some of the following is work in progress ...


Inverted Aerobraking

An Approach to Efficiently use
Extraterrestrial Fuel to
Launch Space Ships from Earth

Pulsed Inverted Aerobraking animation (schematic)   
Pulsed Inverted Aerobraking

Launching a space ship to space is expensive because of the exponential growth of the size of chemical rockets. Bringing up rocket fuel takes rocket fuel. Staging is expensive. But more powerful chemical fuels are impossible. Atomic fuel is environmentally inacceptable. Beaming additional energy from earth to the space ship for higher specific impulse is still an unsolved problem. Using tethers to go to space needs incredible large structures and materials of extraordinary strength. Those materials can not be created yet, and maintaining those large structures will be expensive.

There is another way to break this exponential growth of difficulty to go to space: bridge the gap to space from two sides. Of course I am not talking about building a bridge made of stone, steel, or any other material. Think about it more like setting up an economic supply chain for space travel and travelling up along this supply chain. A key element for this is to deliver part of the fuel and energy needed to go to space from the space side. This needs some space based infrastructure to mine, refine and ship fuel. NASA studies confirm that bringing fuel from space to low earth orbit could be cheaper than launching it traditionally. Using such fuel to create cheaper access to low Earth orbit will make maintaining and expanding this infrastructure even more economic.

The approach proposed here offers the opportunity to achieve specific impulse several times larger than chemical fuels, and therefore bridge a large part of the way to space. And it even gets better: this highly energetic fuel gets most of its energy from falling towards earth, so basically for free.

To give you a quick idea, here is a schematic illustration for an example how inverted aerobraking might be used:

Schematic example for application of Inverted Aerobraking

The following text was written for experts. For better comprehensibility I added an informal description and some (sometimes redundant) comments at the end. Some preliminary results and working notes can be found here.

Problem Description:

Launching a space ship to low Earth orbit (LEO) is expensive, because of the high energy needed to go to LEO. This energy is needed for two purposes:

a) go to orbital height
b) accelerate to orbital speed

Somewhat surprisingly part b) turns out to be the far more expensive part. It is expensive in terms of energy, because if we compare height energy to speed energy, height energy for a 200 km high orbit is only 2 MJ/kg while kinetic energy for such an orbit is 28 to 35 MJ/kg. Of course part a) is a bit more expensive, because additional energy is needed to fight air resistance until the space ship is out of atmosphere and to fight Earths gravity until orbital speed is reached. The last two components depend on the flight trajectory and are more complex to calculate. On the other hand, part b) becomes exponentially more expensive in terms of fuel, because the fuel has to be lifted to orbital height first, using up more fuel.

An example to validate the above conclusion seems to be appropriate here. The Saturn V rocket was able to launch a total mass of 130 metric tons (including ships mass) to LEO, using up two rocket stages and some fuel of the third stage on the way. But the first stage alone would have been more than enough to go to orbital height: when the fuel of the first stage is used up, the Saturn V is at a height of 61 km, i.e. mostly out of atmosphere, and at a speed of 2.7 km/s. Without any further thrust, and assuming this speed vector was created vertically, the space ship would go up for another 270 seconds, i.e. 4.5 minutes, reaching a height of about 420 km, having a total mass of 770 metric tons. I.e. six times more mass could be transported to space than can be transported to Earth orbit. Click here for more details.

Launching payload into Earth orbit is expensive. Launching fuel from Earth is as expensive. Studies have concluded that bringing fuel from extraterrestrial sources like the moon, asteroids or comets will become a cheaper way to supply fuel to Earth orbit in the future, as soon as the necessary infrastructure is installed. This fuel can be used to accelerate space ships from Earth orbit to outer space, but—so far—is no help in launching space ships from Earth to LEO.

And a considerable part of the problem of bringing fuel from outer space to LEO is to slow it down, because coming down from the higher potential of the moon or asteroids will accelerate the fuel to speeds of about 12 to 14 km/s, so about 4 to 6 km/s have to be "killed", which is an energy of 8 to 18 MJ/kg. The energy must either be wasted using aerobraking or it must even be paid for by using up fuel.

