[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
starship-design: A paper on low cost launchers from earth.
Thought you might find this old paper interesting. If your not interested
in the current argument about how to reduce launch costs. Sorry I bothered
A Rocket a Day Keeps the High Costs Away
by John Walker
September 27, 1993
There's a pretty general consensus that one of the greatest barriers
to the exploration and development of space is the cost of launch to
low earth orbit. The incessant and acrimonious arguments among
partisans of the Shuttle, DC-*, NASP, TSTO, Big Dumb Boosters,
bringing back the Saturn V, buying launches from the Russians and/or
Chinese, or of developing exotic launch technologies (laser,
electromagnetic, skyhook, etc.) conceal the common premise of all
those who argue--that if we could launch payloads for a fraction of
today's cost, perhaps at a tenth to a thousandth of today's rates of
thousands of US$ per kilogram, then the frontier would open as the
great railway to orbit supplanted the first generation wagon trains.
The dispute is merely over which launch technology best achieves this
Conventional wisdom as to why industry and government choose not to
invest in this or that promising launch technology is that there
aren't enough payloads to generate the volume to recoup the
development cost and, in all likelihood, there never will be.
How much would it cost to find out if this is true?
What we pay today
Could we take a moment's pause from debating which is the best
successor to the outrageously expensive way we launch now and, as
engineers, ask ourselves just why it is that rockets have to cost tens
or hundreds of million of US$ per shot. Space FAQ space/launchers
gives approximate per-launch costs of representative systems on which
commercial launches can be purchased as:
Vehicle Mission cost, US$ millions
Scout G1 12
Long March 3 33
Titan II 43
Delta 45 - 50
Proton 35 - 70
Atlas 45 - 85
Ariane 4 65 - 115
Titan III 158
Titan IV 315 - 360
I've deliberately not included data on performance, reliability, or
anything else because that would distract us from the most striking
observation about these vehicles; each and every one of them, whatever
the technology, country of origin, original design intent, launch
history, fuel and oxidiser, success or failure in the commercial
launch market, have mission costs in ranging from tens to hundreds of
millions of US$.
Why is this? Why do rockets cost so much?
What's in a launcher?
Let's simplify the problem by focusing entirely on expendable boosters
built with current technologies--those used in the existing launchers
named above. Further, let's consider only pure liquid-fueled
launchers (with the exception of Scout and Pegasus, the core stages of
each of the above launchers are liquid rockets). From an engineering
standpoint, then, what is a rocket?
Well, it consists of a collection, often vertically stacked, of:
Cylindrical fuel and oxidiser tanks
Rocket engines (including turbopumps, gas generators, etc.)
Guidance mechanisms (gimbal joints, hydraulic actuators, APUs)
Guidance and navigation system (IMU, GPS, radio command receiver)
plus other ancillary details like range safety receivers and telemetry
sensors and transmitters and the like, and that's about it, isn't it?
Now the question that comes to mind is this: why should something like
that cost tens to hundreds of millions of US$?
Cylindrical fuel tanks aren't that expensive, and they make up most of
the rocket. (Sure, if you're striving for every last gram of
throw-weight in an ICBM, you can push the tankage cost as high as you
like, but in a commercial launcher?) And rocket engines are finicky,
complicated, and intolerant of defects. Well, yes...but so is a DOHC
4 valve per cylinder turbocharged, intercooled V-8 internal combustion
engine, and nonetheless one can purchase such an engine, integrated
into a ground transportation vehicle, from a number of manufacturers
at a cost three orders of magnitude less than that charged for the
rocket, and expect it to function without catastrophic failures or
extensive maintenance, for five years, tens of thousands of
kilometers, and thousands of mission cycles. Guidance? Again, as
long as we aren't gram-shaving, this is pretty mundane stuff--the
hydraulics can mostly be adapted from airliners, and the electronics
from a PC--"mem'ry for nothin', chips for free". (For an LEO launcher
we don't need radiation-hardened electronics.)
