They intend to do a return to launch site--at least eventually. There will be several intermediate steps to prove the system. If it proves unworkable, always landing on a barge is better than not, I guess.

RTLS will certainly require a good deal of margin (by rocketry standards), but first stages usually have some margin already, and SpaceX will have a nice range of options once the F9H comes on line so that they don't have to fly a rocket "fully loaded" except for certain extra-large payloads.

I imagine they can also optimize most flight profiles for RTLS, by sending the first stage more than usually straight up--which is close to the case anyway. Lateral velocity is mostly the job of the second stage.

While it's never been done before, it's not an unheard-of concept. One of the potential evolutions for the STS was "fly-back boosters".

Barnaby Wainfan patented an interesting alternative: after staging, re-enter, then do a burn and fly ballistically back to launch site. This kills the horizontal speed with air resistance instead of propulsion, potentially saving fuel. You have to regain the altitude though.

http://www.faa.gov/about/office_org/headquarters_offices/ast...

But the extra weight is a concern, it limits payload to orbit. Try flying the first stage back to the launchpad in Kerbal Space Program while delivering a useful payload to orbit! It's instructive. KSP really develops your intuition for rocketry, everyone commenting in this thread ought to buy a copy.

The first video explains the modifications to the game, mainly more realistic aerodynamics, realistic fuels and engine capabilities, a realistic planet and launch site. The second video is him actually performing the mission.

There are a few differences between what he did and what we now know is actually planned, but it's still pretty damn close and demonstrates the feasibility of the concept in a pretty intuitive fashion.

[1] http://www.spacex.com/news/2014/07/22/spacex-soft-lands-falc...

8 Rockomax 48-7S (little orange engine, stuck on with 8 small cube struts) + 1 Rockomax X200-32 Fuel Tank (the large gray tank, one size below the big orange tank) will put a capsule in orbit around Kerbin pretty easily. No such luck in real life.

Typically KSP engines have worse performance than real engines, but Kerbin is pretty small.

That's what I meant by 'the opposite is not necessarily true'. My argument was that if something does not work in KSP, it probably won't work in real life (which was refuted by another commenter because of missing Lagrange points). IMO it still holds true for basic rocket designs and their capabilities for LEO / GEO. To my knowledge, all real world rocket designs have been replicated in-game (albeit with much lower complexity of course) and demonstrated to work.

edit - and for the same reason there are also things that would work in KSP that would fail in the real world.

Do you have an example for that?

> edit - and for the same reason there are also things that would work in KSP that would fail in the real world.

That's what I wanted to say with 'the opposite doesn't hold true'.

Not really applicable to this discussion, but one instance which is highly applicable is that you can't control two ships in KSP at once. If your lower stage needed to do a powered land which an upper stage was still accelerating into orbit, you couldn't replicate that in ksp.

.. Unless your name is Scott Manley ;-). But yes, good point about Lagrangian points [1]. I wasn't aware that KSP is only a two-body-simulation - interesting how the game can hide that with its sphere of influence implementation.

[1]http://forum.kerbalspaceprogram.com/threads/18623-Lagrange-p...

Also, my point with the logic is that when looking at the differences between physics in KSP and the real world, we are dealing with overlapping, rather than nesting sets.

First time I've seen this, thanks. You certainly have a point that feasible designs in KSP vs. real world aren't completely nested. When I was arguing about KSP being an upper limit, I didn't really think about the SSTO case - at least the stock simulation certainly doesn't hold up to be able do any kind of feasibility check for horizontally launched vehicles.

http://en.m.wikipedia.org/wiki/Rotary_Rocket

[1] scare quotes because I know its never easy to land and California or Nevada east of Vandenberg for minimal delta-v requirements.

Interestingly SpacePort America (http://spaceportamerica.com/) has a similar problem, although if you look (https://www.google.com/maps/place/Spaceport+America/@33.0125...) you will see there isn't much east of it.

In the case of the Saturn V, the first stage (S-IC) sequence completed after 2 minutes 41 seconds, at an altitude of 42 miles (68 km) a speed of 6,164 miles per hour (9920 kph). It continued ascending ballistically to an altitude of 68 miles (109 kilometers), ultimately landing 340 mi (550 km) downrange.

https://en.wikipedia.org/wiki/Saturn_V#S-IC_sequence

EVERYTHING.

When you throw away the entire rocket every flight you throw away tens of millions of dollars in hardware. If you can recover and re-fly stages you can get away with a lot of compromises elsewhere. Especially since the first stage, for SpaceX anyway, represents 3/4 of the hardware cost of the launch vehicle.

Even reusing the first stage once means cutting the overall hardware cost of a flight by nearly 1/3. For SpaceX that translates to about $20 million per flight in savings. And that's from one and only one reuse flight of the first stage. In comparison the fuel is nothing. In comparison even a massive payload hit is acceptable. As long as the payload reduction is less than the cost reduction, everything is golden, and the rest is profit margin. For 2 reuses (3 total flights) the hardware cost per launch drops to 1/2 of current costs. For 5 or more uses per first stage the cost drops below 40%.

SpaceX's launches are already cheaper than the competition, dropping below 1/2 of their current cost floor makes it impossible for the competition to keep up and would enable them to own the launch market.

My comment above wasn't shooting down the concept of reuse, but showing that the lateral range of the stage 1 Saturn V booster was pretty significant. To re-land at Kennedy, you'd likely have to:

1. Change the launch trajectory to gain more initial vertical range.

2. (Possibly) reduce the first stage size to decrease its range. This means increasing comparatively the 2nd and 3rd stage sizes.

3. Use of strap-on boosters (themselves independently recoverable) which would reduce the mass of the remaining recoverable stage 1, and hence the momentum that would have to be re-vectored to KSC.

And while reusability is good, it's a bit like Amdhal's Law: your initial gains are the biggest, and likely you're going to see a cost function something like:

    (fixed booster cost + reusability engineering cost) / reuses +
    additional fuel cost + refurbishment costs + launch risk

Which doesn't say that the exercise is futile. Only that past 3-4 reuses you're gaining little for what's likely a large additional expense, or phrased differently, there's a minimum cost if you want to go to space today.

It's still not inevitable that such things will happen with reusability automatically. Most likely the first reusables have high maintenance. But they can be improved.

But risk in general is also a function of use and operational time. This has been borne out in many contexts, including both space flight and aviation. Systems degrade in nondeterministic ways over time, increasing failure risk. Even very minor variations in design -- a few mm of protrusion in a fuel-oil heat exchanger in the Boeing 777, implicated in the British Airways 38 Heathrow crash, given a specific set of circumstances in in-flight ambient temperatures and engine throttle settings resulting in ice-induced fuel starvation -- can have profound impacts.

https://en.wikipedia.org/wiki/British_Airways_Flight_38

It'll be interesting to see how re-use, risk, and cost play out with Space-X.

As for cost, fuel is cheap, it's one of the cheapest things in the whole operation. Think about the fact that every commercial airliner has the capability to make an unpowered landing and to save fuel doing so, why don't they? Because operational complexity adds cost much more than using up fuel does. By the same token, the greatest costs in orbital launch come from throwing away the hardware after every flight and operational complexity. If using extra fuel enables them to more easily reuse the stages and get more launches out of them as well as reduce overall operational complexity, then it makes economic sense.