Date: 17 Jul 91 23:44:55 GMT
From: ssc-vax!bcsaic!hsvaic!eder@beaver.cs.washington.edu  (Dani Eder)
Subject: Re: Cost estimates

In article <33007@hydra.gatech.EDU> ccoprmd@prism.gatech.EDU (Matthew DeLuca) writes:
>Where did I say it was 'not possible'?  I did claim the numbers are pure
>speculation, and the same is true of Boeing.  Until those planes roll out
>the door, nobody knows how much they are going to cost.  

I beg to differ.  I can assure you the people developing the 777 knows
fairly closely how much the airplane is going to cost.  They have to
be able to tell the company board of directors (a) how much we can
expect to sell the planes for, (b) how much we expect to spend on
designing the airplane and building the production line, (c) how much
each airplane will cost to build.  (c) must be sufficiently below (a)
to recoup (b) in a reasonable number of units (300 is the typical
number).  All of this information is required before the board gives
permission to go ahead with the airplane.

(a) is estimated by market research, by comparison with sales of other
aircraft, and by actual sales of the 777 to 'launch customers'.
(we actually had about 50 airplanes on order before we committed
to build the plane)
(b) is estimated by experience with previous airplane programs
(c) is estimated by (1) getting suppliers lined up for parts of the
airplane, here costs are well known as they are quotations, and (2)
for in-house manufacturing, by comparison with other airplanes.

In my judgement, the uncertainties in the estimates are (a) 5%,
(b) 30%, (c) 5%.  So while even Boeing can't predict EXACTLY what
product will cost, we can estimate it closely enough to say with
a high degree of confidence that we will make a profit on each
unit, and have a reasonable degree of confidence of making back
the development costs over time.





-- 
Dani Eder/Boeing/Advanced Civil Space/(205)464-2697(w)/461-7801(h)/#905, 1075
Dockside Dr.,Huntsville,AL35824/Member: Space Studies Institute
Physical Location: 34deg 37' N 86deg 43' W +100m alt.
***THE ABOVE IS NOT THE OPINION OF THE BOEING COMPANY OR ITS MANAGEMENT.***

Date: 6 Jan 92 20:23:03 GMT
From: ssc-bee!bcsaic!hsvaic!eder@beaver.cs.washington.edu  (Dani Eder)
Subject: Re: costly colonies

leech@cs.unc.edu (Jonathan Leech) writes:

>    Since going by the Boeing field on the way to the Seattle SDC and
>seeing several hundred million $ of new airplanes, I've wondered how
>much a 747 would cost to fly had there been only 5 or 6 of them...
>perhaps SR-71 costs would give a clue, as a small number of those were
>built.
>--

I can speak to this question.  Our in-house cost experience says that
the per-unit cost of making a one-of-a-kind aerospace item (like an
airplane) is 5x higher than the cost of the 1000th unit of a
production run (which, by the way, is about how many 747s have
been ordered over the life of the program).  This is empirical
data with quite sound reasons behind it.

For example: we have an enormous milling machine dedicated to
milling wing parts.  We have a big autmated riveter to rivet
wing parts together.  Since Boeing makes a lot of airplanes, it
was economic for us to install computer-controlled machines to
do these things, and reduce manual labor.  When we ship parts,
we have special reuseable containers and carriers.  For example,
the front 40 feet of the 737 is shipped from Wichita Kansas
to Renton Washington on a special rail car.  Again, it is economic
to invest in reuseable containers when you are doing something
many times.  For a one-off design, you would not bother.

It is considerably cheaper to make a batch of identical parts
than to make one of them.  Consider graphite/epoxy moldings that
 are used in some parts of the airplane structure.  The material
comes in fabric layers that are laid up in a special mold, which
is then put in an oven and cured under pressure to make sure
there are no voids in the final part.  Making that mold for
one part makes the cost higher then reusing it for many parts.