The first mass-produced launcher
We've seen from the "standing army" argument for launchers requiring
minimal (airline-scale) ground mission support the impact of fixed
costs on per-mission costs when the number of missions is limited.
But the presence or absence of a "standing army", and the frequency of
flights over which fixed costs are spread, isn't fundamentally linked
to whether the launcher is reusable or expendable. Consider the
following mass-produced expendable rocket.
Number manufactured: 6,240
Number launched: 3,590
Successes: 2,890 (81%)
Failures: 700 (19%)
In inventory: 2,100
Work in progress: 250
Expended in development: 300
Development program cost: US$ 2 billion
Development cost per launcher: US$ 350,512
Total manufacturing cost per launcher: US$ 43,750
Marginal cost, launchers 5000+: US$ 13,000 (Yes, 13K!)
These are actual figures for the first mass-produced rocket vehicle,
the V2 (A4)--fifty years ago. Prices are in US wartime dollars.
Stating the obvious.... The V2 was a suborbital vehicle, intended to
lob high explosive over relatively short distances. Quantity
production of the V2 at Mittelwerk was accomplished with unpaid slave
labour under the brutal rule of the SS. And the failure rate was
unacceptable by current standards.
And yet...consider that this was the very first space-capable rocket
ever built. That it was manufactured under the constraints of a war
that Germany was losing, subject to aerial bombardment by night and by
day, with continual supply shortages. That, as a consequence of Nazi
slave-labour, the desperate war situation, and the state of current
technology, no significant automation was applied to its manufacture.
In February 1945 the underground Mittelwerk V2 factory delivered 800
ready-to-launch V2s; after the war U.S. intelligence expert T. P.
Wright estimated that at full production, unconstrained by wartime
shortages, the Mittelwerk plant could have produced 900 to 1000 V2s
One thousand rockets per month...fifty years ago. Think about that.
A Rocket a Day
Suppose we translate these figures, almost incomprehensible by modern
standards (*three hundred* launch vehicles expended in the development
program!) into quasi-modern terms. Consider an orbital launch vehicle
two-stage, say, clean and green thanks to LH2/LOX propulsion in all
stages. Engines: J2 or RL10s or follow-on uprated versions (we'll
have plenty of opportunity to develop them and phase them in). A
simple two stage cylindrical stack like Titan II, with GPS or
ground-commanded navigation. Payload interface is a big ring with
bolt-holes and a standard fairing with plenty of volume inside.
Sounds a lot like NLS/SpaceLifter, doesn't it? STMEs may have
marginal advantages over sea-level-optimised derivatives of RL10 or
J2, but otherwise what's the difference?
What if we launch one every day?
Three hundred and sixty-five a year.
That would be less than one twenty-fifth the production rate of the V2
under concentrated Allied bombardment in 1945.
How much would each one cost?
Assume we expense the development cost or amortise it over a
sufficiently large number of vehicles that it can be ignored.
Further, assume that our bigger, more complicated (two-stage), and
higher tech (LH2/LOX instead of Ethanol/LOX), launcher costs ten times
as much as the V2, and that 1945 wartime dollars convert into current
dollars at 10 to 1. Then, starting with the US$13,000 marginal cost
of a V2, we arrive at a cost of US$1.3 million per launch vehicle. If
we launch one a day our total vehicle budget will be US$475 million
per year--comparable to a single shuttle flight (no, I don't want to
re-open *that* debate again; let's just say it's the same order of
magnitude, OK?). If our mass produced LH2/LOX launcher equals the
performance of the Delta 6925 by placing 3900 kg in LEO, the cost to
LEO is US$333/kg; if we achieve better throw-weight, this figure goes
down accordingly. If we build the thing so cheap, dumb, and heavy
that its payload is only 1000 kg--one metric ton--the cost rises to
US$1300/kg, which is still a factor of ten lower than the comparable
cost to LEO for Ariane, Atlas, Delta, and Titan.
