AN APPROACH TO ASSESS AIRFRAME WEIGHTREDUCTION PERFORMANCE OF STITCHED COMPOSITE

TECHNOLOGIES

Jason A. Corman∗ , Dimitri N. Mavris∗∗Georgia Institute of Technology

Keywords: Stitched Composites, PRSEUS, Structural Weight Estimation, Conceptual Aircraft Design,Aircraft Technology Performance

Abstract

An approach is presented for assessing perfor-mance of stitched, resin-infused (S/RI) compos-ite airframe technologies. This method considersboth direct and enabling effects of the technolo-gies to reduce structural weight early in the con-ceptual phase of aircraft design. Weight reduc-tion estimates are provided for a S/RI compositetechnology applied to a hybrid wing body (HWB)aircraft. These estimates are found as a functionof: 1) conceptual phase design variables that de-fine the shape, or outer mold line (OML), of theaircraft, 2) structural layout design variables toaccount for design uncertainty, and 3) updatedstructural properties that result from physical ex-perimentation.

1 Introduction

As design in the aircraft industry pushes theboundaries of current capabilities, evolutionaryand revolutionary concepts and technologies areinvestigated to meet market and regulating de-mands of cost, performance, and safety. TheHWB was shown to be an appropriate candi-date for improvements in performance and op-erating costs through the U.S. FAA Continu-ous Low Energy, Emissions, and Noise (CLEEN[14]) and NASA Environmentally ResponsibleAviation (ERA [2]) programs [17]. This con-figuration, however, is outside the scope of tra-ditional conceptual design models that are em-

pirically built with historical production data, es-pecially for airframe weight. Therefore, aircraftperformance estimates are made with a higherlevel of uncertainty and risk, and airframe weightestimation must be implemented through alterna-tive methods.

Comparative improvements for the commer-cial HWB configuration over traditional tube-and-wing (T&W) aircraft were supported bypredicted superiority in aerodynamic efficiency(L/D) [10]. However, internal pressurizationof the cabin section for the HWB presentedstructural challenges for its non-circular cross-section. For example, traditional structural con-cepts would potentially need to be over-sized inthe skin of the HWB centerbody cabin sectionto prevent fatigue of cyclic pressurization whilewithstanding combined bending loads from: 1)lift on the outboard wings and 2) internal pres-sure acting on a relatively flat surface. The re-sulting increased structural weight could counter-act the improvements in aerodynamic efficiency,which justified the need for an appropriate struc-tural technology. Trades to assess the feasibilityand viability of a commercial HWB configura-tion as well as the enabling impact of a structuraltechnology should be made in the early concep-tual design phase. Therefore, a process is neededthat estimates structural technology performancethat considers the effect of conceptual phase ve-hicle design variables. The following sectionsdescribe an approach to characterize the perfor-

1

JASON A. CORMAN , DIMITRI N. MAVRIS

mance of a structural technology in the contextof aircraft structural weight reduction.

2 Structural Technology Performance ofStitched Composites

One of the major benefits of current state of theart (SoA) composites for airframe manufactur-ing is the ability to co-cure stiffened, thin shelledstructural configurations. Co-curing alleviatesthe need for rivets and other fasteners, which cansignificantly reduce structural weight. The draw-back to this configuration, however, is its inabil-ity to arrest damage propagation, effectively re-moving the tear strap implemented between thethin wall and stiffeners in metallic constructedaircraft. Therefore, delamination as a result ofHWB centerbody out-of-plane loads is a criticalfailure mode of SoA aircraft composites.

2.1 PRSEUS Characteristics

A proposed solution for the commercial HWBcenterbody was a structural technology in theclass of stitched, resin-infused (S/RI) compos-ites: the Pultruded Rod Stitched Efficient Uni-tized Structures (PRSEUS), shown in Fig. 1. Thisbreakthrough in manufacturing leverages stitch-ing as a mechanism of damage arrestment andcan therefore use a damage tolerant design phi-losophy rather than the safe-life philosophy re-quired of SoA composites [3]. It also addressedother concerns of traditional composites, such asfastener pull-through and debonding.

Fig. 1 PRSEUS attributes [19]

The following describe PRSEUS technologyfeatures and how they contribute to estimatedstructural weight reduction.

Stitching and Unitization Composite stitchingenables progressive local failures, shift-ing of load paths, and suppression of de-lamination and laminate pull-off failuremodes, which are all mechanisms that con-tribute to structural weight reduction [6].Additionally, unitization of the PRSEUSstructural configurations enables continu-ous load paths at stiffener-to-stiffener in-terfaces with decreased stress concentra-tions while also decreasing the number offasteners required for assembly [19]. Aunitized and stitched stiffener constructionalso creates a panel that can operate locallyin the post-buckling regime. Therefore, lo-cal skin buckling is no longer a means ofcritical failure since the load is transmittedefficiently through the stiffeners [7, 1].

