On 7/12/06, Arthur Smith <apsmith@...> wrote:
> So Zubrin's estimates improved the cost effectiveness of SPS by a factor
> of 16 in just a few years? I suppose that's encouraging...

No, mine did.

> For current communications satellites, launch costs, even to GEO, are a
> small fraction of total expenses, and time is money.

All indications I have are that launch costs are 40% total cost of launching
a commercial satellite, and have been in this ballpark since 1993. The
general trend is that this proportion is going up, not down. Ariane/EELV
class boosters are about as cost effective as they can get (Falcon is trying
to start a new class in terms of cost effectiveness); if Falcon is
unsuccessful, this problem will continue. Satellites themselves are getting
cheaper as S/C busses continue to standardize; A2100 has at least three
different "stretches" analogous to those of airliners.

Also, many of the design features of satelites are due to the continuing
high cost of launching them, if money could be saved in the booster, a lot
of these could be changed to save money. One is to improve the safety
factors to the point were expensive acceptance testing and fracture
propagation analysis for each satellite could be eliminated. Another is to
standardize the payloads by reducing the lifetimes and using smaller
increments in the overall technology. A factor of ten improvement in
booster costs would probably lead to a factor of three or four improvement
in the cost of a GEO transponder lease, to certain limits.

> The numbers you
> have posted suggest for SPS getting to GEO vs just launching to LEO
> would constitute about 90% of total costs; if it took more time but
> saved much of that cost, it would clearly be worth it on a business
> basis that is not there for commsats.

On the model of conventional boosters, a typical conventional
booster...Atlas 551...can launch some 3960kg to GEO and the same booster can
launch 7500kg to a superGTO 1550m/s to GEO, 6875kg to a conventional GTO,
about 20500kg to LEO (Atlas 552, only difference is upper stage has a second
engine.) In each case, the booster costs the same (actually, the LEO
booster is a bit more expensive in this case, extra engine and bigger
fairing.)

On the model of Bluestar/Yellowtug, in addition to the Bluestar ascents that
bring up the satellite (the commercial GEO model that winds up costing
$2500/kg to GEO delivery uses one Bluestar ascent to bring up the satellite,
which then gets its hydrazine load from the station...which in turn was
brought up by another Bluestar. Another Bluestar brings up the YellowFly
upper stage, and then four more to fuel up the tug, that's about six and a
half Bluestar loads for one big GEO satellite. The $4mln. YellowFly top
stage is not reused and forms the biggest expense of this mission.)

> Comm sats also do not have anywhere near the huge solar power to mass
> ratio we'd hope for from an SPS component, so the relative electric
> thrust available would be much less, and the transit would take much
> longer for a typical communications satellite than for an SPS.

Ion engines are heavy, even when the power is there. They also lack thrust
density and require a lot of surface area for not much thrust. It can be
figured out, but I'm not about to do it (I got enough on my plate already;
also, the cost savings needed for the SPS to be successful with only Earth
in the picture are not possible to achieve on electric thrust, even from
LEO.) At the optimistic 200W/kg total satellite system, a theoretical 1N
per 10kg (at 2000W/N) could be attained. That's 0.1m/s2, which is plenty.
The problem is that the spacecraft would be all propulsion system. More
reasonable would be 1N per 200kg. It would take 310,000sec (3 days, 14
hours, 7 minutes) to deliver 1550m/s. With impulse losses and spreading the
thrust over several orbits (thrusting 50% of the time) and 1700m/s (150m/s
impulse losses), it would take about 8 days to get the spacecraft from GTO
to GEO. Not bad, eh?

From LEO there are other problems. Assuming you simply let the sucker
thrust the whole time it was in sunlight, the resulting maneuvers would
raise the apoapsis in the shadow of the Earth, where the poor SPS would then
spend most of its time for a couple of months. It is also harder to keep
the line of apsides and line of nodes straight with this plan. The impulse
losses are far larger. Early on, it would be impossible to thrust more than
50% of the time because of eclipses, and later it would be impractical
because raising the periapsis before the apoapsis is on target is
horrendously inefficient. 30 minutes out of a 5 hour transfer orbit near
the end is 17%. Between eclipse effects and efficient periapsis maneuvers,
it's probably about 25% of the time, and delivers about 2800m/s...I don't
think it's as bad as I thought after all: The total time is 2.24 mln
seconds, which amounts to 26 days.

This set of manuevers would take about 4500m/s and use up 14.9% of the LEO
mass of the SPS. From $250/kg to LEO, this would cost about $294/kg to GEO,
taking a not very onerous 34 days to ascend. Comparing $294/kg GEO ascent
costs to a requirement for $272/kg total costs is not quite as hopeless, but
we are still far, far behind. The satellite hardware under these
circumstances will cost in the ball park of $10,000/kg. That places the
total cost of the satellite at $10,294/kg compared to a requirement of
$272/kg.

