a row. Granted one had slipped from March to April. The pad is already barge maxed for non RTLS missions. So the rate will drop a little months that do not have RTLS missions lined up.

The point is,  the flight rate for the rest of 2023 is not about curve fitting an efficiency model. It is looking at the missions they have stacked up and see how much they reduce (Heavy Dragons) and improve (RTLS) flight rates. The numbers need to model individually each pad.

(I know it won't happen,  but I would love to see four SLC-4E missions this month. Transport and Tracking Layer, Sarah 2/3, Transporter, and a Starlink. It's possible,  but at least one will slip into May)

[kevin-rf]

Well JTRI for starlink 5-10 left on the 25th and came back yesterday.  So 8 days plus a few more for booster processing.

Mileage will vary and CRS SpX-27 did a partial boost back so ASOG would not need to be towed out as far. They mentioned in the webcast it will save them a couple of days on barge cycle time. Not all missions have the extra performance for such a profile. This might be a one off.

[wannamoonbase]

Well JTRI for starlink 5-10 left on the 25th and came back yesterday.  So 8 days plus a few more for booster processing.

Mileage will vary and CRS SpX-27 did a partial boost back so ASOG would not need to be towed out as far. They mentioned in the webcast it will save them a couple of days on barge cycle time. Not all missions have the extra performance for such a profile. This might be a one off.

Regarding the ASDS, there is the ideal cycle times, but then there are things like maintenance cycles, which seem to happen once or twice a year and then there are sea conditions, which have come up.

I think a third East Coast ASDS would help alot but is unlikely.  Short of that maximizing the RTLS and West Coast flights are essential to the goal.

[DanClemmensen]

Well JTRI for starlink 5-10 left on the 25th and came back yesterday.  So 8 days plus a few more for booster processing.

Mileage will vary and CRS SpX-27 did a partial boost back so ASOG would not need to be towed out as far. They mentioned in the webcast it will save them a couple of days on barge cycle time. Not all missions have the extra performance for such a profile. This might be a one off.

Regarding the ASDS, there is the ideal cycle times, but then there are things like maintenance cycles, which seem to happen once or twice a year and then there are sea conditions, which have come up.

I think a third East Coast ASDS would help alot but is unlikely.  Short of that maximizing the RTLS and West Coast flights are essential to the goal.

If SpaceX were to activate a third east-coast ASDS, how many additional F9 launches would it enable in its lifetime? Pick some guesses you are comfortable with and then calculate. Here are my guesses.
No gain  any year with less than about 90 launches/yr, and less than 90-N in years with more than N launches/yr. But the number of F9 launches/yr will decrease when Starship becomes operational and decrease further as Starship become fully operational. This schedule is unknown, so guess:
    2023:   91
    2024: 100
    2025:   60   (SS begins operational flights)
    2026:   40   (SS flights increase)

This says that a third ASDS will enable less than five additional launches during its lifetime. This does not make economic sense, so it won't be activated.

[meekGee]

It's more than that. 

SLC-40 is already maxed.

LC-39A is booked through at least June with Heavy missions, and let's not forget Dragon or Psyche.This essentially throttles the launch rate.We should see an uptick in August and again once Psyche clears the pad in October. LC-39A can match SLC-40's flight rate when it doesn't have payloads that require extra care.

SLC-4E has several RTLS mission,  they greatly help the flight rate.  This April we may even see three RTLS missions in a row. Granted one had slipped from March to April. The pad is already barge maxed for non RTLS missions. So the rate will drop a little months that do not have RTLS missions lined up.

The point is,  the flight rate for the rest of 2023 is not about curve fitting an efficiency model. It is looking at the missions they have stacked up and see how much they reduce (Heavy Dragons) and improve (RTLS) flight rates. The numbers need to model individually each pad.

(I know it won't happen,  but I would love to see four SLC-4E missions this month. Transport and Tracking Layer, Sarah 2/3, Transporter, and a Starlink. It's possible,  but at least one will slip into May)

LC40 is key.
Is it really maxed out?  We've heard that so many times before...  What prevents it from launching every third day for example?

