16.885
Aircraft Systems
Engineering
Cost Analysis
Karen Willcox
MIT
Aerospace Computational Design
Laboratory
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Outline
? Lifecycle cost
? Operating cost
? Development cost
? Manufacturing cost
? Revenue
? Valuation techniques
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Lifecycle Cost
Lifecycle :
Design - Manufacture - Operation - Disposal
Lifecycle cost :
Total cost of program over lifecycle
85% of Total LCC is locked in by the end of
preliminary design.
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Lifecycle Cost
0
20
40
60
80
100
65%
Conceptual
design
Preliminary design, system integration
Detailed design
Manufacturing and acquisition
Operation and support
Disposal
Time
Impact on LCC (%)
85%
95%
(From Roskam, Figure 2.3)
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Operating Cost
? Airplane Related Operating Cost
(AROC)
? Passenger Related Operating Cost
(PROC)
? Cargo Related Operating Cost
(CROC)
? Systems Related Operating Cost
(SROC)
AROC
70%
PROC
18%
CROC
2%
SROC
10%
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Airplane Related Operating
Costs
CASH AIRPLANE RELATED
OPERATING COSTS:
Crew
Fuel
Maintenance
Landing
Ground Handling
GPE Depreciation
GPE Maintenance
Control & Communications
CAROC
60%
40%
Capital
Costs
CAPITAL COSTS:
Financing
Insurance
Depreciation
CAROC is only 60% - ownership costs are significant!
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
CAROC Breakdown per Trip
Landing
6%
Ground
Handling
7%
Control
& Comm
9%
Other
3%
Fuel
20%
Crew
40%
Maintenance
15%
Fuel is roughly 20% of
60% of 70% of Total
Operating Cost
i.e. 8%
typical data for a large commercial jet
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Non-Recurring Cost
Cost incurred one time only:
Engineering
- airframe design/analysis
- configuration control
- systems engineering
Tooling
- design of tools and fixtures
- fabrication of tools and fixtures
Other
- development support
- flight testing
Engineering
Tooling
Other
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Recurring Cost
Cost incurred per unit:
Labor
- fabrication
- assembly
- integration
Material to manufacture
-raw material
- purchased outside production
- purchased equipment
Production support
-QA
- production tooling support
- engineering support
Labor
Material
Support
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Learning Curve
As more units are made, the recurring cost per
unit decreases.
This is the learning curve effect.
e.g. Fabrication is done more quickly, less
material is wasted.
n
x
xYY
0
Y
x
= number of hours to produce unit x
n = log b/log 2
b = learning curve factor (~80-100%)
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Learning Curve
0.55
0
0.2
0.4
0.6
0.8
1
1020304050
Unit number
C
o
s
t
of
uni
t
b=0.9
Every time
production
doubles, cost
is reduced
by a factor
of 0.9
Typical LC slopes: Fab 90%, Assembly 75%, Material 98%
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Elements of a Cost Model
Non-Recurring
Cost
Recurring
Cost
COST
MODEL
0
40
80
120
2000 2010 2020 2030
Year
N
u
m
b
e
r
o
f
p
l
anes
Build Schedule
Plane
Wing
Winglet RibsSkin
Fuselage
Weight
RC/lb
Subparts/lb
NRC/lb
Component Breakdown
0
0.4
0.8
1
0 1020304050
Unit number
Cost
of
u
n
it
Learning Curve
Engineering Data
& Performance
0
1
2
2003 2007 2011 2015
Year
NRC ($B)
NRC Distribution
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Typical Cost Modeling
1. Take empirical data from past programs.
2. Perform regression to get variation with
selected parameters, e.g. cost vs. weight.
3. Apply “judgment factors” for your case.
e.g. configuration factors, complexity
factors, composite factors.
There is widespread belief that aircraft
manufacturers do not know what it actually
costs to turn out their current products.