Chemical rocket engines have exhaust speeds of "only" up to 4 to 5 km/s, i.e. about 10 MJ/kg of chemical energy is transformed to kinetic energy. If we compare the kinetic energy of the fuel dropped from outer space to a space ship with no velocity at all, the fuel has a kinetic energy of 72 to 98 MJ/kg.

If there would be a way to transfer a significant fraction of this kinetic energy of the fuel to the space ship in a non-destructive way as soon as the space ship has reached orbital height (but not orbital speed), this would present a high potential to reduce launch costs. If in addition to that it would be possible to use this fuel as reaction mass, further significant savings would be possible.

Proposed Solution:

At first sight to most people the problem of direct transfer of kinetical energy might seem to be impossible to solve, because any impact of bodies of celestial speed is a major disaster scenario. On the other hand side, a routine flight maneuver for space ships is using such a direct transfer of kinetic energy for negative acceleration: aerobraking.

The space ship flies into the upper atmosphere, where air drag reduces its speed. Even though onstreaming gas molecules have a kinetic energy a magnitude higher than chemical bonding energy, an aerobrake can survive the continued impact of the high energy plasma-like particle stream, building up a protective shield of compressed gas in front of itself. Could it be possible to use the same principle to create positive acceleration? I.e. use inverted aerobraking to "push" a space ship to orbital speed?
The proposed approach breaks this problem down into three sub-problems:

a) Distribute the fuel from outer space over a path, so there will be effectively a continuous stream of fuel to reach the space ship.

b) Navigate the space ship precisely into this stream of fuel and keep it stable during the inverted aerobraking maneuver.

c) Find a way to turn the fuel into gas just before it impacts the space ship, let it create a protective shield of compressed gas or plasma, and use it as reaction mass.

a) could be solved using a swarm of small robotic space crafts carrying the fuel. These robots would distribute themselves on trajectories that will bring them to the space ship one after another just at the right moment. The approximate length of the swarm would depend on the chosen acceleration of the space ship. Assuming an acceleration of 3g, it would be 1000 km long. Acceleration time would be less than 5 minutes. The number of robotic space crafts depend on the way to put the fuel into space and onto the space ship. Even a number of several 10000 robots could turn out to be economical, because then they would be manufactured in large quantities. And they will be reusable. Expensive high precision long distance sensor systems should be necessary only for a small fraction of them. Cheaper local sensors will be used to keep the more simple robots precisely positioned.

Economic considerations for trajectories of the robot swarm will include the fuel costs of an aborted launch (e.g. due to bad weather conditions at launch site), frequency of launch windows and allowable maximum speed of the fuel stream impacting on the aerobrake. A rather simple scenario would place the robotic swarm in a circular LEO. This allows for launch windows every 90 minutes. But it wastes a certain amount of potentially valuable energy by braking the fuel to orbit, and it will not be able to accelerate the space ship to 100% orbital speed, because the smaller the speed difference between fuel stream and space ship, the smaller the transferred energy. When the space ship is as fast as the fuel, no more fuel will reach it. Using the fuel directly when it is coming down from outer space or from a highly eccentric orbit would be potentially more rewarding.

b) Navigating the space ship precisely into this stream of fuel seems to be solvable by today's technology rather easily, considering that differential GPS can achieve positioning accuracies of far less than a meter on a global scale. Although it might turn out that a specialized form of localization needs to be installed for this purpose; existing localization hardware and algorithms may need to be modified to work at high velocities. Landing a rocket vertically is a proven technology. Developing a space ship able to "land" precisely in time "on top" of the fuel stream released by the robotic swarm will be more difficult, but appears to be possible.

Re-entry capsules of space missions do not need complex actuators to keep a stable orientation during aerobraking, but to stay on a precalculated trajectory is a bigger challenge. There are geometries that would be naturally stable on a focused jet of fuel, but they may turn out to be too heavy. Depending on the shape and functionality of the aerobrake it might be possible to deflect part of the fuel stream in a controlled way to different directions. Otherwise doing active steering with rocket engines arranged like an orbital maneuvering system would be necessary.

