Rocket scientist here, and I regret to inform you that your numbers are ludicrously optimistic. First, the assumption that you can get away with 40 kg of rocket dry mass per passenger is not something I can ever see happening in this universe. To try to get slightly better numbers, I took a 737 MAX8 (technically the BBJ version, but close enough) which gives me about 575 lb/255 kg of empty weight per passenger if you load it up to the regulatory limit. Call it 300 kg total at a minimum with passengers. Now in fairness, that includes weight for things like engines and wings (with, you know, fuel tanks) that we really should split out if we were doing things properly, but I'm not going to bother. Atomic rockets has the more complex equations to scale that stuff.
Second, there's no way that you can ignore all of the other costs. For modern airlines, fuel is about 15% of operating costs, half of employee salaries.
Also, no, you won't use solid rockets for this. Those are frankly a bad idea if you want to put people onboard because you can't shut them down and they have potentially very bad failure modes that most liquid rockets don't. (Although methane/LOX is also a potential bomb, but at least there, you have more mitigation options.) And they're likely to be inherently more expensive because you have to handle specialized materials that are basically explosives. Solid is great when you need easy storage, but that's mostly a military requirement.
Yeah, looking back at Starship specs, assuming a 100 tonne payload of people we get 120 kg of rocket per person. Plugging that in I get an optimal speed of 1.4 km/s and propellant costs of $164. A little slower but still a similar price as a plane ticket. I'll add a note on the post.
Staffing:
Since rocket takeoff and landings are already performed by onboard computers I think rocket planes could have very minimal staff, particularly because the trip is short and would be tricky to have e.g. a beverage service in the boost phase or at zero-g. So that means most of the staff would be on the ground, and their attention can be split over multiple flights, making them a smaller fraction of the overall cost (particularly since the fuel costs are higher relative to airplanes).
Of course, in a world where governments allowed rockets to land on their territory, there would probably be some legally-mandated staff aboard.
Capital costs:
This is where the optimistic analysis comes in, because I wasn't sure how to model this properly so I ignored it. I would guess that with rockets safe enough to be reused many times, the capital costs (as a percent of ticket price) would be similar to airplanes (and I think it's less than 10% in that case?).
If you have ideas for how to improve the model (particularly for capital costs and dry mass per person) it would be great to do a follow-up post!
I don't think most airline staff is on the planes, either. A typical domestic narrowbody has 2 pilots and 3 or 4 cabin crew. On the ground, you have the gate agents, checkin staff, baggage handlers and plane handlers, as well as the mechanics in the background. And yes, those are split across multiple flights, but there are also more of them. Even for mechanics, I would ballpark 2-3 hours per flight hour. I'd expect it would be at least as much as things today, because while you may not have onboard crew, I expect the labor of turning the thing to be nontrivial.
Capital costs are going to be heavily dependent on how quickly you can turn the thing after each flight. A modern narrowbody flies something like 4000 hours/year. The advantage is that you can do a flight in minutes instead of hours. The disadvantage is that I really doubt you'd be able to come anywhere near matching that utilization, because this is going to be a higher-stress device with much nastier failure modes. At the same time, modern rockets are really not optimized for quick turns. I have no idea what the investment to fix that would look like, sadly.
I do have suggestions for the dry mass model. Specifically, I'd look at this modified rocket equation:
M = R * ( Mpl / (1 - (Pf * (R-1)) - (Pi * R)) )
Mpl is payload mass, in this case the weight of the people and the specific stuff that supports them (pressure vessels, chairs, life support, etc). Call it 150 kg/person all up. (Pressure vessels that you plan to use repeatedly are heavy.) R is the mass ratio. Pf is the ratio of stuff that scales with fuel, which in practice means the percentage weight of the tanks. And Pi is the fraction of things like engines and dry structure which scales with total weight. I don't have excellent numbers for either, but if I take 2 km/s of delta-V, Ve of 3.7 km/s, .05 for Pf and .1 for Pi, I get a total non-fuel mass of 190 kg/person. I suspect it's actually higher than that, and .2 Pi takes it to 240 kg/person. (Obviously, this isn't infinitely scalable, but it's an easy metric to work in.) The big thing is that the structural drivers for this will be quite a bit different than a normal rocket, which sees a lot less time in use than this would have to. SpaceX's lead Falcon 9 booster is at 20 flights, which is going to be absolutely trivial on the scale of this thing if you have any hope of making a profit.
Fair point on the staffing. I still think its fair to say that staff would add less than an order-of-magnitude to the cost estimate? I tend not to worry about stuff like that when I'm using such a crude model.
