Effects of alternative wing structural concepts on high speed civil transport life cycle costs

AIAA-96-1381; 37th AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Conference, Salt Lake City, UT, April 15-17, 1996.

1American Institute of Aeronautics and Astronautics

EFFECTS OF ALTERNATIVE WING STRUCTURAL CONCEPTS ONHIGH SPEED CIVIL TRANSPORT LIFE CYCLE COSTS

William J. Marx* Dimitri N. Mavris† Daniel P. Schrage‡

Aerospace Systems Design LaboratorySchool of Aerospace EngineeringGeorgia Institute of Technology

Atlanta, GA 30332-0150

Abstract

An integrated design and manufacturing approachallows economic decisions to be made that reflect theentire system design as a whole. In order to achievethis objective, integrated cost and engineering modelswere developed and utilized within a focused designperspective. A framework for the integrated product andprocess design of an aircraft system with a combinedperformance and economic perspective is described inthis paper. This framework is based on the concept ofDesign Justification using a Design for Economicsapproach. A production cost model is developed that issensitive to material and product/process selections forthe wing structure. The use of cost/time analyses isdescribed and applied for evaluation of process trades atthe sub-component level of design. Results of anIntegrated Product/Process Development (IPPD) casestudy are presented for potential High Speed CivilTransport (HSCT) wing structural concepts. Costversus performance studies indicate that an aircraft witha hybrid wing structural concept, though moreexpensive to manufacture than some homogeneousconcepts, can have lower direct operating costs due to alower take-off gross weight and less mission fuelrequired.

List of Symbols

x production unit number

γ x xth unit cost

Q0 a pre-specified production quantity

ν learning curve slope

ω weight sizing curve slope

* Ph.D. Candidate, School of Aerospace Engineering.NASA LaRC GSRP Fellow. Student member AIAA.email: [email protected]://www.cad.gatech.edu/asdl/bill

† Manager, ASDL. Assistant Professor, Member AIAA. ‡ Co-Director, ASDL. Professor, Member AIAA.

Copyright © 1996 by the American Institute ofAeronautics and Astronautics, Inc. All rights reserved.

w material weight

w0 a pre-specified finished-material weight

ρl labor rate

f average annual percentage inflation rate

y0 base year

y current year

σm raw material cost

φm material buy-to-fly ratio

ψm material burden rate

γ xl xth unit labor cost

λ xm xth unit material cost

γ cxl cumulative labor cost through xth unit

λcxm cumulative material cost through xth unit

ξl generalized labor theoretical first unit cost

ξm generalized mat'l theoretical first unit cost

xb learning curve break-point unit

Introduction

Product designs can be evaluated with several wellknown and common metrics. Aircraft take-off grossweight, component weights, wing tip deflections, andflutter boundaries are common examples of productmetrics used to evaluate alternative structural designconcepts. The integration of manufacturing knowledgeinto the design process is a technical challenge thatimplicitly proposes a new problem: What metricsshould be used to evaluate designs in terms of theirprocess characteristics? The most logical process metricis cost. The term "cost" implies many concepts, eachof which may be best suited to evaluate a particular partof a production model.

Aerospace manufacturers today are searching fortechniques to gain a sustainable, competitive advantagein the global marketplace. As methods are beingdeveloped for integrating design and manufacturing, the



area itself is broadening in scope. Several disciplines inboth the fields of design and manufacturing separatelybecome involved with any integration development.Many techniques and systems have been developed forproviding producibility guidance to the designer,1-6 fordetermining the cost implications of structural designdecisions earlier in the design process,7-9 and fordesign-to-cost applications.10-14 Current efforts arefocusing on making cost an independent design variablein the design and optimization process itself.15

The idea of designing an aircraft with substantialearly regard to manufacturing cost is a relatively newconcept. Most systems designs of the past have beendesigned with performance being the main objectivefunction; however, this is no longer possible in today'smarketplace. The emphasis on lowering all aspects ofLife Cycle Costs (LCC) is leading to the developmentof new techniques to address LCC early in the designprocess. The use of cost models concurrently withdesign models brings with it many technical challenges,including the issues of data fidelity and the mappingbetween product and process design variables.

Approach

Traditional justification procedures often fail tocapture the relevant information and factors necessary tojustify investment in a particular technology forproduction. An approach developed by Noble &Tanchoco16 proposes the consideration of the systemjustification concurrently with its design. With such anapproach, it should possible to design a system and

justify its development and implementation. Thetechnique was originally proposed for the design ofmanufacturing systems, but is applied to the integrateddesign and manufacturing of an aircraft majorcomponent for this research.

In general, the economic justification of a design isthe determination of whether a particular action ordecision is justifiable based upon its economicconsequences. Design Justification is a term used todescribe a design process in which the economicramifications of design decisions are consideredconcurrently with design development and are used toguide the design process to result in an economicallyfeasible solution that meets the [mission and]performance requirements of the design.17

Addressing cost concurrently with the performancecharacteristics of the design can alleviate a commonproblem for economic justification. In theimplementation used for this research, the mapping ofsystem cost variables to performance design variables isnot done by the cost analyst after the design iscompleted. Instead, the use of integrated engineeringand cost models allows the cost data to be generated andanalyzed as the physical system is being designed. Themapping occurs through the analysis code integration.

Typically, design economics are not evaluated untila design concept is relatively mature. At this point,most of the design's incurred costs have been fixed. TheDesign Justification process requires simultaneousperformance and economic evaluations for the aircraftsystems, as illustrated in Figure 1.

DevelopSpecifications

and Constraints

Knowledge GuidedSynthesis of Design

Concurrent Evaluationof Design Performance

and Economics

Design ModificationStrategy

Figure 1 Design Justification Process16

Design Justification is characterized by fourconcepts: it is simultaneous, goal-directed, decision-maker centered, and knowledge-guided.16 These fourconcepts separate the approach from the traditional,sequential approach that has been used to design mostexisting aerospace systems. The four elements arebriefly described in the following sections. The goal-directed nature of the Design Justification process andits relation to Integrated Product/Process Development

(IPPD) is described in more detail because of itsimportance for structuring the concept evaluation tradestudies.

Simultaneous Design EffortIntegrated engineering and cost models are required

for Design Justification . The simultaneousconsideration of the performance (or product design)characteristics of a design and its associated cost (or



process design) characteristics allows D e s i g nJustification as the design matures. Detailed costanalyses conducted during the design process allow thekey cost drivers to be discovered and addressed before thedesign matures to a point beyond which re-design costsare too expensive to be implemented.

Goal-Directed DesignA goal-directed design process is one in which the

goals or objectives of the design are specified, and canbe modified, by the designer and are used during thedesign process to direct the analysis. At phases duringthe design process, an analysis of the possible benefitsfrom variables that could improve performance andeconomic characteristics must be conducted. Thesimultaneous consideration of the design's product andprocess design characteristics during D e s i g nJustification is known as IPPD. Figure 2a shows aconceptual IPPD framework for aerospace systemsdesign.

Integrated Product and Process DevelopmentFigure 2a illustrates the decomposition activities

from conceptual design (system level) to preliminarydesign (component level) to the detail design (part level)and manufacturing process level in a clockwise manneron the outer circle. Continuing clockwise leads to therecomposition activities, from part to component tosystem level. The inner loops on the top portion of thefigure represent integrated product/process trades at thesystem level. The middle loop represents the same,except for component level product/process trades.Similarly, the lower loop corresponds to part leveltrades for products/processes. An iteration around thelong outer loop represents what has typically been donein past sequential system design; re-design was oftenrequired due to product design incompatibilities withmanufacturing processes.

SYSTEMPROCESS

RECOMPOSITION

SYSTEMFUNCTIONAL

DECOMPOSITION

COMPONENTFUNCTIONAL

DECOMPOSITION

COMPONENTPROCESS

RECOMPOSITION

PARTPROCESS

RECOMPOSITION

PARTFUNCTIONAL

DECOMPOSITION

ProductTrades

ProcessTrades

ProductTrades

ProcessTrades

PRELIMINARYDESIGN

(PARAMETER)

PRELIMINARYDESIGN

(PARAMETER)

DETAILDESIGN

(TOLERANCE)

DETAILDESIGN

(TOLERANCE)

MANUFACTURINGPROCESSES

CONCEPTUALDESIGN

(SYSTEM)