It could be possible to lay out kinetic fuel in a configuration that keeps the space ship stable within the stream of kinetic fuel, so complex controlled actuators on the space ship are not necessary. This has to be payed for with a certain amount of kinetic fuel passing by without hitting the space ship or with a larger and more heavy heat shield. To avoid this, the layout conficuration could be adapted in real time to counteract and control undesired movements of the space ship. See NoteOnStability.

c) Turning the fuel into gas and letting it interact with a specialized kind of aerobrake could turn out to be the most difficult and the most critical of the sub-problems. Several approaches should be investigated. The fuel could be a gas at the moment of release. Or it could be a liquid, sprayed to the vacuum, where it evaporates. Or maybe it can be turned into a gas by the exhaust stream of the space ship, potentially aided by chemical energy from the fuel itself.

The nature of the shield of compressed gas in front of an aerobrake must be studied in detail to find out under what conditions it will form, especially in combination with the chosen method of releasing the fuel to vacuum. Computational fluid dynamics should be used to study this. What are the implications of a pulsed fuel stream? How fast must it be pulsed? Can the aerobrake be built "bullet proof" to withstand an occasional tiny drop of not yet gasified fuel? What is the possible range of fuel impact speed that a single aerobrake can use? I.e. will it be possible to use impact speeds up to 14 km/s? What is the lowest economically usable impact speed? How much of the fuel stream can be used for inverted aerobraking, and how much impulse is transferred?

Since kinetic energy grows squared with velocity, but impulse just grows linear, the physical behavior of high speed impact is very different from everyday experience. Being a computer scientist and roboticist myself, I lack some expertise in this field. But apparently it might turn out that to generate impulse from the high energy impact economically, additional on-ship fuel might be useful. (It could be a coolant for the aerobake at the same time.) Another, not mutually exclusive approach could be to capture the fuel in an aerobrake the shape of a big rocket nozzle, where it (hopefully) would form a good protective gas layer, which is heated up by the incoming gas stream and will (again hopefully) build up an exhaust stream, which multiplies the resulting impulse. The following figure illustrates this (very optimistic) idea. It displays by no means a real solution. It is given here as a mere starting point for further discussions and research. Click for a larger image.

Note that in the above Saturn V example for going to orbital height, time of free fall is larger than the time needed to accelerate to orbital speed at 3g. Doing inverted aerobraking purely with horizontal forces while being in free fall in respect to the vertical direction is therefore a valid option.


I would like to thank Dr. Anthony Zuppero for his encouragement and his very helpful comments on my ideas about the concept of inverted aerobraking.


For comments, discussion, or funding (e.g. to present this at a conference) please contact me.

Axel Walthelm -


26. Nov. 2003 - 31. March 2004

In simple words (and therefore sometimes somewhat simplified):

The big problem is not to go up to space. The big problem is to go to orbital speed, so the space ship stays in space and doesn't fall down to earth. Going up is only about 7% to 17% of the effort. More than 83% of the effort to launch a space shuttle is for going to orbital speed.

Inverted aerobraking deals with this bigger part only. To go out of the atmosphere, normal rocket techniques will be used. You can see inverted aerobraking as an alternative propulsion system for the second stage of a new space shuttle. But this does not mean that the shuttle will have multiple stages at lift off, because the fuel for inverted aerobraking comes from space and is delivered "just in time".

I'm not talking about the next space shuttle. (Note: some people prefer to use the term "Reusable Launch Vehicle" or RLV for any space shuttle that is not the current space shuttle.) Inverted aerobraking needs a good space based infrastructure, so it will be many years before it becomes useful. But to decide just when the right time for it has come, scientific and technical studies should be done as early as possible.

Some people reading the general outline of the concept seem to believe I am talking about a large scale big size concept. It is possible to use inverted aerobraking to launch very big payloads. On a per kilogram base this may turn out to be cheapest. But my current personal opinion is that inverted aerobraking should be possible too for small space crafts the size of the current space shuttle or even smaller.

