Re: Mojave airport is not a spaceport

From: Andrew Nowicki (andrew_at_nospam.com)
Date: 06/24/04 

Date: Thu, 24 Jun 2004 19:10:50 +0200

John Carmack wrote:

JC> Why do you keep bringing up wings? Most of the people on this thread
JC> are partial to VTVL.

JC> A BDB can do a powered landing with little more than software changes
JC> and the addition of landing struts (you could even avoid that if you
JC> are REALLY confident in your terminal positioning and land on a
JC> special ground structure, but I don't reccomend that). Yes, you need
JC> to cart some more propellant around, but these are big, easy to
JC> fabricate vehicles that can be "made in shipyards", right? Adding
JC> some size doesn't cost much, and the operational win would be dramatic
JC> compared to a splashdown recovery.

Yes, VTVL (vertical takeoff and vertical landing on dry land without
parachute and without wings) is the cheapest option and probably the
best one if the rocket is sturdy enough to survive hard landing
without catastrophic damage. (Russian landing retrorockets malfunctioned
several times.)

JC> The major point of contention is splashdown versus a powered landing.

This issue got more coverage in this thread than it deserves.
Reusability is much more important.

JC> We have done some work with big parachutes, and I'm not a fan. Taking
JC> a boat out to fish your rocket out of the ocean is going to suck.
JC> Landing the booster is feasible, economical, and gives the best
JC> operational characteristics, as long as you are willing to accept some
JC> limitations on stage trajectory and aspect ratio. I contend that
JC> these are worthwhile tradeoffs.

We could go on for a long time if you enjoy this topic.

AN> ...The best design for the second stage is my engine cluster... Its
AN> description is posted at:
AN> http://www.islandone.org/LEOBiblio/SPBI101.HTM#engine_cluster

JC> Qantity and replication are easy in a spread*** or CAD program. It
JC> is a little more troublesome in the real world. We recently made out
JC> lives much, much better by moving from four differentially throttled
JC> engines to a single larger engine with jet vanes. At some point mass
JC> production effects can kick in, but it isn't in the development stage.

There is huge difference between making the rocket by hand and
making it by a robot.

Before the Civil War guns were made by gunsmiths. Their parts
were not interchangeable because the gunsmiths could not make
identical parts by hand. Making the rockets by hand takes lots
of time, and quality control is difficult -- you have to inspect
every weld because one bad weld can ruin your rocket.

Most of the Agena rocket engine was made by a robot -- coolant
passages were drilled in a monolithic slab of aluminum alloy.
My engine cluster has similar design -- it can be made by a
milling robot. The robot is cheaper than the rocket plumber,
and the engine is so sturdy that it may survive the hard VTVL
landing. You do not have to worry about weak welds -- there are
very few of them. Fabrication quality is determined by your CAD
drawing rather than by the robot. Even if you make just one
rocket, it is cheaper to make it by the milling robot than by
hand. I do not know if standard robots can make the narrow
injection nozzle holes, but they can certainly make all the
other holes.

JC> You can't design a high aspect ratio vehicle, but again, it doesn't
JC> really matter for a booster stage. Make it squat, and let the upper
JC> stage be a sphere if it wants to. Go ahead and be highly non-optimal
JC> in the aerodynamics and staging fraction if it gets you good
JC> operability. A booster like this would be a cargo elevator to 100km
JC> or so. Up and down on the hour if you wanted to.

Can you comment on stacking the engines sideways?
(http://www.islandone.org/LEOBiblio/SPBI1010.JPG)
I believe that having lots of engines covering
large area of the rocket is a good idea because
it improves thrust, specific impulse, or both.

JC> Most of the fundamental complexity of a rocket stage is independent of
JC> stage performance. Lots more stages will give lots more problems.
JC> Pushing performance requirements to the edge can easily give even more
JC> problems, which is why I'm not an advocate of a completely SSTO
JC> design, but two stages is going to be both more reliable and easier to
JC> develop and test than more stages.

I have seen lots of comments about problems caused by large
number of stages, but I do not understand these problems.
It seems to me that when you stack identical rocket stages
like Lego blocks, your only problem is designing the
explosive bolts which hold the stages together.
 
JC> An upper stage from us would probably use 98% peroxide and kerosene.

Other good choices are H2O2/RP-1 and H2O2/propylene.
They are easy to store, but, to the best of my knowledge,
their critical pressures and critical temperatures are high.
If the coolant pressure is lower than its critical pressure,
bubbles may form in the coolant.

If you use oxygen/methane instead of H2O2/RP-1, the payload
mass will increase by about 50% and it will be easier to
cool the engine. Methane can be extracted from natural gas,
so it is cheap and easily available. Propellant tanks holding
liquid oxygen and liquid methane are covered with natural
thermal insulation in the form of frost. Oxygen and methane
have similar boiling point temperatures, so there are no
problems with a propellant freezing in a pipe.

JC> Probably still pressure fed, but at a tank pressure of only
JC> 100 psi or less, which doesn't hurt it much in vacuum operation.

100 psi = 6.9 bars
That is pretty low pressure. (Russian RD-170 engine has the chamber
pressure of 245 bars). Low pressure means high mass ratio, but the
engine must have large exhaust nozzle exit area to produce high
thrust and high specific impulse. Again, the engine cluster looks
like the right choice.
 
JC> A gas-and-go RLV would be a huge advance even if it used TEN TIMES
JC> the propellant that a conventional rocket used for a given amount of
JC> propellant.

I agree.