16.885 Aircraft Systems Engineering Cost Analysis Karen Willcox MIT Aerospace Computational Design Laboratory AEROSPACE COMPUTATIONAL DESIGN LABORATORY Outline • Lifecycle cost • Operating cost • Development cost • Manufacturing cost • Revenue • Valuation techniques AEROSPACE COMPUTATIONAL DESIGN LABORATORY 9/19/2004 16.885 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. AEROSPACE COMPUTATIONAL DESIGN LABORATORY 9/19/2004 16.885 Lifecycle Cost 100 95% 85% 80 65% 60 40 Disposal Operation and support Impact on LCC (%) 20 design Detailed design Manufacturing and acquisition Preliminary design, system integration Conceptual 0 Time (From Roskam, Figure 2.3) AEROSPACE COMPUTATIONAL DESIGN LABORATORY 9/19/2004 16.885 Operating Cost ¾ Airplane Related Operating Cost SROC (AROC) 10% CROC ¾ Passenger Related Operating Cost 2% (PROC) PROC Cargo Related Operating Cost ¾ 18% AROC (CROC) 70% ¾ Systems Related Operating Cost (SROC) AEROSPACE COMPUTATIONAL DESIGN LABORATORY 9/19/2004 16.885 Airplane Related Operating Costs CAPITAL COSTS: CASH AIRPLANE RELATED Financing OPERATING COSTS: Insurance Capital Crew Depreciation Fuel Costs CAROC Maintenance 40% 60% Landing Ground Handling GPE Depreciation GPE Maintenance Control & Communications CAROC is only 60% - ownership costs are significant! AEROSPACE COMPUTATIONAL DESIGN LABORATORY 9/19/2004 16.885 CAROC Breakdown per Trip Other Control & Comm 3% Ground Fuel is roughly 20% of Handling 9% 60% of 70% of Total 7% Operating Cost Landing Crew i.e. 8% 6% 40% Maintenance 15% Fuel 20% typical data for a large commercial jet AEROSPACE COMPUTATIONAL DESIGN LABORATORY 9/19/2004 16.885 Non-Recurring Cost Cost incurred one time only: Engineering - airframe design/analysis - configuration control - systems engineering Engineering Tooling - design of tools and fixtures - fabrication of tools and fixtures Other Tooling - development support - flight testing Other AEROSPACE COMPUTATIONAL DESIGN LABORATORY 9/19/2004 16.885 Recurring Cost Cost incurred per unit: Labor - fabrication - assembly - integration Labor Material to manufacture -raw material - purchased outside production - purchased equipment Production support Material -QA - production tooling support - engineering support Support AEROSPACE COMPUTATIONAL DESIGN LABORATORY 9/19/2004 16.885 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 Y x Y 0 x Yx = number of hours to produce unit x n = log b/log 2 b = learning curve factor (~80-100%) AEROSPACE COMPUTATIONAL DESIGN LABORATORY 9/19/2004 16.885 Learning Curve 1 0.8 Every time production 0.6 doubles, cost is reduced 0.4 b=0.9 0.55 by a factor Cost of unit unit of Cost of 0.9 0.2 0 0 10 20 30 40 50 Unit number Typical LC slopes: Fab 90%, Assembly 75%, Material 98% AEROSPACE COMPUTATIONAL DESIGN LABORATORY 9/19/2004 16.885 Elements of a Cost Model Engineering Data Build Schedule & Performance 120 lanes lanes 80 40 0 Number of p of Number 2000 2010 2020 2030 Year Recurring Cost Component Breakdown Plane Wing Fuselage COST Winglet Skin Ribs MODEL RC/lb NRC/lb Weight Subparts/lb Non-Recurring Learning Curve NRC Distribution Cost 1 2 0.8 0.4 1 Cost of unit 0 NRC ($B) 0 10 2030 4050 Unit number 0 2003 2007 2011 2015 Year AEROSPACE COMPUTATIONAL DESIGN LABORATORY 9/19/2004 16.885 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. AEROSPACE COMPUTATIONAL DESIGN LABORATORY 9/19/2004 16.885 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 AEROSPACE COMPUTATIONAL DESIGN LABORATORY 9/19/2004 16.885 Cost Modeling Plane Centerbody Wing Landing Propulsion Systems Payloads Final Gear Assembly At this level, the degree of detail can Winglet Outer Inner … range from e.g. “wing” Wing Wing to “rivet”. Identifier Weight Area RC per Subparts NRC per NRC time pound per pound pound distribution Labor Material & Support Tooling Engineering Other Equipment AEROSPACE COMPUTATIONAL DESIGN LABORATORY 9/19/2004 16.885 Development Cost Data Baseline QA Baseline Dev. Labs Baseline QA QA Baseline Dev. Dev. Labs Labs Baseline Tool Tool Fab. Fab Baseline ToolFab.Tool Fab. Baseline Tool Tool Design Design Baseline M.E. M.E. Baseline