Warp-Knit Fabric and CAPRI Carbon fibercomposite materials have much higherstiffness-to-density and strength-to-densityratios and are less prone to fatigue com-pared to conventional aluminum aircraftmaterials [15]. These attributes, withoutother considerations of the design process,allows for more weight efficient design.Therefore, composite construction that be-haves like damage arresting conventionalaluminum construction takes full advan-tage of these better performing ratios. Thedesign of the warp-knit fabric enablesstitching and the benefits associated withit, and the controlled atmospheric pressureresin infusion (CAPRI) process eliminatesthe need for autoclave curing. Withoutthe need for an autoclave, much largerpanels can be fabricated, which savesweight through decreasing the number ofrequired panel joints, and a reduction ofmanufacturing costs can also be achieved[19].

Foam Core Frame The frame acts similar to

2

AN APPROACH TO ASSESS AIRFRAME WEIGHT REDUCTION PERFORMANCE OF STITCHEDCOMPOSITE TECHNOLOGIES

a hat stiffener, increasing the moment ofinertia of the cross section to counteractbending. A foam core is used for both man-ufacturing and performance functionality.The foam expands during cure, providingan inner surface to contain resin and main-tain shape. A secondary function is to pro-vide stability to the frame web by resistinginward deflections during loading.

Rod Stringer The rod stringer provides signif-icant bending stiffness compared to tra-ditional integral blade stiffeners or J-stiffeners. Its pultruded rod is dominatedby 0-degree fibers to increase stiffness andstrength in the stringer direction, and thepultrusion process ensures straightness ofthe rod to aid in stability by maintainingstringer shape.

These characteristics all provide detailed struc-tural information as to why a stitched, resin-infused composite may be more structurally ef-ficient; however, they offer little insight into theactual structural weight reduction that can beachieved through PRSEUS implementation. Atransfer function is needed that connects struc-ture details to this performance metric, which isdescribed next.

2.2 Performance Estimation

To ensure consistency with conceptual designmodels, technology performance at the aircraftlevel is tied to the aircraft structural weight.Characterization of this performance, i.e. struc-tural weight reduction, requires a basis of com-parison for the technology. The baseline for com-parison is an important definition because it alsomust be consistent with conceptual design mod-els, especially empirical models built from histor-ical data like NASA’s Flight Optimization Sys-tem (FLOPS) [12]. The following convention foraircraft level structural technology performanceis defined as a weight reduction:

∆WS =WS,B −WS,T (1)

where WS,B is the airframe structural weight ofan aircraft implemented with baseline structure,

and WS,T is the airframe weight of an aircraft im-plemented with the technology. This conventionwas used so that “increased performance” couldbe denoted as an increase to the value for ∆WS,and therefore, positive values of performance de-note a decrease in structural weight.

Implementation of performance impact inconceptual design is defined as a scale factor onweight, or:

kST = 1− ∆WS

WS,B=

WS,T

WS,B(2)

In FLOPS, historical data points for traditionalT&W aircraft regression models encompass air-frames that were built with some type of thin-shelled skin-stiffened aluminum. Therefore,proper definition of baselines is important toavoid potential mishaps in bookkeeping. For ex-ample, consider a scenario in which the base-line chosen for higher fidelity technology per-formance estimation was a different compos-ite structure, which was the case in the bench-mark PRSEUS performance estimation study[19]. However, the conceptual design model wasgenerated with data from aluminum-based air-craft configurations. In this case, technology per-formance is defined as:

∆WS,total =WS,CD −WS,T

= (WS,CD −WS,B)+(WS,B −WS,T )

= ∆WS,B +∆WS,T (3)

where WS,CD is the structural weight estimatedby the traditional conceptual design model, and∆WS,B represents performance of the compositebaseline over the aluminum “baseline” in the tra-ditional conceptual design model.

If structural technology performance is tobe characterized for conceptual design, thereare other factors that contribute to the potentialweight reduction. For instance, variation in outermold line (OML) or structural layout (SL) for agiven aircraft could lead to variation in structuralweight due to: external aerodynamic load distri-butions, fuel distribution, internal load paths, etc.If the variation in structural weight (∂WS/∂X) is

3

JASON A. CORMAN , DIMITRI N. MAVRIS

BenchmarkImplementation

NotionalPerformanceFunction

BenchmarkImplementation

NotionalPerformanceFunction

Fig. 2 Potential for error in technology perfor-mance in a scalar vs. functional approach

different for the baseline structure and the tech-nology, then by definition, performance wouldvary throughout the design space, X.