Capital costs: now that you say it, the launch cadence and lifetime of the engines is probably the key factor here. If a rocket can get reused a lot per year, you can pay back your loans pretty quickly, if not, capital will be very expensive. It seems like the rocket motors are the part with the critical combination of short lifetime and high costs. SpaceX probably has good data on lifetime and the difficulty of replacing an engine, I'll see what I can find.
Rocket equation: neat! You can almost use Pf and Pi as figures-of-merit for a rocketplane design, though maybe they should be scaling exponents?
I think the counterintuitive thing with this model is that mass per person doesn't change fuel costs that much because the delta-V goes way down to adapt. For example, when I put in 240 kg / person (link below) I get an optimal speed of only 1.2 km/s and the fuel costs increase to $189. It's fair to point out that at this speed, its not much better than proposed supersonic airliners!
So my assumption that the rocket can get reused a lot (and quickly) is probably the issue and capital costs are much larger relative to fuel. To make that concrete, if a rocket costs $100 mil to build, flies 10 times over its lifetime, and carries 1000 passengers (?) that's $10K per ticket (ignoring the fact that if the launch cadence is low, interest payments will increase this a bit).
Ack. It has clearly been far too long since I did this because I completely overlooked that this is really not how delta-V works. Taking the absolute easiest (and dumbest) analysis possible (flat earth, vacuum, all delta-V instantly at 45 deg), I get a range of 146.7 km for 1.2 km/s and 408 km for 2 km/s. These are definitely not right, but at least provide a sanity check. For the 9000 km you need to get from LA to Paris, this method suggests a minimum delta-V of 9.4 km/s. That's probably a bit too high (the assumption of a flat earth is going to break down hard here) but even if I am fairly generous and assume you can do it on, say, 7 km/s, I get a dry mass of 2716 kg/person with Pf 0.05 and Pi 0.1. In practice, that's on the very edge of what you can manage without staging. I actually wonder if these might not be transatlantic/transcontinental instead of going straight from LA to Paris.
I knew there was something wrong with the numbers from earlier. That explains it. Sorry for zeroing in on the wrong part.
As for capital, I'm not sure those are good SWAG numbers. Worth noting that ~$100 million is a 737 that carries ~150 people. And you're not going to do that on a rocket of similar size, to say nothing of a thousand people. But 10 flights is also really low for something that has an operational model closer to an airliner than even what SpaceX is doing now. You could probably swap number of flights and number of passengers and get something vaguely reasonable, although even that is probably optimistic about what your manufacturing is going to cost.
Oh interesting, this point about delta-V is much more important.
You're right that if dV requirements are high, then shorter hops would be necessary, but at that point, the time savings are too small to justify the costs.
The larger point being that if you can't fly outside the atmosphere at a variable speed, then the model breaks down.
Though I don't really get why rockets and ICBM's have to follow a ballistic/orbital trajectory. Naively, couldn't a rocket boost out of the atmosphere and make many small ballistic hops (setting the difficulty of doing that aside) instead of reaching orbit?
It depends on how short the hops are. If you can go LA>New York in 20 minutes, spend an hour or so in the lounge and then get to Paris in another 20 minutes, that seems obviously better than a direct flight from LA to Paris.
Edit: On a normal jetliner, that is.
>Though I don't really get why rockets and ICBM's have to follow a ballistic/orbital trajectory. Naively, couldn't a rocket boost out of the atmosphere and make many small ballistic hops (setting the difficulty of doing that aside) instead of reaching orbit?
I mean, you could, it would just be worse. The effects here are sharply nonlinear. The "flat earth, no atmosphere" model has range scaling with the square of delta-V. Obviously, reality is more complicated, but to a first approximation, "a series of ballistic hops" just means "using thrust to fight gravity at points along the way". The basic idea of an orbit is that you are going so fast that the ground is falling out from below you as fast as you are falling. If you go slower, you need to use thrust to counter your falling, and it's strictly better to not do that.
If you want a better intuitive sense of how this works, Kerbal Space Program is excellent and fun.
It’s sad that we don’t have more people thinking big like this. Supersonic travel was once seen as the natural progression of air travel and yet here we are…in 2024…and there is not a single operational supersonic jet.
Fuel costs seem to be the primary issue. Naturally, the faster we travel the greater the drag. Getting the price down seems to be a matter of getting out of the atmosphere…but that implies rocketry.
Rocket scientist here, and I regret to inform you that your numbers are ludicrously optimistic. First, the assumption that you can get away with 40 kg of rocket dry mass per passenger is not something I can ever see happening in this universe. To try to get slightly better numbers, I took a 737 MAX8 (technically the BBJ version, but close enough) which gives me about 575 lb/255 kg of empty weight per passenger if you load it up to the regulatory limit. Call it 300 kg total at a minimum with passengers. Now in fairness, that includes weight for things like engines and wings (with, you know, fuel tanks) that we really should split out if we were doing things properly, but I'm not going to bother. Atomic rockets has the more complex equations to scale that stuff.
