Economics of Cheap Space Travel

[MKSheppard]

[NOTE: I am not a rocket engineer, I just play one on television, and am using crude approximations here.]

NASA calculated in 1962 using the then-level of fatalities and commercial availability that in order for space travel to be as common as air travel, you needed:

LOW EARTH ORBIT:
50 x Flights a Year, each with 50 passengers, with one un-successful flight.

LUNAR TRAVEL
25 x Flights a Year, each with 10 passengers, with one un-successful flight.

Costs required to make this happen would have been:

$5/lb to LEO in 1962 Dollars. ($40/lb in 2011 Dollars)
$25/lb to Moon in 1962 Dollars ($190/lb in 2011 Dollars)

Later in mid 1970 at a speech at the University of Michigan, Lockheed's Max Hunter mentioned that if costs per pound went below $50/lb ($285 in 2011), then Texas Instruments would do manufacturing in space. He also mentioned that if costs went below $5/lb ($28.5/lb in 2011), then it would become feasible for Hilton to build space hotels.

Right now, it's pretty hard to figure out launch costs, since little if any actual costing data is released by the companies involved in the business (all proprietary trade information, natch).

But there are a few datapoints:

SpaceX in order to attract people to their SLVs has to release public data, and until 2004, Boeing/ULA was selling Delta IV commercially, and had to release public data.

I've collated the datapoints into this table using information found in:

Falcon 1 User's Guide
Falcon 9 User's Guide
Delta IV Payload Planners Guide - September 2007



Red Background is extrapolated data based off Falcon 9 performance reduction as you go higher.

To help you understand this a bit better, here's some explanation of the orbits -- Low Earth Orbit (LEO) is not monolithic.

200 km @ 28.7 degrees: Basically Gemini/Mercury manned flights. You stay up there for a couple weeks. Air drag is still a significant influence (relatively).

400 km @ 28.7 degrees: High altitude LEO. Stay there for years (with appropriate yearly reboosts). Not reachable by Russia directly due to poor geographic location.

400 km @ 51.6 degrees: High altitude LEO. Stay there for years (with appropriate yearly reboosts). Reachable by Russia. Currently used for ISS, previously used for Salyut, Mir and Skylab (50 deg).

Ok, onto our next point:

One of the most common refrains used as a way to lower launch costs is to build the vehicles en-masse, and lower costs via mass production.

Brief Description of Production Cost Curves by Me

In August 1965 at the Conference on the Exploration of Mars and Venus at the Virginia Polytechnic Institute they put together costs for a 1975 Manned Venus Flyby and a 1978 Manned Mars Flyby using the following assumptions:

Saturn V: 90 percent learning curve, 62 prior units built; $70 million unit cost.
Saturn IB: 90 percent learning curve, 52 prior units built, $22 million unit cost.
Apollo CSM: 95 percent learning curve, 70 prior units built, $72 million unit cost.

This gives us an idea of what kind of PCC/LCC we can realistically expect from SPACE [tm].

Here are the final computed prices for Falcon 1e/9/Delta IVM/IVH using a PCC/LCC of 90%:



As you can see, mass production can reduce the cost of sending something into LEO by up to 50%. But it still remains a steep wall to be climbed.

I honestly don't think that mass production of space launch vehicles past the range of 50-100 launchers to a Block will ever happen.

For one, by the time you hit that range of production, you would be ready to move onto the next Block, requiring a lot of retooling.

Finally, each individual flight would see little increments of improvement creep in until rocket #56 off the production line has so many differences that it can't credibly be compared to rocket #3.

If you wanted to lower launch costs to maybe $700 a pound for Falcon 9, you would have to produce 8,200 of them to the same basic design with a LCC/PCC of 90%.

That clearly isn't happening. Not unless we suddenly discover we have to build a moonbase in five years to fight aliens on the moon to save the earth in Space War I.

So let's get into costing this per individual seat.

SpaceX's Dragon capsule will likely weigh about 18,000 pounds when it's in the 7 person configuration. That's about $30 million dollars roughly to launch it with the 50th Falcon 9 off the production line.

Figure a basic cost of $30 million for a Dragon Capsule, and that it flies for ten flights before retirement, and you have about $5 million basic cost per flight for the capsule (recovery, refurbishment, and allocating for replacement added into basic cost).

With six paying passengers, that comes out to about $5.83 million per passenger. That's almost nine times less than what Russia is charging for a Soyuz flight right now as of 2011 ($52~ million USD).

Even if SpaceX manages to develop a larger 12-person capsule, the seat prices only drop to $3.18 million. ($30m basic launch cost + $5m capsule amortized cost)

Clearly, we are not going to get the price down to what the average person is willing to pay with expendable launch systems.