The barges have transit times as hard limiters but what limits the pad turn-around time that can't be incrementally improved?

[Robotbeat]

None of these things are yet at some fundamental turnaround limit. Every limit can be pushed. Get the booster off the barge a little faster. Tow the barge a little faster. Add more RTLS to the mix. Exercise all 3 pads more evenly. …and yeah, do a few Boca Chica flights.

There’s plenty there to get to 100 flights this year.

[wannamoonbase]

Well JTRI for starlink 5-10 left on the 25th and came back yesterday.  So 8 days plus a few more for booster processing.

Mileage will vary and CRS SpX-27 did a partial boost back so ASOG would not need to be towed out as far. They mentioned in the webcast it will save them a couple of days on barge cycle time. Not all missions have the extra performance for such a profile. This might be a one off.

Regarding the ASDS, there is the ideal cycle times, but then there are things like maintenance cycles, which seem to happen once or twice a year and then there are sea conditions, which have come up.

I think a third East Coast ASDS would help alot but is unlikely.  Short of that maximizing the RTLS and West Coast flights are essential to the goal.

If SpaceX were to activate a third east-coast ASDS, how many additional F9 launches would it enable in its lifetime? Pick some guesses you are comfortable with and then calculate. Here are my guesses.
No gain  any year with less than about 90 launches/yr, and less than 90-N in years with more than N launches/yr. But the number of F9 launches/yr will decrease when Starship becomes operational and decrease further as Starship become fully operational. This schedule is unknown, so guess:
    2023:   91
    2024: 100
    2025:   60   (SS begins operational flights)
    2026:   40   (SS flights increase)

This says that a third ASDS will enable less than five additional launches during its lifetime. This does not make economic sense, so it won't be activated.

Except a few weeks ago some SpaceX exec, responsible for F9/FH (I forget his name) alluded to more than 100 flights, I believe he reference 200 a year. 

2024 maybe the peak F9 year, after 6.5 years of following Starship and the 20-ish months since Elon said Starship would launch next month, we are seeing that superheavy sized rockets are a lot harder to handle, prepare and fly than their smaller relatives.  That is to say, SH&SS may be longer than we think before they are hauling the mail for SpaceX.

The time to decide on adding a 3rd ASDS was the week after ASOG was delivered.  It won't happen now, but the ASDS' will limit the F9 flight rate.  That is if 100-ish flights per year is seen as a limit.

[RedLineTrain]

None of these things are yet at some fundamental turnaround limit. Every limit can be pushed. Get the booster off the barge a little faster. Tow the barge a little faster. Add more RTLS to the mix. Exercise all 3 pads more evenly. …and yeah, do a few Boca Chica flights.

There’s plenty there to get to 100 flights this year.

You can do some partial boostbacks as well, if you really need to optimize barge use.

[meekGee]

None of these things are yet at some fundamental turnaround limit. Every limit can be pushed. Get the booster off the barge a little faster. Tow the barge a little faster. Add more RTLS to the mix. Exercise all 3 pads more evenly. …and yeah, do a few Boca Chica flights.

There’s plenty there to get to 100 flights this year.

I agree on almost all.

I think barge towing speed may be limited by things outside their control - there has to be a max practical speed limit and they may be there already.

And weather is weather.

(That's why I'm looking at barge turn-around as a possible limiter)

Agreed about RTLS, and about possible partial boostback. 

I definitely think pad turn-around can be improved. Same with payload integration.

I'm not at all saying 100/yr is a problem, just looking for things that may change the curve

[Redclaws]

Yeah, speed for a ship and especially for a barge shaped thing being towed, is heavily limited by sea state, among other things.  The scaling of required power for greater speeds in water is also an exponential (a complex one that depends on many factors and I won’t claim to fully understand, but required power is generally a cube or greater of the speed for “normal” hull shapes, a barge might be worse).  So they probably can’t speed the barges themselves up much - they probably already tow them at or near what they consider the comfortable/safe limit, because that’s such an obvious and simple time waster.

But more barges and more towing ships is hardly an extravagance if they’re needed.