ProcessTrades

INTEGRATEDPRODUCTPROCESS

DEVELOPMENT

ProductTrades

Figure 2a A Conceptual Framework for Integrated Product and Process Development



PR

OD

UC

T D

ES

IGN

DR

IVE

N

PR

OC

ES

S D

ES

IGN

DR

IVE

N

METHODOLOGY FOR INTEGRATED PRODUCT AND PROCESS DEVELOPMENT

REQUIREME NTS & FUNCTIONAL

ANALYSI S

SY ST EM DECOMP OSI TION&

FUNCTIONAL ALLOCAT ION

SY ST EM SY NTHESI STHROUGH MDO

SYS TEM A NALYSIS&

CONTROL

ES TABLISHTHE NEED

DEFI NE THE P ROBLEM

ESTABLISH VA LUE OBJECTI VE S

GENERATE FE AS IB LEALTERNATIVES

EVALUATEALTERNATIVE S

7 M& P TOOLS A NDQUALITY FUNCTIONDEP LOYME NT (QFD)

ROBUST DES IGNASSES SM ENT &OPTI MIZAT ION

ON-LINE QUALITYENGINEERING &

STATI STI CAL P ROCESSCONTROL (SPC) MAK E DECISI ON

SYSTEMSENGINEERING METHODS

QUALITYENGINEERING METHODS

COMPUTER-INTEGRATED ENVIRONMENT

TOP-DOWN DESIGNDECISION SUPPORT PROCESS

Figure 2b Interaction of the Four Key Elements to Implement IPPD

While Figure 2a represents the activities desired andinformation flow required for IPPD, it does not providea methodology for integrating the tools required toimplement IPPD. The practice of IPPD requires thesimultaneous use of several techniques integrated in adecision support process. A generic methodology thatis being used to implement IPPD at Georgia Tech isshown in Figure 2b. This methodology can be appliedfor the system, component, or part level tradesillustrated in Figure 2a, or for an integratedcombination. Illustrated in Figure 2b are theinteractions of the four elements necessary for parallelproduct and process trades to be made at the appropriatelevels of decomposition and recomposition. The fourkey elements are a top-down design decision supportprocess, systems and quality engineering methods, and acomputer integrated environment.

The heart of the implementation is a top-downdesign decision support process. Decision support is anessential element that can support a trade-off process andcan be used to focus efforts on design goals. It suppliesa logical, rational means for including factors that mustbe considered when making a decision. For the wingstructural design of the HSCT, the decision supportprocess logic starts with an identified HSCT conceptualdesign at the system level and an overall evaluationcriterion (OEC). For the wing major component (orsub-system level), the HSCT system level requirementsare deployed down to a preliminary design level. Areview of candidate materials and processes, as well aspanel and substructure concepts is a prerequisite for thegeneration of alternative wing concepts. The feasibilityof the concepts is determined with a detailed structuralanalysis to satisfy both static and dynamic load cases.Evaluation of the alternatives is typically based only onproduct metrics such as the primary wing structureweight. If cost is included as a product metric, it is also

weight-based. In this paper, the evaluations of thefeasible concepts will also be based on time/costprocess metrics such as fabrication, assembly, andmaterial costs. With the developed integrated designsystem, the structural concept selection can be basedupon evaluation of both the product and process metricsthat characterize the concepts.

The systems engineering methods on the right sideof Figure 2b are decomposition-oriented and productdesign driven. The flight conditions of the HSCTmission profile determine the aerodynamic and structuralrequirements for the wing structure. Data from a finiteelement structural analysis and optimization of the wingstructure are used to re-size aircraft for different wingconcepts and constitutes the system synthesis. Systemanalysis is typically based on product design metrics,but, in this paper, also includes process metrics.

The quality engineering methods illustrated on theleft side of Figure 2b are recomposition-oriented,statistically-based, and process design driven. Therecomposition process starts at the bottom of the figurewith use of quality or value engineering estimates of themanufacturing process step times to manufacture thewing structure. A process-based assessment of thedesign in terms of its production costs at the majorcomponent level is used in the evaluation of the feasiblealternatives. For top-level IPPD studies, a QualityFunction Deployment (QFD) technique is very useful tohelp define the design problem once the need has beenestablished, and track and deploy the decision-makingprocess. However, it has not been included for thisIPPD trade-off process for the wing structure.

The top of Figure 2b illustrates how the threepreviously described elements function within acomputer integrated environment. This environmentallows the interactive involvement of the threeelements, indicated by the arrows between elements.



The IPPD trade studies associated with the evaluation ofalternative wing structural concepts required theintegration of several design tools into a functionaldesign system.

Decision-Maker Centered Design ApproachThe emphasis of the Design Justification procedure

is information processing and presentation in anappropriate manner to aid the designer. A design-process centered approach requires the analysis of adesign against a set of weighted criteria, resulting in arecommended solution. A decision-maker centeredapproach, on the other hand, requires the data generatedduring the design to be received by the designer andprocessed with guidance. The designer can conductsensitivity analyses to determine the advantages ordisadvantages of allowing the design to progress in onedirection over another.

Knowledge-Guided Design ApproachAs opposed to many current systems design

techniques, Design Justification is knowledge-guided.With a non-knowledge-guided approach (Figure 3), theconceptualization of the design is done strictly by thedesigner. A problem is modeled to a particular level offidelity, then analyzed and interpreted. The designer canmodify the model until it meets the design objectives.

Conceptual(designer)input output

Analytical

results definition

Figure 3 Non-Knowledge-Guided Approach16

While such a process is sufficient for many cases,design problems that require interpretation of heuristicknowledge can not be handled. The knowledge-guideddesign approach (Figure 4) has an additional componentto handle heuristic problems: explicit knowledge. Thisapproach allows the designer to model his system andguide the design process utilizing the knowledgecomponent.

Conceptual(designer)input output

Analytical

results definition

Knowledge

decisions guidance

Figure 4 Knowledge-Guided Approach16

The knowledge component could theoretically existin many forms; a Knowledge-Based System (KBS) wasused for this project.

Implementation of Design Justification

The implementation of the Design Justificationprocess required three areas of research and development.First, it involved the acquisition/development of a bodyof manufacturing knowledge. This was based on severalprevious supersonic transport studies,18-28 as well asstudies of candidate materials and processes to be usedfor the HSCT.24,29-32 The most relevant knowledgewas embodied in a Knowledge-Based System, CLIPS,33

(C Language Integrated Production System) which wasused for this research.

Second, a production cost model for the wing thatcaptures the effects of alternative materials andmanufacturing processes was developed. This alsoincluded determining the most appropriate techniques topresent the product and process trade results to thedesigner.

Third, the process design tools were integrated withproduct design tools to form a functional design system(corresponding to the computer integrated environmentof Figure 2a). Several pre- and post-processors andintegration scripts were written to link the KBS with atop-level LCC model and an aircraft sizing code.FLOPS34 (Flight Optimization System) was used tosize the aircraft and the production cost model became anew module within ALCCA (Aircraft Life Cycle CostAnalysis).35 The integration enables designers toquantify where and how the processing costs producesignificant effects and propagate through the aircraft'slife cycle. The integration of the sizing code into theevaluation framework and the automated data mappingand passing alleviates distortions in data due tocommunication of results between analysis processes.

Details of all three of these components are beyondthe scope of this paper. The KBS is described by Marxet al36 and the multi-level LCC model developed withALCCA is discussed in Marx et al.37 Details of thewing production cost model are presented in thefollowing section.

Wing Production Cost ModelIt was necessary to develop and implement a new

production cost module for the LCC code. The form ofthe previous weight-based wing cost equation inALCCA was:

Costwing = a ⋅cf ⋅ Weightwing( )b⋅ef

(1)where a and b are based on a historical database ofaluminum aircraft (wings) and cf and ef are complexityfactors that can be used to adjust the costs for titaniumor composite wings.



For a complex, integrated structure like the HSCTwing, such a model is not sufficient. Cost EstimatingRelationships (CERs) developed for estimating wingcosts are usually based on a historical database ofexisting wings. Virtually all of those wings wouldhave been manufactured from aluminum. In addition,most of the wings undoubtedly have a much higheraspect ratio than the HSCT wing. Hence, applicationof the existing wing CER in ALCCA was not afeasible option for the HSCT wing.

Since lower weights do not always signify reducedcosts, a model based only on component weights willnot show the correct trends for many new advancedmaterials and manufacturing processes. Severalexamples of this trend exist, including the use ofthermoplastics versus thermosets. Thermoplasticmaterials are often selected by the product design teams,but are replaced with thermosets once more realisticcost/time manufacturing information is obtained,especially for large thermoplastic components.

The new model includes both labor and materialcosts for the wing. Calculation of the production costsboth in terms of hours and dollars allows sensitivitystudies to be conducted based not only on componentweights versus costs, but also production times, laborrates, raw material costs, buy-to-fly ratios, and alternateprocessing methods versus many cost and timeelements.

Using the typical exponential function form of thelearning curve, the xth unit cost, γ x , as a function of a

known Q0th unit cost, γ Q0 , and learning curve slope

can be written as:

γ x = γ Q0(Q0 )−β xβ

(2)where:

β = ln(ν)ln(2) (3)

and ν is the decimal value of the slope of the givenlearning curve. If γ x represents the xth unit cost for apre-specified finished material weight, equation (2) canbe modified as:

γ x = γ Q0

1Q0

βw

w0

η

xβ

(4)where:

η = ln(ω )ln(2) (5)

with ω as the decimal value of the slope of the weightsizing curve, w is the material weight, and w0 is thepre-specified finished material weight upon which theunit costs are based.