Figure 2 shows the potentially incorrect es-timation of technology performance if a scalarcharacterization is performed, i.e. for a singlebaseline outer mold line

(X∗

OML), rather than a

functional characterization throughout the OMLdesign space. A similar phenomena could oc-cur in the structural layout design space

(X∗

SL).

The research presented in this paper describes amethodology for a complete, traceable character-ization of an S/RI composite technology, includ-ing effects of the design space itself. This ap-proach also provides insights to the probability oferror in performance estimation when comparinga functional versus baseline scalar performanceestimation technique.

3 Approach for Performance Estimation

Obtaining traceable estimates of structural tech-nology performance as a function of the designspaces mentioned in the previous section requiresmore advanced structural weight estimation tech-niques than empirical models built from histor-ical data. Physics-based weight estimation wasused in the benchmark approach for scalar char-acterization of PRSEUS weight reduction perfor-mance [19]. The global-local approach describedby Velicki considers a single baseline aircraftstructural model to which assessments in technol-ogy performance are made. Achievement of ∆WSestimation as a function of the OML and SL de-sign spaces is critically enabled by a fully para-

Weight Function

Failure ModeMargin Function

Benchmark Estimate

XSL*xSL

Hypothesis(Notional)

BenchmarkImplementation

Out

er M

old

Lin

e(C

once

ptua

l Des

ign)

Str

uctu

ral L

ayou

t(C

once

ptua

l/P

reli

m D

esig

n)

Tech

nolo

gyP

reli

m/D

etai

led

Des

ign

Fig. 3 Hierarchy of structural technology perfor-mance characterization

metric finite element modeling framework, likethat described in Refs. [9, 5].

A decomposition of this functional charac-terization problem is shown in Fig. 3. The toplevel shows XOML design space, which directlyconnects to conceptual design tools. The middlelevel is the structural layout design space, XSL,which is typically fluid and a source of epistemicuncertainty in the conceptual design phase. How-ever, a definition of XSL is required for the gen-eration of a global finite element model (GFEM)representation of the aircraft in a physics-basedweight estimation process. The lowest level is thetechnology design space, XT , or the componentlevel. In physics-based structural weight estima-tion, these design variables exist for every dis-crete sizing component in the GFEM, and theirvalues are calculated through a constrained opti-mization problem, i.e. structural sizing.

Setting up this mapping is the first step indeveloping a systematic, traceable, and repeat-able process to characterize performance for anyweight-reducing structural technology. Begin-ning at the technology level, if the characteris-tics of structural technology performance, ∆WS,at a given design space level indicate a poten-tially significant relationship with the conceptualdesign space, XOML, then performance should becharacterized at the next higher level. There-

4

AN APPROACH TO ASSESS AIRFRAME WEIGHT REDUCTION PERFORMANCE OF STITCHEDCOMPOSITE TECHNOLOGIES

ChooseModel

Define Design Spaces

TechnologyLevel

StructuralLayoutLevel

Outer MoldLineLevel

ImplementPerformance

CharacterizeExperiments

Tollgate Framework

Step 1: Step 2: Step 3: Step 4: Step 5: Step 6: Step 7:

Performance Evaluation ( )

i: Establish Requirements

ii: Evaluate Characteristics

iii: Assess Availability

iv: Make Decision

A: Sample Design Code

B: Evaluate Potential Indicators

C: Create Surrogate Model

D: Assess Aggregate Design Space

E: Explore Design Space

F: Down-select Design Variables

G: Make Decision

a: Enumerate Potential Objectives

b: Identify Metric Trends

c: Determine Causality in Design Space

d: Trace Causality to Root Source

Fig. 4 A tollgate approach for structural technology performance characterization and implementation

fore, evidence supporting or refuting hypothesesat each design space level is built on results gen-erated at the current design space level and allthose below it.

This mapping establishes a tollgate formula-tion for the structural technology to mitigate un-necessary modeling and to generate observationswhich help decide at each step if moving forwardis appropriate. On the right side of Fig. 3, movingup the design space levels requires a significantlylarger turnaround time, both in model setup andexecution. This mapping also enables traceabil-ity and allows for a decoupling of potential ef-fects when moving from one design space levelto the next. For example, the impact of structurallayout level design variables, XSL, on weight andperformance metrics, WS,T or ∆WS, are the re-sult of a summation of effects from all consideredstructural components in the entire aircraft. Un-less the structural layout is held constant betweenthe baseline configuration and the technology-infused configuration, then single-panel effectscannot be compared between configurations. Ad-ditionally, local loading conditions cannot be ex-plicitly controlled in a full aircraft model - theyare a function of the global aircraft load cases andstructural properties.

