Proceedings of a conference held in

Seattle, Washington

Review for general release October 29-November 1, 1990

January 31, 1993
 NASA Conference Publication 3104
Part 1

First NASA
Advanced Composites
Technology Conference

Compiled by
John G. Davis, Jr.
NASA Langley Research Center
Hampton, Virginia

Herman L. Bohon
Lockheed Engineering & Sciences Company
Hampton, Virginia

Proceedings of a conference sponsored by the

National Aeronautics and Space Administration,
Washington, D.C., and held in
Seattle, Washington
October 29-November 1, 1990

IXl/_A
National Aeronautics and
Space Administration

Office of Management
Scientific and Technical
Information Division

1991
 PREFACE

The First NASA Advanced Composites Technology Conference was sponsored

by the NASA Langley Research Center to review recent advances in research and
development of advanced composites technology for applications to military and
commercial aircraft. The NASA Advanced Composites Technology (ACT) Program is a
major new multiyear research initiative to achieve a national goal of technology
readiness before the end of the decade. This initiative is carried out through a
cooperative program between industry, universities, and the Government conducting
research in materials processing, analysis development, innovative designs and
manufacturing methodology.

The First NASA Advanced Composites Technology Conference was held in

Seattle, Washington, October 29 through November 1, 1990. The conference provided
a forum for the composites community to exchange information and an opportunity to
observe recent progress in advanced composites technology. Fifty-two papers were
organized into sessions that emphasized composite transport technology development,
advances in design and manufacturing, and research in materials and structural
mechanics. In addition, a session sponsored by the Department of Defense
emphasized lessons learned from current applications programs. This conference
publication is a compilation which contains the papers presented in these sessions.

The use of trademarks or manufacturers' names in this publication does not

constitute endorsement, either expressed or implied, by the National Aeronautics and
Space Administration.

John G. Davis, Jr.

Herman L. Bohon

°°.

III
 CONFERENCE ORGANIZATION

Structures Technology Program Office

Structures Directorate
NASA Langley Research Center
Hampton, VA

General Chairman Technical Chairman Administrativ_ Assistant

John G. Davis, Jr. Herman L. Bohon Stuart E. Pendleton
NASA Langley Lockheed Engineering Lockheed Engineering
& Sciences Company & Sciences Company

SESSION ORGANIZERS

John G. Davis, Jr. NASA Langley Research Center

James H. Statues, Jr. NASA Langley Research Center
Norman J. Johnston NASA Langley Research Center
David Beeler Wright Research and Development Center

iv
 SESSION OUTLINE

FIRST NASA ADVANCED COMPOSITES TECHNOLOGY

CONFERENCE

Session [

Composite Transport Technology

Chairman" Michael F. Card, Chief, Structural Mechanics Division
NASA Langley Research Center

Session I/

Advances in Design and Manufacturing

Chairman: Darrel Tenney, Chief, Materials Division
NASA Langley Research Center

Session I1_[

DoD Composites Applications

Chairman: David Beeler, Head, Nonmetals Branch
Wright Research and Development Center

Session IV

Materials Research
Chairman: Norman J. Johnston, Chief Scientist, Materials Division
NASA Langley Research Center

Session v

Work in Progress - Materials and Structures

Chairman: W. Tom Freeman, ACT Structures Technology Program Office
NASA Langley Research Center

Session VI

Structural Mechanics Research

Chairman: James H. Starnes Jr., Head, Aircraft Structures Branch
Structural Mechanics Division, NASA Langley Research Center

V
Consolidation of Graphite/Thermoplastic Textile Preforms for Primary
Aircraft Structure ............................... 293
J. Suarez and J. Mahon

Cost Studies for Commercial Fuselage Crown Designs ............. 339

T. H. Walker, P. J. Smith, G. Truslove, K. S. Willden, S. L. Metschan, and
C. L. Pfahl

A Unified Approach for Composite Cost Reporting and Prediction in the

ACT Program ................................. 357
W. Tom Freeman, Louis F. Vosteen, and Shahid Siddiqi

Session III

F-15 Composite Engine Access Door ...................... 371

Ramaswamy L. Ramkumar and James C. Watson

Fabrication of the V-22 Composite Aft Fuselage Using Automated

Fiber Placement ................................ 385
Robert L. Pinckney

Lessons Learned for Composite Structures ................... 399

R. S. Whitehead

Part 2*

Session IV

Preliminary Properties of a Resin From Ethynyl Terminated Materials ...... 419

Paul M. Hergenrother and John W. Connell

Infiltration/Cure Modeling of Resin Transfer Molded Composite Materials

Using Advanced Fiber Architectures ...................... 425
Alfred C. Loos, Mark H. Weideman, Edward R. Long, Jr., David E. Kranbuehl,
Philip J. Kinsley, and Sean M. Hart

Powder Towpreg Process Development ..................... 443

Robert M. Baucom and Joseph M. Marchello

Effects of Intra- and Inter-Laminar Resin Content on the Mechanical

Properties of Toughened Composite Materials ................. 455
Dodd H. Grande, Larry B. Ilcewicz, William B. Avery, and Willard D. Bascom

Studies of Fiber-Matrix Adhesion on Compression Strength ........... 477

W. D. Bascom, J. A Nairn, and D. J. Boll

*These papers are presented in NASA CP-3104, Part 2.