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Cost Modeling
? Aircraft is broken down into modules
– Inner wing, outer wing, …
– Modules are classified by type
? Wing, Empennage, Fuselage, …
? Cost per pound specified for each module type
– Calibrated from existing cost models
– Modified by other factors
? Learning effects
? Commonality effects
? Assembly & Integration: a separate “module”
? 2 cost categories: development & manufacturing
Production run: a collection of modules
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Cost Modeling
Plane
Centerbody
Landing
Gear
Propulsion
Systems
Final
Assembly
Payloads
Winglet
Outer
Wing
Inner
Wing
Wing
…
WeightIdentifier
RC per
pound
Subparts
per pound
Area
Labor
Material &
Equipment
Support
At this level, the
degree of detail can
range from e.g. “wing”
to “rivet”.
NRC per
pound
Tooling
Engineering Other
NRC time
distribution
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Development Cost Data
Baseline Engr.
Baseline M.E.
Baseline Tool Design
Baseline ToolFab.
Baseline Dev. Labs
Baseline QA
Baseline QA
Baseline Dev. Labs
Baseline Tool Fab.
Baseline Tool Design
Baseline M.E.
Baseline Engr.
Baseline QA
Baseline Dev. Labs
Baseline Tool Fab
Baseline Tool Design
Baseline M.E.
Baseline Engr.
non-dimensional labor hours
non-dimensional time
Boeing data for large commercial jet
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Development Cost Model
? Cashflow profiles based on beta curve:
? Typical development time ~6 years
? Learning effects captured – span, cost
11
)1()(
E D
tKttc
0
0.01
0.02
0.04
0.05
0.06
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53
normalized time
Support
Tool Fab
Tool Design
ME
Engineering
normalized cost
(from Markish)
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Development Cost Model
Wing
20%
Empennage
9%
Fuselage
37%
Landing Gear
1%
Installed Engines
8%
Systems
17%
Payloads
8%
Representative non-recurring cost breakdown by parts for
large commercial jet (from Markish).
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Development Cost Data
For your reference: $/lb assembled from public domain
weight and total cost estimates plus representative NRC
breakdown by aircraft part (see Markish).
Engineering ME
Tool
Design Tool Fab Support Totals
40.0% 10.0% 10.5% 34.8% 4.7% 100.0%
Wing $7,093 $1,773 $1,862 $6,171 $833 $17,731
Empennage $20,862 $5,216 $5,476 $18,150 $2,451 $52,156
Fuselage $12,837 $3,209 $3,370 $11,169 $1,508 $32,093
Landing Gear $999 $250 $262 $869 $117 $2,499
Installed Engines $3,477 $869 $913 $3,025 $408 $8,691
Systems $13,723 $3,431 $3,602 $11,939 $1,612 $34,307
Payloads $4,305 $1,076 $1,130 $3,746 $506 $10,763
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Manufacturing Cost Model
? Aircraft built ? modules required
? Modules database
– Records quantities, marginal costs
– Apply learning curve effect by module, not by aircraft
Labor Materials Support
85% 95% 95%
time
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Manufacturing Cost Model
Representative recurring cost breakdown by parts for large
commercial jet (from Markish).
Wing
27%
Empennage
10%
Fuselage
28%
Landing Gear
3%
Installed Engines
9%
Systems
6%
Payloads
11%
Final Assembly
6%
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Manufacturing Cost Data
For your reference: $/lb values assembled from public
domain weight and total cost estimates plus representative
RC breakdown by aircraft part (see Markish).
Labor Materials Other Total
Wing $609 $204 $88 $900
Empennage $1,614 $484 $233 $2,331
Fuselage $679 $190 $98 $967
Landing Gear $107 $98 $16 $221
Installed Engines $248 $91 $36 $374
Systems $315 $91 $46 $452
Payloads $405 $100 $59 $564
Final Assembly $58 $4 $3 $65
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
NASA Cost Models
Online cost models available at
http://www.jsc.nasa.gov/bu2/airframe.html
e.g. Airframe Cost Model
- simple model for estimating the development and
production costs of aircraft airframes
- based on military jet data
- correlation with empty weight, max. speed, number of
flight test vehicles, and production quantity
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Revenue Model
Revenue model must predict market price
and demand quantity.