The success of this mapping is dependent onthe ability to systematically identify whether theconceptual design space has a consequential im-pact on structural technology performance and,in turn, determine whether that impact introduces

risk in conceptual design studies and experimentdesign for technology development and demon-stration. Development of this systematic ap-proach follows a top-down decision support pro-cess described in Ref. [11]. This was found to bean appropriate framework because the mappingin Fig. 3 is set up as a series of decision points,e.g. the decision to move forward to a more com-prehensive characterization of structural technol-ogy performance.

The resulting methodology is shown in Fig. 4,and is described in further detail in Ref. [4]. Thisapproach was used in a functional characteriza-tion of the PRSEUS technology, which is de-scribed in the following sections.

4 Test Case

For this research, structural technology perfor-mance is estimated for the PRSEUS technol-ogy on a commercial configuration HWB air-craft. Two baseline OML configurations wereconsidered, NASA’s generic HWB301 [13] andthe NASA/Boeing N2A [8], and their planformsare shown in Fig. 5. Design space explorationwas considered for the latter.

The design space considered for the XOMLand XSL levels of the mapping is shown in Ta-ble 1. A down-selection was performed and itwas determined that the outboard wing parame-ters contributed to a significant amount of vari-ability across all sections of the HWB, e.g. cen-

5

JASON A. CORMAN , DIMITRI N. MAVRIS

Boeing/NASA N2A NASA HWB301

Fig. 5 Boeing/NASA N2A and NASA HWB301planforms

terbody (cb), rear centerbody (rcb), trapezoidalwing (tw), and outboard wing (ow). To pro-vide a basis of comparison similar to the bench-mark technology performance estimation process[19], orthogrid stiffened sandwich composites(12 variables in XT,B) and integrally blade stiff-ened composites (6 variables in XT,B) were ap-plied to the centerbody and wing sections (tw andow) of the baseline vehicle (XB). The same lami-nate stack was used for both PRSEUS (17 designvariables in XT,T ) and baseline composite config-urations; however, knockdown and non-optimalfactors on allowables for strength and stabilitywere appropriately tuned for the PRSEUS con-cept.

Table 1 OML and structural layout design vari-ables considered for N2A outboard wing

Variable Description

S Planform areaAR Aspect ratioT R Taper ratioΛLE Leading edge sweepcFS Front spar % chordcRS Front spar % chordprib Rib spacing (pitch)

5 Results and Observations

The tollgate process shown in Fig. 4 was fol-lowed for this test case. In Step 1, the mod-eling and simulation environment developed byLaughlin was used [9], and Step 2 was performedfor the test case setup in the previous section. At

the technology level, a simple non-FEM flat platewas modeled in Collier Research’s Hypersizerin order to test different loading scenarios [16].The two baseline HWB configurations, N2A andHWB301, were used in for the structural layoutlevel.

At the XOML level, different assumptions re-garding the treatment of XSL were tested. First,an “Uninformed” approach kept the same arbi-trary structural layout for both the baseline andtechnology vehicle configuration. A “Bench-mark” approach was also tested, in which thestructural layout was optimized for a baselineOML configuration, XOML,B, and kept con-stant for both technology and baseline structuresthroughout the OML design space. Finally, the“STEED” treatment, representing the StructuralTechnology Evaluation for Experiment Designmethodology [4], optimized the structural layoutfor every point in the XOML,B and XOML,T for thebaseline and technology configurations, respec-tively. With this approach, enabled by a para-metric FEM-based weight estimation frameworkand surrogate modeling, technology characteris-tics that enable more efficient structural layoutscould be examined.

5.1 Technology Level

The objective of Step 3 was to assess structuraltechnology performance as an aggregate of thetechnology level design space that was consid-ered. Surrogates were developed for structuralweight of each of four panels that were modeledat this level, single panel baseline and technologyrepresentations of the centerbody and wing, andperformance for the representations of the wingand centerbody were calculated through:

∆WS,w =WS,blade(xT∼ f (OML,SL)

)−WS,PRSEUS,w

(xT∼ f (OML,SL)

)∆WS,cb =WS,orthogrid

(xT∼ f (OML,SL)

)−WS,PRSEUS,cb

(xT∼ f (OML,SL)

)(4)

respectively, where xT∼ f (OML,SL) represents thetechnology level variables that directly map to