°,,

VIII
The Initiation, Propagation, and Effect of Matrix Microcracks in Cross-Ply
and Related Laminates ............................ 497
John A. Nairn, Shoufeng Hu, Siulie Liu, and Jong Bark

Comparison of Impact Results for Several Polymeric Composites Over

a Wide Range of Low Impact Velocities ..................... 513
C. C. Poe, Jr., M. A. Portanova, J. E. Masters, B. V. Sankar, and
Wade C. Jackson

Delaminations in Composite Plates Under Impact Loads ............. 549

Scott R. Finn and George S. Springer

Micromechanics of Fatigue in Woven and Stitched Composites ......... 579

B. N. Cox, M. S. Dadkhah, R. V. Inman, M. R. Mitchell, W. L. Morris, and
S. Schroeder

Characterization of Multiaxial Warp Knit Composites .............. 589

H. Benson Dexter, Gregory H. Hasko, and Roberto J. Cano

Development of Stitching Reinforcement for Transport Wing Panels . . . .... 621

Raymond J. Palmer, Marvin B. Dow, and Dona!d L. Smith

Session V

Development of Resins for Composites by Resin Transfer Molding ........ 647

Edmund P. Woo, Paul M. Puckett, and Shawn J. Maynard

Advanced Fiber/Matrix Material Systems .................... 659

J. Timothy Hartness

Mechanical and Analytical Screening of Braided Composites for

Transport Fuselage Applications ...................... 677
Mark J. Fedro, Christian Gunther, and Frank K. Ko

Ultrasonic Detection and Identification of Fabrication Defects

in Composites ................................. 705
Edward R. Long, Jr., Susan M. Kullerd, Patrick H. Johnston, and
Eric I. Madaras

Developments in Impact Damage Modeling for Laminated Composite

Structures .................................. 721
Ernest F. Dost, William B. Avery, Gary D. Swanson, and Kuen Y. Lin

Multi-Parameter Optimization Tool for Low-Cost Commercial Fuselage

Crown Designs ................................ 737
Zelda Zabinsky, Mark Tuttle, Douglas Graesser, Gun-In Kim,
Darrin Hatcher, Gary Swanson, and Larry Ilcewicz

Comparison of Hand Laid-Up Tape and Filament Wound Composite

Cylinders and Panels With and Without Impact Damage ............. 749
Dawn C. Jegley and Osvaldo F. Lopez

ix
Effects of Scale in Predicting Global Structural Response ............ 761
R. B. Deo and H. P. Kan

Design and Analysis of Grid Stiffened Concepts for Aircraft Composite

Primary Structural Applications ........................ 779
Damodar R. Ambur

Optimization of Composite Sandwich Cover Panels Subjected to

Compressive Loadings ............................ 791
Juan R. Cruz

A Comparison of Classical Mechanics Models and Finite Element Simulation

of Elastically Tailored Wing Boxes ....................... 809
Lawrence W. Rehfield, Richard D. Pickings, Stephen Chang, and Michael Holl

Advanced Fiber Placement of Composite Fuselage Structures .......... 817

Robert L. Anderson and Carroll G. Grant

Process and Assembly Plans for Low Cost Commercial Fuselage

Structure 831
Kurtis Will'den,' Siephen" IV{etschan," and" Val "Stark,ey ................

Session VI

Progressive Failure Methodologies for Predicting Residual Strength and Life

of Laminated Composites ........................... 843
Charles E. Harris, David H. Allen, and T. Kevin O'Brien

Multiple Methods Integration for Structural Mechanics Analysis and

Design .................................... 875
J. M. Housner and M. A. Aminpour

Probabilistic Composite Analysis ........................ 891

C. C. Chamis and P. L. N. Murthy

A Rayleigh-Ritz Analysis Methodology for Cutouts in Composite

Structures .................................. 901
Steven G. Russell

Effects of Bolt-Hole Contact on Bearing-Bypass Damage-Onset Strength ..... 921

John H. Crews, Jr. and Rajiv A. Naik

Prediction of Stiffener-Skin Separation in Composite Panels ........... 939

Han-Pin Kan, Mary A. Mahler, and Ravi B. Deo

Experimental Behavior of Graphite-Epoxy Y-Stiffened Specimens Loaded

in Compression
P. D. Sydow anci I_1._J."Sl_uart.......................... 953

Structural Response of Bead-Stiffened Thermoplastic Shear Webs ........ 969

Marshall Rouse

X
Evaluation of Some Scale Effects in the Response and Failure of
Composite Beams ............................... 979
Karen E. Jackson and John Morton

An Overview of the Crash Dynamics Failure Behavior of Metal and

Composite Aircraft Structures ......................... 1005
Huey D. Carden, Richard L. Boitnott, Edwin L. Fasanella, and Lisa E. Jones

Tailored Composite Wings With Elastically Produced Chordwise Camber ..... 1037
Lawrence W. Rehfield, Stephen Chang, Peter J. Zischka, Richard D. Pickings,
and Michael W. Holl

xi
Part 1
 Luncheon Address
* * *

COMPOSITES: A VIABLE OPTION

John E. McCarty

While it sounded great to be asked to talk about composites, I found it difficult to

select subject areas that would be of real interest. My choice is based on saying
some things about where the maturity of the composite aircraft structures is today
and what that means in terms of future criteria for application. This focus was the
basis for my title selection. The other issue that will be addressed was requested by
NASA and focuses on composites structures cost. This fits well with the state-of-
the-art interpretations I will discuss first, since the cost issue must be viewed from
both the current status and future points of view. The difficulty in presenting
something in these areas is not in the subjects themselves but in trying to present a
real world viewpoint to an audience of composite experts. So, with recognition of the
expertise of the audience, I hope you will see something in this presentation about
how to view composite aircraft structure.

Introduction

• NASA - introduction and vision

• Composite aircraft- option validation

• Aircraft structure

• Composite cost considerations

• Composite potential- commercial

3
 As noted, my initiation into the composite field and my association with NASA
began at basically the same time. My first trip to NASA Langley was to take over
contract management of Boeing's first composite contract with NASA that Reid June
had won and was managing. Boeing management had decided that Reid was to help
out on SST and I was to replace him on the NASA contract. My flight to NASA was
one of those that I am sure all of you who have traveled a lot have experienced. I
arrived at Newport News at 4:00 o'clock in the morning with the scheduled review
on the contract at 8:00 o'clock. As the picture shows, about a minute after Reid
introduced me as the new contract manager and the lights were dimmed for the
presentation, I caught up on my missed night's sleep. There were and are those
at NASA that have taken many opportunities to remind me of this, their initial
impression of me.

NASA- Introduction

I
0

iii_i..."+

_:;_ii::ii:.::::i::ii:._ii::_i::::ii_i;:
.....