Logistics and Ground Support
Okay, you say, suppose mass production in these absurd quantities
could actually drive the hardware cost down to less than a million and
half per bird, we still haven't accounted for the standing army that
launch operations require. If it takes thousands or tens of thousands
of people to launch tens of vehicles per year, won't it take hundreds
of thousands to launch one every day?
Well, why should it? Again consider the V2. In the two weeks from
September 18-30 1944, a total of 127 V2s were launched from five
different launch sites. That's an average of almost ten a day. This
was accomplished by two mobile groups totaling about 6,300 men and
1600 vehicles, forced to relocate frequently due to the Allied
advance, and subjected to frequent aerial bombardment. It was
estimated that, given adequate supply, one hundred V2s could be
launched per day in a "maximum effort" by the mobile units, and that a
rate of half that, 350 per week, was sustainable.
Parkinson's law notwithstanding, why, after fifty years of
technological progress and experience in launch operations, should it
take tens of thousands of people and hundreds of millions of dollars
to achieve a launch rate one fiftieth that of a V2 group launching the
very first operational ballistic missile from a launch site with tanks
and infantry advancing toward it and airplanes flying over dropping
bombs on them?
Yes, LH2 is trickier to handle; a multistage rocket requires a more
complicated launch and service facility, and so on. But if we design
up-front for a sustained launch rate of one per day, can we not find
ways around these problems? Perhaps a mobile transporter / erector /
launcher like SS24 or Pershing II, with fuel and oxidiser delivered by
underground pipes that attach to the launch truck. Or something....
Let's tell the engineers to go figure it out and see if they come up
with something that works.
It can't be impossible; the Soviet R-7 series launchers (Vostok /
Voskhod / Soyuz) almost furnish an existence proof. These launchers,
despite their mechanical complexity (4 liquid boosters and 20 first
stage engines), are typically launched one to two days after
horizontal delivery to the pad. On several occasions beginning in
1962, two manned launches were made from the same pad less than 24
hours apart. On October 11-13 1969, three manned missions (Soyuz 6,
7, and 8) were launched from the same pad within 48 hours.
If we use contemporary sensors and computers to automate the fueling
and checkout, why does the "launch team" need to be huge? Bob drives
the launcher out to the middle of the circle of concrete, hooks up the
hoses, then goes back to the blockhouse and presses the green "Start"
button. An hour later, or so, the "Ready" light comes on, and at High
Noon he pushes the red "Go" button. Sitting immediately to his right
Fred, in the blue suit, follows the proceedings on a laptop computer
with his index finger on the orange "Oops" button.
Assuming things go OK, ten minutes after the ship lifts, Bob goes out
and drives the launch truck back to the garage where it's reloaded
with the next rocket (assume we have ten trucks, or so, to pipeline
the setup process and account for attrition). Then it's off the
cafeteria for lunch.
Every proposal, prosaic or exotic, for a high-capacity,
fast-turnaround launch system immediately runs into the objection,
"There just aren't enough payloads to make the system pay. Other than
a few established markets for satellites, there just aren't that many
profitable, useful, or interesting things to do in space right now,
and we already have too many launchers chasing too few launch
This is the heart of the chicken-and-egg problem that is blocking the
development and exploration of space.
As long as launches cost tens or hundreds of millions of US$ each,
only governments and the very largest corporations will be able to
afford them, and only for the most obvious and essential purposes,
such as communication, earth resource, navigation, and reconnaissance
satellites. And as long as the number of such payloads is less than a
hundred per year, who is realistically going to pay to develop a
launcher capable of sustained rates many times as great, however cheap
it ends up being? You'd just end up with a huge pile of rockets
gathering dust waiting for payloads, wouldn't you?