[Robotbeat]

The barge is not near its hull speed limit. 8 knots is till enabling for pretty fast turnaround. 600km each way (typical for Starlink) means about 3.5 day turnaround in principle possible per barge even at just 8 knots average speed. That would give a limit of 300 per year for all 3 droneships, not counting RTLS or Boca Chica.

So again, I don’t think those are the hard limits we think they are.

[FLHerne]

One way to reduce barge cycle time would be unloading at ports further downrange.

Charleston, for instance, is roughly half the distance of Port Canaveral from the ~50-55° landing site for CRS/crewed/some Starlink.

We saw them do it with the fairing recovery vessels a few times, dropping off at Morehead City for a quick turnaround. Boosters would be a greater logistical challenge but perhaps cheaper than a third ASDS.

[Robotbeat]

One way to reduce barge cycle time would be unloading at ports further downrange.

Charleston, for instance, is roughly half the distance of Port Canaveral from the ~50-55° landing site for CRS/crewed/some Starlink.

We saw them do it with the fairing recovery vessels a few times, dropping off at Morehead City for a quick turnaround. Boosters would be a greater logistical challenge but perhaps cheaper than a third ASDS.

That is a super good idea. If you have a lot of boosters but not a lot of droneships, it could pencil out. What is the distance from port of more hood city to where the boosters land? 300km?

Halving the time would increase the tempo by a factor of 2 but it means now you need more port loading infrastructure and having to truck the boosters further. Probably would be better to build another droneship, BUT it does give SpaceX some possible flexibility if they need to just crank up the throughput this year. Adding a crane and a booster stand is probably a lot easier than a building another droneship if you are trying to do it in just a few months.

… I guess that’s just another advantage to F9 being roadable.

[ZachS09]

SpaceX once used a port in Jacksonville when they were experimenting with drone ship landings in the mid-2010s.

That could be another place to unload.

[meekGee]

The barge is not near its hull speed limit. 8 knots is till enabling for pretty fast turnaround. 600km each way (typical for Starlink) means about 3.5 day turnaround in principle possible per barge even at just 8 knots average speed. That would give a limit of 300 per year for all 3 droneships, not counting RTLS or Boca Chica.

So again, I don’t think those are the hard limits we think they are.

That's good, I thought it was tighter than that.

Between this margin and the possibility of small boostback burns and the remote offloading idea - yeah, this feels like barges can support the existing curve.

Weather remains, but it's not as big a impactor. It's frustrating when it happens but it's rare.

So - payloads?  With the transition to V2 mini, probably not an issue.

So if no bottlenecks, then the sum influence of continuous small improvements will support continuing improvements.. 

So what will be the highest n for 30 consecutive days this year?  Will there be a 10 towards end-of-year?

[meekGee]

One way to reduce barge cycle time would be unloading at ports further downrange.

Charleston, for instance, is roughly half the distance of Port Canaveral from the ~50-55° landing site for CRS/crewed/some Starlink.

We saw them do it with the fairing recovery vessels a few times, dropping off at Morehead City for a quick turnaround. Boosters would be a greater logistical challenge but perhaps cheaper than a third ASDS.

That is a super good idea. If you have a lot of boosters but not a lot of droneships, it could pencil out. What is the distance from port of more hood city to where the boosters land? 300km?

Halving the time would increase the tempo by a factor of 2 but it means now you need more port loading infrastructure and having to truck the boosters further. Probably would be better to build another droneship, BUT it does give SpaceX some possible flexibility if they need to just crank up the throughput this year. Adding a crane and a booster stand is probably a lot easier than a building another droneship if you are trying to do it in just a few months.

… I guess that’s just another advantage to F9 being roadable.

I like it too, and the main upside is that there's almost no lead time (in comparison with building a new drone ship).

So unless there's a barge already being built, I'd easily go with this idea.

[Robotbeat]

This tweet shows a nice map. Looks like the distance to the nearest Carolina port is about a factor of 2.5x closer? Jacksonville doesn’t look close enough to make a big difference.

https://twitter.com/SpaceOffshore/status/1356007758911438848/photo/1