Engr. Engr.. Baseline Tool Design Baseline M.E. Baseline Engr. non-dimensional labor hours non-dimensional time Boeing data for large commercial jet AEROSPACE COMPUTATIONAL DESIGN LABORATORY 9/19/2004 16.885 Development Cost Model • Cashflow profiles based on beta curve: t c (t ) Kt D 1(1 t )E 1 • Typical development time ~6 years • Learning effects captured – span, cost 0.06 Support 0.05 Tool Fab 0.04 Tool Design ME 0.02 Engineering (from Markish) normalized cost 0.01 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19normalized 20 21 22 23 24 25 26 27 28 29 time 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 AEROSPACE COMPUTATIONAL DESIGN LABORATORY 9/19/2004 16.885 Development Cost Model Payloads 8% Wing 20% Systems 17% Empennage 9% Installed Engines 8% Landing Gear 1% Fuselage 37% Representative non-recurring cost breakdown by parts for large commercial jet (from Markish). AEROSPACE COMPUTATIONAL DESIGN LABORATORY 9/19/2004 16.885 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). Tool Engineering ME 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 AEROSPACE COMPUTATIONAL DESIGN LABORATORY 9/19/2004 16.885 Manufacturing Cost Model • Aircraft built Æ modules required • Modules database – Records quantities, marginal costs – Apply learning curve effect by module, not by aircraft Labor Materials Support 85% 95% 95% time AEROSPACE COMPUTATIONAL DESIGN LABORATORY 9/19/2004 16.885 Manufacturing Cost Model Final Assembly 6% Wing Payloads 27% 11% Systems 6% Installed Engines 9% Empennage Landing Gear 10% 3% Fuselage 28% Representative recurring cost breakdown by parts for large commercial jet (from Markish). AEROSPACE COMPUTATIONAL DESIGN LABORATORY 9/19/2004 16.885 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 $65 AEROSPACE COMPUTATIONAL DESIGN LABORATORY 9/19/2004 16.885 NASA Cost Models Online cost models available at http://www.jsc.nasa.gov/bu2/airframe.html e.g. Airframe Cost Model - simple model for estimating the development and production costs of aircraft airframes - based on military jet data - correlation with empty weight, max. speed, number of flight test vehicles, and production quantity AEROSPACE COMPUTATIONAL DESIGN LABORATORY 9/19/2004 16.885 Revenue Model Revenue model must predict market price and demand quantity. 180 90 160 80 140 767-200ER 70 A300 767-300ER 120 A310 777-200 60 A330 777-300 100 50 A340 747-400 747 80 MD-11 40 A300 767 60 30 A330 777 price ($M) 20 40 A340 deliveries MD-11 10 20 0 0 1980 1985 1990 1995 2000 1985 1990 1995 2000 year year Historical wide body data from Markish. No correlation found between price and quantity. AEROSPACE COMPUTATIONAL DESIGN LABORATORY 9/19/2004 16.885 Aircraft Pricing Cost-Based Pricing Market-Based Pricing Performance Cost + Profit = Price Operating Cost Market Competition Value Passenger Appeal Personal aircraft Commercial transport Business jets? Military aircraft Source: Schaufele AEROSPACE COMPUTATIONAL DESIGN LABORATORY 9/19/2004 16.885 Commercial Aircraft Pricing • Total Airplane Related Operating Costs are fairly CAROC constant. • Aircraft price must balance CAROC. PRICE Total AROC COST/WEIGHT (Capital costs) TRADE-OFF AEROSPACE COMPUTATIONAL DESIGN LABORATORY 9/19/2004 16.885 Business Jet Empirical Data Figure A7 in Roskam: -1 AMP1989 = log {0.6570 + 1.4133 log WTO} AMP1989 is predicted airplane market price in 1989 dollars Take-off weight: 10,000 lb < WTO < 60,000 lb BUT Gulfstream GIV and 737 BJ versions do not fit the linear trend. AEROSPACE COMPUTATIONAL DESIGN LABORATORY 9/19/2004 16.885 Commercial Jet Empirical Data Figure A9 in Roskam: -1 AMP1989 = log {3.3191+ 0.8043 log WTO} AMP1989 is predicted airplane market price in 1989 dollars Take-off weight: 60,000 lb < WTO < 1,000,000 lb AEROSPACE COMPUTATIONAL DESIGN LABORATORY 9/19/2004 16.885 Military Aircraft Empirical Data Figure A10 in Roskam: -1 AMP1989 = log {2.3341+ 1.0586 log WTO} AMP1989 is predicted airplane market price in 1989 dollars Take-off weight: 2,500 lb < WTO < 1,000,000 lb AEROSPACE COMPUTATIONAL DESIGN LABORATORY 9/19/2004 16.885 Revenue Model: Price • Assumption: market price based on 1.