Consider the following scenario. The Agency announces a procurement
in which bidders are invited to provide launches, one per day, of 2000
kg or more to a standard Low Earth Orbit, mating with a specified
payload and shroud interface and to a prescribed set of services on a
flat concrete pad. A suitably derated payload is specified for polar
orbit. Bids of more than US$1.25 million per successful launch will
be returned unread. The winner of the bid will be awarded a
fixed-price contract for 1000 launches at the agreed price. The first
100 launches will be considered development flights and will be
purchased at the bid price regardless of success or failure; afterward
only successful launches will be purchased. The procurement will be
re-competed every 1000 launches; if a new vendor wins with a
substantially lower cost per launch, they will be granted the same
development period for the first 100 flights. The vendor retains all
rights to the launcher design and is free to offer it on the open
market independent of the Agency.
Immediately the launch contract awarded, the Agency announces the
availability of daily flights of 2000 kg to LEO or 1500 kg to polar
orbit. Commercial enterprises may purchase launches for whatever
purpose they wish at a price equal to the Agency's cost per launch
plus 25%. Unsold flights are offered on a first-come, first-served
basis to researchers, government agencies, and individuals. In the
event of excess demand, non-commercial proposals will be selected by a
peer review process similar to that used to allocate telescope time at
astronomical observatories. All risks of launch failure are borne by
the provider of the payload; clients should note historical failure
rates and build appropriate spares. Provider of the payload assumes
all liability for it once it separates from Agency's rocket. Payloads
shall be delivered by truck to the loading dock of the Agency's Rocket
Garage. All payloads must be supplied with adequate documentation to
verify their content and safety. The payload interface specification
handbook is available for US$5 from the Agency's toll-free order line;
payload test and integration jigs are available in the Agency's
regional centres and many major universities around the world. Plans
for building your own are available for US$5.
Payloads delivered to the Rocket Garage are inspected to ensure they
are not nuclear bombs, sacks of gravel, or otherwise unacceptable.
Payloads containing propulsion hardware are reviewed especially
closely. Assuming no big no-nos, the payload is bolted to the top of
the next free rocket, the requested orbit inclination is dialed into
the rocket's guidance system, and it moves down the queue toward the
The adventurous will recall that the Project Mercury capsule had a
launch weight of 1935 kg.
If fewer than one payload a day arrives at the Rocket Garage (as is
certain at the outset), the Agency will store the excess rockets in
the Rocket Warehouse out back, while continuing to launch at least one
per week with an inert concrete payload (in a rapidly decaying orbit)
to maintain launch team proficiency and verify the continuing quality
of rockets supplied by the vendor.
This procurement and offering of launch services is explicitly
intended to punch through the chicken-and-egg problem. In essence,
the Agency would be spending US$475 million a year on a flock of 365
hens, then waiting to see if eggs started to show up. This runs the
risk, of course, of ending up with egg all over one's face.
Suppose it isn't possible to build a rocket that will orbit half the
payload of a Delta, launched 50 times less frequently than the V2, at
a cost ten times greater than that primitive fifty year old missile.
In that case nobody responds seriously to the Agency's bid, and the
Agency goes and blows the money on something else, vowing to try again
in ten years.
Now suppose the rockets do start showing up one a day, and departing
on schedule with a success rate that makes the supplier's profit
margin juicy enough to fund further R&D, but the payloads don't
appear. The Agency rapidly becomes the butt of every stand-up comic
and a motion is introduced in the Legislature to re-name it the
"Orbital Ready-Mix Delivery Agency". Well, if that's how it plays
out, I guess we all ought to pack up and go home then, shouldn't we?
Because that would demonstrate, in a real-world test, than there
really aren't very many useful things to do in space, after all. That
even if we push the marginal cost of launches down to zero, nobody
will be able to think of anything to use them for, not for Venus probe
science fair projects, personal spysats, hypersonic surfing
demonstration/validation flights, nor microgravity research, material
processing, life sciences, remote sensing, VLBI radio astronomy,
optical astronomy, or anything else. That other than the existing
big-market space applications, there's no earthly reason to leave the
Earth, that much of the "space age" was based on faulty premises, that
the "final frontier" isn't worth exploring.
Is this likely to be the case?