6

AN APPROACH TO ASSESS AIRFRAME WEIGHT REDUCTION PERFORMANCE OF STITCHEDCOMPOSITE TECHNOLOGIES

the OML and SL design spaces, e.g. com-ponent length, component width, and boundaryconditions like loads. A 50,000-case MonteCarlo simulation (MCS) was performed witheach of the surrogates – WS,blade, WS,PRSEUS,w,WS,orthogrid , and WS,PRSEUS,cb – to represent per-formance distributions, ∆WS,w and ∆WS,cb, for theXT∼ f (OML,SL) design space. The functions rep-resented in Eqn. 4 were generated with cases inwhich all other variables in XT for each structureconcept were optimized for weight.

Aggregate results representative of the cen-terbody concept are shown in Fig. 6 for a flatpanel pressure loading condition, P. Each plotshows a set of histograms with these single vari-able influences. Each histogram shows resultsfrom a Monte Carlo simulation with a uniformdistribution of a single variable of the corre-sponding color: Red - panel length, Green - panelwidth, and Blue - local pressure magnitude. Thebounds for each of these variables was ±50%of their nominal values. Each plot representsa different default value of the other two vari-ables in XT∼ f (OML,SL) corresponding to a per-centage level of its respective range above itslower bound: Top - 25% and Bottom - 50%.

The distribution of histograms in Fig. 6 isunique to the bounds placed on the variables inXT∼ f (OML,SL), and therefore, the following ob-servations are made with these bounds in mind.The first and obvious observation is that the vari-ation of weight reduction performance is decid-edly non-zero in each histogram. Any variabil-ity in performance as a function of XT∼ f (OML,SL)warrants investigation of the next level since thetechnology level is the furthest removed fromOML.

Variation in centerbody performance remainsrelatively consistent for each level of the other de-faulted parameters. This trend means that a simi-lar gradient exists for the orthogrid and PRSEUSconcepts, i.e. ∂WS/∂x for x in XT∼ f (OML,SL).However, a slight increase in spread of the his-tograms exists in the bottom plot. Therefore, athigher levels of pressure, there is a greater chancethat panel sizes exist in which PRSEUS perfor-mance becomes negative, i.e. structural weight is

" WS,i,cb

(P) for XT,j

-40 -20 0 20 40 60 80 100

Pro

babi

lity

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18

0.2

Panel Length (lpanel

)

Panel Width (wpanel

)

Load, P

" WS,i,cb

(P) for XT,j

-40 -20 0 20 40 60 80 100

Pro

babi

lity

0

0.05

0.1

0.15

0.2

0.25

0.3

Panel Length (lpanel

)

Panel Width (wpanel

)

Load, P

Fig. 6 Effect of panel dimensions on ∆WS undera panel pressure load

increased by implementing PRSEUS rather thanthe baseline orthogrid concept. Increased struc-tural weight is likely for smaller panels in whichthe pressure loading has less of an impact onthe maximum bending stresses. This trend isan important consideration for design parameterslike centerbody bay width and rib spacing, whichdefine panel dimensions for pressure containingcomponents and have a larger affect on perfor-mance variability within the bounds examined.

5.2 Structural Layout Level

A 3-level full factorial design of experiments(DoE) was performed on the structural layoutparameters identified in Table 1. While frontand rear spar locations are historically defined byconsiderations for high lift and control devices,it was assumed that an appropriate configurationcould be achieved for each design in XSL. Fig-ures 7 and 8 show resulting data from the DoE forthe N2A and HWB301 configurations, respec-tively. Each plot shows front spar and rear spar

7

JASON A. CORMAN , DIMITRI N. MAVRIS

40

Front Spar Location

3020

N2A Technology Performance

105560

Rear Spar Location

6570

75

18

17

16

15

14

13

21

20

19

22

80

Tot

al"

W(%

)

2 ft

Rib Spacing

3 ft

4 ft

40

Front Spar Location

3020

N2A Baseline

105560

Rear Spar Location

6570

75

1. 02

1

0. 98

0. 96

1. 12

1. 1

1. 06

1. 08

1. 04

80

Tot

al W

/Wno

m

2 ft

Rib Spacing

3 ft

4 ft

40

Front Spar Location

3020

N2A PRSEUS

105560

Rear Spar Location

6570

75

0. 82

0. 88

0. 86

0. 85

0. 87

0. 84

0. 83

80

Tot

al W

/Wno

m,B

2 ft

Rib Spacing

3 ft

4 ft

Fig. 7 Total ∆WS (XSL = XSL,B = XSL,T ) forN2A configuration enabled by PRSEUS

location plotted along the x-axis and y-axis, re-spectively, and each surface represents a differentrib spacing: prib = 2 ft (blue), prib = 3 ft (green),and prib = 4 ft (red). Values of the design vari-able ranges for this experiment were chosen toreflect the conventional wing structure reviewedin Ref. [18].