'' %_i_i:!!?
I ii!il

4
 I feel it is only fair play that I give my friends at NASA the similar type of re-
minder they have often given me on my introduction to them. This chart is to
remind them about the NASA/ACEE program initial bid phase that was won by
Lockheed. As all company managements will, Boeing's wanted to know why they
lost the contract bid. Since Reid and I were the ones preparing the bid, we of course
went with our bosses to NASA to find out what we did wrong. There were several
reasons expressed by NASA for Boeing's loss. A simple one was that Lockheed just
had a better overall proposal. But one point that NASA focused on and that they
said was one of the keys for our loss was our use of the 737 vertical tail component,
since the 737 represented an aircraft with little future sales potential. My only re-
sponse to NASA now is to say I hope your vision into the future is better now.

NASA: Vision

5
 With this discussion of the past, between those I hope I can still call friends, let's
take a look at the status of composites technology from the point of view of the pre-
sentation title. I selected the title because I had a boss that said, "When a material
is considered during the preliminary design and product development phase of an
aircraft program on an equal basis with other material options, it has become a
viable option." I believe this is the status of composite materials today. What does
this mean in terms of the selection and potential usage on future commercial trans-
port aircraft. To understand this you need to understand the evolutionary changes
that take place in commercial aircraft and the role the service usage plays in the
design approaches taken. The selection of a viable material also means that it is
not selected because it is the fad or to start the technology learning process. The
scale on the chart emphasizes the equal weighing of all the structure material op-
tions on their merits. This means no bias criteria for or against a material systems
selection. I am sure all the members of the audience that have been promoters of
composites have experienced the imposition of special criteria on composites by those
not wanting to change from metals. Finally, the things that are needed to make the
change are a need to improve the product, an inhouse champion at the appropriate
management level, and a need in the marketplace for a new aircraft in a competitive
environment. The key here again is the acceptance of composites on an equal basis
and weighed on a scale with the same acceptance parameters.

Composite Structures Selection

• Commercial aircraft design environment

• Evolution
• Service time

• Selection from the options

• Criteria bias
• Weighing of options

• Making the change

• Need
• Champion
• Market

6
 This chart simply displays that we now have a significant history of composite
applications in military, civil, and commercial aircraft. This history also suggests
that with this background the technology base is such that the material is truly a
viable option for future designs.

Composite Structure Applications

Development _ _ 1_..-

material sub..._._S _f "t"

......._ ....

Fuselage / Milita_;_10 years ='/ _Bship

O Starship _tr_ --_,uLarger!

] applications -_=.--/ w ,,-.- / wlng.ane

Structural
Wing/ - AV-eB _ / 15 years ._ / Tuse,age=,.
weight, %
/ F-l* ,_/ ATR-1 _Transport
Stabilizers/ F-18_ AIH-I? and civil

Control / F-15 J A-320/41

surfaces ] /" /

fairings / F-14_ 0.I//"

F-111 ............. 767 j

Calendar years

7
 One of the questions often asked by management when stepping up to the selec-
tion of a new material system is "Do we know how to certify it?" This list of civil
and commercial aircraft components and aircraft that have been certified should em-
phatically answer that question in a positive manner. The certification process has
been experienced by all the large commercial airframe manufacturers. The FAA now
has the experience to deal with composite structural systems and to feel confident in
their certification process. This certification question can no longer be a basis for not
selecting composites on an equal basis with metals.

Certified - Applications
Key Primary Structure

• Transport category
• Aerospatiale/Aeritalia outboard wing 11/15/89
• Airbus A300-600 vertical stabilizer 3/28/88
• Airbus A310-300 vertical stabilizer 6/10/87
• Airbus A320 vertical and horizontal stabilizer 12/15/88
• DC-10 vertical stabilizer 6/O3/86
• Boeing 737 horizontal stabilizer 11 / 14/84

• Normal (civil) category

• Windecker Eagle entire aircraft 12/18/69

• Beech 2000 (Starship) entire aircraft 6/14/88

• Piaggo P-180 empennage, canard, and aft fuselage 5/07/90

8
 I show this picture of the Beech Starship since it is certified and in production.
It represents what can certaidy be ca!!ec! ar? a! composite aircraft. The issues of
fuel in the wing and a fuselage pressurized shell were both accounted for in its
certification process. With the production underway it will be the aircraft by which
all future composite civil aircraft will be measured. My salute to Beech Aircraft
Company for the development and certification of the Starship.

Starship
I

9
 Almost all my general or overview presentations have shown this picture, since
I believe it clearly dramatizes the size considerations that must be part of the step-
ping up to commitment of composites to commercial aircraft. The risk from a manu-
facturing and cost point of view is directly related to this size effect. The structural
mechanics are not different between fighters and commercial transports, but the
manufacturing differences are as great as the differences shown in this picture.
Those who have worked on large aircraft production readily recognize this size
effect.

Aircraft Size Considerations

G nlmrn
-5A

15

10
 This is just another way of expressing the effect of size. It simply says that as
expected there are major structural differences in the 767 rudder and the AV-8A
wing. ‘These differences are simply reflected in their weight differences, even though
their dimensions are very similar. The difference in their design requirements
obviously makes significant difference in their design details and the manufacturing
cost of those details.

Application Considerations

J
767 rudder
433 Ib / AV-8A wing box
1,078 Ib

I 28 f t a d
I 34 f t a d

11
 These two pictures show that the technology for design of composite components
covers not only the fighter size airframes but now covers wing spans that are similar
to large commercial aircraft. Therefore, to continue the argument that industry does
not have the background in large airframes is no longer a valid view of composite
airframe structures. This preceding discussion states simply and clearly that it is
now time and the industry is ready to complete the technology steps to the large
commercial aircraft of composites, a viable material option.

B=2 Bomber

Transport Size Transition to Composites

Northrop
flying wing 707 6-16 0-2

172
(52

1 _.
.
( l e lm)

t 45 years t
12
 The next step in assessing if the large commercial aircraft industry should step
up to the design and development of the final major components of wing and fuse-
lage lies with the need for new aircraft. The cycle for new commercial aircraft (those
not government subsidized) continues to grow in time. The past time cycle was 7
to 10 years and has more recently expanded to 12 to 15 years. This chart simply
shows that even with the new aircraft coming along now, there should be a market
for another new round of aircraft in the period 2005 to 2015. Therefore, the final de-
velopments needed for cost risk reduction in producing a commercial aircraft should
be addressed now. The current NASA ACT program, which this meeting is about,
has the right timing to aid the US commercial airframe companies to compete with
the rest of the world.