Other readers told me that the term "fuel" is misleading, because usually fuel is something you can burn. This is not necessarily true for fuel for inverted aerobraking. The fuel for inverted aerobraking comes from space. It is different from conventional rocket fuel, because it does carry its energy as kinetic energy (speed energy), not as chemical energy. A more general and slightly better term might be "propellant". Kinetic propellant carries at least 3 to 7 times more energy than the best chemical rocket fuels. And if done right, most of this energy comes "for free" from letting the fuel fall down towards earth.

Do you remember how small the return vehicle from the Apollo moon landings was? Compared to the big Saturn V it was really tiny. It is much easier to bring fuel from the moon down to earth than to bring fuel from earth up to space. Another source of fuel could be asteroids and comets. (Some info about that can be found at Asteroids may even take less energy in some cases, but it takes more time. Going to the moon is a matter of days. Going to an asteroid is a matter of months.

As long as you stay out of the so called "gravity well" of earth, space travel is cheap. Going up the gravity well is the problem. But the term "gravity well" is a little misleading, because once you are out of atmosphere, height can be treated off for speed. As soon as you travel fast enough into a direction where you can go around the center of the gravity well, which is occupied by earth and its atmosphere, you are in orbit. Close to earth a speed of 7.9 kilometers per second is needed. If you go as fast as 11.2 kilometers per second, you will leave the gravity well. That is: a velocity change (physicists like to call it delta-vee or v) of about 7.9 km/s gets you to orbit. Another velocity change of only 3.3 km/s will bring you anywhere in the solar system or beyond. You see, the important part of space travel is velocity, not height.

To lift off from the moon and go to outer space, only 2.4 km/s are necessary. That is really cheap, in comparison. Do you remember that kinetic energy grows squared with speed? A car at 80 km/h needs about four times longer to brake to a full stop than if it drives at 40 km/h (that's 50 vs. 25 miles per hour). So 2.4 km/s means only 5% the energy of 11.2 km/s. Going to space from the moon takes only 5% of the energy! And there is no atmosphere on the moon you have to get out of in the beginning and a lot less gravity has to be fighted during launch. On earth that can easily add another 2 km/s to the total speed difference the space craft has to generate. On the moon it takes only about 0.04 km/s to fight gravity. Take this into account and launching from the moon is 30 times easier in terms of energy. If you furthermore take into account that the fuel is much more energetic when used for inverted aerobraking, it is potentially more than 100 times cheaper to use fuel from the moon than from earth.

The basic idea of inverted aerobraking is to transfer kinetic energy from the fuel / propellant to the spaceship as directly as possible. The propellant will "push" the space shuttle like the wind pushes a sailing ship. But the magnitude of speed difference is so big, that this comparison does not carry very far. No sail or parachute can stand the direct impact of any big number of molecules with a speed difference of several kilometers per second, no matter how strong the material it is built of. Chemical bonding energy (the forces that keep molecules of any solid substance together) is much less than the kinetic energy of any molecule at a speed of several kilometers per second. Kinetic fuel is really hot. So at first glance one would conclude, that such a direct transfer of kinetic energy is impossible.

But there is a standard maneuver of space crafts that does such a direct transfer: aerobraking. The space ship goes into the upper atmosphere until "air drag" slows it down. Again, "air drag" is a misleading term, because a wind speed of several kilometers per second turns wind into something completely different. It behaves like plasma. Maybe you have seen a photo of the current space shuttle in orbit with a glow around it. This sometimes happens when the upper layers of atmosphere hit the shuttle. The force applied is still very small, but it already creates a visible effect. When the current space shuttle does the aerobraking maneuver, astronauts say it is like flying through a neon tube. But of course it gets much hotter outside.

A space craft doing aerobraking needs a heat shield, which is called an aerobrake. The current space shuttle uses ceramic tiles to protect itself from the heat of aerobraking. These tiles are reusable. Considering what I said about chemical bonding energy above, this is kind of surprising. How do these tiles manage to survive aerobraking? Apparently it is the air compressed in front of the space shuttle, which provides additional protection to the space ship, like a protective cushion. This layer of extremely hot gas could be called a "plasma shield".