If we took something conceptually similar to Max Faget's DC-3 Concept:



Essentially, it's a vertically launched, horizontally landing fully reusable system. There's a very good reason for vertical launch. The heaviest plane in the world right now is the An-225 with a 650 tonne MTOW. A lot of TSTO concepts are in the range of 2,000+ tonnes.

That's a lot of weight the landing gear and structure have to take when it taxis to takeoff. By contrast, launching like a rocket means you can size the landing gear for it's empty weight, saving a lot of weight and money.

The first stage (booster), would have weighed 1,600 tonnes fully loaded and 273 tonnes dry, for a propellant loading of 1,327 tonnes. It would have been powered by no less than 10-12 SSMEs, and landed with the assistance of four jet engines at a conventional runway.

The orbiter would have weighed 395 tonnes fully loaded and 121.5 tonnes dry, for a propellant loading of 273.5 tonnes. Power would have been through two SSMEs. It would have had four jet engines for conventional runway landing. Crossrange would have been 200+ miles. By contrast Orion will have about 100 miles, and the Shuttle 1,200~ miles.

Crew would have been 2, with ten passengers somewhere in there.

The payload bay of the orbiter would have been 8 x 30 feet (versus Shuttle's 15 x 60 ft bay); and the complete system would have carried 15,000 lbs into a 498 km high orbit with an inclination of 55 degrees.

There were several reasons why the DC-3 concept died.

• The USAF demanded 1,200 mile crossrange, so orbiters on a polar orbit could land at Vandenberg AFB when they were brought into the program to "spread costs". DC-3 only had 200~ mile CR.
• Fully-Reusable was expensive. They were looking at easily $9.9 to $7.7 billion for R&D for that, compared to $6.4 billion for a partially reusable. It went even lower to like $5B cost for the shuttle configuration we know and sort of like.
• The DC-3 used titanium. At the time, only a few closed machine shops working on classified military programs had experience in making stuff on a big scale (SR-71). If it was aluminium, the whole aerospace industry could work on STS. This was why USAF preferred aluminum.
• Von Braun and his MSFC team were dubious of the viability at present (1970) technological levels of a large cryogenic tank system that could repeatedly be chillled/warmed and pressed for a semi-expendable system.

But back to costing a DC-3 style concept.

Assuming the orbiter/booster each cost $1.25 billion each and fly 100 missions before replacement, that's about $30 million cost per flight for the hardware. Maybe $50 million every 25th flight as it gets overhauled.

Propellant costs in a fully reusable system would quickly dominate:

LOX: $198.41/tonne (2001)
RP-1: $1,235/tonne (2010)
RP-1: $2,435/tonne (If gas prices hit $7.50)
LH2: $5,490/tonne (2010)
PBAN: $10,846/tonne (Derived from $2.50/lb in 1985 for PBAN with 4mlb production/yr).

A full tank-up of a DC-3 would cost about $8.7 million.

(Yes, I know you would have a lot of LOX, bringing the price down; but lets assume that it's 100% LH2 to provide a margin of cost safety.)

We'll assume for simplicity that total launch costs are $45m to be on the really safe side budgetwise.

An 8 x 15 ft cargo bay can accomodate about five rows of airliner seats in a 2+1 configuration, giving you a 15 passenger module. Add that to the 10 guys up front in the orbiter, and you have 25 passengers paying about $1.8 million a seat.

If you went with a non-traditional seating system, in which people stood up (remember that during launch, they will be on their backs, and while they wait for re-entry, they'll be in mostly zero gee); you could get a lot more people in; maybe about 45. That drives prices down to $1m.

If you went for a passenger-only shuttle, eliminating the cargo bay and making one long passenger compartment, you could get even more people in. Probably about 65. Prices go down to $690k.

At this point fundamentally, the best we can get things down to with a rocket based TSTO system is about $150-200k per seat. That's if we assume a near-totally depreciated asset in which the only reoccurring cost is filling up the fuel tank(s) and the parts installed will last until retirement.

That's more than enough to get a space-based economy going. The US military already pays the equivalent of about $100k for a soldier's gear these days, and companies would be willing to pay $300k to send someone up into orbit to spend a year or two working at a space based research laboratory.

But it's still far below what the average person is willing to pay.

Some historical comparisons of airline tickets:

1940: Delta Airlines Atlanta-Fort Worth on DC-3: $38.50 ($608.60 in FY11)
1960s: Cleveland-Washington DC: $75 ($526.93 in FY11)
2011: Dulles to Toyko with 1.5 months' warning: $1,000~

For the common man, the best seat price he could realistically afford would be about $4 grand.