Naturally, things aren't as easy to accomplish in the real world as
they are to bandy about on paper. Special relativity limits the
velocity with which one can wave one's arms, and the UNDO button
doesn't remove a hole you've just bored the wrong place into an
expensive piece of metal. Many things might go wrong in an attempt to
jump-start the exploitation of space this way. The two real biggies
are discussed above: "it won't work", or "space isn't worth it". Here
are some others I'm concerned about as well.
Range Capacity. Given current low launch rates, configuring a range
is complicated and takes a long time which couldn't accommodate daily
launches, especially to a variety of inclinations. And most existing
spaceports can't handle both equatorial and polar launches. Maybe we
should plan on Hawaii or Cape York from the outset and get the
paperwork started to declare an appropriate air and sea exclusion zone
(for two hours per day around the scheduled launch time). Any rocket
that meets the launch rate and cost criteria cannot require complex or
expensive ground infrastructure.
Environmental Issues. One reason for insisting on LH2/LOX rather than
Kerosene/LOX, hypergolics, or solids/hybrids is that it's clean. We
could launch one every minute and contribute less to global warming,
ozone layer depletion, and other varieties of atmospheric pollution
than 747s crossing the Atlantic every day. Also, exhaust and/or
fluffy white clouds resulting from the occasional really bad day
aren't harmful to anybody who happens to be downwind. On the solid
waste issue, clearly dropping big chunks of aluminum and steel into
the ocean every day isn't a particularly elegant way to break the
bonds of gravity, not compared to all those sleek paper spaceplanes on
the magazine covers. But I suspect if one were to compare the total
mass wasted in expended stages to that of non-recycled aluminum cans
and automobile engines, it would be an insignificant percentage. It's
worth noting that what we're throwing away every day consists
basically of aluminum and iron with a dash of silicon, and that these
are three of the four most abundant elements in the Earth's crust.
Besides, outside the two-hour launch period, salvage boats are welcome
to recover the expended stages and sell them for scrap.
Space Junk. So many launches may run the risk of unacceptably
polluting the near-Earth environment. Clearly, as noted above, care
will be required not to launch payloads likely to explode or otherwise
misbehave in orbit. Payloads will probably have to be released in
orbits which guarantee the timely decay and burn-up of expended upper
stages. We need to make sure the upper stage always burns up
completely, leaving no chunks to go "thump" in the night. Payloads
intended for high-traffic or high-risk final orbits will require
special certification that they will dispose of themselves in a
Fuel cost. It may be that if we succeed in pushing the hardware cost
down, we'll end up with an airline-like situation where fuel cost
becomes a major component of the expense. I don't know how much
liquid hydrogen goes for today, and I haven't tried to predict what
it would cost when purchased in the quantities a launch a day would
require. This needs to be worked out. Even daily launches should be
a minor consumer in the market for liquid oxygen.
Payload pyrotechnic servicing. In the discussion of payload delivery
and integration, I confess to glossing over the issue of pre-launch
payload servicing. You can't just take a satellite with a solid kick
motor and a hundred kilograms of hydrazine on board down to the DHL
counter and ship it to the spaceport. The hazardous aspects of
payload processing must be done in a thoroughly professional manner at
a facility close to the launch site, and the design of these aspects
of payloads must be subjected to design reviews comparable to those
currently used for commercial launches. This increases the payload
cost, but not the launch cost. It will probably promote the emergence
of standard spacecraft buses which provide these components of the
payload, which can be serviced for launch for a flat fee by their
Tracking and control. The daily launch rate envisioned here would
overwhelm existing ground control facilities. Yet the experience of
AMSAT and UOSAT proves that sophisticated and expensive gear isn't
required to manage a satellite, at least in LEO. Without access to
TDRSS or a global tracking network, most satellites are going to have
be very autonomous, communicating with their makers in occasional
high-bandwidth gabfests as they pop above the horizon. Since it's
very likely that one or more manufacturers will offer a standard
satellite bus compatible with the launcher, providing power,
communications, etc., perhaps they will also market access to an
uplink and downlink as a value-added service. From your nearest ISDN
jack or Internet site, you could send and receive packets to your
satellite and let the bus vendor worry about how and when they were
delivered. Deep space missions are a problem; those who propose them
are going to have to obtain time on a big dish as part of their grant
proposal. One hopes that if many missions with clear scientific merit
are proposed, money might be forthcoming to expand the existing deep
space communication facilities.