Market Forecast- Share by Airplane Size

Units and Dollars Seats

Total Market 1992 to 2005
• Less than 120 seats = 241 to 350
120 to 170 More than 350
3,000 300 171 to 240
!iiiii!i:;i::_
Units
140
Dollars
120
!ii!ii!i!ii!i_
2,000 !iiiiii; iiiiiiiili!
iii 200 100
I Historical __ _.F°recast_ .

1989
8O
dollars in
UI_ils iiiili!_iiii!ii ii!iiiiiiiili billions
Number
::::..:: of seats in 60
iiiiiii_;iiill
i i!iiii_!!iiii thousands
1,000 100
40

i_i!iiiiiii_
iiiiiiii!iiiii
_&_._
0
Less 120 171 241 More 1975 80 85 90 95 2000 05
than to to to than
120 170 240 350 350

13
 This chart shows that the cost of aircraft has followed a fairly well defined pat-
tern and that to compete in the future the cost of the aircraft must meet or beat this
established pattern. No company will risk the initiation of a new aircraft within its
own funding capabilities if it cannot be sure of its ability to not only project the cost
but to control the manufacturing cost after commitment to production. This scenario
leads directly to the reason that cost of composite structures is now the issue, not
structural technology. The evaluation and design of the structure will require the
continued development of better composite structural analysis tools. However, the
technology to manufacture that structure in an economical manner is the key to the
risk acceptance for the manufacturers. The measuring stick for assessing the suc-
cess of development of the manufacturing processes and tools will be the comparison
of the cost of large composite structures against the 50 to 60 year base for metals
structures, a tough comparison for composite structures.

Aircraft Cost Trends

Aircraft Costs 1988 to 1989

1B

/ c-1_ _ii_
TM

100M _ / / 7%_A00-
F 14 "/ .//_. A310
Unit cost,
/ n 1:]F-15 .,,"_'A325
1989 dollars AVaBQ/ "F-18 /_j/
/ i._ =,/_"
_ 737,'_,MD 82

10M
/ / /_, DHC'-8 _, s
/,/_ _ // ATR Current Aircraft Prices '88
..,_"" .\'o 2 H'60 t.,_'/2 Interavia Publishing

/// co,,
/ _ _#o#2_,, Summa_
o,.ajo,
ooowoa.on
S,.ste,.,s
/ / /// Oerense
N_,,s./17/89
2.0 1.0 0.5 0.4 0.3 Aviation Week 3/20/89
Cost, $K/Ib

1M I I I
1,000 10,000 100,000 1M
Empty weight, Ib

14
 This chart shows what we must consider in a plan to assess how to meet that
tough comparison with the manufacturing experience base of metal structures and
has been seen by many of you at other presentations and conferences, I show it here
because I believe in what it tells us so strongly. Our ability as structural engineers
and manufacturing engineers to save cost and weight decreases with program time.
As the program progresses to each following phase, opportunities for cost and weight
savings are lost. The first person to have the largest chance at cost and weight
savings is the configuration engineer as he draws the first three view drawings
and establishes the aircraft arrangement. From the aircraft arrangement comes
the structural arrangement against which various material systems and structural
concepts are assessed. There is complete freedom for structural considerations in
this stage of design development within the constraints of the airplane mission.
As can be expected, each following phase is constrained by decisions made in the
preceding phase. This scenario clearly suggests that cost and weight be a significant
consideration at the initial step as well as performance or mission requirements.

Program Schedule and Cost Savings

Decreasing Opportunities for Cost Savings

1
Dollars

Preliminary Product Production Production

design development design

Time

15
 This chart shows how the idea of the value of weight savings varies with the
same development cycle. During the initial phase it is not a very large number
since the optimism is that the design will easily meet weight projections and per-
formance requirements. When the cycle reaches the production drawing release, the
real weight numbers begin to appear and panic sets in to meet weight targets and
performance specifications. This raises the value of weight savings, and this is when
all the awards for weight savings are given. These two charts together suggest that
the awards in the production drawing release phase may add to the weight savings
needed to solve the overweight problem when recognized. However, since the sav-
ings are evaluated on a dollars/pound basis, this is a cost issue as well as a weight
issue and perhaps the award for both cost and weight saving should be issued in the
preliminary design phases as well. Simply stated, we may be giving awards at the
wrong time.

Program Schedule and Value of Weight Savings

Dollars
I

\
per Ib /,

Preliminary Product /
design developmen!t/
............................
"_ I Production

I design Production .....................

Time I=

16
 This chart (upper section) is one that I have used many times to focus on the
issues of static strength and damage tolerance of composites. The lower half of the
chart addresses the cost, considering both the manufacturing costs and the in service
maintenance cost. The bottom four items are of direct interest to this presentation.
The material cost is one of the issues that is being discussed at this conference.
The other is the so called "designed-in cost". Those are the costs that the designer
causes to occur due to the design he selects. The influence of the design process on
cost is the single greatest influence on product cost. The facilities that are available
and the manufacturing techniques selected by manufacturing rate second to the
designer effect on cost. The designer must be aware of and understand how his
design selection is affected by the available facilities and manufacturing methods.
There has been considerable written and said about the design and manufacturing
interface required on composites, and there is no need to repeat that here. I will,
however, add that I believe the need for this interface is much greater for composites
than for metal structures. However, as the next chart illustrates, the designer is the
key to real cost savings.

Structural Design Drivers

( Static • Loads - external and internal

strength • Material and structural allowables
• Failure criteria and analysis

Damage • Flaw growth

Safety
tolerance • Residual strength
• Inspection

Flutter • Stiffness
margin • Flutter analysis

Maintenance • Durability - fatigue

cost • Inspection cost
• Repair cost
Economics
• Material cost
Production • Designed-in cost
cost • Manufacturing cost
• Quality control cost

17
 To allow the designer to make effective assessments of the cost of his design, he
requires tools to do it in the same manner that he meets the other requirement
displayed on this chart. This does not mean he must become a cost estimator, but
he needs tools that allow him to make cost one of the tradeable items as he goes
through the thought process for his design development. I believe that the post
design cost evaluation by an estimator does not allow the same degree of design
innovation as allowing the designer to work cost as one of his design parameters.
Since the design process that selects from many options occurs at various design
stages and detail levels, cost tools are needed at each stage and level. Real design
cost improvement can be made if we provide the designer with real cost evaluation
tools. These tools do not have to represent the dollar cost of the design, only the
relative cost of one design approach to another. With such tools in the hands of
the designers we will see considerable improvement in the "designed-in cost" of
composite airframe structures.