Alienware makes money selling $3,999 desktops to the performance market. I'd imagine a flight to space would be in even more demand than street cred at gaming conventions.

When LH2 scramjets with an ISP of about 1,250 become practical to about Mach 6 or 7, it means that our first stage booster that weighs 1,650 tonnes suddenly shrinks to only about 350 tonnes.

Likewise, by then large LH2 engines with an ISP of about 460 will be around, meaning that the orbiter can shrink a little.

All this would end up with a system capable of offering seats to orbit for about $70-80k when totally depreciated.

Eventually as scramjets become more advanced, each one going higher and faster than the previous generation; we can push staging faster and higher.

Instead of having to stage at Mach 6, we can wait to stage until Mach 8+ or more, and spend more time running on scramjet power before we switch to rocket power for the last push to orbit.

At some point, we'll know enough to integrate our scramjets with a rocket engine to create a RBCC (rocket based combined cycle) system that weighs less than having both engine types on the launcher, making some sort of SSTO feasible, saving a lot of money on operating costs.

By this point, there would be enough of a market for aerospace companies to seriously propose building large spaceplanes that could carry upwards of 300 or more people into orbit, allowing for space travel to undergo the same revolution that happened when 747s and other widebodies began to appear from 1967 onwards.

The biggest cost would be involved in getting out of earth's gravity well. Once you were docked at the Alan B. Shepard Orbital Hub, you would only spend a couple hundred more to go to Armstrong City on the moon.

-----------------

One last thing before I sign off:



At the flight cost of $25 million that STS was supposed to have during the planning phase, these 72 peeps would be paying only $347k.

Even if we went with the current $500m per STS launch of today, the cost per passenger would be $6.9 million, beating out SpaceX and the Russians.

[Agent Sorchus]

Nicely done, however it would appear that either your 6.9 million dollars per person for the STS is wrong, the Data for the hypothetical Falcon 9 dragon capsule's 5.83 million per person, or your conclusion that the shuttle is cheaper than the Dragon is wrong. Or are you saying that it would be more profitable since the shuttle's have large lifespans?

[MKSheppard]

Nicely done, however it would appear that either your 6.9 million dollars per person for the STS is wrong

It's been stated many times that the reoccuring cost of launching the Shuttle is about $450 million dollars. Up that to $500m to be a bit on the safe side cost-wise.

With the shuttle stack as it is currently now, you could have five passengers plus two crew. That means that the five passengers would have to pay $100 million each to make up the launch costs.

But if you put a pressurized cargo module like the one shown in that GRAPH; then with 72 peeps to split the $500m cost, each individual seat would cost only $6.9m.

the Data for the hypothetical Falcon 9 dragon capsule's 5.83 million per person

I've stated my basic assumptions -- SpaceX has not yet stated how much it actually has cost them to make the capsule, so I've assumed a $30m basic cost for each one fresh off the production line.

SpaceX has also stated they plan to refly each capsule. There were studies for this to be done in the 1960s for Apollo Command Modules up to ten times. This lets you split the cost of the capsule over more than one mission.

$30m / 10 flights = $3 million cost per flight. I upped that to $5 million to be on the safe side, since you have to pay for a ship to recover the capsule at sea, transport costs to bring it to the factory, refurbishment/replacement labor, etc.

The actual launch cost is entirely dependent on the Falcon 9's cost.

Space X is quoting a $56 million price right now for F9. With six paying passengers on a Dragon, and my assumption of $5 million per mission cost for a Dragon, it would cost each person $10.1 million to go to the ISS.

But if you were going into orbit on the 25th Falcon 9; you would only pay $6.6 million a seat.

Likewise, by the 50th Falcon 9; the cost would be down to $6~ million.

Of course, the X factor in this is how well Space X can mass produce the Dragon and refurbish them between flights. That would also affect the per-seat price a lot.

[Agent Sorchus]

I've stated my basic assumptions -- SpaceX has not yet stated how much it actually has cost them to make the capsule, so I've assumed a $30m basic cost for each one fresh off the production line.

SpaceX has also stated they plan to refly each capsule. There were studies for this to be done in the 1960s for Apollo Command Modules up to ten times. This lets you split the cost of the capsule over more than one mission.

$30m / 10 flights = $3 million cost per flight. I upped that to $5 million to be on the safe side, since you have to pay for a ship to recover the capsule at sea, transport costs to bring it to the factory, refurbishment/replacement labor, etc.

The actual launch cost is entirely dependent on the Falcon 9's cost.

Space X is quoting a $56 million price right now for F9. With six paying passengers on a Dragon, and my assumption of $5 million per mission cost for a Dragon, it would cost each person $10.1 million to go to the ISS.

