20 thoughts on “More Thoughts On Neil Stephenson And Rockets”
The base fuel cost to LEO is around $10/kg of payload. For most mature transport industries total cost is around 3-5 times fuel cost, and there seems little reason to think space transports are inherently any different. A mature space transport industry should be able to deliver payload to LEO for under $50/kg. The industry is currently about a thousand times more expensive than it should be and something needs to change.
I noted in comments that someone thought there was a substantial small scale limitation, hence large rockets were required. However the primary small scale limitation is actually the payload mass of a person, which infers a minimum GLOW in the 10 ton range. Everything else can scale down sufficiently, including payloads (I do advocate some degree of air launch to reduce aero losses and launch bureaucracy). 10-20 ton GLOW (1-2 person) seems to be about the optimal size for rocket vehicle development and operation. Sadly there are not actually that many companies pursuing this rocket vehicle scale.
Five fleets of five 10 ton GLOW rocket vehicles each flying on average three times a day would collectively serve something like five times the current global demand. It will be a very long time before large rockets are needed.
I should have stipulated that five fleets of five rockets (the reusable kind), is about the minimum necessary for fleet redundancy and open commercial competition. Less than that and one is getting into monopoly territory – back to where we are now.
it is clearly true that the reusability of rocket engines can be improved significantly and that the vehicles they power can be made fully reusable.
A fully reusable vehicle to LEO is a big step, some claiming it’s not even possible (with which I disagree.) But a fully reusable vehicle in LEO we could do yesterday. Some entrepreneur just needs to think of it as a charter service.
What do we need to test reusability of rocket engines? How about the same thing… a fully reusable vehicle in LEO?
Did I mention we could have a fully reusable vehicle in LEO now for practical experience for incremental improvements? …and that the cost could be recovered in support of future missions not yet specified (and would encourage the development of future missions since risk is reduced by an operational component?)
Best of all, the cost is known because the components have already flown which contrasts with the overblown costs of projects that do not include a fully reusable vehicle in LEO.
Oh and it also creates a brand new market for supply of fuel to LEO (and beyond.)
…or we could continue to design one use specific vehicles instead of general purpose vehicles.
Costs to orbit come down with volume, right? So why not give volume a reason to be with something we could use and build now for very little cost and a huge return? If it were operational, NASA would be forced to include it in most mission plans which would make it profitable almost immediately.
If Bigelow can’t find a buyer for the idea he should do it himself.
“The base fuel cost to LEO is around $10/kg of payload. For most mature transport industries total cost is around 3-5 times fuel cost, and there seems little reason to think space transports are inherently any different. A mature space transport industry should be able to deliver payload to LEO for under $50/kg. The industry is currently about a thousand times more expensive than it should be and something needs to change.”
Any data on this? Is there an existing rocket where you can add up the payload weight, the pounds of propellant (oxidizer counts as “fuel” until someone has a working airbreather), what the propellants cost delivered to the rocket tanks, and verify the $50/kg?
Some back of the envelope numbers (not terribly realistic, but should be within an order of magnitude):
Space shuttle:
payload: 24,950 kg
LOX/LH mix: 730,000 kg @ $1.60/kg (84% LOX at $0.01/kg, 16% LH at $10/kg)
SRB propellant mass: 1,008,000 kg @ $20/kg
Total cost: ~$22M, or about $1,000/kg
(note that almost the entire cost is in the solid propellant)
Titan II GLV:
payload: 3,600 kg
GLOW: 154,000 kg (probably a very high proportion is propellant)
cost: @ $10/kg: ~$1.5M, or $400/kg
(they used a pretty pricey propellant combo)
What most people think is that you need about 20 times your payload’s mass in propellant, which costs only a few dollars per pound. There are two problems: first, the payload is a small fraction of the rocket mass; and second no one is using inexpensive propellants. A better guesstimate is that you need about 50x the payload in propellant. So if you used inexpensive propellant, $50/kg or so is probably possible.
LOX/Kerosene costs almost nothing, but no one uses it to go to orbit. They probably should, though…
From the SpaceX web site:
F9 uses Lox Kero… 23050 lbs to LEO
735000lb glow.
Assume its 95% propellant (I know I can find the real number, I’m lazy)
Call it a$1 /lb for propellant. (I think Lox kero can be less than 0.5/lb)
That’s propellant cost of $30/lb to leo
With a 5X cost that’s 150/llb or $75 per kg.
Falcon 9 call it 333ton GLOW and 10 ton payload to LEO – 3% payload mass. Propellant mass is probably about 90% of GLOW.