Designer's Tools

Design Designer's Design

parameter tools incentives

Mc fb Regulators
Safety f -1.0
(strength) b- I M.S. Fb (FAA)

Durability fmax, DFR, N reg Warranty

Service life

Weight Density x volume = weight Performance

Cost 9 Sales
Profit

18
 This weight and cost chart is an old one that we put together in the early 70's
and has been used repeatedly by many people at Boeing. What it displays does
not present anything different from what you would expect to see. The two major
components each make up about the same percentage of the total structural weight.
In a similar manner, the costs that are shown are as expected because of the size
of these components; they are a high percentage of the total cost. The fuselage
costs more because of its higher part count. However, when used in connection
with the next chart, one can begin to assess where to work the weight savings the
hardest and where to make cost the key design driver. These charts can tell you
how to approach design solutions such as damage tolerance as well. The approach
to each area of design can be initially established with this level of information. The
distribution may vary in small percentages--commercial transport to transport--but
not significantly enough to not allow for this information to provide guidance of some
up-front design approach decisions to be made.

Weight and Cost Distribution

Commercial Aircraft

50 -

Weight
4O Cost

30
Percentage
of total
20

10

0
Wing Fuselage Nacelle* Empennage Other
and strut

*Powerplant weight and cost excluded

19
 A simple and perhaps extreme example of cost as the key driver, is the interme-
diate ribs. These ribs do not support any external attachments or form fuel bays.
They simply maintain box shape and support the skin/stringers surface panels. They
usually have their web gages set by fuel slosh and must maintain a reasonable stiff-
ness to support the surface panels. These requirements limit their potential weight
savings. Therefore, the chart assumption was only a 10% weight savings, which
results in a negligible weight saving to the total airframe. However, in a study we
did in the same time period that we generated these weight/cost charts, we found
that on a dollar per pound basis these ribs were the most costly element of the wing.
Therefore, some simple review of the weight and cost history can guide you, as it
would for this example, to make cost the key design driver for intermediate ribs.
This can be done in those early pay-off design phases previously discussed. Simi-
larly, very obvious is the selection of the surface panels of the wing for weight saving
potential. Since it is a large contributor to total cost, the design must also reflect the
best cost options while allowing the maximum weight saving to be gained. Again,
many early inputs to the design process can be made from simple information and
design histories regardless of the material system. To most of you good composite
designers in the audience this is nothing new.

Weight-Saving Areas Selection

Structural Total
element, % Wing, % Fuselage, % aircraft, %
Wing
• Skin and stiffeners 25 15.5 5.7
• Spars 20 3.0 1.1
• SOB 15 1.2 0.4
• Special ribs 20 1.0 0.4
• Intermediate ribs 10 0.4 0.0
• Center-section beams 20 0.4 0.0
21.5 7.6
Fuselage
• Skin, stringers, and frames 25 10.8 4.0
• Keel and wheelwell 20 3.2 1.2
• Floors and (Ioorbeams 20 2.4 0.9
• Doors 15 1.7 0.6
• Bulkheads 20 2.0 0.7
• Windows 15 0.8 0.3
20.9 7.7

• Aircraft total (wing and fuselage): 15.3%

2O
 There has been and continues to be much discussion about composite designs
being "black aluminum" designs. The discussions say that if the designs were not so
black aluminum looking we could save more cost and weight. This chart shows we
should stop looking for the non-black aluminum. The geometry of the cross section
of aircraft structures looks the way it does because those are the most efficient struc-
tural shapes. How those shapes are made out of composites is a separate issue from
the shapes. Ply orientations and lay-up sequences are the real design differences
with composite structures. These shapes were taken from an old paper, Optimiza-
tion of Multirib and Multiweb Wing Box Structure Under Shear and Moment Loads,
by Donald H. Emero and Leonard Spunt of North American. I have not shown all
the cross sections displayed in that paper on this chart. If you can review that paper
and come up with some additional cross sections of significant differences, I would
like to hear from you. Let's focus our development time on the real issues remaining
to produce cost effective composite structures and not waste it looking for the non-
black aluminum shape or geometry.

Structural Concepts

Internally stiffened Trap corr semisandwich

j-- j--
Zee stiffened Straight Y-tee stiffened Truss core semisandwich

Straight Y-stiffened Semitrap corr semisandwich

_m m .&._,& EL.EL
Integral tee Curved Y-tee stiffened Hat section stiffened

21
 These two photo charts are shown to add to my just discussed issue of “black
aluminum.” The key surface panels of commercial aircraft are made of basically long
skinny members and panels. I believe that again to look for something different is
a total waste of development funds. The real question to be addressed with develop-
ment funds is how to design these shapes for the low manufacturing cost while se-
lecting those that will perform well structurally and save weight. The development
of manufacturing methods along with the design of these shapes needs to be the
focus. The manufacturing methods development must be focused not on just those
methods that are simply the lowest cost (without the recognition that these methods
must have the ability to produce these two shapes). Those manufacturing methods
that do not, from the start, recognize the effects of commercial airframe size and
these shape requirements will be a waste of development effort and funds. I want
to be careful here and not to forget t o say that as the manufacturing methods for
these two factors are developed, there is a requirement for continuing development
to improve and enhance the structural data base and analysis tools. They offer the
means of opening the options door to the fullest for the design engineer.

Structural Elements

Wing Skin and Stringers

Fuselage Skin, Stringers,
and Frames

23
 I have just made my case for not looking for the non-black aluminum designs and
the recognition of the key structural elements. These charts validate that industry
is effectively doing that in spite of continuing discussion of looking for non-black
aluminum designs. We have developed the tape laying machine that does in effect
the same job that the skin mill does. In the next chart we are developing pultrusion
machines we hope will make long skinny members like we machine on spar mills. I
present these pictures to show you have already recognized what I just previously
said. To improve on that you should continue to look further at the processes that
can produce these required shapes. I believe we have to take the correct steps
while allowing the non-knowledgeable t o continue to look for the wrong thing, non-
black aluminum designs. We need t o focus our effort on continuing t o explore new
manufacturing methods that can reduce the cost of making the real structural
shapes required.