Second, there's no way that you can ignore all of the other costs. For modern airlines, fuel is about 15% of operating costs, half of employee salaries.
Also, no, you won't use solid rockets for this. Those are frankly a bad idea if you want to put people onboard because you can't shut them down and they have potentially very bad failure modes that most liquid rockets don't. (Although methane/LOX is also a potential bomb, but at least there, you have more mitigation options.) And they're likely to be inherently more expensive because you have to handle specialized materials that are basically explosives. Solid is great when you need easy storage, but that's mostly a military requirement.
Thanks for the feedback! I always err towards an optimistic analysis of speculative tech (https://splittinginfinity.substack.com/p/optimistic-cost-benefit-analysis) as a way to rule out certain ideas and combat pessimism.
On to the other points:
Engine mass:
Yeah, looking back at Starship specs, assuming a 100 tonne payload of people we get 120 kg of rocket per person. Plugging that in I get an optimal speed of 1.4 km/s and propellant costs of $164. A little slower but still a similar price as a plane ticket. I'll add a note on the post.
Staffing:
Since rocket takeoff and landings are already performed by onboard computers I think rocket planes could have very minimal staff, particularly because the trip is short and would be tricky to have e.g. a beverage service in the boost phase or at zero-g. So that means most of the staff would be on the ground, and their attention can be split over multiple flights, making them a smaller fraction of the overall cost (particularly since the fuel costs are higher relative to airplanes).
Of course, in a world where governments allowed rockets to land on their territory, there would probably be some legally-mandated staff aboard.
Capital costs:
This is where the optimistic analysis comes in, because I wasn't sure how to model this properly so I ignored it. I would guess that with rockets safe enough to be reused many times, the capital costs (as a percent of ticket price) would be similar to airplanes (and I think it's less than 10% in that case?).
If you have ideas for how to improve the model (particularly for capital costs and dry mass per person) it would be great to do a follow-up post!
I don't think most airline staff is on the planes, either. A typical domestic narrowbody has 2 pilots and 3 or 4 cabin crew. On the ground, you have the gate agents, checkin staff, baggage handlers and plane handlers, as well as the mechanics in the background. And yes, those are split across multiple flights, but there are also more of them. Even for mechanics, I would ballpark 2-3 hours per flight hour. I'd expect it would be at least as much as things today, because while you may not have onboard crew, I expect the labor of turning the thing to be nontrivial.
Capital costs are going to be heavily dependent on how quickly you can turn the thing after each flight. A modern narrowbody flies something like 4000 hours/year. The advantage is that you can do a flight in minutes instead of hours. The disadvantage is that I really doubt you'd be able to come anywhere near matching that utilization, because this is going to be a higher-stress device with much nastier failure modes. At the same time, modern rockets are really not optimized for quick turns. I have no idea what the investment to fix that would look like, sadly.
I do have suggestions for the dry mass model. Specifically, I'd look at this modified rocket equation:
M = R * ( Mpl / (1 - (Pf * (R-1)) - (Pi * R)) )
Mpl is payload mass, in this case the weight of the people and the specific stuff that supports them (pressure vessels, chairs, life support, etc). Call it 150 kg/person all up. (Pressure vessels that you plan to use repeatedly are heavy.) R is the mass ratio. Pf is the ratio of stuff that scales with fuel, which in practice means the percentage weight of the tanks. And Pi is the fraction of things like engines and dry structure which scales with total weight. I don't have excellent numbers for either, but if I take 2 km/s of delta-V, Ve of 3.7 km/s, .05 for Pf and .1 for Pi, I get a total non-fuel mass of 190 kg/person. I suspect it's actually higher than that, and .2 Pi takes it to 240 kg/person. (Obviously, this isn't infinitely scalable, but it's an easy metric to work in.) The big thing is that the structural drivers for this will be quite a bit different than a normal rocket, which sees a lot less time in use than this would have to. SpaceX's lead Falcon 9 booster is at 20 flights, which is going to be absolutely trivial on the scale of this thing if you have any hope of making a profit.
Fair point on the staffing. I still think its fair to say that staff would add less than an order-of-magnitude to the cost estimate? I tend not to worry about stuff like that when I'm using such a crude model.
Capital costs: now that you say it, the launch cadence and lifetime of the engines is probably the key factor here. If a rocket can get reused a lot per year, you can pay back your loans pretty quickly, if not, capital will be very expensive. It seems like the rocket motors are the part with the critical combination of short lifetime and high costs. SpaceX probably has good data on lifetime and the difficulty of replacing an engine, I'll see what I can find.