0
10
20
30
40
50
60
70
80
90
1980 1985 1990 1995 2000
year
deliveries
A300
A310
A330
A340
747
767
777
MD-11
0
20
40
60
80
100
120
140
160
180
1985 1990 1995 2000
year
price ($M)
767-200ER
767-300ER
777-200
777-300
747-400
MD-11
A300
A330
A340
Historical wide body data from Markish. No correlation found
between price and quantity.
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Aircraft Pricing
Personal aircraft
Business jets?
Military aircraft
Cost-Based Pricing Market-Based Pricing
Cost + Profit = Price
Performance
Operating Cost
Competition
Passenger Appeal
Commercial transport
Market
Value
Source: Schaufele
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Commercial Aircraft Pricing
? Total Airplane Related
Operating Costs are fairly
constant.
? Aircraft price must
balance CAROC.
COST/WEIGHT
TRADE-OFF
CAROC
PRICE
(Capital costs)
Total AROC
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Business Jet Empirical Data
Figure A7 in Roskam:
AMP
1989
= log
-1
{0.6570 + 1.4133 log W
TO
}
AMP
1989
is predicted airplane market price in 1989
dollars
Take-off weight: 10,000 lb < W
TO
< 60,000 lb
BUT Gulfstream GIV and 737 BJ versions do not fit
the linear trend.
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Commercial Jet Empirical
Data
Figure A9 in Roskam:
AMP
1989
= log
-1
{3.3191+ 0.8043 log W
TO
}
AMP
1989
is predicted airplane market price in 1989
dollars
Take-off weight: 60,000 lb < W
TO
< 1,000,000 lb
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Military Aircraft Empirical
Data
Figure A10 in Roskam:
AMP
1989
= log
-1
{2.3341+ 1.0586 log W
TO
}
AMP
1989
is predicted airplane market price in 1989
dollars
Take-off weight: 2,500 lb < W
TO
< 1,000,000 lb
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Revenue Model: Price
? Assumption: market price based on
1. Range
2. Payload
3. Cash Airplane-Related Operating Cost (CAROC)
? Regression model:
? Note that speed does not appear. No significant
statistical relationship between price and speed
was found in available data.
)()()(
21
CAROCfRangekSeatskP
D
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Revenue Model: Price
0
10
20
30
40
50
60
70
80
0 1020304050607080
Actual price ($M)
E
s
tim
a
te
d
p
r
i
c
e
($
M
)
y=x
Airbus
Boeing
Narrow bodies
Estimated price ($M)
Actual price ($M)
Narrow bodies:
1.91
0.735( ) 0.427( ) ( )P Seats Range f CAROC
Model from Markish, price data from Aircraft Value News, The
Airline Monitor, 2001.
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Revenue Model: Price
Wide bodies:
2.76
0.508( ) 0.697( ) ( )P Seats Range f CAROC
0
20
40
60
80
100
120
140
160
0 20 40 60 80 100 120 140 160
Actual price ($M)
E
s
tim
a
te
d
p
r
ic
e
($
M
)
y=x
Airbus
Boeing
Wide bodies
Estimated price ($M)
Actual price ($M)
Model from Markish, price data from Aircraft Value News, The
Airline Monitor, 2001.
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Revenue Model: Quantity
? Demand forecasts
– 3 sources: Airbus; Boeing; Airline Monitor
– Expected deliveries over 20 years
– Arranged by airplane seat category
? Given a new aircraft design:
– Assign to a
seat category
– Assume a
market share
– Demand forecast ?
20-year production
potential
0
500
1000
1500
2000
2500
3000
3500
4000
100 125 150 175+ 200 250 300 350 400 500+
Seat Category
Q
u
a
n
tity
Airbus
Airline Monitor
Boeing
Seat Category
Quantity
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Revenue Model: Dynamics
? Expected aircraft deliveries: forecasted
? Actual deliveries: unpredictable
? Observe historical trends: growth rate, volatility
y = 94.571e
0.0228x
R
2
= 0.3724
y = 52.776e
0.0167x
R
2
= 0.4092
0
50
100
150
200
250
300
350
400
1 4 7
10 13 16 19 22 25 28 31 34 37 40
6-mo. period
uni
t
s
narrow
wide
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Valuation Techniques
The top 5 investor questions:
? How much will I need to invest?