Metal and Composite = Skins

Skin Mills

24
Tape Layup Machine

25
 No additional discussion is needed for this chart in addition to information pre-
sented previously. Let’s just keep a continued focus on the right manufacturing
methods development. The following charts will show that we have a t least as many
options as metals, if not more.

Metal and Composite - Stringers

Spar Mills

Pultrusion Machine

26
 In continuing to address the cost issues the question often arises as to the struc-
tural and manufacturing options available with composites. They are still viewed
by some as a new structural system, but I am certain not by those of you in this
audience. A reduced set of structural and manufacturing options seem to be im-
plied. This may be limiting the usage of composite structural applications. These
two charts are presented to show simply that I don't see it that way and that the
options in both cases are equal. We have yet to see all that might become available
for composites. We may still have some development or learning to apply to what I
show here, but I believe they are all possible.

Structural Options

Options Metal Composite

• Joining
• Mechanical
• Bonding
• Integral * * (Cocure)
• Configurations
• Skin and frame
• Skin, frame, and major
Iongerons
• Skin, frame, and stringer
• Honeycomb and frame
• Honeycomb shell
• Grid stiffened and frame
• Grid shell

27
 This chart makes a similar statement about the fabrications options available for
composites. Neither of these charts is said to be complete in terms of all possibilities
but to be reasonably representative of the current development status.

Form and Fabrication Options

Metal Composite
• Element forms
• Sheet and plate Tape layup and filament winding
• Extrusion Pultrusion
• Forging Resin infusion
• Shaping
"k
• Roll forming
• Hydropress Hot forming and thermoplastic
• Machining Hot forming and thermoplastic
• Stretch forming Aligned discontinous fibers and thermoplastic
• Joining
_r
• Mechanical fastening
• Bonding
• Diffusion bonding Cocuring
• Welding Thermoplastic fusion

28
 From the two previous charts I tried to show that the composite designer has
the same degree of options open to him as the metal designer. The manufacturing
methods offer the same level of options. I did not say that the manufacturing meth-
ods were equal in cost or efficiency. Metals manufacturing has a considerable head
start in both the development and applications aspects of the methods. This chart
addresses an applications aspect that may sometimes be overlooked when making
cost comparisons. This chart shows that the learning curve, I believe, is not straight
as often depicted by the dashed line, but is more like the solid line. The solid line
shows that in the production of the first 30 to 40 aircraft there is little so-called
learning going on. This is due to the high number of changes that usually occur as
the aircraft goes into production and both engineering and manufacturing correct
their mistakes. The steep slope indicates that changes are reduced and the real
people-learning takes place. This is followed by a lower but continuing reduction in
man-hours per aircraft. This slope does not represent any continued people-learning
improvements but those improvements due to manufacturing methods and tooling
changes that are made to reduce cost. These changes are only made when the cost
savings can be shown to be effective over the remaining production run. These
changes and ideas are carried forward to the next design. The point of the chart
is that this carry over is not zero for composites but it does not have the years of
background that metals have available. While the composites are not at zero as I
show (relative to the final slope), they are so close to that for commercial transport
major components that it represents the real position relative to the metals for 50 to
60 years. This number of years' difference makes it very difficult for composites to
compete on cost alone.

Learning Curve

Labor-
hours

Composites _'"',,,,,,,,i_"----... Metals

Units produced

29
 This chart suggests one of those areas that can be addressed now, and I do not
believe it is being addressed in competition with metal. If I need a clip or short
stiffener of composite material, it is usually manufactured for that specific location
and usage. In metals I have standards books of extruded shapes and possibly a
set of rolled shapes. From these I can often select what I need. If one looks at the
many uses of clips or brackets and web stiffeners used in aircraft, one wonders
why a supplier has not come forward for this structural element. A large number
of metal extrusion dies have been developed over the years just noted. Maybe it’s
time to start something similar in composites. I made a very rough estimate of
the number of short stiffeners or attachment clips on a commercial transport. My
estimate is that more than a million feet per year would be needed based on 10,
767 size airplanes a month. If a simple angle were produced in a standard lay-up
and angle, say 90 degrees, there are many locations where it could be used. It is
possible that if that angle were made of thermoplastic, it could be used for angles
other than 90 degrees by placing it in a heated die to change the shape angle. If
there are other items t o be addressed as composite structural standards, I leave that
to this audience.

Stiffeners - Requirements

Spar Stiffeners

30
Floorbeam Stiffeners

31
 We are hearing a lot about the cost of composite materials and that it is difficult
for composites to compete with aluminum on the basis of just the basic material
cost alone. This chart is based on some data from a NASA document looking at
the cost of composite material relating to type of composite material. I have taken
that data and ratioed it to the cost of aluminum. The displayed ratios agree that
composite material costs are high. When we have ratios this high in comparison
with aluminum it is difficult for composites to compete on a cost basis. However,
I offer the right hand graph on this chart as another look at this cost ratio. If we
consider the out-the-door cost of the materials rather than just the in-the-door cost,
we get a different picture. For wing skins, we order special large roll taper skins,
and we often machine away 60 to 70 percent of the material. For certain airlines
that like shiny fuselage skin, we order premium sheet. Neither of these elements,
that are part of the previously shown major weight contributors, are purchased at
the base price of aluminum. Second, when we consider the 60 to 70 percent chips in
the machined skins (even with the 50-a-pound resale on the chips) the cost ratios
are significantly altered for the out-the-door cost ratios. I should note that my
comparison on material cost only addresses the material cost and amount used;
no labor is included. The general utilization of composite materials is in the 120%
range. That is, we need 120% of the out-the-door weight in-the-door to make the
parts. This comparison changes the ratio to about 20 to 30% of the in-the-door ratio
between aluminum and composites. Let's be sure we recognize this aspect before we
say how expensive composite materials are as compared to metals.