Kerosene to LOX mass ratio is typically around 1:2.5, kerosene requirement is 8.5kg/kg of payload to LEO and LOX 21.5kg
Kerosene used to cost around $0.50/kg, maybe it is a bit higher than that now. LOX requires around one kWhr/kg to produce, in bulk it is something like $0.10-$0.20/kg. Say we assume $0.75/kg for kerosene and $0.15/kg for LOX I get just under $10 of propellant per kilogram to LEO.
LH2/LOX is also similar in cost, say 15% LH2, 70% LOX and 5% payload. LH2 I think used to cost around $4/kg for the shuttle, however this was perhaps not the cheapest supply. The equivalent energy of 1kg of hydrogen in natural gas (from which LH2 is derived) costs around $0.60, liquefaction need not be that expensive. The LH2 is proportionately much less and the payload fraction higher than for kerosene (due to greater ISP, though greater dry mass, the expensive part, does cut into this a bit). At $2.50/kg of LH2 I still get just under $10 of LH2/LOX propellant per kilogram of payload to LEO.
However existing rockets are so expensive that propellant cost is in the noise – propellant performance matters far more than propellant cost.
Use liquefied natural gas (~$0.25/kg) and LOX and you get down to around $5 of propellant per kilogram to LEO. LNG has slightly better ISP than kerosene and the fuel ratio is also much lower, like around 1:3.5. It does require slightly more tankage though due to its lower density. A number of CH4/LOX engines have been tested.
Reusability of rocket *engines* can be tested on the ground, on a test stand.
I am thinking that payload mass fractions in staged rockets is maybe 1:30, more like 1:100 if they ever get an SSTO going. The SSTO is probably “wasteful” of fuel, but it is supposed to cut costs by making the thing more like an airliner rather than a small fleet of aircraft.
Does this 3-5 times fuel cost work, like, for a trip on Southwest? On Amtrak? In your car?
I don’t know if we should go for full resuability. Space-flight is a very hostile environment that chews up things like heat shields and engines. I think a better architecture would focus on reusing the airframe, and being able to quickly swap out certain components that need to be checked or rebuilt, such as the engines and heat-shield. Don’t try to save everything, because you’ll end up tearing it apart and replacing it every flight anyway.
If, rather than having to tear the shuttle apart every time it needed to go up, you could simply unbolt and pull the main engines and bolt the new ones on (which you have prepared and ground tested, while the shuttle was flying), and rather than replace hundreds of individually crafted ceramic tiles every-time it flies, simply bolt on a one-piece phenolic expendable shield, then I think operational efficiency could be greatly improved.
Add to that the fact that drop-tanks aren’t all that expensive, and the only remaining lemon with the shuttle-type concept is the SRBs.
The shuttle isn’t inherently a worthless idea, it just needs to be done a whole lot better.
I would regard a two stage system where everything flies back and lands though as optimum.
Oh, and the double-preburning configuration of the SSMEs, designed to wring the last second of Isp out of LOX/LH2 probably isn’t worth it in terms of moving parts, operating temperature and pressure, and complexity, vs. simply carrying extra propellant.
3x is I think a good guide for air freight – a third capital, a third operations and maintenance, and a third fuel. 5x for passengers (they require a lot more looking after). I think cars work out roughly the same as do ships, trains may be a little off if tracks are under utilized.
This relationship is not coincidental. If fuel costs go up then more efficient and expensive designs are used, ships go slow to save fuel, etc., if fuel costs go down the reverse happens, people start using SUVs more, and so forth.
2% payload fraction is a figure I have often seen suggested for SSTO vehicles. But a lot has happened in recent decades and dry mass is something that is subject to incremental reduction.
There are also some out there tricks possible to greatly reduce dry mass, like self supporting inflatable external tanks (a bit like the inflatable Bigelow modules but for propellant). This can greatly reduce tank mass (perhaps less than 1% of GLOW), and make it far less of a constraint on vehicle design. Another trick is electric quadrotor air launching from within a say helium filled aeroshell – cheaper than using rocket, reduced aero losses, no tank insulation requirements, no engine altitude compensation, etc. There are a few more (including very light weight vertical landing solutions and higher T/W engines), but the difficult problem yet to find a really good solution is highly reusable lightweight reentry shielding.
Does this 3-5 times fuel cost work, like, for a trip on Southwest? On Amtrak? In your car?
I work for an airline, and our fuel costs are 40% of our operational costs. (Well, it was last time I checked, and admittedly fuel prices have fallen since then.)
I think its time to put a stake in the ground though, to cool down all the RLV proponents : Armadillo has been in the “business” for longer than a decade, and they have been moderately successful. They havent flown to 10 miles altitude yet ..