? How much will I get back?
? When will I get my money back?
? How much is this going to cost me?
? How are you handling risk & uncertainty?
Investment Criteria
? Net present value
? Payback
? Discounted payback
? Internal rate of return
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Net Present Value (NPV)
? Measure of present value of various cash flows in different
periods in the future
? Cash flow in any given period discounted by the value of a
dollar today at that point in the future
– “Time is money”
– A dollar tomorrow is worth less today since if properly
invested, a dollar today would be worth more tomorrow
? Rate at which future cash flows are discounted is
determined by the “discount rate” or “hurdle rate”
– Discount rate is equal to the amount of interest the
investor could earn in a single time period (usually a
year) if s/he were to invest in a “safer” investment
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Calculating NPV
? Forecast the cash flows of the project over Its
economic life
– Treat investments as negative cash flow
? Determine the appropriate opportunity cost of capital
(i.e. determine the discount rate r)
? Use opportunity cost of capital to discount the future
cash flow of the project
? Sum the discounted cash flows to get the net present
value (NPV)
NPV C
0
C
1
1 r
C
2
1 r
2
!
C
T
1 r
T
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
NPV example
Period Discount Factor Cash Flow Present Value
0 1 -150,000 -150,000
1 0.935 -100,000 -93,500
2 0.873 +300000 +261,000
Discount rate = 7% NPV = $18,400
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Discount Rate
? One of the problems with NPV: what discount rate
should we use?
? The discount rate is often used to reflect the risk
associated with a project:
the riskier the project, use a higher discount rate
? Typical discount rates for commercial aircraft programs:
12-20%
? The higher the discount rate, the more it does not
matter what you do in the future...
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Payback Period
? How long it takes before entire initial investment is
recovered through revenue
? Insensitive to time value of money, i.e. no discounting
? Gives equal weight to cash flows before cut-off date & no
weight to cash flows after cut-off date
? Cannot distinguish between projects with different NPV
? Difficult to decide on appropriate cut-off date
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Discounted payback
? Payback criterion modified to account for the time
value of money
– Cash flows before cut-off date are discounted
? Surmounts objection that equal weight is given to all
flows before cut-off date
? Cash flows after cut-off date still not given any weight
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Internal rate of return (IRR)
? Investment criterion is “rate of return must be greater
than the opportunity cost of capital”
? Internal rate of return is equal to the discount rate for
which the NPV is equal to zero
? IRR solution is not unique
– Multiple rates of return for same project
? IRR doesn’t always correlate with NPV
– NPV does not always decrease as discount rate
increases
NPV C
0
C
1
1 IRR
C
2
1 IRR
2
!
C
T
1 IRR
T
0
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Decision Tree Analysis
? NPV analysis with different future scenarios
? Weighted by probability of event occurring
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Dynamic Programming
? A way of including uncertainty and flexibility in the
program valuation
? Key features:
? Certain aspects of the system may be uncertain, e.g. the
demand quantity for a given aircraft = UNCERTAINTY
? In reality, the decision-maker (aircraft manufacturer) has
the ability to make decisions in real-time according to
how the uncertainty evolves = FLEXIBILITY
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Dynamic Programming:
Problem Formulation
? The firm:
– Portfolio of designs
– Sequential development phases
– Decision making
? The market:
– Sale price is steady
– Quantity demanded is unpredictable
– Units built = units demanded
? Problem objective:
– Which aircraft to design?
– Which aircraft to produce?
– When?
9/19/2002 16.885
AEROSPACE COMPUTATIONAL DESIGN LABORATORY
Dynamic Programming:
Problem Elements
1. State variables s
t
2. Control variables u
t
3. Randomness
4. Profit function
5. Dynamics
? Solution:
? Solve iteratively.
>