Assume equation (4) represents costs in terms ofdirect labor hours. It can be transformed to dollar costsby multiplying by a labor rate, ρl . Inflation effectsmust be accounted for to scale CERs developed for aspecific year's dollars. Recognizing that inflation has a

compounding effect similar to interest, equation (4) cannow be written as:

γ xl = γ Q0

1Q0

βw

w0

η

ρl (1+ f )(y−y0 ) xβ

(6)

where f represents the average annual percentageinflation rate over the period of years from y0 to y.

An equation for unit x material costs can bedeveloped similarly:

λ xm = wσmφmψm1

Q0

βw

w0

η

(1+ f )(y−y0 ) xβ

(7)where w is the material weight, σm is the raw materialcost, φm is the material buy-to-fly ratio (correspondingto a scrap rate), and ψm is the material burden rate.

The cumulative labor and unit costs are found bymultiplying equations (6) and (7) by x, the cumulativenumber of units produced to yield:

γ cxl = γ Q0

1Q0

βw

w0

η

ρl (1+ f )(y−y0 ) xβ +1

(8)

λcxm = wσmφmψm1

Q0

βw

w0

η

(1+ f )(y−y0 ) xβ +1 (9)

For simplification purposes, now let ξl and ξm

represent generalized labor and material Theoretical FirstUnit Costs (TFUCs), respectively.

ξl = γ Q0

1Q0

βw

w0

η

ρl (1+ f )(y−y0 ) (10)

ξm = wσmφmψm1

Q0

βw

w0

η

1+ f( ) y−y0( )(11)

Rewriting equations (6) and (7) with ξl and ξm gives

γ xl = ξl xβ

(12)λ xm = ξmxβ

(13)which now look like the more familiar form of thelearning curve equation.

This production cost model formulation so farimplicitly implies a single learning curve function.However, production functions are not always bestrepresented by a single learning curve. The followingformulation further develops the production cost laborequations for a double learning curve.

A double learning curve formulation implies twodistinct learning curve slopes, ν1 and ν2 . This leads to

two values for β ; let

β1 =ln ν1( )ln 2( )

β2 =ln ν2( )ln 2( ) (14)

For a learning curve breakpoint, xb , equation (12) isnow written as:

γ xl = ξl xβ1 x ≤ xb (15)



γ xl = ξl xb( )β1 x

xb

β2

x ≥ xb (16)Since the decrease in hours or costs due to a cumulativedoubling in production is now following the secondlearning curve slope, ν2 , with respect to a new firstunit point, namely xb , equation (16) must include the

ratio x

xb

in order to have the doubling occur with

respect to the cumulative number produced after xb .Equation (16) can be simplified to:

γ xl = ξl xb( )β1−β2 xβ2x ≥ xb (17)

Using a basic property of the natural log function,equation (18) can be written:

γ xl = ξl eln xb( )( )β1−β2

xβ2

x ≥ xb (18)or:

γ xl = ξleδ xβ2

x ≥ xb (19)where:

δ = β1 − β2( ) ln xb( ) (20)Equation (19) now represents the corrected form of theequation for the cumulative labor cost after theproduction breakpoint, xb , following the secondlearning curve slope, ν2 . To summarize, the labor costequations with a production breakpoint are:

γ xl = ξl xβ1

x ≤ xb (21)γ xl = ξle

δ xβ2x ≥ xb (22)

and the cumulative labor costs are now:

γ cxl = ξl xβ1+1

x ≤ xb (23)γ cxl = ξle

δ xβ2 +1x ≥ xb (24)

Equation (13), the material costs, can also be similarlymodified for a double learning curve.

The implementation of this model required actualdata for several of the variables in the equations. A

study by Resetar et al38 provided data for recurringmanufacturing costs, tooling costs, raw material costs,buy-to-fly ratios, and labor rates for a variety ofadvanced airframe materials.

Cost/Time AnalysisAs stated previously, the emphasis of the Design

Justification process is information processing andpresentation in an appropriate manner to aid thedesigner. With the wing production cost model asformulated above, the designer can determine themanufacturing costs, tooling costs, inspection costs,and material costs for the wing structure. By perturbingcertain variables, an almost endless combination ofsensitivity analyses can be conducted. The analysis andinterpretation of the cost and time data output from thewing production cost model is a complex task in itself.The Design Guide for Producibility39 suggests the useof cost/time analysis plots for interpreting productioncost estimates.

Figure 5 shows a generic plot of a cost/timeanalysis for a theoretical production operation. The x-axis is the unit cost while the y-axis represents thecumulative production time. The steps that comprise aprocess model for a manufacturing function are time-based, and translate to respective costs. For a particularproduction quantity, the sum of the steps traces a pathon the plot, shown with the dashed line, to the endpoint, represented with a small circle.

Learning curve effects will be reflected in thecost/time analysis plot. The farthest point on the lowerright of the cost/time curve represents the TFUC.Intuitively, the TFUC has the highest cost per unit. Asthe production quantity increases and the learning curveeffects are realized, the cost per unit decreases along thecost/time curve. Different process steps, learningcurves, labor rates, and production quantities willproduce noticeable effects on the cost/time curves.Families of cost/time curves can be generated to showthe effects of process alternatives at any level of theproduction, to determine the most significant costdrivers so that efforts can be focused to lower productioncosts, and to serve as process constraint curves forproduct/process design trades.



FinishingOperations

Material Cost

Cumul. time

Cost / Unit

Production

Setup

DesignTools

PurchaseMaterial

Tool Design Cost

SetupCost

Production Cost Largest RunFinishing Operations Cost Largest Run Production Cost Smallest Run

Finishing Operations Cost Smallest Run

Cost/Time Curve

End Points for WideRange of ProjectedLot Sizes

Largest Run

Smallest Run

Theoretical First Unit Cost(TFUC)

Figure 5 Cost/Time Analysis for Theoretical Production39

System and Model Verification

The implementation is verified in this section beforeproceeding to the product/process trades associated withintegrating design and manufacturing for a modern HighSpeed Civil Transport (HSCT). Limited data existsfrom previous industry and NASA HSCT studies.Economic data for model verification is particularlyscarce since most companies consider their data relatedto production costs highly competition sensitive. Thissection presents a summary of the detailed system andmodel verification that was completed before the HSCTcase study was started.

Summary of the Lockheed SST Evaluation ProcessAnalytical design studies were performed by the

Lockheed-California Company in the 1970s to assessthe relative merits of several structural arrangements,concepts, and materials applicable to an arrow-wingSupersonic Transport (SST) configuration, shown inFigure 6. The Lockheed SST concept carried 234passengers, had a cruise Mach of 2.7, and a baselinewing area of 10,900 square feet. The SST studies18-22

resulted in a large amount of published data. Muchinformation exists in the reports from the variousdisciplines, including performance and weights. The

published production cost data, though notcomprehensive, is believed to be the most completenon-proprietary set of data that could be found forverification of the production and economic models.

Figure 6 Lockheed SST Arrow-Wing Concept21

An initial structural analysis screening wasconducted by Lockheed on many potential wingstructural concepts. These included the chordwisestiffened, spanwise stiffened, and biaxially stiffened(monocoque) concepts shown in Figures 7-10. Unitweights were calculated for point design regions on thewing for each concept. There were three point design



regions: the forward strake area, the inboard wing box,and the outboard wing box. By interpolating from theanalysis point design regions, the unit weights derivedduring the initial screening were applied over all designregions to derive the total box structural weight for eachprospective concept. Results of Lockheed's initialstructural concept screening led to a more detailed

analysis of five structural concepts: a chordwise-stiffened, convex beaded concept; a spanwise hat-stiffened concept; a mechanically fastened, biaxially-stiffened, honeycomb concept; a welded, biaxially-stiffened, honeycomb concept; and a boron-polyimidereinforced chordwise-stiffened concept.

Figure 7 Chordwise Stiffened SST Wing Concepts21

Figure 8 Composite Reinforced SST Wing Concepts21

Figure 9 Spanwise Stiffened SST Wing Concepts21

Figure 10 Monocoque SST Wing Concepts21

The wing weights were categorized as primarystructure (variable weight) or secondary structure (fixedweight). The primary structure consists of thatstructure which carries the primary load and isinfluenced by the structural concept being evaluated, orin other words, the upper and lower skin panels andstiffeners, spars, and ribs. The secondary structure isthat structure which is not as directly affected by thestructural concept choice (such as flaps and ailerons).

The structural mass of each aircraft configurationwas estimated first based on the premise of a fixedvehicle size and taxi weight. This permits thedetermination of the allowable fuel for the aircraft,which ultimately yields the range capability. The costswere subsequently determined for each configuration.The disadvantage to using this approach was that,though the relative flyaway costs could be compared,the concepts could not be compared on an equivalentperformance basis, since their respective missioncapabilities (i.e., range, in this case) were different.