But if you were going into orbit on the 25th Falcon 9; you would only pay $6.6 million a seat.

Likewise, by the 50th Falcon 9; the cost would be down to $6~ million.

Of course, the X factor in this is how well Space X can mass produce the Dragon and refurbish them between flights. That would also affect the per-seat price a lot.

I see now what you are talking about now. It didn't occur to me to add the capsule and launch vehicle together to get a per person cost of both. Of course you could have been more clear originally, but it is all good.

[MKSheppard]

A lot of this is sort of baseline research to help me in defining the parameters of space travel in the near future for a SF universe I'm working on.

Everyone likes to romanticize the Golden Age of Aviation, which closed with the Boeing Clipper Model 314s flying across the Pacific to exotic destinations.

What is usually left out of the rose-tinted rememberances is how much that cost.

The San Francisco - Hong Kong route cost $760 one way and $1,300~ round trip in 1940. That's equvalent to $12k and $20.5k in today's money.

Likewise, a ticket going from Germany to Lakehurst on the Hindenberg was $400 USD in 1936. ($6,368 in 2011).

Going further back to the Golden Age of Passenger Liners...

When Titanic sailed, the prices were crudely:

One of two 1st Class parlour suites on B deck: $4,350 USD ($97,240~ in 2011).

Min. 1st Class Fare -- £26/$130 ($2,906 in 2011)
Min. 2nd Class Fare -- £10/$50 ($1,117 in 2011)
Min. 3rd Class Fare -- £7/$35 ($782 in 2011)

[someone_else]

One of the most common refrains used as a way to lower launch costs is to build the vehicles en-masse, and lower costs via mass production.

Another very common refrain about lowering launch costs is "build space infrastructure". That usually means some kind of orbital fuel depot (either refueled from the ground or from the Moon) and using space tugs to catch payloads sent to very low orbits to have cheaper boosters.
Or use moment-exchange tethers to do the same without fuel depots and space tugs.
It's like building roads or railroads IN SPAAACE!!! Hefty price to make cheaper the travel afterward.
Look at This

Also, you have not talked about the launch pad's and ground operation cost per year (apart from refurbishment of reusable stuff for DC-3 things). That's another point usually trumpeted as good reason for relatively smaller rockets launched more frequently being cheaper than a few bigger rockets on the long run.

If you wanted to lower launch costs to maybe $700 a pound for Falcon 9, you would have to produce 8,200 of them to the same basic design with a LCC/PCC of 90%.

There will probably be some kind of lower limit before employing child-slave-labour or outsourcing everything to countries where child-slave-labour is the norm.

Everyone likes to romanticize the Golden Age of Aviation, which closed with the Boeing Clipper Model 314s flying across the Pacific to exotic destinations.
What is usually left out of the rose-tinted rememberances is how much that cost.

Cost-benefit should be what you're looking at. Going in America (like Titanic) was a very sought-after thing by rich and poor alike, worth even high prices. Technically, most illegal immigrants arriving in Italy pay around a thousand euros to the people smugglers for their ride to get here.

Going in space where there is little besides your capsule doesn't have that kind of appeal, unless you have cool stuff[TM] to do in space.

[someone_else]

This blog contains links to better papers than the one I linked above.

[Samuel]

Shep, do your figures include profit? Because it looks like they are only the costs of space travel and companies are going to have to charge more than that to be in the business.

[MKSheppard]

Shep, do your figures include profit?

No. But I can do some more research to try and find profit margins for common industries:

Here are some five year averages and a description of each from Investopedia.com:

Gross Profit Margin

The gross profit margin - or gross margin for short - tells us the profit a company makes on its cost of sales, or cost of goods sold. In other words, it indicates how efficiently management uses labor and supplies in the production process.

Gross Profit Margin = (Sales - Cost of Goods Sold) / Sales

Let's say a company has $1 million in sales and the cost of its labor and materials amounts to $600,000. Its gross margin rate would be 40% ($1,000,000 - $600,000/$1,000,000).

United Parcel Service: 33.4%
PEPCO (my power company): 28.01%
Southwest Airlines: 24.72%
Massey Energy: 19.45%
Boeing: 18.62%
Lockheed Martin: 9.54%

Average: 22.3%

Net Profit Margin

Net profit margins are those generated from all phases of a business, including taxes. In other words, this ratio compares net income with sales. It comes as close as possible to summing-up in a single figure how effectively managers run the business:

Net Profit Margins = Net Profits after Taxes / Sales

If a company generates after-tax earnings of $100,000 on its $1 million of sales, then its net margin amounts to 10%.