For the N2A, the top plot shows a total rangeof performance from 13.1% structural weightsaved by PRSEUS to a potential 21.4%, result-ing in a potential difference of 15,207 lbs usingthe benchmark definition of performance. Weightreduction at the nominal design point was es-timated at 16,371 lbs, and therefore, the totalrange of difference throughout the design spaceis nearly 67.5% of the benchmark predicted per-formance value. An error of this magnitude issignificant and supports the consideration of vari-ation in structural layout for structural technol-ogy performance estimation as a function of theOML.

Performance in the top plot is the direct sub-traction of the bottom right plot from the bottomleft plot, which show the total structural weightof the PRSEUS and baseline structural con-cepts, respectively. These plots are normalizedby the nominal total baseline structural weight,WS,B (cFS = 25%, cRS = 67.5%, prib = 3 ft) for

40

Front Spar Location

3020

HWB301 Technology Performance

105560

Rear Spar Location

6570

75

17

16

15

14

22

21

19

20

18

80

Tot

al"

W(%

)

2 ft

Rib Spacing

3 ft

4 ft

40

Front Spar Location

3020

HWB301 Baseline

105560

Rear Spar Location

6570

75

1. 01

1

0. 99

0. 98

0. 97

0. 96

1. 04

1. 03

1. 02

1. 05

80

Tot

al W

/Wno

m

2 ft

Rib Spacing

3 ft

4 ft

40

Front Spar Location

3020

HWB301 PRSEUS

105560

Rear Spar Location

6570

75

0. 88

0. 8

0. 78

0. 76

0. 74

0. 86

0. 84

0. 82

80

Tot

al W

/Wno

m,B

2 ft

Rib Spacing

3 ft

4 ft

Fig. 8 Total ∆WS (XSL = XSL,B = XSL,T ) forHWB301 configuration enabled by PRSEUS

ease of comparison. PRSEUS total structuralweight, shows a higher degree of nonlinearity incomparison to the baseline, and the surfaces crossat multiple points in the design space, signifyingfurther nonlinearity as a function of rib pitch.

Also noted is the minimum weight de-sign point among any configuration forboth baseline and technology-infused N2Aaircraft: the PRSEUS configuration withxSL,T = [cFS = 10%, cRS = 55%, prib = 4 ft].PRSEUS performance for the total aircraftstructural weight at this point is ∆WS = 17.9%,which shows that the best structural design doesnot necessarily correspond to the configurationwith best technology performance. In this case,the technology enabled a better performingstructural layout compared to the most optimalbaseline configuration of xSL,B = [cFS = 10%,cRS = 67.5%, prib = 2 ft]. Under the assump-tions of the model, both configurations seemedto show better performance for the technologyand lower structural weight for the entire aircraftat a front spar location of 10% chord.

The other OML design point considered wasthe HWB301, and the same set of results areshown in Fig. 8. The first observation of the per-formance surfaces is the separation of surfacesat the region xSL,T = [cFS = 10%,cRS = 80%],

8

AN APPROACH TO ASSESS AIRFRAME WEIGHT REDUCTION PERFORMANCE OF STITCHEDCOMPOSITE TECHNOLOGIES

showing nonlinearity in performance as a func-tion of rib spacing. This region is also theset of designs with largest baseline and technol-ogy structural weights and the smallest reduc-tion in weight for the PRSEUS technology on theHWB301. The minimum weight solution for theHWB301 exists at the point of greatest PRSEUSperformance, a departing trend from the N2A.Using the benchmark approach in this case, solong as topology optimization was performed indesigning the structural layout, would generatethe same estimate as an approach in which theeffort was take to optimize both baseline andtechnology configurations separately. The differ-ences between the two baselines demonstrate thevalue of examining XSL independently of XOMLin the 2nd level of the mapping. These observa-tions help delineate SL performance trends fromOML performance trends and make a case for ex-amining the OML space with optimization of thestructural layout at each design point, xOML.

5.3 Vehicle Level

The last step (Step 5) in the tollgate processshown in Fig. 4 represents the XOML level inframework mapping. At the beginning of thissection, three different treatments of structurallayout in the OML design space were described.Figure 9 shows a comparison of the aggregatetechnology performance (top) and the aggregatestructural weights of baseline and PRSEUS con-figurations (bottom) with the three treatments.The histograms are the result of a 100,000-case MCS of uniformly distributed parameters inXOML (also XSL for the “Uninformed” approach)and are shown with transparency for clarity on alldistributions.