This led to a re-sizing of all configurations for aconstant payload/range requirement. The payload washeld constant, the range was constrained, and the grossweight of the aircraft was allowed to vary to meet thedesired range goal. Following the re-sizing, the conceptevaluation could be performed on a minimum-total-system-cost (with regards to operating cost variations)for concepts that had the same mission capability.

A relatively detailed description of the productionhours required for producing the wing structuralconcepts was published.19 The estimations includedmaterial dollars and fabrication (including subassembly)and assembly (including tank seal) hours. However, the"bottom line" of Lockheed's production cost analyseswas a "value-per-pound" factor for each concept that wassubsequently used for the economic evaluations. Withthe use of these "value-per-pound" factors, the economicsubroutines within ASSET (Advanced SystemSynthesis and Evaluation Technique, a Lockheed sizingcode from the 1970s) were used to estimate the flyaway



costs for five aircraft, based on a total production of 300units.

Weight-Based Cost Model LimitationTo prove the limitations of current weight-based

cost models concerning process variations, the results ofLockheed's SST sizing analysis were used in a weight-(only)-based LCC analysis (i.e., ALCCA, without theproduction data from CLIPS). The calculated systemcosts do not reflect the process-dependent material,fabrication, tooling, and assembly costs. Figure 11shows the incorrect weight-based system flyaway costsfor the constant-payload aircraft generated in thisfashion.

0.950

1.000

1.050

1.100

1.150

1.200

1.250

1.300

1.350

1.400

1.450

Re

lati

ve

V

alu

e

chordwise stiffened

spanwise stiffenedmonocoque-weldedmonocoque-mechanicalchordwise, B/PI reinf.

Concept Evaluation Metric

wingweight

process-based cost

OEW weight-based cost

TOGW

Figure 11 Weight-Based LCC Model Limitation

The highest weight concept, the chordwise-stiffenedconfiguration, is also the highest cost concept using aweight-based LCC model (the fourth column in Figure11), which is incorrect in this case. The twomonocoque (biaxially stiffened) concepts are actually themost expensive concepts (the fifth column in Figure11), even though their wing weights, OEWs, andTOGWs are substantially lower than those of thechordwise and spanwise stiffened concepts.

Hence, the product decomposition and processrecomposition (Figures 2a and 2b) required for IPPD

must occur if the LCC estimates are to reflect processvariations. This must include a preliminary structuralanalysis to generate data beyond the major componentweights. If the LCC analyses are based on onlycomponent weights, process variations will have littleeffect on cost.

Production Cost Model VerificationAll costs calculations in ALCCA for this

verification were based on a total production of 300units. The costs for the SSTs were originally calculatedusing 1973 dollars; those calculated in ALCCA are in1992 dollars. The results can be presented on anequivalent economic basis using either of two simplemethods. First, the flyaway costs estimated in 1973dollars could be inflated to 1992 dollars, or vice versa.Second, the relative flyaway costs could be presented.Since the relative costs (both flyaway and operating), asfunctions of wing structural concepts, will be moreuseful for eventual HSCT cost/benefit studies, therelative SST costs will be discussed in the followingsections for both the constant TOGW and constantpayload-range SSTs.

Constant TOGW SSTThe system costs, normalized with respect to the

composite-reinforced concept, as predicted with theintegrated system developed for this research, are givenin the Table 1 for the constant 750,000 pound TOGWaircraft. The relative differences of the predicted values,with respect to Lockheed's original estimates are alsoincluded for comparative purposes.

All of the predicted trends are correct with respect tothe composite reinforced concept. The wing conceptswith the highest areal weights (the chordwise andspanwise stiffened all titanium concepts) displayflyaway costs lower than the composite reinforcedconcept. However their range capability was far belowthat of the boron/polyimide concept. The biaxiallystiffened concepts had higher flyaway costs than thecomposite reinforced concept, even though their range

Table 1 SST Concept Relative System Costs for Constant TOGWstructural chordwise spanwise monocoque monocoque chordwise

arrangement units (mechanical) (mechanical) (mechanical) (welded) (B/PI reinf)Constant TOGW (Lockheed):flyaway cost $1973M 0.966 0.951 1.119 1.117 1.000DOC c/sm 1.010 1.000 1.057 1.063 1.000IOC c/sm 1.044 1.033 1.011 1.011 1.000Constant TOGW (ALCCA):flyaway cost $1992M 0.977 0.971 1.095 1.020 1.000DOC c/sm 1.025 1.023 1.041 1.018 1.000IOC c/sm 1.027 1.024 1.014 1.014 1.000Relative differences:flyaway cost 1.12% 2.11% -2.09% -8.73% 0.00%DOC 1.42% 2.31% -1.52% -4.17% 0.00%IOC -1.66% -0.93% 0.24% 0.24% 0.00%



capabilities were less than that of the compositeconcept. This is due to the fact that significantly moreproduction hours were required to fabricate the sandwichpanels than the more conventionally stiffened panels.The composite reinforced concept was considered ahigher technology concept than the biaxially stiffenedconcepts, but had a lower wing areal weight and, thus, alower TOGW, tended to offset the costs incurredthrough an increase in technology level.

Though the flyaway cost of the welded monocoquestructure was correctly estimated to be greater than thatof the composite reinforced concept, the magnitude ofthe difference was not as great as that predicted withASSET. This discrepancy, which is only evident forthis particular concept, can be explained as follows.The "value-per-pound" factors used with ASSET in theLockheed-California studies accounted for material cost,fabrication and assembly hours, and tooling hours. Forthe welded, biaxially stiffened concept, the material andtooling "value-per-pound" factors were not significantlydifferent. However, their predicted production hoursrequired for the mechanically fastened, monocoqueconcept were three times greater than what was predictedfor the welded, monocoque concept. This does notcorrespond to the published values for the flyaway costsof these concepts being almost equal. ALCCA'spredictions for the flyaway costs for the biaxiallystiffened concepts were consistent with the number ofhours required to produce the concepts, an advantage tousing a model based on time (hours), as opposed toapplying factors to weight-based algorithms only.

The operating cost calculations correlate withASSET's predictions as well. A minor difficulty in

verifying the models was the realization that thepublished operating costs for the constant TOGWaircraft were for operation at the calculated design range.This again, is another difficulty in attempting to do aconsistent economic concept evaluation with conceptsthat do not have the same mission capability.However, once the appropriate stage lengths were usedin the DOC and IOC calculations, the ALCCA-predictedvalues correlated well with the ASSET values, againwith the exception of the welded monocoque concept.Both of the predicted values for the welded monocoqueconcept's DOC are consistent with the estimated valuesof the flyaway costs. Or, in other words, an increase (ordecrease) in flyaway cost (or acquisition price for theairlines) should lead to an increase (or decrease) in theDOC for the aircraft because of the cost of ownershipportion of the DOC (depreciation, interest, andinsurance costs). Lockheed's predicted value for theDOC is consistent with their predicted value for theflyaway cost for the welded concept; ALCCA's isconsistent as well.

Constant Payload-Range SSTThe relative system costs, as calculated with the

integrated system, are given in Table 2 for the constantpayload-range aircraft. These aircraft were all required tomeet the 4,200 nmi range requirement. Though theTOGWs were significantly different, the aircraft canbetter be compared on an equivalent economic basissince they all have the same mission capability. Again,the original estimates by Lockheed are included forcomparison.

Table 2 SST Concept Relative System Costs for Constant Payload-Rangestructural chordwise spanwise monocoque monocoque chordwise

arrangement units (mechanical) (mechanical) (mechanical) (welded) (B/PI reinf)Constant Payload-Range (Lockheed):flyaway cost $1973M 1.089 1.054 1.135 1.155 1.000DOC c/sm 1.109 1.083 1.067 1.093 1.000IOC c/sm 1.044 1.044 1.011 1.011 1.000Constant Payload-Range (ALCCA):flyaway cost $1992M 1.093 1.072 1.082 1.010 1.000DOC c/sm 1.115 1.097 1.048 1.033 1.000IOC c/sm 1.034 1.030 1.010 1.007 1.000Relative differences:flyaway cost 0.38% 1.72% -4.69% -12.55% 0.00%DOC 0.57% 1.31% -1.84% -5.52% 0.00%IOC -1.02% -1.34% -0.10% -0.43% 0.00%

The flyaway costs for the constant payload-rangeconcepts again all exhibit the same trends as thosepredicted by Lockheed. In this case, the composite-reinforced concept is simply the lowest cost alternative.Though the non-composite reinforced, chordwisestiffened concept and the spanwise, hat-stiffened concepthad the lowest flyaway costs in the case for constant

TOGW, their flyaway costs are now higher than that ofthe composite-reinforced concept. This is due to thefact that, during the re-sizing for constant payload-range,their OWEs and TOGWs increased significantly,resulting in an increase in their flyaway costs. Themonocoque welded concept exhibited the same



characteristics in this case as it did when constrained toa maximum TOGW of 750,000 lbs.

The operating costs were easier to verify for thisconstant payload-range case. All stage lengths (4,200nmi) were equal. The ALCCA-predicted values wereconsistent with those predicted by ASSET.