Material Cost Comparison

Composite Versus Metal (Aluminum)

Metals (Aluminum) and

Composites Cost Ratios Material Usage

-
- I
Current materials
omposites
Thermoplastic T-300/AS-4
fibers

Toughened resins
Cost
Thermoplaslics
Ratio 1
t T-800/M-7
fibers
Toughened resins tal (Aluminum)

Metal (aluminum) I
0 5 10 15 20 25 30 In the Usage Out the
door door
Material cost ratios in the door

32
 While I have just defended the cost of composite materials, I would like to send
a message to the material suppliers here by again using a chart from a NASA
report. If this chart is right, and I believe the trend it shows is correct, then the
suppliers need to examine their marketing approach and recognize the business
potential out there for them if the price of composite materials can be reduced.
The reduction in cost can make the application expand to the major structural
elements of the airframe, namely the wing and fuselage. I believe continuing effort
on the part of the suppliers is needed and should be part of all future research and
development. The airframers need to continue to clearly define their needs and
their manufacturing approaches, while the supplier needs to identify the items in
the airframer's requirements that are the key cost drivers in the product. They,
of course need to, on their own, continue to look for ways to reduce their cost so
they can pass it on in the reduced price of composite materials. The market target
potential is so large with the development of an all composite commercial transport
that this cost aspect is as important as any needing attention.

Composite Material Usage

Projection of Potential Usage

20 - Includes - _ 20.0
transport wing .:_iiii!ili
and fuselage $
:'_"_'_"_
.:ii_ii!i!iiiiii_
_

structure :,_
:::::::::::::::::::::::::::::::::

Composite
usage, ,_iiiiiiiiiiiiii!iiiiiiiii!iiiiiiiiiiii_iiii!iii!
Ib x 10 6 10

i Excludes
transport wing
and fuselage
1.0 structure

0
1977 1985 2OOO

33
 All of us here, as well as those involved in the development of composite struc-
ture, have, I am sure, tried to dream up as many reasons possible that composites
produce benefits to both the airframe manufacturer and the aircraft user. The ones
I have selected here I believe are real. They are either well defined today for com-
posites or will be in the future when fully applied to the commercial transports.
We have shown over and over again we can save weight, and you have heard that
again at this conference. The performance resulting from that saving certainly can
produce fuel saving and again, in today's world climate, that is becoming an issue
for the airlines. In terms of maintenance the fatigue characteristics of composites
will continue to reduce the fatigue maintenance requirements significantly. The
area that composite structures has not received enough benefits credit for is the
area of corrosion. Many dollars go into the airframe manufacturing process for the
addition of corrosion protection of the structure. Similarly, the airlines spend many
dollars maintaining this corrosion protection in their fleets. The use of composites
eliminates a major portion of this requirement for corrosion protection. The chart
on the right shows only a small part of the corrosion protection applications applied
to Boeing aircraft. Finally, I believe as we continue our learning process in how
to produce large commercial transport composite structures, we will see more and
more opportunities to reduce the cost of composite structures and accrue additional
benefits.

Composite Application Benefits

• Performance improvements - weight saving

• Fuel savings
• Improved DOCs

• Reduced maintenance
• Large reduction in fatigue problems
• Corrosion maintenance greatly reduced

• Potential manufacturing cost reduction

• Better material usage
• Reduced assembly (more monolithic parts)

34
 As noted, this chart does not show all the corrosion protection areas for commer-
cial transport (which is a significant cost item in the manufacture and maintenance
of the aircraft).

Corrosion Protection and Maintenance

Upper surface inspar

skin corogard

Lower surface
inspar skin
corogard ---Passenger floor
Corogard
upper and
lower stabilizer
skin

BB Enamel overcoat
Corogard

Rudder and Front spar and

elevator leading edge

L
lower surface panel
(similar to Trailing edge (detail A)
detail A) cove areas
including rear spar
Processes used (all areas)
(detail A)
Enamel in detail
APU compartment
BB Enamel overcoat
IBIB Yellow primer
overcoat

35
 Where are the key development needs? As I see it, this list presents my best
assessment of the continuing development needs as they relate to commercial trans-
port aircraft. I am sure that not everyone will agree with this list, but I believe it
encompasses some of the key needs. Many at this conference have talked about
the cost of composite structures, and I have briefly addressed it today. Cost may
represent the largest stumbling block to the application of composites to commercial
aircraft. The large pressure shell considerations are the key concerns for the appli-
cation of composite to commercial aircraft fuselages. The Aloha Airlines incident
of a couple of years ago keeps before us the fatigue and pressure issue in metals
and reminds us not to short change those same issues in composites. If we expand
the use of composites, we will surely want to raise the strain levels used with time,
and in doing so, the low strain level of usage we have employed previously will not
provide the error protection of the past. Therefore, we need to continue to expand
our knowledge on the issues of fatigue and flaw growth, even if we plan to design to
a "no growth" approach with composites. Finally, for commercial transport aircraft
certifications we need to continue to address the issue of the effect of environment
on composite structure full scale validation tests. Also, the issue of proof of the "no
growth" approach needs very careful review. I believe the enhanced loads approach
to be totally wrong; therefore, I included this issue in the needs list of areas for
continuing development support.

Composites Continuing Development

Development Needs

• Material cost reduction

• Manufacturing cost reduction

• Fuselage pressure shells - damage tolerance

• Industry accepted - failure theory

• Effects higher strain utilization

• Fatigue
• Flaw growth

• Certification - additional validation of -

• Ambient full-scale testing
• Flaw "no growth" approach

36
 This chart is one that an oldtimer has the privilege of presenting. The whys and
wherefores of these dates are many and well developed by my personal prejudices
on composites, but I thought I would leave them with you as reasonable targets and
challenges for the commercial airplane people in the audience.

I again want to thank NASA for inviting me to make this presentation and to say
thanks to those many friends at NASA who have helped me over the years. Also my
thanks goes to those of you in the audience that I have worked with on committees
and contracts that have helped me and been my friends for these many years in
composites. And, of course, to many associates and friends at Boeing who are here,
thanks for your great support and help over the years. I hope that all the goals of
the ACT program are achieved and the benefits well recognized by the non-technical
community as well as by the technical community.