Another part of the cost equation is that it typically costs
ten times as much to design and develop aerospace hardware
as to produce the first unit.  If you build one unit, then
you have to pay (say) $5 billion for development plus $500
million for the first airplane.  In reality, we have perhaps
spent twice that over the life of the 747 program (because
it has gone through several upgrades, and there is both
passenger and cargo versions), say $10 billion, but the
AVERAGE unit cost (not 1000th unit cost), is $120 million
and the development cost spread over the 1000 units comes
to $10 million per unit.  Thus we have an actual $130
million per 747 (the current price), versus a hypothetical
$5.5 billion if you wanted just one.  By the way, I don't have
access to the real historical cost data on the 747, but I
worked backwards from the price and knowing typical cost trends.

 
-- 
Dani Eder/Boeing/Advanced Civil Space/(205)464-2697(w)/232-7467(h)/
Rt.1, Box 188-2, Athens AL 35611/Member: Space Studies Institute
Physical Location: 34deg 37' N 86deg 43' W +100m alt.
***THE ABOVE IS NOT THE OPINION OF THE BOEING COMPANY OR ITS MANAGEMENT.***

Date: 28 Aug 92 14:26:19 GMT
From: eder@hsvaic.boeing.com
Subject: Sizing of launch vehicles (was Saturn Class)
Newsgroups: sci.space

jbh55289@uxa.cso.uiuc.edu (Josh 'K' Hopkins) writes:

>Allen said...
>is), how do you plan to use an HLV?   The baseline plans I've seen put Ariane 5
>at about the right size, and if you increase the payload, you decrease the 
>number of flights to the point where I begin to wonder if it's commercially
>viable.  Maybe we need to agree on a definition of Heavy Lift.


Okay, time for a lesson in the economics of launch vehicles:

The Air Force performed a study in 1989 for the Secretary of Defense
on launch systems for SDI.  The study was executed by the Space Systems
Division of the AF Systems Command.  These are the folks who procure
rockets for the AF.  In the study they surveyed all existing and
proposed domestic launchers, and evaluated their costs at different
flight rates.

I plotted the data at high flight rates and found that the cost per
kilogram ran as $8,000 x (mass in tons)^-0.3.  Thus, as the launch
vehicle got larger, the cost per kilogram goes down, approximately
dropping by half for each factor of ten in payload increase.

Now, this is purely recurring costs, and does not include development.
In 1986 I was the cost engineer for an earlier space transportation
study, funded jointly by NASA and the AF.  We found that for a clean
sheet expendable rocket, the development cost ran ( in today's dollars)
as $1.5 million x (liftoff weight)^0.553.  In other words, the development
cost ran roughly as the square root of size.  The payload relationship
ran Payload wt = -10,427lb + (0.0436 x Gross liftoff weight).  So,
for particular sizes we have:

Payload (lb)	Liftoff weight (lb)	Development cost ($)

10,000		468,500			$2.05 billion
20,000		697,900			$2.55 billion
30,000		927,200			$2.99 billion
40,000		1,156,600		$3.38 billion
50,000		1,386,000		$3.74 billion

Now, for the space station delivery job alone, you have a need to
deliver 75 tons per year (5 Shuttle loads), for say 10 years.  Thus
you have a total delivery requirement of 750 tons, or 750,000 kg.
To find total cost, we need to add recurring and development costs,
with the development cost spread over the launched mass:

Payload (kg)	Launch $/kg	Develop $/kg	Total $/kg
4,536		5,082		2,733		7,816
9,072		4,128		3,400		7,528
13,608		3,655		3,987		7,642
18,144		3,353		4,506		7,860
22,680		3,136		4,987		8,122

So, considering the Space Station delivery job only, the optimum
size seems to be around 10 tons payload.  Now, if you consider
other launch jobs that exist, and look at longer time horizons,
you tend to get larger vehicles as the optimum, but you are a
long way from HLLV numbers unless you have massive traffic
requirements.