The results presented in this section displayexcellent correlation with the most comprehensive setof non-proprietary data available for SST model andsystem verification. All of the relevant trades, for bothperformance and economics, have been verified. Withthe use of a powerful LCC code like ALCCA,integrated with a production cost KBS and a sizing code,it is possible to show the effects of alternative structuralconcepts in cost/benefit studies. In addition, withautomated data passing, file preparation, and pre- andpost-processing utilities, it is now possible to show theeffects of alternative concepts or process implicationson several constituents of the Life Cycle Cost at thedesired level of fidelity. This will be demonstrated forthe HSCT wing structural concept.

The use of a Knowledge-Based System "in the loop"with an integrated sizing code and multi-level LifeCycle Cost analysis model appears to be an excellentway to determine both the performance and economicconsequences of conceptual and preliminary structuraldesign decisions.

Hence, the integration scheme and system levelwing areal weight modeling assumptions are valid forshowing product/process trades for wing productioncosts. The source of the production data for thisverification exercise was the SST studies. However, theadvanced materials and processes that are projected forHSCT airframe production required a new set ofproduction cost data. Details of the IPPD trade studyconducted for the HSCT wing structure are presented inthe next section.

HSCT Concept Description

The HSCT was chosen as the case study aircraft forthis research for several reasons. It is currently understudy by several of the U.S. aerospace industry primeand sub-contractors, as well as NASA and manyuniversities. The technological challenges associatedwith this aircraft provide many topics for both industrialand academic research.

The modern HSCT concept is a 300-passengeraircraft with a design range of 5,000 nautical miles. Itscruise speed is currently set at Mach 2.4. Weightestimates range from an optimistic 750,000 pounds toover 900,000 pounds, depending on the specificconfiguration. The structural design life of the aircraftis targeted at 20,000 flight cycles and 60,000 flighthours. Take-off and landing field lengths are specifiedto be no longer than 11,000 feet.

The HSCT was modeled in FLOPS with relevantgeometry, mission parameters, aerodynamic data, andpropulsion data. The baseline HSCT modeled inFLOPS had the characteristics defined in Table 3.

Table 3 HSCT Configuration DescriptionRange 5,000 nmiTOGW 825,000 lbsNumber of passengers 300Cruise ceiling 70,000 ftApproach speed 155 kts.Fuselage length 330 ft.Wing area 9,000 ft2

Wing span 130 ftHorizontal tail area 1,000 ft2

Vertical tail area 600 ft2

Horizontal tail sweep angle 35˚Vertical tail sweep angle 41˚

Off-the-shelf aerodynamic design codes40-46 wereused to generate the user-defined induced dragcoefficients for FLOPS. The optimum twist andcamber distribution for the wing and the area-ruledfuselage were used in generating the drag polars thatwere input into FLOPS. A fixed cycle for a mixed-flowturbofan engine was selected from the FLOPS-suppliedengine cycles.

The mission used for sizing the aircraft in FLOPSwas a nominal 85% supersonic and 15% subsonic (bydistance) mission. A plot of a representative splitsub/supersonic mission profile generated by FLOPS isshown in Figure 12. As illustrated in Figure 12, boththe subsonic and supersonic cruise segments were flownat a fixed Mach number, with the optimum altitude forspecific range calculated by FLOPS.

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

0.0 50.0 100.0 150.0 200.0 250.0 300.0 350.0

Mac

h

nu

mb

er

0

10000

20000

30000

40000

50000

60000

70000

Alt

itu

de

Mach number

Altitude

Mission Time (min)

Figure 12 Mission Profile

Wing Region Descriptions

The wing is divided into three main design regions:the forward strake, the inboard (or main) wing box, andthe wing tip box. Figure 13 shows a CATIA solidmodel of the baseline HSCT aircraft that illustrates the



three design regions of the baseline HSCT conceptevaluated in this paper.

Also illustrated on the starboard side of the aircraft isthe optimum twist and camber distribution of theaerodynamic grid generated for this HSCT wing

planform (with the z-ordinates scaled by a magnitude of4). The area-ruled fuselage is also clearly visible fromthis perspective.

douglas2.gifForward Strake

– PMC face and polyimide honeycomb core sandwich skins

– PMC sub-structure

Inboard Wing Box– SPF/DB Ti skin structure

– Ti sub-structure

Wing Tip Box– PMC face and polyimide

honeycomb core sandwich skins

– PMC sub-structure

leading edge structure– SPF/DB Titanium

trailing edge structure– discretely-stiffened PMC twist and camber

distribution

Figure 13 CATIA Solid Model Showing the Baseline Hybrid HSCT Wing Structural Concept

The baseline wing concept is representative ofcurrent concepts under study by industry. Compositehoneycomb skin panels and Superplastically Formed /Diffusion Bonded (SPF/DB) titanium skin panels areviable candidates for HSCT wing structure.2 4

Academically, a hybrid concept provides a good baselinefor IPPD trade studies. Three potential HSCT wingstructural concepts are evaluated in this paper using theapproach and computer implementation describedpreviously. These substantially different concepts werechosen to show a variety of trends in cost andperformance.

Concept one has SPF/DB titanium skin panels andtitanium substructure in all three wing designregions.47 Concept two has Z-stiffened PMC skinpanels and PMC substructure in all wing regions.47

Concept three has the baseline hybrid wing structure24

illustrated in Figure 13. While the structural productand process characteristics of the concepts were different,the planform shape and size remained constant.

In order to conduct IPPD trades, the product andprocess characteristics of the individual design regionswere identified. The product characteristics are similarfor all three concepts, but the processing characteristicsare quite different. Table 4 shows the product andprocess characteristics for the baseline structural concept(concept three, the hybrid structure).



Table 4 Wing Region Product and Process CharacteristicsWing Region Product Characteristics Process Characteristics

Forward Strake The forward strake regionexperiences light or moderatebiaxial loads with respect to wingbending. Much of the forwardstrake is designed to meetminimum gage requirements.

The loft-to-loft distance is large in this section; accessfor automation is possible up to a depth of about 24" (thelength of a human arm). As opposed to the conventionalbirdcage assembly approach, in which the internalstructure is assembled first and then the skin panels aresubsequently attached, this section of the wing will beassembled in reverse order. First the lower skin will beplaced in its respective tool, the spar caps will then be(automatically) fastened to it. The spar webs will befastened next, followed by the top spar caps. The upperpanel, which is tool coordinated, is then placed andfastened. With this sequence of assembly steps, thesubstructure will be floated, as opposed to the wing panelsbeing floated. The loft dimensions are very critical on thistype of a wing. Therefore, the tools will be made for theskin panels to achieve the necessary tolerances. At depthsof more than 24" from the skin panel, assembly will bemanual (which increases the fastening cycle time). It isassumed that the spar caps and spar webs were cocured,requiring no additional stitching.

Main Wing Box The major load path of thewing is in the inboard wing box.This section experiences highspanwise load intensities due towing bending, and variablechordwise load intensities due tobending and torsional effects.This section has the highestaverage areal weights. Theengine pylons are attached to therear spar of this section. Theinboard flap and spoiler will alsobe attached to the rear spar of thissection.

A four-sheet SPF/DB titanium skin panel is used inthis highly loaded section of the wing. The substructureis also all titanium, with extruded spar caps and sheetmetal webs. The two inner sheets are first intermittentlyroll-seam welded with a cross pattern while the two outersheets are periphery welded. The four sheets are heated tothe processing temperature, and put in the SPF/DB press.Gas is injected to inflate/form the sheets; the temperatureis held for diffusion bonding, followed by a cooling andunloading period. Then the panel is trimmed. No jig isneeded for this procedure, since there are no stringers. Thepanels are integrally stiffened, hence there is a reducedfastener count. There is also no spring-back of thestructure. A disadvantage in using this process is the hightooling cost for the individual dies.

Wing Tip Box The wing tip box is the moststiffness-critical design region ofthe wing. This sectionexperiences high load intensities,and is driven by stiffnessconstraints to prevent flutter andto ensure aileron controleffectiveness. An outboard flapand spoiler, flaperon, and aileronwill be attached to the rear spar ofthe wing top box.

A conventional assembly approach will be used forthis section of the wing. The substructure will beassembled first, followed by the skin panels. One of thepanels will be put on first, then drilled and removed. Theother skin will be placed on the substructure and fullyfastened. The first skin panel will then be tacked in placeand mechanically fastened. Blind fasteners will be used onparts of the wing tip structure since the tip is relativelythin (approximately 6" loft-to-loft distance).

Relative Concept CostsA recent study from McDonnell Douglas provided

the relative data given in Table 5.48 The objective ofthe study was to determine the effects of alternativestructural concepts on the manufacturing (fabricationand assembly labor) hours and tooling costs for each of

three design regions of the wing. The processcharacteristics given in Table 4 were used to generatethe relative numbers given in Table 5. The causes ofthe high and low numbers due to the specific processcharacteristics of the design region and/or the structuralconcept are listed below the table.