> The fact that you can only thrust for 50% of the time in LEO doesn't
> seem a huge disadvantage - the more serious one would be whether SPS
> components would need sufficient additional mass to withstand the
> temperature variations that launch to GTO saves costs overall.

I doubt that's a big deal. The question really is how long you are willing
to spend in the LEO environment. Assuming like most SPS advocates do, that
the SPS is assembled on LEO, it has to deal with those temperature
variations anyway.

> I don't think a reusable tug makes any sense for SPS construction; the
> components should be self-propelled with a small engine and propellant
> store, once in orbit.

_Only_ if you use SEP. Also, there is the possiblity of using a half-tug.
Dig this:

The space tug has big electric engines, but not enough solar power to run
them. To propel an SPS it uses the mother of all electrical harnesses (i.e.
the sort that would run a scrapyard shredder.) To propel anything else, it
has an SPS receiver, when using it, it can thrust in the shade. The SPS, of
course has its own ion engines for stationkeeping, and say those take 30
days to complete delivery from GTO, while the tug takes 30 days to do the
GTO delivery. While pushing the SPS, its using its power (unless its a
follow on SPS, then it can use the power of an earlier one via its
receiver.) After separation, it aerobrakes back to a LEO station to refill,
or pick up another payload. There could be the problem of the SPS receiver
and ion engines being incompatible with aerobraking...that's a pretty easy
problem to fix though: beam it some power, and then it can use its
propulsion system instead.

This leaves an anti-SPS option though. Beaming power to electric tugs from
the ground could also work. Efforts to do this sort of thing are already
being done in a space elevator contest. I think this is a pretty good
idea...the microwave space tug?

> The cost differential between LEO and GEO with chemical propulsion may
> also drop if on-orbit propellant is available, as has been proposed
> elsewhere. It could come from the Moon or from ultra-cheap Earth rocket
> launches (big dumb boosters that are just putting up rocket fuel, so we
> don't care if they blow up 20% of the time).

We do actually...the environmental mess of losing big boosters that often?
Also, the propellant would have to be managed for balance, and so would not
be a whole lot cheaper than a repeat satellite launch (excluding the
satellte's processing and acceptance testing of course.) Also, the only
cost estimate numbers I have up that have any hope of making the SPS
actually work are already based on a reusable vehicle with similar
benefits. It works like this:

Bluestar has a bit of overkill in its orbiter's ascent engines. Also, the
OMS can be fired in combination to add some thrust. The normal payload
limitation of 8000kg is due to entry and landing considerations (the
Bluestar has a worse glide ratio than the Shuttle for landing, but it has
much lower entry wing loading...more crossrange, less glide distance, but
more time and turning ability kind of wash it out, so basically it is just
as much of a pain as the Shuttle is to land. The Ascent version will be a
bullet with parachutes.)

The tanker load of Bluestar is compatible with Bluestar's OMS (MON3/MMH for
the Series I and oxymethane for Series 2). With a hardware load of 8000kg,
the Bluestar is put to the limits during an abort (where it does not have
time to deploy the payload), but with a tanker load of much higher, say
12000kg, the tug propellants can be dumped out the OMS nozzles in the event
of a main engine failure. During a normal ascent, some OMS is fired early
during ascent to assist. In this manner, a Bluestar could deliver 10000kg
of tug propellants instead of an 8000kg module.

> I don't particularly care what technology is used - I'm just looking at
> the physics end of it to understand what's possible, and it seems
> Zubrin's and your estimates are ignoring some pretty obvious ways to
> save money by taking advantage of the physics available. Remember, if we
> actually do this on a reasonable scale, we're talking about hundreds of
> GW-scale satellites that will each have masses of hundreds to thousands
> of tons, total expenses will be at least in the hundreds of billions of
> dollars, so such a system could afford to do all sorts of custom things
> that might not work on a smaller scale. Whatever's been done up to now
> for communications satellites, ISS, etc. is not terribly relevant on
> that scale.

The SPS will not be able to start up on that scale. I think if one
seriously (i.e. as seriously as Bigelow's inflatables or SpaceX Falcon) took
on an SPS, the first one would probably wind up being a 10kWe model that has
a mass of perhaps 100kg (comparable commsats mass at 2500kg) and then beamed
it somewhere it would be used...like MDRS. It would ride up as a secondary
payload on an Ariane 5G, then use its own ion engine to place itself on GEO,
and cost millions instead of hundreds of billions. Alas, once this exciting
little project reaches fruition, it is very unlikely to inspire full scale
SPS, just like ATP and DC-X failed to inspire their full scale dreams
because of the technicals and economics.

Terry


[Non-text portions of this message have been removed]