The top plot shows that an “Uninformed”treatment of the SL comparatively over-predictsperformance on average and has a much higherspread on the predicted weight reduction capa-bility of the PRSEUS technology. This treatmentis somewhat unrealistic but may be required withlimited parametric modeling capabilities. Com-paring the “Benchmark” and “STEED” treat-ments shows that holding the structural layout

"WS,total

(%)-20 -15 -10 -5 0 5 10 15 20 25 30

Pro

babi

lity

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

UninformedBenchmarkSTEED

WS,T,total

(lb) #1050.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7

Pro

babi

lity

0

0.01

0.02

0.03

0.04

0.05

0.06

Uninformed BaselineUninformed PRSEUSBenchmark BaselineBenchmark PRSEUSSTEED PRSEUS

Fig. 9 Technology performance (top) and totalstructural weights (bottom) in XOML

constant between the baseline and PRSEUS con-figurations limits the performance of the technol-ogy throughout the design space, even thoughthe structural layout has been optimized for thebaseline at each design point. Only by optimiz-ing the structural layout for both configurationsin the “STEED” treatment was the full weightreduction potential of PRSEUS achieved. Thistrend is further displayed by the left shift of the“STEED” histogram (yellow) compared to the“Benchmark” (green) in the bottom plot, whereeach histogram represents a different baseline orPRSEUS structural weight distribution within thedesign space.

An alternative depiction of PRSEUS per-formance throughout the OML design space isshown in Fig. 10. Each scatter plot showsPRSEUS performance (y-axis) as a function ofthe total weight of the PRSEUS configured air-craft. The top plot shows the comparison ofthe “Uninformed” (grey) vs. “Benchmark” (red)treatments of the structural layout. By optimiz-ing the structural layout for the baseline config-

9

JASON A. CORMAN , DIMITRI N. MAVRIS

Fig. 10 Technology performance vs. total air-craft structural weight

uration in the “Benchmark” treatment, unrealis-tic performance estimates are mitigated. If theSL was set arbitrarily for each design point inthe OML for both baseline and PRSEUS config-urations, it could be interpreted that decreasedPRSEUS performance contributes to increasedaircraft structural weight. However, the “Bench-mark” and consequently “STEED” (bottom plot)treatments of the structural layout show other-wise.

By considering separate optimization ofstructural layouts for both baseline and tech-nology configurations (“STEED” treatment), itis shown that: 1) structural weight of the ag-gregate design space is decreased, 2) increasedPRSEUS performance does not translate to theminimum weight design in the XOML space, and3) there is a general increase in performancethroughout the design space, which substantiatesthe observations from Fig. 9. These observa-tions warranted the “STEED” treatment of the SLin characterization of technology performance,and therefore the weight reduction estimates with

0 = "WTO

(XOML

,kS-k

B) (%)

-8 -6 -4 -2 0 2 4 6 8 10 12

Pro

babi

lity

0

0.005

0.01

0.015

0.02

0.025

0.03

0.035

0.04

0.045

0.05

Fig. 11 Error in Benchmark estimates of WTO byneglecting ∆WS ∼ f (XOML)

the STEED approach were fit with a surrogatemodel: kST ∼ f (XOML), where kST is definedin Eqn. 2.

This functional PRSEUS performance esti-mate was implemented in a FLOPS model forthe N2A, which was given the same missionand explored different designs in XOML to assesschanges in conceptual design metrics, e.g. fuelburn (WF), takeoff gross weight (WTO), etc. Fig-ure 11 shows the estimated error in WTO through-out the design space by implementing a scalarPRSEUS performance value and neglecting thefunctional relationship of performance with theOML.

6 Conclusions

This paper provided an approach for S/RI com-posite performance estimation that is traceable tothe lowest level technology design space. Theapproach provides insights to the source of tech-nology weight reduction capability and how thetechnology performs in each level of the designspace mapping. This information can be lever-aged in the implementation of the technology inconceptual design trade studies, which reducespotential error and risk in aircraft performanceestimates. Additionally, the methodology couldbe used in physical experiment planning for tech-nology development and demonstration to iso-late particular phenomena that contribute to un-certainty in technology performance.

10

AN APPROACH TO ASSESS AIRFRAME WEIGHT REDUCTION PERFORMANCE OF STITCHEDCOMPOSITE TECHNOLOGIES

References

[1] A. C. Bergan, J. G. J. Bakuckas, A. E. Lovejoy,D. C. Jegley, J. Awerbuch, and T.-M. Tan. As-sessment of damage containment features of afull-scale prseus fuselage panel through test andteardown. In 27th ASC Technical Conference,2012.