Table 5 Relative Concept Costs48

Aluminum Titanium Titanium PMC PMCDesign Region Z SPF/DB H/C H/C ZForward Strake baselinefabrication labor hours 1.0 0.67h 0.81 0.97 1.13b

fabrication tooling cost 1.0 1.22d 0.30f 0.36g 0.34c

assembly labor hours 1.0 0.86 0.86 0.86 1.00assembly tooling cost 1.0 0.95 0.95 0.95 0.97Tip Box baselinefabrication labor hours 1.0 0.65h 0.87 1.13 1.46b

fabrication tooling cost 1.0 1.80d 0.53f 0.61g 0.61c

assembly labor hours 1.0 0.95 0.95 0.95 1.07assembly tooling cost 1.0 0.92 0.92 0.92 0.94Inboard Wing Box baselinefabrication labor hours 1.0 0.77h 0.86 0.94 1.15b

fabrication tooling cost 1.0 0.98e 0.33f 0.35g 0.38c

assembly labor hours 1.0 1.24a 1.24a 1.24a 1.32a

assembly tooling cost 1.0 0.95 0.95 0.95 0.97

a relatively high number of fasteners neededb includes stitching zee's onto the panelc stitched/RFI is a low-cost tooling approachd high cost of tooling (die) for the SPF/DB panelse the die/jig cost is nearly that of aluminum, due to

high compound curvature in this regionf no stretch form of skin requiredg one-sided toolingh not a time-intensive process step

Concept 1 is represented by the Titanium SPF/DBcolumn in the table. Concept 2 data is the PMC Zcolumn in the table. The baseline hybrid concept datais marked. The knowledge contained in Tables 4 and 5was embodied in the rule bases in CLIPS. It was usedto adjust the production cost estimates for the specificmaterials and processes used to manufacture thestructure in each section of the wing. This way, thecost estimates generated are representative of the actualproduction processes and materials used to manufacturethe wing structure.

Results

The results presented in this section are based onseveral assumptions:

1. The process-dependent parameters in the model arefor the wing only.

2. The manufacturer's module in the LCC code is nota factory simulation model, but a model of theproduction of a single aircraft. The deliveryschedule is assumed to be equivalent to theproduction schedule, and the order scheduleproceeds the delivery schedule by the number ofmonths in the RDT&E phase.

3. Distribution of the RDT&E costs, themanufacturing and sustaining costs, and revenue(income) are as outlined by Marx et al.37

4. All estimates are based on a production of 600units, unless indicated otherwise.

5. The airline module in ALCCA provides ananalysis of the costs and revenues an airline willexperience in purchasing and operating oneaircraft. Schedules for financing, depreciation,interest, tax, and cash flow are also found in Marxet al.37

6. Fuel costs were assumed to be $0.61 per gallon.The economic mission was 3,800 nmi and aircraftutilization was 5,000 hours per year. The loadfactor was 0.65. The aircraft was assumed to befinanced at an interest rate of 8% over a 20 yeareconomic lifetime for the aircraft.

7. All cost estimates are in 1992 dollars.8. Propulsion costs are not eroded with a learning

curve. They were considered a fixed cost from theairframer's perspective.

9. The effects of the structural concept and processalternatives on the maintenance of the aircraft werenot considered. Hence, variations in DOC for theconcepts are due only to differentials in financing,interest, depreciation, and insurance costs for theaircraft.

Cost versus Performance StudiesThe first plots are normalized with respect to

Concept 1. Figures 14-17 show a series ofcomparisons of product and process metrics for thewing, the aircraft, and the operating costs. These plotsrepresent product and process trades that can be used toevaluate feasible alternative wing structural concepts, asrelated to Figures 2a and 2b.



0.85

0.95

1.05

1.15

1.25

1.35

1.45

Re

lati

ve

V

alu

e

SPF/DB TiZ-stiffened PMCSPF/DB Ti & PMC H/C


primarywing wt

totalwing wt

wingareal wt

OEW fuel wt TOGW

Figure 14 Component and System Weights

0.85

1.05

1.25

1.45

1.65

1.85

2.05

2.25

2.45

2.65

Re

lati

ve

V

alu

e



primarywing wt

wingTFUC

airframeTFUC

aircraftTFUC

aircraftprice

600thwing cost

Figure 15 Component and System Costs

0.85

1.35

1.85

2.35

2.85

3.35

3.85

4.35

4.85

5.35

Re

lati

ve

V

alu

e



primarywing wt

wing TFUmfg cost

wing TFUQA cost

wing TFUtool cost

wing TFUmatl cost

total wingTFUC

Figure 16 Sub-component and Component Costs

0.99

1.00

1.01

1.02

1.03

1.04

1.05

1.06

1.07

1.08

1.09

Re

lati

ve

V

alu

e



DOC IOC TOC $/RPM

Figure 17 Operating Costs

Figure 14 shows several standard product evaluationmetrics that are used to evaluate concepts duringconceptual design. The three wing concepts show theintuitive trend: the concept with the lowest weight(concept 3) corresponded to the aircraft with the lightestOWE, least mission fuel required for the design range,and lightest TOGW. The heaviest wing concept(concept 2) resulted in the heaviest OWE, most missionfuel required, and heaviest aircraft TOGW. One mightexpect the composite wing concept to weigh less thanthe titanium concept. However, since much of theHSCT wing structure experiences light to heavy bi-directional loads, the discretely-stiffened PMC conceptis much heavier due to its inability to resist the biaxialloads as well as the SPF/DB panels or the PMChoneycomb panels in concepts 1 and 3.

Figure 15 shows some interesting trends. Thehybrid concept, which had the lowest wing weight, hasa TFUC approximately 40% higher than the slightlyheavier SPF/DB concept. The Z-stiffened PMC conceptis substantially higher than both of the other concepts.The airframe and aircraft TFUCs exhibit the sametrends, but with smaller relative magnitudes. However,when the various manufacturing costs, tooling costs,material costs, and learning curves are all consideredsimultaneously, the prices required for the manufacturer

to make a 12% ROI on the production investment for600 units are much closer in magnitude. The Z-stiffened PMC concept price is about 15% more thanthe SPF/DB concept. The price for the hybrid conceptwas slightly higher than that of the SPF/DB concept,even though it weighed less. A significant consequenceof this will be noted later.

The TFUCs for the wing labor and materials areshown in Figure 16. The relative magnitude of theTFU tooling costs for the concepts is quantified in thefigure. The TFU tooling costs for the heavy PMC Z-stiffened concept are extremely high, when compared tothe other concepts.

The interesting effect of the differences in pricebetween the all SPF/DB titanium wing concept and thehybrid concept are evident in Figure 17. Recall that theprice of the lighter aircraft with the hybrid wingstructure was higher than that of the aircraft with the allSPF/DB titanium structure. One might expect that,according to assumption (9), the operating costs for thehybrid aircraft would be more than that of the other. Asshown in Figure 14, the mission fuel required by theaircraft with the hybrid wing was less than that for theaircraft with the all SPF/DB wing. The higher price ofthe hybrid concept was more than offset by the lowerfuel cost to fly the mission. Hence, the DOC of the



aircraft with the hybrid wing structure is less than thatof the aircraft with the all SPF/DB titanium wing.This type of result is precisely that which can be usedfor the economic justification of a major designdecision. It illustrates the cost trade-off betweenproduction and O&S costs.

Cost/Time AnalysesThe next set of plots are cost/time analysis plots

(refer to Figure 5). Cost/time curves are plotted for the

constituents of the total direct recurring labor(fabrication and assembly, quality assurance orinspection, and tooling). The sum of the three laborcomponents yields the total labor curve, shown as thedotted line in Figures 18-20.

The plots show the costs for production quantitiesfrom the 10th unit to the 1000th unit. The farthestpoint on the lower right end of each curve is the 10thunit; the highest point on the left end of each curverepresents the 1000th unit.

0

20

40

60

80

100

120

140

160

180

0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0

Cu

mu

lati

ve

Lab

or

Ho

urs

(m

illi

on

s) manufacturingquality assurancetoolingtotal labor

Unit Labor Cost ($1992M)

Figure 18 Concept 1 Labor Cost/Time Analysis

0

50

100

150

200

250

300

350

400

0.0 25.0 50.0 75.0 100.0 125.0 150.0 175.0 200.0

Cu

mu

lati

ve

Lab

or

Ho

urs

(m

illi

on

s) manufacturingquality assurancetoolingtotal labor



0

50

100

150

200

250

0.0 15.0 30.0 45.0 60.0 75.0 90.0 105.0 120.0C

um

ula

tive

L

abo

r H

ou

rs

(mil

lio

ns) manufacturing

quality assurancetoolingtotal labor



0

50

100

150

200

250

300

350

400

0.0 25.0 50.0 75.0 100.0 125.0 150.0 175.0 200.0

Cu

mu

lati

ve

Lab

or

Ho

urs

(m

illi

on

s) SPF/DB Ti & PMC H/CZ-stiffened PMCSPF/DB Ti


Figure 21 Labor Cost/Time Analysis: All Concepts

The use of the Cost/Time analyses can help identifythe key cost drivers that are not evident when onlyconsidering the individual component manufacturingcosts or cost of the entire aircraft. For all threeconcepts, the quality assurance costs, both in terms ofdollars and hours, are the smallest cost element of thewing labor for any production quantity.