So the key to inverted aerobraking is to reproduce this protective effect of compressed gas between the space shuttle and the impacting fuel. And the impacting fuel must be turned into a gas before it impacts the heat shield. How to do this exactly in the most efficient and economic way is an open question which needs some research. I have outlined a few options in the text above, but detailed studies and simulations are necessary to choose between them.

Other results from Project Orion indicate, that heat shields can also be designed to withstand short pulse-like impacts of plasma, because the protective plasma shield builds up almost immediately with only ablating a little material from the top of the heat shield. Since depositing kinetic propellant in a perfectly uniformous density along the path is difficult, high frequency pulses of plasma impact seem to be a valid alternative. Project Orion was designed for a pulse frequency of about one plasma impulse per second. Therefore about 1/4th of the projected ship consisted of a big shock absorber system. Inverted aerobraking could easily go for pulse frequencies of 100 per second at a pulse intensity of 1/100th. That way a much smaller shock absorbing system or even none at all would be sufficient. (Remember: 1 Hz is a shaking movement you can feel; 100 Hz is a humming vibration you can hear. (Although without some sound-damping it would probably be loud enough to be felt in the stomach too.))

Click here to see a schematic animation how pulsed inverted aerobraking could work.

To place the fuel in space on a precise precalculated trajectory is a technological challenge, which can be solved using a swarm of small robotic space crafts. The fuel of each robotic space craft will be released only seconds before it hits the space shuttle. If the fuel is released as a gas, it would be shot towards the space shuttle. This will result in a propulsion of the robotic space craft. In fact this could be done with traditional low power rocket engines. If the fuel is released as a fluid, it will remain concentrated for a longer time, and can therefore be set free at lower speed and earlier. If the fuel is released as a solid substance (e.g. as snow flakes), it can stay in space even longer, which gives more time to navigate the robotic space ship away from the space shuttle. But then the protective layer of gas between space shuttle and fuel must be much stronger, because it has to turn the fuel into a gas (or plasma).

How many of these robotic space crafts are necessary to accelerate a space shuttle to orbital speed? It depends on the way to place the fuel in space. My first estimates are in the range of hundreds to ten thousands. That may sound very expensive, but economic solutions are possible. Satellites are expensive, because they are built as unique specimen. Building them in large quantity is much cheaper. Satellites are expensive because they must not fail, so they must be built redundant and with extra high quality technology. If some of the robotic space crafts fail, it is only necessary to find out in time to separate the malfunctioning ones from the swarm. As soon as space infrastructure is installed, capturing and repairing malfunctioning satellites will become much cheaper too. Anyway, the robotic space crafts are completely reusable. Satellites are typically operational for many years. The initial costs to bring the robotic swarm to space can be distributed over many launches.

I am a computer scientist and roboticist. So I am quite confident that the robotic swarm is a realistic option. The main reason for this web site is to find other specialists willing to crosscheck and refine the concept of inverted aerobraking.

10. Jan. 2004 - 10. Jul. 2004

Some Quantitative and Qualitative Estimates and Related Projects

The Saturn V example

Kinetic fuel is really hot!

How much kinetic fuel is needed?

How much does on-board fuel help?

What Size will the Swarm of Robotic Spacecrafts be?

What Project Orion can tell us...

The Impact Reaction Engine

An Animation to Illustrate Pulsed Inverted Aerobraking

Notes on stability of the space ship during inverted aerobraking

Excerpts from Discussions

Lucio de Souza Coelho (Jan 13 2004):

By the way, maybe the horrible scifi-flavoured expressions "meteor propulsion" or "meteor sailing" are more insightful than "inverted aerobraking". :-)

Suggestions for better names are welcome.
My first working title (many years ago) was "milky way launching", because I imagined the fuel to be a path of snow flakes, looking like a milky way. Then I did consider "blow by" (by analogy with the so called swing by maneuver). But currently I think that "inverted aerobraking" is the best descriptive name, even though it may not be the most intuitive one, at first.

Kevin Parkin, authoer of an interesting proposal to use beamed microwaves to
launch space ships, wrote (Feb 21 2004):
Thanks for your e-mail!  Congratualations on putting some original
thought into your inverse aerobraking scheme. [...]
You might also want to consider the use of pellets onto a pusher plate,
similar to project Orion:

thanks for your reply. [...] Last weekend I read "Project Orion". Really interesting and fun to read.
(Although it contains more about nuclear and space politics than I ever wanted to know.)