Rocket equation: neat! You can almost use Pf and Pi as figures-of-merit for a rocketplane design, though maybe they should be scaling exponents?
I think the counterintuitive thing with this model is that mass per person doesn't change fuel costs that much because the delta-V goes way down to adapt. For example, when I put in 240 kg / person (link below) I get an optimal speed of only 1.2 km/s and the fuel costs increase to $189. It's fair to point out that at this speed, its not much better than proposed supersonic airliners!
So my assumption that the rocket can get reused a lot (and quickly) is probably the issue and capital costs are much larger relative to fuel. To make that concrete, if a rocket costs $100 mil to build, flies 10 times over its lifetime, and carries 1000 passengers (?) that's $10K per ticket (ignoring the fact that if the launch cadence is low, interest payments will increase this a bit).
Optimal speed: https://www.wolframalpha.com/input?i=minimize+%280.028%29*%2810000*1000%29%2F%28x%29+%2B+%289%29*240*%281%2F5%29*%28exp%28x%2F%28350*9.81%29%29+-+1%29
Fuel cost: https://www.wolframalpha.com/input?i=%289%29*240*%281%2F5%29*%28exp%28x%2F%28350*9.81%29%29+-+1%29%2C+x%3D1244
Ack. It has clearly been far too long since I did this because I completely overlooked that this is really not how delta-V works. Taking the absolute easiest (and dumbest) analysis possible (flat earth, vacuum, all delta-V instantly at 45 deg), I get a range of 146.7 km for 1.2 km/s and 408 km for 2 km/s. These are definitely not right, but at least provide a sanity check. For the 9000 km you need to get from LA to Paris, this method suggests a minimum delta-V of 9.4 km/s. That's probably a bit too high (the assumption of a flat earth is going to break down hard here) but even if I am fairly generous and assume you can do it on, say, 7 km/s, I get a dry mass of 2716 kg/person with Pf 0.05 and Pi 0.1. In practice, that's on the very edge of what you can manage without staging. I actually wonder if these might not be transatlantic/transcontinental instead of going straight from LA to Paris.
I knew there was something wrong with the numbers from earlier. That explains it. Sorry for zeroing in on the wrong part.
As for capital, I'm not sure those are good SWAG numbers. Worth noting that ~$100 million is a 737 that carries ~150 people. And you're not going to do that on a rocket of similar size, to say nothing of a thousand people. But 10 flights is also really low for something that has an operational model closer to an airliner than even what SpaceX is doing now. You could probably swap number of flights and number of passengers and get something vaguely reasonable, although even that is probably optimistic about what your manufacturing is going to cost.
Oh interesting, this point about delta-V is much more important.
You're right that if dV requirements are high, then shorter hops would be necessary, but at that point, the time savings are too small to justify the costs.
The larger point being that if you can't fly outside the atmosphere at a variable speed, then the model breaks down.
Though I don't really get why rockets and ICBM's have to follow a ballistic/orbital trajectory. Naively, couldn't a rocket boost out of the atmosphere and make many small ballistic hops (setting the difficulty of doing that aside) instead of reaching orbit?
It depends on how short the hops are. If you can go LA>New York in 20 minutes, spend an hour or so in the lounge and then get to Paris in another 20 minutes, that seems obviously better than a direct flight from LA to Paris.
Edit: On a normal jetliner, that is.
>Though I don't really get why rockets and ICBM's have to follow a ballistic/orbital trajectory. Naively, couldn't a rocket boost out of the atmosphere and make many small ballistic hops (setting the difficulty of doing that aside) instead of reaching orbit?
I mean, you could, it would just be worse. The effects here are sharply nonlinear. The "flat earth, no atmosphere" model has range scaling with the square of delta-V. Obviously, reality is more complicated, but to a first approximation, "a series of ballistic hops" just means "using thrust to fight gravity at points along the way". The basic idea of an orbit is that you are going so fast that the ground is falling out from below you as fast as you are falling. If you go slower, you need to use thrust to counter your falling, and it's strictly better to not do that.
If you want a better intuitive sense of how this works, Kerbal Space Program is excellent and fun.
Rockets for VIPs will be here in a few years
It’s sad that we don’t have more people thinking big like this. Supersonic travel was once seen as the natural progression of air travel and yet here we are…in 2024…and there is not a single operational supersonic jet.
Fuel costs seem to be the primary issue. Naturally, the faster we travel the greater the drag. Getting the price down seems to be a matter of getting out of the atmosphere…but that implies rocketry.
Yeah and for all of these engines there's probably going to be more wear and tear than airplane engines. Replacement costs could add a lot too.
Still, we can dream.