Close investigation of the tooling andmanufacturing cost/time curves in the figures revealsseveral interesting trends. For concept 1, the allSPF/DB titanium concept plotted in Figure 18, themanufacturing (fabrication and assembly) cost and timeare greater than the tooling cost and time for allproduction quantities.

As indicated previously in Figure 16, the TFUC ofthe tooling for the Z-stiffened PMC wing is much

higher than that of the other two concepts. The hightooling cost (in terms of dollars) for low productionquantities for this concept is evident in Figure 19. Forthe lower production quantities, the tooling cost ismuch greater than the manufacturing cost in dollarswhile being approximately equal in hours. As theproduction quantity increases, the dollar costs of thetooling decline to be nearly equal to the manufacturingdollar costs.

The hybrid baseline concept has the most interestingCTA plot, shown as Figure 20. For the lowerproduction quantities, the tooling dollar costs exceed themanufacturing dollar costs with the hours being nearlyequal. However, as the production quantity increases,the tooling costs fall below the manufacturing costsboth in terms of dollars and hours.



The total labor cost/time curves are plotted in Figure21 to show the relative total labor costs for the threewing concepts. Concept 2 has the highest labor costfor all production quantities while concept 1 has thelowest total labor cost for all production quantities.

As suggested by Resetar et al,38 distinct learningcurves were used for each component of the labor and

material costs. This was implemented in ALCCA, andthe results are clearly visible when plotted in log-logformat in Figures 22 and 23. The figures present thewing costs and some of the aircraft major componentcosts for the hybrid wing concept relative to productionquantity.

1

10

100

1000

1 10 100 1000

Co

st

($1

99

2M

)

airframemanufacturingQAtoolingmaterialtotal wing

Production Quantity

Figure 22 Concept 3 Wing and Airframe Costs

10

100

1000

1 10 100 1000

Co

st

($1

99

2M

)

airframetotal wingpropulsionunit mfg costaverage

Production Quantity

Figure 23 Concept 3 Component and System Costs

The effect of a learning curve is to reduce theproduction cost by a specific fraction each time thecumulative production doubles. When plotted in log-log format, the result is a straight line. The steeper thelearning curve, the higher the initial costs, but the morethe costs decrease as cumulative production increases.The tooling labor costs have the steepest slope inFigure 22.

The airframe and wing costs are shown again inFigure 23 for comparison to the total unit

manufacturing cost and average unit cost (including allamortized RDT&E, production, and sustaining costs).The effect of assumption 8 (the fixed propulsion costs)is shown by the propulsion costs not changing as theproduction quantity increases. This results in the non-linear curve for the unit manufacturing costs.

Figures 24 and 25 show the annual and cumulativemanufacturer cash flow distributions for HSCTproduction with the baseline wing structural concept.

-10000

-5000

0

5000

10000

15000

20000

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

Year

Do

lla

rs

($1

99

2M

)

RDT&Emanufacturingsustainingincomenet cash flow

Year

Figure 24 Concept 3 Annual Cash Flow

-100000

-50000

0

50000

100000

150000

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

Year

Do

lla

rs

($1

99

2M

)

RDT&Emanufacturingsustainingincomenet cash flow

Year

Figure 25 Concept 3 Cumulative Cash Flow

Figure 24 shows the annual cash flows over theassumed 60 month period of pre-production and the 10year production period. RDT&E costs are distributedover the first 72 months of the program. The plot ofthe revenue clearly shows the order payments anddelivery payments corresponding to the delivery

schedule. The annual net cash flow is negative for thefirst six years of the program, but is positive for theremaining nine years of the program.

The cumulative annual cash flows are shown inFigure 25. The most interesting line on the plot is thecumulative net cash flow. For a 600 unit production



run, with the aircraft selling at a price that correspondsto a 12% ROI for the manufacturer, the break-evenpoint is approximately nine and a half years into theproduction; and the break-even unit was 365.

While these plots are interesting, the comparisons ofdiscounted ROI versus price show more of an effect thata designer or production manager may like to know.The cash flows must be calculated to determine theinternal rate of return. Figure 26 shows a plot ofmanufacturer's internal rate of return versus the sellingprice of the aircraft for aircraft with the three wingconcepts.

0

5

1 0

1 5

2 0

2 5

3 0

3 5

4 0

4 5

5 0

0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7

Dis

co

un

ted

R

OI

(%)

SPF/DB Ti & PMC H/CZ-stiffened PMCSPF/DB Ti

Relative Price

Figure 26 Manufacturer's ROI vs. Aircraft Price

The horizontal line in the figure is placed at 12%ROI for the manufacturer. As shown, the price of theaircraft with the hybrid wing concept is slightly higherthan that of the aircraft with the all SPF/DB wingconcept. The price for concept 2 is almost 10% higherthan that of concept 1.

Conclusions

In terms of both the product and process trades, theZ-stiffened PMC wing concept is not a viable structuralconcept. Its inability to support the bi-directional loaddistribution resulted in it having the highest wingweight, OWE, TOGW, and mission fuel. Its largeweight and relatively expensive production costs yieldan aircraft with the highest selling price and operatingcost of the three concepts

Comparing and contrasting the remaining twoconcepts is not as simple. While the hybrid wingconcept weighed less than the all SPF/DB titaniumwing, all components of its total wing production costswere somewhat higher. However, when all RDT&E,manufacturing, and sustaining costs are amortized overthe production of the aircraft, the resulting cash flowsresult in relatively similar selling prices for the aircraftto achieve a pre-specified ROI for the manufacturer.Though the aircraft with the hybrid wing had a higherselling price, its lower weight and required mission fuelreduce its direct operating costs to be slightly lowerthan those of the aircraft with the all SPF/DB titanium

wing. This is an excellent example of a Cost versusPerformance trade study result showing the balancebetween acquisition (production) costs and operationalcosts.

In terms of the functionality of the approach andimplementation, the system was able to accuratelyreproduce the published performance18 and economic19

(including manufacturer and air carrier) results for theSST studies conducted by the Lockheed-CaliforniaCompany in the 1970s. The values predicted for HSCToperating costs and required revenue are consistent withthe limited published HSCT economics data.23

Future Work

The case study presented in this paper showed theLife Cycle Cost implications of alternative structuralconcepts for the wing of the HSCT. The remainder ofthe work required to complete this research project willbe to conduct sensitivity analyses due to perturbationsof particular process variables for the baseline concept.These will be process trades that show the effects ofunder- or over-predictions for the fabrication, assembly,and tooling costs; the effects of different values for thelearning curves and buy-to-fly ratios; and the effects ofraw material costs, production quantity, and possiblyproduction breakpoints on Life Cycle Costs.

Acknowledgments

This research is being funded under a NASA LangleyGraduate Student Researchers Program (GSRP)Fellowship (NGT-51101). NASA Langley direction isprovided by S. M. Dollyhigh at Langley ResearchCenter. Tom Galloway of NASA Ames providesadditional consultation for Design for Affordabilitystudies.

Alex Velicki and Keith McIver of McDonnellDouglas provided much help with characterizing theconcept product and process characteristics, respectively.

References

1 Fisher, D. C., "Design and Analysis Aid forEvaluating Aircraft Structures," Proceedings ofthe 2nd International Conference on Industrial andEngineering Applications of ArtificialIntelligence and Expert Systems, Vol. 1,Tullahoma, TN, June, 1989.

2 Messimer, S. L., and Henshaw, J., "CompositesDesign and Manufacturing Assistant,"International Journal of Materials and ProductTechnology, Volume 9, Nos 1/2/3, 1994.

3 Zumsteg, J. R., Pecora, D., and Pecora, V. J., "APrototype Expert System for the Design and



Analysis of Composite Material Structures,"Proceedings of the ASME InternationalComputers in Engineering Conference, Boston,MA, 1985.

4 Pecora, D., Zumsteg, J. R., and Crossman, F. W.,"An Application of Expert Systems toComposite Structural Design and Analysis,"Proceedings of the ASME Winter AnnualMeeting, Miami, FL, 1985.

5 Rao, K. P., Viswanath, S., Sridhara Murthy, S.,and Jayatheertha, C., "An Expert SystemApproach for the Design of CompositeLaminates," Journal of the Indian Institute ofScience, Vol. 70, 1990.

6 Warburton, L. M., and Glatfelter, J. W.,"Development and Utilization of a Knowledge-Based Environment on a Production HelicopterProgram," Proceedings of the AHS Vertical LiftAircraft Design Conference, San Francisco, CA,January, 1995.