[2] F. Collier, R. Thomas, C. Burley, C. Nickol, C.-M. Lee, and M. Tong. Environmentally respon-sible aviation: Real solutions for environmen-tal challenges facing aviation. In 27TH INTER-NATIONAL CONGRESS OF THE AERONAU-TICAL SCIENCES, 2010.

[3] R. C. J. Combes. Design for damage tolerance.Journal of Aircraft, 7:18–20, 1969.

[4] J. A. Corman. A methodology for structuraltechnology performance characterization to en-able reduction of structural uncertainty. PhDthesis, Georgia Institute of Technology, 2017.

[5] J. A. Corman, N. Weston, C. Friedland, D. N.Mavris, and T. W. Laughlin. Rapid airframedesign environment (rade): A parametric, mod-ular, and multidisciplinary framework for con-ceptual phase aiframe design. In 2018 AIAAModeling and Simulation Technologies Confer-ence, page 1930, 2018.

[6] E. Glaessgen, I. Raju, and J. P. C.C.Fracture mechanics analysis of stitchedstiffener-skin debonding. In 39thAIAA/ASME/ASCE/AHS/ASC Structures,Structural Dynamics, and Materials Confer-ence and Exhibit, 1998.

[7] D. C. Jegley and A. Velicki. Status ofadvanced stitched unitized composite aircraftstructures. In 51st AIAA Aerospace SciencesMeeting including the New Horizons Forum andAerospace Exposition, 2013.

[8] R. Kawai. Acoustic prediction methodology andtest validation for an efficient low-noise hybridwing body subsonic transport. Technical report,NASA ARMD Subsonic Fixed Wing ProjectContract NNL07AA54C, 2011.

[9] T. W. Laughlin. A parametric and physics-based approach to structural weight estimationof the hybrid wing body aircraft. Master’s the-sis, Georgia Institute of Technology, 2012.

[10] R. Liebeck. Design of the blended wing body

subsonic transport. Journal of Aircraft, 41:10–25, 2004.

[11] D. N. Mavris, A. P. Baker, and D. P. Schrage.Ippd through robust design simulation for an af-fordable short haul civil tiltrotor. 1997.

[12] L. McCullers. Flight optimization system, re-lease 6.12, user’s guide. NASA, 2004.

[13] C. L. Nickol. Hybrid wing body configurationscaling study. In 50th AIAA Aerospace SciencesMeeting including the New Horizons Forum andAerospace Exposition, 2012.

[14] P. Niskode, R. Stickes, B. Allmon, and R. De-Jong. Faa cleen consortium open session. Pre-sentation CLEEN Consortium Open Session,Georgia Institute of Technology, Atlanta, GA,October, 27, 2010.

[15] M. C.-Y. Niu. Composite airframe structures:practical design information and data. AdasoAdastra Engineering Center, 1992.

[16] C. Research. HyperSizer Rod/Bulb StiffenedPanel Family, User’s Manual Edition 5.9, 2010.

[17] J. S. Schutte and H. Jimenez. Technology ass-esment of nasa enviromentally responsible avi-ation advanced vehicle concepts. In 49th AIAAAerospace Sciences Meeting, 2011.

[18] M. D. Sensmeier and J. A. Samareh. A studyof vehicle structural layouts in post-wwii air-craft. In 45th AIAA/ASME/ASCE/AHS/ASCStructures, Structural Dynamics and MaterialsConference, 2004.

[19] A. Velicki. Damage arresting composites forshaped vehicles: Phase 1 final report. Techni-cal report, NASA/CR-2009-215932, 2009.

Acknowledgments

The authors would like to graciously thank Dr.Fayette Collier, Ms. Dawn Jegley, and Mr. AlexVelicki for their support and guidance with re-spect to the NASA ERA program and PRSEUSconcept.

Contact Author Email Address

jason.corman@gatech.edu

11

JASON A. CORMAN , DIMITRI N. MAVRIS

Copyright Statement

The authors confirm that they, and/or their companyor organization, hold copyright on all of the origi-nal material included in this paper. The authors alsoconfirm that they have obtained permission, from thecopyright holder of any third party material includedin this paper, to publish it as part of their paper. Theauthors confirm that they give permission, or have ob-tained permission from the copyright holder of thispaper, for the publication and distribution of this pa-per as part of the ICAS proceedings or as individualoff-prints from the proceedings.

12