Dani Eder

-- 
Dani Eder/Boeing/Advanced Civil Space/(205)464-2697(w)/232-7467(h)/
Rt.1, Box 188-2, Athens AL 35611/Member: Space Studies Institute
Physical Location: 34deg 37' N 86deg 43' W +100m alt.
***THE ABOVE IS NOT THE OPINION OF THE BOEING COMPANY OR ITS MANAGEMENT.***

Date: 1 Sep 92 18:34:40 GMT
From: Dani Eder <eder@hsvaic.boeing.com>
Subject: Sizing of launch vehicles (was Saturn Class)
Newsgroups: sci.space

fcrary@ocf.berkeley.edu (Frank Crary) writes:

>In article <1579@hsvaic.boeing.com> eder@hsvaic.boeing.com (Dani Eder) writes:

>>Now, for the space station delivery job alone, you have a need to
>>deliver 75 tons per year (5 Shuttle loads), for say 10 years.  Thus
>>you have a total delivery requirement of 750 tons, or 750,000 kg.
>>To find total cost, we need to add recurring and development costs,
>>with the development cost spread over the launched mass:

>This neglects  the use of pre-existing systems (e.g. ones where the
>development costs have already been paid for.)

>But more importantly, you are treating the Space Station as a constant-mass
>payload, which need only be delivered to orbit. Assembling the pieces
>will also be required, and clearly assembly of 75 10-tonne pieces is
>much harder than assembly of (say) 15 50-tonne pieces. 

>I suspect the difference in costs (which was only around a 10% difference
>between 10 and 25 tonne payloads) would be overshadowed by the changes
>in on orbit assembly costs.

I'm sorry I wasn't clear enough in my original posting.  The mass of the
Space Station at assembly complete is about 200 tons.  The rest of the
mass represents supplies, experiments, and orbit maintenance propellant.
The five flights per year is the figure being planned for by the space
station logistics people.

My intent in the posting was was to show that the development cost has
to be considered in any launch vehicle selection.  At the traffic rates
for the space station support job, the optimal new vehicle size is
pretty small.  Note that in the 1986 study I worked on, we considered
F1 and F1-A engines as booster engines.  They turned out to be un-economic
for the job we were designing for - launching a combined SDI and civilian
traffic model.  The actual answer we got was that for mostly eastern
launches you actually want a fully reuseable two stage shuttle, sized
so it flies about 20-25 times per year (take your total annual traffic
and divide by 20-25, and that gives you the payload size).  This rule
holds across a range of vehicle sizes for this reason:

As you go to larger and larger rockets, you still have to develop them,
build them and launch them.  The test facilities, assembly buildings,
and launch pads all get bigger the larger the rocket gets, and thus cost
more.  You want to spread these fixed costs over a large number of 
launches to minimze them.  On the other hand, launch operations costs
tend to go with the number of launches, not the size of the vehicle.
The point where the incremental facilities cost balances the incremental
operations cost comes out to 200-250 launches.  If you project a 10
year operating life, then you size for 20-25 launches per year.

Another reason for getting to rates of 2 a month is production efficiency.
The 737 production line takes 20-30 days to run the aircraft from delivery
of sections (fuselage sections, wings, landing gear), through installation
of systems (engines, avionics), to rolling a near-complete aircraft out
the door (it can fly, but painting, and customer-specific items get put
in elsewhere).  During this time the aiplane moves through several 
'stations' in the assembly line, spending a few days at each station.

So, the assembly stations, which are all needed, have natural rates
of several a month.  If you build at a much lower rate, like 6 times
a year, your factory is not being used efficiently.  I maintain that
a rocket is inherently simpler than an airplane to build, and therefore
the efficient factory production rate will be even higher.

I agree with your comment about simplicity of assembly.  It is even
stronger than you state. A Station built of larger pieces actually 
will mass less than one built of smaller pieces, since you eliminate
the connectors (electrical, hatches, etc) between sections.  This tends
to make larger launch vehicles look more attractive than otherwise.
To my knowlege, no one has ever taken a combined look at Space Station
and launch vehicle costs to look for a global optimum (after all, the
programs are under different associate administrators).  The Space
Station has mandated from the beginning that they will use the Space
Shuttle for delivery, so the question never was asked.

Dani Eder

-- 
Dani Eder/Boeing/Advanced Civil Space/(205)464-2697(w)/232-7467(h)/
Rt.1, Box 188-2, Athens AL 35611/Member: Space Studies Institute
Physical Location: 34deg 37' N 86deg 43' W +100m alt.
***THE ABOVE IS NOT THE OPINION OF THE BOEING COMPANY OR ITS MANAGEMENT.***