7 Williams, M. J. R., "Automated Aircraft EngineDevelopment Cost Estimating Using ArtificialIntelligence," Oral Paper presented at the 27thAIAA Joint Propulsion Conference, Sacramento,CA, 1991.

8 Mabson, G. E., Flynn, B. W., Ilcewicz, L. B., andGraesser, D. L., "The Use of COSTADE inDeveloping Composite Commercial AircraftFuselage Structures," AIAA-94-1492,Proceedings of the 35thAIAA/ASME/ASCE/AHS/ASC Structures,Structural Dynamics, and Materials Conference,Hilton Head, SC, April, 1994.

9 Metschan, S. L., Wilden, K. S., Sharpless, G. C.,and Andelman, R. M., "Cost ModelRelationships Between Textile ManufacturingProcesses and Design Details for TransportFuselage Elements," Proceedings of the ThirdNASA ACT Conference, Long Beach, CA, June,1992.

10 Lee, P., "A Process Oriented Parametric CostModel," AIAA-93-1029, Proceedings of theAIAA/AHS/ASEE Aerospace DesignConference, Irvine, CA, February, 1993.

11 King, N. E., "Heuristics as a Design Tool forBrilliant Eyes Cost Engineering," AIAA-92-1055, Proceedings of the AIAA/AHS/ASEEAerospace Design Conference, Irvine, CA,February, 1992.

12 Apgar, H., "Design-to-Life-Cycle-Cost inAerospace," AIAA-93-1181, Proceedings of theAIAA/AHS/ASEE Aerospace DesignConference, Irvine, CA, February, 1993.

13 Meisl, C. J., "The Future of Design IntegratedCost Modeling," AIAA-92-1056, Proceedings ofthe AIAA/AHS/ASEE Aerospace DesignConference, Irvine, CA, February, 1992.

14 Solverson, R. R., "Design to Cost with PRICEH," AIAA-93-1030, Proceedings of theAIAA/AHS/ASEE Aerospace DesignConference, Irvine, CA, February, 1993.

15 Proceedings of the Defense ManufacturingCouncil/Industry Affordability Task ForceWorkshop, National Center for AdvancedTechnologies, Washington, D. C., March, 1995.

16 Noble, J. S., and Tanchoco, J. M. A., "Design forEconomics," Chapter 16, ConcurrentEngineering: Automation, Tools, andTechniques, edited by Andrew Kusiak, JohnWiley & Sons, Inc., 1993.

17 Noble, J. S., and Tanchoco, J. M. A., "ConcurrentDesign and Economic Justification inDeveloping a Product," International Journal ofProduction Research, Vol. 28, No. 7, pp. 1225-1238.

18 Jensen, R. N., and Mijares, R. D., "MassAnalysis," Section 15, Substantiating Data forArrow-Wing Supersonic Cruise AircraftStructural Design Concepts Evaluation, NASA-CR-132575-4, October, 1975.

19 Sakata, I. F., "Production Costs," Section 16,Substantiating Data for Arrow-Wing SupersonicCruise Aircraft Structural Design ConceptsEvaluation, NASA-CR-132575-4, October,1975.

20 Sakata, I. F., "Concept Evaluation and Selection,"Section 17, Substantiating Data for Arrow-WingSupersonic Cruise Aircraft Structural DesignConcepts Evaluation, NASA-CR-132575-4,October, 1975.

21 Sakata, I. F., and Davis, G. W., "Evaluation ofStructural Design Concepts for an Arrow-WingSupersonic Cruise Aircraft," NASA-CR-2667,May, 1977.

22 Wright, B. R., and Ross, R. L., "TechnologyAssessment Studies Applied to SupersonicCruise Vehicles," NASA CR-145133,September, 1976.

23 "High Speed Civil Transport Study," NASA-CR-4233, Boeing Commercial Airplanes, 1989.

24 Brunner, M., and Velicki, A., "Study of Materialsand Structures for High Speed Civil Transports,"NASA-CR-191434, McDonnell DouglasAerospace, September, 1993.

25 Röhl, P. J., "A Multilevel DecompositionProcedure for the Preliminary Wing Design of aHigh Speed Civil Transport Aircraft," Ph.D.Thesis, School of Aerospace Engineering,Georgia Institute of Technology, Atlanta, GA,June, 1995.

26 "Study of High Speed Civil Transports," NASA-CR-4235, Douglas Aircraft Company, 1989.

27 Bhatia, K. G., and Wertheimer, J., "AeroelasticChallenges for a High Speed Civil Transport,"AIAA-93-1478, Proceedings of the 34th



AIAA/ASME/ASCE/AHS/ASC Structures,Structural Dynamics, and Materials Conference,La Jolla, CA, April, 1993.

28 Tzong, G., Baker, M. Yalamanchili, K., andGiesing, J., "Aeroelastic Loads and StructuralOptimization of a High Speed Civil TransportModel," AIAA-94-4378, Proceedings of the 5thAIAA/NASA/USAF/ISSMO Symposium onMultidisciplinary Analysis and Optimization,Panama City Beach, FL, September, 1994.

29 Markus, A., "Resin Transfer Molding for AdvancedComposite Primary Wing and FuselageStructures," Proceedings of the 2nd NASA ACTConference, Lake Tahoe, NV, November, 1991.

30 Markus, A., Thrash, P., and Rohwer, K., "Progressin Manufacturing Large Primary AircraftStructures Using the Stitching/RTM Process,"Proceedings of the Third NASA ACTConference, Long Beach, CA, June, 1992.

31 Hatakeyama, S. J., "Wing Structures Developmentfor the HSCT," Aerospace Engineering, May,1995.

32 Suarez, J., and Dastin. S., "Comparison of ResinFilm Infusion, Resin Transfer Molding, andConsolidation of Textile Preforms for PrimaryAircraft Structure," Proceedings of the 2ndNASA ACT Conference, Lake Tahoe, NV,November, 1991.

33 Giarratano, J. and Riley, G., Expert Systems:Principles and Programming, PWS PublishingCompany, Boston, MA, 1994.

34 McCullers, L. A., "FLOPS: Flight OptimizationSystem," Release 5.7, User's Guide, NASALangley Research Center, Hampton, VA,December, 1994.

35 Bobick, J. C., Braun, R. L., and Denny, R. E.,"Documentation of the Analysis of the Benefitsand Costs of Aeronautical Research andTechnology Models - Volume 1: TechnicalReport," NASA-CR-152278, July, 1979.

36 Marx, W. J., Schrage, D. P., and Mavris, D. N.,"An Application of Artificial Intelligence forComputer-Aided Design and Manufacturing,"Proceedings of the International Conference onComputational Engineering Sciences, MaunaLani, HI, July, 1995.

37 Marx, W. J., Mavris, D. N., and Schrage, D. P.,"A Hierarchical Aircraft Life Cycle Cost

Analysis Model," AIAA-95-3861, Proceedings ofthe 1st AIAA Aircraft Engineering, Technology,and Operations Congress, Los Angeles, CA,September, 1995.

38 Resetar, S. A., Rogers, J. C., and Hess, R. W.,"Advanced Airframe Structural Materials: APrimer and Cost Estimating Methodology,"Project AIR FORCE Report R-4016-AF, RANDCorporation, 1991.

39 Military Handbook: Design Guidance forProducibility, MIL-HDBK-727, April, 1984.

40 Carlson, H. W., and Walkley, K. B., "NumericalMethods and a Computer Program for Subsonicand Supersonic Aerodynamic Design andAnalysis of Wings With Attainable ThrustConsiderations," NASA-CR-3808, 1984.

41 McCullers, L. A., "AWAVE: User's Guide for theRevised Wave Drag Analysis Program," NASALangley Research Center, Hampton, VA,September, 1992.

42 Harris, R. V., Jr., "An Analysis and Correlation ofAircraft Wave Drag," NASA TMX-947, March,1964.

43 Middleton, W. D., and Lundry, J. L., "A Systemfor Aerodynamic Design and Analysis ofSupersonic Aircraft," Part 1 - GeneralDescription and Theoretical Development, NASACR-3351, 1980.

44 Middleton, W. D., Lundry, J. L., and Coleman, R.G., "A System for Aerodynamic Design andAnalysis of Supersonic Aircraft," Part 2 - User'sManual, NASA CR-3352, 1980.

45 Middleton, W. D., Lundry, J. L., and Coleman, R.G., "A System for Aerodynamic Design andAnalysis of Supersonic Aircraft," Part 3 -Computer Program Description, NASA CR-3353, 1980.

46 Middleton, W. D., and Lundry, J. L., "A Systemfor Aerodynamic Design and Analysis ofSupersonic Aircraft," Part 4 - Test Cases, NASACR-3354, 1980.

47 Velicki, A. "Materials and Structures for theHSCT," Aerospace Engineering, April, 1995.

48 Phone conversations with K. McIver, McDonnellDouglas Aerospace, Long Beach, CA, January,1996.

Effects of alternative wing structural concepts on high speed civil transport life cycle costs

Documents

Transcript of Effects of alternative wing structural concepts on high speed civil transport life cycle costs