Lockheed Martin: 6.92%
United Parcel Service: 5.53%
Southwest Airlines: 4.79%
Boeing: 4.14%
PEPCO (my power company): 2.48%
Massey Energy: 1.28%

Average: 4.2%

So it appears on average that charging 20% over your base cost will net you about 4% after-everything-else profit.

[MKSheppard]

Right now here are some random prices circa 2010:

Platinum: $55.8 million/tonne
Gold: $38.6 million/tonne
Silver: $587,394/tonne
Tin: $21,300/tonne
Nickel: $20,800/tonne
Copper: $7,500/tonne
Lead: $2,200/tonne
Aluminum: $2,182/tonne
Zinc: $2,150/tonne
Iron Ore: $143/tonne
Uranium: $42/tonne

Platinum is very much the hypothetical space beaver pelt. It's something in high demand, yet scarce -- only a couple hundred tonnes are mined each year.

Estimates have been done on the amount of platinum a typical asteroid would have using meteorite samples. They've sampled meteorites and found percentages from 0.55% to 1.5%.

Obviously, in the real world [tm] asteroidal percentages will be much lower. A good catchall might be 0.05%, giving you about 100 tonnes of stuff from a 62~m diameter asteroid massing 214,000 tonnes.

The price of the platinum itself would be $5.5 billion; giving you a lot of money to play with regarding setting up your space mining company.

You'd need to keep the total costs below about $4.5 billion if we go by the 20% profit margin needed to break even on all costs.

The biggest problems I see are:

• There's only so many people who can mine the stuff and still be viable. In 2007 only 184 tonnes of platinium were mined on Earth, making it so expensive that this undertaking is actually feasible. What happens when the market gets at least 100 tonnes of asteroidal platinium dumped on it every other year? You'd see quite a bit of asteroid mining companies go bankrupt or be forced to sell off their extra-terrestial assets when the spot price of platinium drops, ruining their business plan.
• Transport costs. You need to find some way of moving at least 1,600 tonnes of material to an asteroid (Tunnel Boring Machine, nuclear power plant to run all this, automated fab factory, electric arc furnace). A lot of stuff can be built on site, like parts for the tunnel boring machine to reduce your mass costs. To make this work, transport costs to the asteroid belt need to be on the order of less than $400/lb -- or $1.4 billion for 1,600 tonnes, leaving you enough to buy the equipment, operate it, and pay for the finished material to be transported back to earth.

Practically, I don't see asteroid mining (or even lunar mining) becoming possible until we have developed nearly completely automated mining systems here on earth. If you don't have to keep a group of people alive in the asteroid belt, you can save on a lot of complexity in your plans and transport costs.

[PeZook]

Shep, couldn't asteroid mining become profitable when it's needed for space-based manufacturing?

Otherwise you'd need to ship it from Earth.

In this sense, space tourism might lead to space industry: space tourism means people go up to have vacations/live in space. Facilities for them to do that get bigger to provide better services, but need maintenance. Companies start looking for cost reductions after a certain level of tourism is reached, start mining to make at least some spare parts on site, bam, space industry.

Of course it still means we need space travel cheap enough to get more tourists than the occasional millionaire every five years or so.

[MKSheppard]

Shep, couldn't asteroid mining become profitable when it's needed for space-based manufacturing?

Otherwise you'd need to ship it from Earth.

You would need a nearly complete automated factory to make that feasible. Something that could input raw ore, smash it up, smelt it, and then play with the final material to process it into whatever forms are needed (I beams, etc etc).

We're almost there in a sense with the recent interest in 3D material printers. But right now it's still cheaper to have humans around in a factory or mine in many places than a robot which needs expensive programming time to work.

But I can see automated factories/mines taking off around 2030-40 due to the graying of the world's population, along with the increasing demands of union workers in both developing and developed countries.

[MKSheppard]

Right now this is what we can do regarding launch vehicles:

Link to C3 Performance

It takes like the graph notes -- 40 km2s-2 of C3 to put you on a trajectory to cross with Ceres in about 1.3 years.

There are other ways you can get there with less energy, but they need a lot of complicated gravity slingshots.

I have the numbers from that table (it's actually scores of datapoints) on what masses you can inject to that specific energy:

Delta IVH: 11,400 lbs (5.17 tonnes)
Ares V: 58,600 lbs (26.5 tonnes)
Ares V/Centaur: 71,000 lbs (32.2 tonnes)

These masses are OK if you just want to make a quick flyby and zip past the place ala Voyager 1/2.

But if you want to actually orbit the target, you need to generate a delta vee of about 4.6-5 km/sec as part of a Ceres Orbit Insertion (COI) burn.