George Dyson, author of the book "Project Orion", wrote (Mar 27 2004):
If and when there really is some kind of high-velocity space 
transportation network, then your idea becomes even more important in
solving the problem that is usually neglected: once you get where you
are going, how do you slow down?
Now, this aspect is new to me. If you go to a planet with some 
atmoshphere, aerobraking would be the typical answer. But yes, if you
go really fast, say 20 km/s, aerobraking will become difficult. So doing
some inverted inverted aerobraking could be a solution to that.

And sometimes one would like to travel fast to some place without
atmosphere. Say for example some asteroid which is mined for propellant.
Having highly skilled personnel sitting on the space ship for months
doing nothing would be quite an incentive to build some faster means of
transportation. Not to mention that you need to bring less food,
exercising facilities etc. Going faster may be cheaper. Good point!

Christopher M. Hirata published a very interesting chart of delta-v necessary
when travelling between Earth, Moon, and Mars:
According to this chart 9.7 km/s are necessary to go to LEO. This is 1.8 km/s more
than orbital speed and confirms my estimate that only 20% of the effort to go
to orbit are spent on going up to space.
When I asked him how he got this number, he wrote (Apr 18 2004):
The amount of Delta V required to reach orbit is somewhat dependent on the
launch vehicle (due to the issues you discussed below); the 9.7 km/s
figure is typical (I computed it by adding up the numbers for one of the
launch vehicles).
For large launch vehicles such as the Shuttle or Saturn V, more Delta V is
lost to gravity than to atmospheric drag.

Graham Cowan (Jul 06 2004):
A system like this was part of story that I'm almost certain
was by Donald Kingsbury. Oxygen from the moon was released as vapour
from many points along a rail, and the vessel needing to be
boosted to orbital speed came in along this rail and hit the
oxygen. It was called an "imp", short for impact-something.

Donald Kingsbury (Jul 10 2004):
Dear Axel,
The book was "The Moon Goddess and the Son."  Relevant bibliography follows
which is obtainable on my website
I was very aware that energy could be tapped by dropping stuff from the
moon to the earth, but I might have done it slightly differently than
your proposal.  I was converting kinetic energy into electrical
energy.  When a ship from earth was captured by the "leoport", the ship
was accelerated and the leoport decelerated.  When a ship came in from
the moon, the moonship was decelerated and the leoport accelerated. 
There were a lot of ideas kicking around in that novel.  There was an
orbiting tether attached to the lunar system with interesting energy
transfer characteristics -- not nearly as difficult as an earth tether,
and especially so because the tether wasn't fastened to the moon's
surface, but orbited the moon.  The "elevator" delivered the ship to
the moon at much less than lunar escape velocity.  I recollect it was
by electrical capture -- electricity being generated by the landing,
and used by the take-off.  I remember that I was toying with the idea
of reverse aerobraking, but I don't recollect whether the idea worked
its way into the novel or not, I'll have to reread it myself to find
out!  There is certainly a lot of merit in your proposal if the
engineering challenges can be met.

You have certainly defined the problem properly.  We don't need to go
into space -- what we need is orbital velocity!

The trick is to exchange energy and momentum between the people going
up and the people going down!

My novel "The Moon Goddess and the Son" is a bit hard to come by after
20 years, but I still have copies and will send you one if you give me
an address.

PS I liked your site. The Internet is wonderful!

That book would be the first price awarded to my proposal (but in the meantime
I got myself a copy from a British book store).

Andrew Nowicki, who compiled a list of alternative ways to space (which is
pretty much biased towards launch guns, but very interesting nonetheless) was
kind enough to send me a copy of a few pages from the december 1979 issue of
"Analog Science Fiction Science Fact" which is related to "The Moon Goddess
and the Son" and describes an early variant of inverted aerobraking I haven't
been aware of. This proposal seems to be ignored in all of the compiled
lists available on the internet. I wonder why...

Visitors Count