Extrapolate from here ..
They may very well reach 10 miles this year and 72 miles the year after that.
Armadillo has been in the “business” for longer than a decade, and they have been moderately successful. They havent flown to 10 miles altitude yet ..
Extrapolate from here ..
Dang, that baby’s been gestating for almost nine months now, and hardly anything has happened…
Extrapolate from here.
“I think its time to put a stake in the ground though, to cool down all the RLV proponents : Armadillo has been in the “business” for longer than a decade, and they have been moderately successful. They havent flown to 10 miles altitude yet.”
Most engineering professions have a mantra of sorts that suggests you can engineer anything with the following criteria:
Make the product…
…sooner
…cheaper
…reliable
.
NASA showed us that if you throw cost as a consideration (“waste anything but time”) that you can build just about anything once you set your mind to it.
In defense of John Carmack, he is throwing the concept out that it has to be ready to fly yesterday out the window, and using some of the design principles that have long been used in the computer software industry of rapid prototypes of small test concepts and dozens or hundreds of iterations of the design cycle to perfect the overall design.
I defy anybody to show a team that has more design experience and has fired more rockets with more types of fuel and designs than Armadillo…. with perhaps the exception of Wherner Von Braun and his team from Peenemünde. John Carmack has also done that at a fraction of the cost of other rocket development programs and largely out of his own pocket.
In defense of John Carmack, he is throwing the concept out that it has to be ready to fly yesterday out the window
In a sense he has taken it close to its extreme, in that something has to fly almost immediately, just not all the way to orbit. This is a very powerful strategy in software development, and it has shown itself useful for aerospace development too. Carmack was not the first to do it that way of course, neither for software nor for rockets. But he was going against a lot of conventional wisdom, showing it to be wrong in the process.
the difficult problem yet to find a really good solution is highly reusable lightweight reentry shielding.
Maybe that’s because everyone is thinking of using atmospheric drag for the bulk of the job of reentry delta vee. It’s easy and it works but it requires that reentry shielding. A powered descent stage would not necessarily need high-maintenance shielding at all. It isn’t something that makes sense if you have to launch all your propellant along with you every time, but if you have a propellant depot in orbit then powered descent stages become possible. Such a vehicle could bleed off most of the speed before reaching the upper atmosphere, thus requiring no more shielding than SpaceShipOne.
The base fuel cost to LEO is around $10/kg of payload. For most mature transport industries total cost is around 3-5 times fuel cost, and there seems little reason to think space transports are inherently any different. A mature space transport industry should be able to deliver payload to LEO for under $50/kg. The industry is currently about a thousand times more expensive than it should be and something needs to change.
I noted in comments that someone thought there was a substantial small scale limitation, hence large rockets were required. However the primary small scale limitation is actually the payload mass of a person, which infers a minimum GLOW in the 10 ton range. Everything else can scale down sufficiently, including payloads (I do advocate some degree of air launch to reduce aero losses and launch bureaucracy). 10-20 ton GLOW (1-2 person) seems to be about the optimal size for rocket vehicle development and operation. Sadly there are not actually that many companies pursuing this rocket vehicle scale.
Five fleets of five 10 ton GLOW rocket vehicles each flying on average three times a day would collectively serve something like five times the current global demand. It will be a very long time before large rockets are needed.
I should have stipulated that five fleets of five rockets (the reusable kind), is about the minimum necessary for fleet redundancy and open commercial competition. Less than that and one is getting into monopoly territory – back to where we are now.
it is clearly true that the reusability of rocket engines can be improved significantly and that the vehicles they power can be made fully reusable.
A fully reusable vehicle to LEO is a big step, some claiming it’s not even possible (with which I disagree.) But a fully reusable vehicle in LEO we could do yesterday. Some entrepreneur just needs to think of it as a charter service.
What do we need to test reusability of rocket engines? How about the same thing… a fully reusable vehicle in LEO?
Did I mention we could have a fully reusable vehicle in LEO now for practical experience for incremental improvements? …and that the cost could be recovered in support of future missions not yet specified (and would encourage the development of future missions since risk is reduced by an operational component?)
Best of all, the cost is known because the components have already flown which contrasts with the overblown costs of projects that do not include a fully reusable vehicle in LEO.
Oh and it also creates a brand new market for supply of fuel to LEO (and beyond.)
…or we could continue to design one use specific vehicles instead of general purpose vehicles.
Costs to orbit come down with volume, right? So why not give volume a reason to be with something we could use and build now for very little cost and a huge return? If it were operational, NASA would be forced to include it in most mission plans which would make it profitable almost immediately.