This is due to Ceres having little if any gravity and zero atmosphere, so your spacecraft has to do all the work, instead of letting the target's gravity or atmosphere do it for you, like Mars.

So if you're launching straight to the target without a parking orbit; these are your options for Ares V/Centaur:

Option A: Chemical insertion

You use a 22.25 tonne C-OIS (chemical orbital insertion stage) with 20 tonnes of propellant and an ISP of 450 to brake a 5.7 tonne payload into Ceres orbit.

Cost is about $51,740/lb with the following numbers:

$500m for Ares V
$50m for Centaur Upper Stage
$100m for the C-OIS.

Option B: Nuclear insertion

You use a 17 tonne N-OIS (nuclear orbital insertion stage) with 12 tonnes of propellant and an ISP of 950 to brake a 10.25 tonne payload into Ceres orbit.

Cost is about $33,199/lb with the following numbers:

$500m for Ares V
$50m for Centaur Upper Stage
$200m for the N-OIS

Option C: VASMIR insertion

You use a 15.5 tonne V-OIS (VASMIR Orbital Insertion Stage) with 5.5 tonnes of propellant and an ISP of about 3,000 to brake a 16.5 tonne payload into Ceres Orbit.

Cost is about $20,623/lb with the following numbers:

$500m for Ares V
$50m for Centaur Upper Stage
$200m for the V-OIS (I'm assuming you need some sort of small nuclear reactor to power it that far from the sun -- that needs to be flight qualified, which can be done cheaper than a nuclear engine.)

---------------

Now; eventually, I can foresee a unified interplanetary transfer stage being developed that combines VASMIR with something like P&W's proposed Triton NTR engine, which can provide large amounts of electrical power during coast phases, along with providing high thrust and high ISP (900+) during injection manouvers in gravity wells.

Basically, the interplanetary transfer stage is switched to classical NTR mode to provide the 5 km/sec of delta vee + high thrust needed to accelerate something in a parking orbit around Earth to a 40 km2s-2 of C3.

Once it's separated from the cargo, it can then switch to nuclear reactor mode to power several VASMIR engines which slowly push the interplanetary transfer stage onto a trajectory back to an earth or lunar orbit, where it can be refuelled again.

You can even try to allocate extra fuel to the VASMIR powered stage that is pushed to escape velocity, so that once it's done braking it's payload around Ceres, it can very slowly return to earth, allowing you to have a fully reusable transportation system.

The big problem with reusable everything is that costs to LEO need to become cheap enough to refuel it rather than expending the stage; to make up for the difficulties involved in making something last longer than it needs to.

Some crude costs all baselined around the amount of fuel you needed to impart the following delta-v onto 10 tonnes of payload:

Trans-Lunar Injection: 10,040 fps
Lunar Orbit Insertion: 2,730 fps
Trans-Earth Injection: 3,190 fps
Earth Orbit Insertion: 10,050 fps
Midcourse Corrections: 50 fps
Performance Reserve (0.75% of total): 195 fps

Total: 26,255 fps (8,002 m/sec)

(Delta V came from Nuclear Flight System Definition Study: Phase III Final Report -- April 1971)

LH2 Conv. (420 ISP): $78/lb <-- J-2
LH2 Conv. (460 ISP): $178/lb <-- Latest RL-10
LH2 Nuke. (825 ISP): $1,016/lb <-- NERVA
LH2 Nuke. (925 ISP): $1,396/lb <-- Timberwind Lower End
LH2 Nuke. (1000 ISP): $1,680/lb <-- Timberwind Upper End

VASMIR/Argon (3000 ISP): $7,693/lb <--VASMIR range

----

Assumptions made were that conventional stages were $100 million, nuclear stages $200, and VASMIR $250 million (reactor to power it).

It was also assumed further that
the loaded/dry ratios of the stages were:

Conventional: 10 (S-IVB's ratio essentially)
Nuclear: 5 (RIFT's S-N was this)
VASMIR: 2.5 (all that mass for radiators/reactor, etc).

Another assumption was made that 2.5% of all the fuel would remain un-used due to various factors.

I then divided the mass of the fuel needed to fill the stage by the cost of the stage minus 30% to get the $/lb to orbit needed to make it economical.

[someone_else]

Regarding the cost of mining stuff in space, have you considered rocket fuel? The satellite industry needs to send satellites up to GEO at weird inclinations (delta-vee intensive), and could use fuel mined in space to do it. That's somewhat easier to do than mining platinum, and since you're producing your own fuel....