NASA/Congress will never do it. Who said anything about NASA or the
U.S. Congress? A total budget of US$475 million per year is within
the reach of many industrialised nations, especially at a time when
defence spending is being curtailed, aerospace companies are suffering
from excess capacity, engineering and manufacturing people are
suffering lay-offs, and policy makers worry about how to convert
defence industries without harming readiness by eroding the industrial
base. US$475 million per year represents the following percentage of
the early 1990's defence budgets (CIA World Factbook 1992) of the
Country % Defence Budget
South Africa 13.6%
China 4.0% (approx)
United Kingdom 1.1%
United States 0.15%
Any country whose government became convinced that a scheme like this
might give it a long-term (literal) leg up in the world and beyond,
eventually, could implement it by reprogramming a small percentage of
its existing military spending, much of which would flow right back
into its own industries and economy and might be seen to have military
value it its own right. For that matter, US$475 million is just about
what Microsoft will spend on R&D in fiscal year 1993 and a third of
their pre-tax profit, and it's less than 3% of Motorola's sales for
the same year, so well-heeled and forward-looking companies (or
consortium of such) could play as well.
The near-term development of space is constrained by excessive costs
of launching payloads to low Earth orbit. The development of
innovative launch technologies is discouraged by an apparent over
capacity of existing launchers, "where will the payloads come from?",
while development of payloads for new space applications isn't
affordable given current launch costs.
Rocketry was originally developed as a branch of artillery.
Proponents of various reusable launch technologies argue that as long
as an artillery-like model is maintained, affordable launches will
never be possible. But to be effective, artillery must not only have
adequate throw-weight, it must also provide a rapid rate of fire while
minimising the cost of expended rounds. Today's space launch
"artillery" costs tens to hundreds of millions of US$ per shot and
fires at intervals measured in weeks or months. Yes, expendable
launchers are artillery, and the ones we have today are, as artillery
pieces, extremely overpriced and under-performing.
The last time liquid rockets were truly treated as artillery was the
very first time they were used in war, the A4/V2, fifty years ago.
Despite an increasingly desperate war situation, constant supply
problems, and aerial bombardment, V2s were manufactured at rates of up
to 800 per month, launched at a comparable pace, and produced at a
marginal cost of US$13,000 (1945 dollars) for each additional rocket
after the first 5000.
Making allowances for all the differences between Nazi Germany and the
modern world, between a not very militarily useful nor reliable weapon
and a viable space launcher, between a one-stage Ethanol/LOX missile
and a multistage LH2/LOX launcher, between 1945 wartime dollars and
current currency, still one must ask why, after 50 years of
technological progress and rocket experience, our current rockets cost
not five, not ten, not twenty times as much as a V2, but between one
hundred (Pegasus) and two thousand four hundred (Titan III/SRM) times
as much. Is what a Delta 6925 does, lobbing 3900 kg into LEO,
fundamentally three hundred times more expensive than what a V2 did
fifty years ago?
It is interesting to observe that current launchers are bought and
launched in quantities about a thousand times less than those of the
V2 at peak production. In no sense are they mass-produced, and
therefore they do not benefit from either the means of mass production
(investment in highly-automated manufacturing), nor from the learning
curve that results when one builds hundreds and thousands of an
identical product. Could it be that a large component of the present
unacceptably high launch cost is both cause and effect of the present
low rate of launches?
Kelly Starks Internet: firstname.lastname@example.org
Sr. Systems Engineer
Magnavox Electronic Systems Company
(Magnavox URL: http://www.fw.hac.com/external.html)