If Bigelow can’t find a buyer for the idea he should do it himself.
“The base fuel cost to LEO is around $10/kg of payload. For most mature transport industries total cost is around 3-5 times fuel cost, and there seems little reason to think space transports are inherently any different. A mature space transport industry should be able to deliver payload to LEO for under $50/kg. The industry is currently about a thousand times more expensive than it should be and something needs to change.”
Any data on this? Is there an existing rocket where you can add up the payload weight, the pounds of propellant (oxidizer counts as “fuel” until someone has a working airbreather), what the propellants cost delivered to the rocket tanks, and verify the $50/kg?
Some back of the envelope numbers (not terribly realistic, but should be within an order of magnitude):
Space shuttle:
payload: 24,950 kg
LOX/LH mix: 730,000 kg @ $1.60/kg (84% LOX at $0.01/kg, 16% LH at $10/kg)
SRB propellant mass: 1,008,000 kg @ $20/kg
Total cost: ~$22M, or about $1,000/kg
(note that almost the entire cost is in the solid propellant)
Titan II GLV:
payload: 3,600 kg
GLOW: 154,000 kg (probably a very high proportion is propellant)
cost: @ $10/kg: ~$1.5M, or $400/kg
(they used a pretty pricey propellant combo)
What most people think is that you need about 20 times your payload’s mass in propellant, which costs only a few dollars per pound. There are two problems: first, the payload is a small fraction of the rocket mass; and second no one is using inexpensive propellants. A better guesstimate is that you need about 50x the payload in propellant. So if you used inexpensive propellant, $50/kg or so is probably possible.
LOX/Kerosene costs almost nothing, but no one uses it to go to orbit. They probably should, though…
From the SpaceX web site:
F9 uses Lox Kero… 23050 lbs to LEO
735000lb glow.
Assume its 95% propellant (I know I can find the real number, I’m lazy)
Call it a$1 /lb for propellant. (I think Lox kero can be less than 0.5/lb)
That’s propellant cost of $30/lb to leo
With a 5X cost that’s 150/llb or $75 per kg.
Falcon 9 call it 333ton GLOW and 10 ton payload to LEO – 3% payload mass. Propellant mass is probably about 90% of GLOW.
Kerosene to LOX mass ratio is typically around 1:2.5, kerosene requirement is 8.5kg/kg of payload to LEO and LOX 21.5kg
Kerosene used to cost around $0.50/kg, maybe it is a bit higher than that now. LOX requires around one kWhr/kg to produce, in bulk it is something like $0.10-$0.20/kg. Say we assume $0.75/kg for kerosene and $0.15/kg for LOX I get just under $10 of propellant per kilogram to LEO.
LH2/LOX is also similar in cost, say 15% LH2, 70% LOX and 5% payload. LH2 I think used to cost around $4/kg for the shuttle, however this was perhaps not the cheapest supply. The equivalent energy of 1kg of hydrogen in natural gas (from which LH2 is derived) costs around $0.60, liquefaction need not be that expensive. The LH2 is proportionately much less and the payload fraction higher than for kerosene (due to greater ISP, though greater dry mass, the expensive part, does cut into this a bit). At $2.50/kg of LH2 I still get just under $10 of LH2/LOX propellant per kilogram of payload to LEO.
However existing rockets are so expensive that propellant cost is in the noise – propellant performance matters far more than propellant cost.
Use liquefied natural gas (~$0.25/kg) and LOX and you get down to around $5 of propellant per kilogram to LEO. LNG has slightly better ISP than kerosene and the fuel ratio is also much lower, like around 1:3.5. It does require slightly more tankage though due to its lower density. A number of CH4/LOX engines have been tested.
Reusability of rocket *engines* can be tested on the ground, on a test stand.
I am thinking that payload mass fractions in staged rockets is maybe 1:30, more like 1:100 if they ever get an SSTO going. The SSTO is probably “wasteful” of fuel, but it is supposed to cut costs by making the thing more like an airliner rather than a small fleet of aircraft.
Does this 3-5 times fuel cost work, like, for a trip on Southwest? On Amtrak? In your car?
I don’t know if we should go for full resuability. Space-flight is a very hostile environment that chews up things like heat shields and engines. I think a better architecture would focus on reusing the airframe, and being able to quickly swap out certain components that need to be checked or rebuilt, such as the engines and heat-shield. Don’t try to save everything, because you’ll end up tearing it apart and replacing it every flight anyway.