I mean, a proton can lift 22 tons up to LEO, but to get stuff in GEO it must waste 15+ tons of payload for the vehicle able to move up to GEO the satellite. That vehicle is going to be at least 12 tons of fuel. If a proton launch costs around 85 millions, those 12 tons of fuel are worth 3.9 $ million per ton.

A falcon 9 heavy can lift 32 tons to LEO, but only around 9-10 tons to GEO, so that's again another 15+ tons at 3 millions per ton.

Ok, it's not as much as platinum or gold, but you don't have to make reentry pods for them, and it's much easier to manufacture. And isn't going to saturate the market like with platinum/gold/stuff, but possibly decrease the price of reaching space, allowing more customers to get in space paying your services. (Profit$$$$$$$$)

If you offer the LEO-to-GEO boosting services too (i.e. you use your own space tug instead of having them launch a new vehicle each time), that may pay off even more (less mass must be lofted by the customer).

This would also be juicy for NASA and other science-oriented agencies.

[PeZook]

Actually rocket fuel costs next to nothing compared to the rest of the launch ; Most launch costs are in the infrastructure, engines and trained personnell to run it all.

That's actually part of the reason why manned launches are more expensive than unmanned ones: they incorporate all sorts of extra checks, countdown holds etc that are manpower-intensive.

[someone_else]

Actually rocket fuel costs next to nothing compared to the rest of the launch. Most launch costs are in the infrastructure, engines and trained personnell to run it all.

Don't know if you understood my point.
Fuel costs here on Earth = next to nothing
Fuel cost in LEO when needed to put payloads to GEO and beyond = around 3 millions per ton
It's another payload (necessary to bring its main payload to GEO). If you use this payload mass to carry fuel, then you cannot say, make a multi-payload launch to split the launch costs on multiple customers (or launch more stuff for the same customer, the same).

This means for both launchers in my post above, you can deliver them 30-60 millions worth of LEO rocket fuel from moon or asteroid ISRU.

Can you keep the fuel price competitive? Don't know. But sounds easier to do than setting up a platinum mining base on Ceres for sure.

Considering that a falcon 9 heavy costs 95 millions, shaving off around half the cost of such LEO-to-GEO fuel (around 30 millions if you want to send 9-10 tons in GEO) isn't so bad for a customer that has stuff to deliver in GEO.
If the rocket can receive enough fuel in orbit to bring its whole LEO payload to GEO (yes, it needs more fuel from Earth, but that's cheap as we all know so it isn't an issue), you can make a multi-payload launch, splitting launch costs between customers.

That is, with a Falcon 9 heavy you could bring to GEO three times its current GEO payload, without building, checking and launching 3 Falcon 9 heavies.
Do you see the profit$$$ now?

[MKSheppard]

The satellite industry needs to send satellites up to GEO at weird inclinations (delta-vee intensive), and could use fuel mined in space to do it. That's somewhat easier to do than mining platinum, and since you're producing your own fuel....

No actually, that's where you use a solar-electric tug.

Take one VASMIR engine. Connect it to a really big solar array.

Use VASMIR tug to push satellites from LEO into GEO over weeks. Best part is because VASMIR is so efficient -- 3000 to 5000 ISP, you can easily keep the tug supplied with fresh fuel through even a a Falcon 1e.

[someone_else]

No actually, that's where you use a solar-electric tug.
Take one VASMIR engine. Connect it to a really big solar array.

VASIMRs are good for SCIENCE!!!! payloads. Being only 70% efficient and somewhat power-hungry, there is the issue of radiator weight. Its thrust sucks balls so it will take friggin ages to get in the right orbit. That's not a problem for a decade-long space mission, but for commercial use it's a show-stopper.

My spreadsheet (made with equations of Atomic Rockets and some eyeballed structural mass requirements, it's not anywhere near Gospel, feel free to point out where I'm wrong) tells me that one of such engines with 30 km/s delta-v at the exhaust velocity stated in the link above (50 km/s, that is VASIMR's "low gear", so to speak, big thrust, more fuel consumption. Can reach better fuel efficiency at the expense of its already-sucky thrust) and with a payload of around 14 tons would weight 28 tons empty and 50 tons with fuel.
The critter needs 200 kW of power, provided by a 6.7 ton solar array (already part of the total mass above) with panels producing 0.03 kw per kg (30 kg per kw).
Being VASIMR only 70% efficient, it must reject 60 kW of waste heat from the engine.
Atomic rockets page on radiators is too limited to be useful.
If you use the same radiators as the ISS, an HRS panel weighting 2470 pounds (1.12 metric tons) can reject 11.8 kW of heat. It would need around 6 or so such panels. Not a killer (now the payload decreases to 7 tons), but it's somewhat unwieldy, considering the solar panel's surface too.
With heat pipe radiators it would be way better, but I don't know if they got past the paper-only stage.

Anyway, its acceleration would be 0.0001 m/s, and being the delta-v needed to go from LEO to GEO 3000-4000 m/s, the time it will take will be between a year and a year and 97 days. Not that practical.

With ten such engines and a craft weighting 200 tons (110 without fuel) with 18 tons of payload and an impressive 66,7 ton solar panel, the acceleration doubles (an extremely powerful 0.00025 m/s), allowing the same trip in 6-8 months. Slightly more practical
But now there are 600 kW of waste heat that require 50 HRS panels, weighting 57 tons. You may get away with less than a ton if using some of the (speculative) heat pipes in the (old) papers I read (like this).

Next step is a craft with one hundred more engines to get only three times the original acceleration (0,00031 m/s), weighting 1600 tons (934 tons without fuel) and a ridiculous 666 tons solar panel. 15 tons of payload, 6 MW of waste heat to dump somewhere, somehow. LEO-GEO travel times are now a whopping 3-5 months. Too massive to be cost-effective with current launch costs.

On top of that, all three crafts are a self-propelled space debris hazards (lots of surface that can be hit by small-ish space debris generating more of such debris).

Don't know about you, but I'd still place my chips on chemical rockets and fuel from lunar ISRU.

[someone_else]

Damn, missed the post edit time .

I realized that 30 km/s of delta-v is a little too much. 10 km/s is more than enough for a round-trip with enough fuel for any contingency.
The text below is a correction to the above figures.

First craft: payload of around 7 tons would weight 16 tons empty and 20 tons with fuel. Acceleration would be 0.0002 m/s, need 6-8 months to go and slightly less to come back. Not that practical imho.

Second craft, ten such engines and a craft weighting 120 tons (100 without fuel) with 18 tons of payload and an impressive 66,7 ton solar panel, the acceleration doubles (an extremely powerful 0.00042 m/s), 3-4 months to go and slightly less to come back. Slightly more practical, although its mass starts to become a problem.

Third craft, one hundred engines to get slightly less than three times the original acceleration (0,0005 m/s), weighting 950 tons with a ridiculous 666 tons solar panel. 7 tons of payload, 6 MW of waste heat to dump somewhere, somehow. LEO-GEO travel times are now 1-2 months (plus another 1-2 months to come back in LEO). Too massive to be cost-effective with current launch costs.

[someone_else]

Fuel cost in LEO when needed to put payloads to GEO and beyond = around 3 millions per ton

This is for example where most fuel depot believers think you can make money from in the short term by launching fuel depots with other (possibly cheaper) rockets. For example, if you launch a fully loaded Falcon 9 heavy to LEO that must bring everything to GEO, now it only needs around 20 tons of fuel when it is in LEO.
Another falcon 9 heavy bringing to LEO 30 tons of fuel costs 95 $ Millions, and leaves in orbit 5-7 tons of fuel in a fuel depot for any other needing it.

But those 20 tons of fuel allow you to bring 30 fucking tons to GEO, which is three times a falcon 9 heavy GEO capacity, at only 2/3 the cost since you launched only two such rockets.

And every few launches or so, you have a "free ride" to GEO using the fuel left in orbit from other rockets (if you transfer it in a suitable LEO fuel depot for medium-ish term storage).

You can do more or less the same trick with all rockets. You launch two of the same kind, and the second has the fuel to bring a full LEO payload loaded on the first up to GEO at only 2/3ds of the price.

VASIMRs are good for SCIENCE!!!! payloads.

To get around the Earth Sphere (a fictional name for everything in Earth Orbit, Lagrange points and moon), rockets are more than enough delta-v wise, their thrust is awesome and their complexity is pretty low compared to solar-electric or even nuke-electric, making development somewhat less painful.

Most proposals I've read around talk about using EML1 as a fuel depot, that is placing a fuel depot in the Earth-moon lagrange point 1.
That's around a "middle point" delta-v wise, that would allow rockets to take off from the moon, bring it fuel as their payload, and get down again (now half-empty) without staging. And with fuel depots in LEO too (either from the Earth or from slowly aerobraking full EML1 fuel depots), rockets can get to LEO, refuel, head for EML1, refuel, go wherever they want in GEO (and back to EML1) or to the Moon (and back to EML1).

[aussiemuscle308]

The San Francisco - Hong Kong route cost $760 one way and $1,300~ round trip in 1940. That's equvalent to $12k and $20.5k in today's money.

Interesting since we see Indiana Jones jumping on planes to fly around the world (set in the late 30s). Lucky those Universities and Museums have plenty of cash.