If, rather than having to tear the shuttle apart every time it needed to go up, you could simply unbolt and pull the main engines and bolt the new ones on (which you have prepared and ground tested, while the shuttle was flying), and rather than replace hundreds of individually crafted ceramic tiles every-time it flies, simply bolt on a one-piece phenolic expendable shield, then I think operational efficiency could be greatly improved.
Add to that the fact that drop-tanks aren’t all that expensive, and the only remaining lemon with the shuttle-type concept is the SRBs.
The shuttle isn’t inherently a worthless idea, it just needs to be done a whole lot better.
I would regard a two stage system where everything flies back and lands though as optimum.
Oh, and the double-preburning configuration of the SSMEs, designed to wring the last second of Isp out of LOX/LH2 probably isn’t worth it in terms of moving parts, operating temperature and pressure, and complexity, vs. simply carrying extra propellant.
3x is I think a good guide for air freight – a third capital, a third operations and maintenance, and a third fuel. 5x for passengers (they require a lot more looking after). I think cars work out roughly the same as do ships, trains may be a little off if tracks are under utilized.
This relationship is not coincidental. If fuel costs go up then more efficient and expensive designs are used, ships go slow to save fuel, etc., if fuel costs go down the reverse happens, people start using SUVs more, and so forth.
2% payload fraction is a figure I have often seen suggested for SSTO vehicles. But a lot has happened in recent decades and dry mass is something that is subject to incremental reduction.
There are also some out there tricks possible to greatly reduce dry mass, like self supporting inflatable external tanks (a bit like the inflatable Bigelow modules but for propellant). This can greatly reduce tank mass (perhaps less than 1% of GLOW), and make it far less of a constraint on vehicle design. Another trick is electric quadrotor air launching from within a say helium filled aeroshell – cheaper than using rocket, reduced aero losses, no tank insulation requirements, no engine altitude compensation, etc. There are a few more (including very light weight vertical landing solutions and higher T/W engines), but the difficult problem yet to find a really good solution is highly reusable lightweight reentry shielding.
Does this 3-5 times fuel cost work, like, for a trip on Southwest? On Amtrak? In your car?
I work for an airline, and our fuel costs are 40% of our operational costs. (Well, it was last time I checked, and admittedly fuel prices have fallen since then.)
I think its time to put a stake in the ground though, to cool down all the RLV proponents : Armadillo has been in the “business” for longer than a decade, and they have been moderately successful. They havent flown to 10 miles altitude yet ..
Extrapolate from here ..
They may very well reach 10 miles this year and 72 miles the year after that.
Armadillo has been in the “business” for longer than a decade, and they have been moderately successful. They havent flown to 10 miles altitude yet ..
Extrapolate from here ..
Dang, that baby’s been gestating for almost nine months now, and hardly anything has happened…
Extrapolate from here.
“I think its time to put a stake in the ground though, to cool down all the RLV proponents : Armadillo has been in the “business” for longer than a decade, and they have been moderately successful. They havent flown to 10 miles altitude yet.”
Most engineering professions have a mantra of sorts that suggests you can engineer anything with the following criteria:
Make the product…
…sooner
…cheaper
…reliable
.
NASA showed us that if you throw cost as a consideration (“waste anything but time”) that you can build just about anything once you set your mind to it.
In defense of John Carmack, he is throwing the concept out that it has to be ready to fly yesterday out the window, and using some of the design principles that have long been used in the computer software industry of rapid prototypes of small test concepts and dozens or hundreds of iterations of the design cycle to perfect the overall design.
I defy anybody to show a team that has more design experience and has fired more rockets with more types of fuel and designs than Armadillo…. with perhaps the exception of Wherner Von Braun and his team from Peenemünde. John Carmack has also done that at a fraction of the cost of other rocket development programs and largely out of his own pocket.
In defense of John Carmack, he is throwing the concept out that it has to be ready to fly yesterday out the window
In a sense he has taken it close to its extreme, in that something has to fly almost immediately, just not all the way to orbit. This is a very powerful strategy in software development, and it has shown itself useful for aerospace development too. Carmack was not the first to do it that way of course, neither for software nor for rockets. But he was going against a lot of conventional wisdom, showing it to be wrong in the process.
the difficult problem yet to find a really good solution is highly reusable lightweight reentry shielding.
Maybe that’s because everyone is thinking of using atmospheric drag for the bulk of the job of reentry delta vee. It’s easy and it works but it requires that reentry shielding. A powered descent stage would not necessarily need high-maintenance shielding at all. It isn’t something that makes sense if you have to launch all your propellant along with you every time, but if you have a propellant depot in orbit then powered descent stages become possible. Such a vehicle could bleed off most of the speed before reaching the upper atmosphere, thus requiring no more shielding than SpaceShipOne.