Application Projections
Large Commercial Transports

• Complete composite structures

• Wing - before year 2000

• Fuselage - by year 2010

37
Page intentionally left blank
 ADVANCED COMPOSITE STRUCTURAL CONCEPTS AND MATERIAL

T .........
_U_T_T _
G_ES
T
_
F_, PRIMARY T _^_
A_R_,_'_ STRUCTURES

Anthony Jackson
Lockheed Aeronautical Systems Company

SUMMARY

Structural weight savings using advanced composites have been demonstrated for
many years. Most military aircraft today use these materials extensively and Europe
has taken the lead in their use in commercial aircraft primary structures. A major
inhibiter to the use of advanced composites in the United States is cost. Material
costs are high and will remain high relative to aluminum. The key therefore lies in
the significant reduction in fabrication and assembly costs. The largest cost in
most structures today is assembly.

As part of the NASA Advanced Composite Technology Program, Lockheed

Aeronautical Systems Company has a contract to explore and develop advanced
structural and manufacturing concepts using advanced composites for transport
aircraft.

Wing and fuselage concepts and related trade studies are discussed. These
concepts are intended to lower cost and weight through the use of innovative
material forms, processes, structural configurations and minimization of parts. The
approach to the trade studies and the downselect to the primary wing and fuselage
concepts is detailed. The expectations for the development of these concepts is
reviewed.

INTRODUCTION

The Lockheed program consists of two phases. Phase 1 is currently underway and
Phase 2 is an option scheduled to start in 1992. Phase 1 consists of five tasks:
Task i, Design/Manufacturing Concept Assessment, is the subject of this paper; Task
2, Structural Response and Failure Analysis, involves structural analysis methods
development; Task 3, Advanced Materials Concepts, covers the development of new
polymeric matrix systems for HSCT; Task 4, Assessment Review, is the phase final
review leading to a decision on whether to exercise the option for Phase 2; and Task
5, Box Beam, the subject of another paper at this conference, involves the
fabrication and assembly of the C-130 wing center box developed under a previous
NASA contract.

The goals of this program are to identify emerging technologies which will lead
to a 25 percent cost saving, a 40 to 50 percent weight saving, and a 50 percent
reduction in parts count to validate the low cost manufacturing and to verify both
the structural response and the weight savings.

APPROACH

Four wing and three fuselage concepts were selected for this program. These
concepts are shown in Figure I. The concepts were selected based on their potential
for meeting the criteria, with a moderate risk. The Lockheed L-1011 was selected as

39
Z
W
r",
(.5
Z
Z
W
W
5
I--
I--
£
< o.,
u
o
L)
4
o
40
 I--a
zLU
LUa
m
(/)
LU
UJ <<
..i
iii z
U._
(/)
!!
o
LU
Z
111
U.
U.
i--
(/)
W
0
W
o
o
o
0
0
Z
W
ffl
z
U. v
LU
Z
<[
(/)
41
the baseline airplane for this study. A wing location was selected which is
relatively highly loaded and will yield test panels which can be tested in existing
test machines and fixtures. The upper surface has to be buckling critical. The
fuselage location was selected in a similar manner. Figure 2 shows the location of
the wing section, and Figure 3 shows the location of the fuselage section. The
design criteria for the wing are shown in Figure 4 and for the fuselage in Figure 5.

WING CONCEPTS

Four wing concepts were selected for the trade studies. Concept #I is the
Modular Wing. This concept is built up of various components each of which is
fabricated using a different process. The stiffeners are pultruded, the skins are
automatic tape placed, the ribs are press formed and the spars are filament wound.
Concept #2 is the Resin Transfer Molded Wing. This concept is made from woven
stitched preforms and is molded in two pieces. Concept #3 is the Advanced Tow
Placement Wing. This concept also involves other fabrication processes, but the
covers are made by automatic tow placement (ATP). Concept #4 is the Braided Wing.
This concept is fabricated mainly by 2D and 3D braiding.

CONCEPT #I - MODULAR WING

This design is shown in Figure 6. The covers, spars and ribs are fabricated
separately and are assembled by conventional methods. The covers are blade
stiffened. The stiffeners are fabricated from dry preforms which are resin infused
and either B-staged or fully cured Tee sections. The skins are fabricated in two
parts. The inner skin is discontinuous at the stiffeners. It can be laid up by
automatic tape dispenser and can be cut into strips by a waterjet cutter or by a
Gerber cutter. The outer skin is laid up over a tool containing the stiffeners and
inner skin strips by automatic tape dispenser. The fabrication sequence is shown in
Figure 7.

LOCKHEED L-1011 TRISTAR WING

_" A FWD SPAR

. t--- 9987 "t

'°-4/. REAR
SP
"RZ -
TOP VIEW WING BOX

FWD
.184
SPANWISE SKIN SPLICE

/ I ..... - .96 115 |

15 _ _ J19

.192 SECT A - A OWS 151.1

Figure 2. Baseline Wing and Study Location

42
 LOCKHEED L-1011 TRISTAR FUSELAGE

19 FT 7 IN.

6.0

-_ 178FT75 N- 1
fS
3.0_
I] 0

WINDOW

,9.o cAB,, J--

5.0"

SECTA-A FS750.00

Fisure 3. Baseline FuselaKe and Study Location

NX 14,000 LB/IN

NXY 2,000 LB/IN

10.38 PSI BURST

7.82 PSI CRUSHING

ECCENTRICITY .1%

Gt MIN. .70 E6 PSI

UPPER COVER LOADS O.W.S. 151.00

• COVERS BUCKLING RESISTANT

• MAX. ALLOWABLE STRAIN .006 IN/IN TENSION

.0044 IN/IN COMPRESSION

Fisure 4. Wing Desisn Criteria

43
 CONDITION CROWN SIDEWALL KEEL

MAXIMUM TENSION (LB/IN) 1307 432 318

489 705
MAXIMUM COMPRESSION (LB/IN) 943

SHEAR (LB/IN) 150 600 300

LOADS AT F.S. 750.00

• SKINS SHEAR BUCKLING RESISTANT BELOW 1G

• MAX. ALLOWABLE STRAIN .006 IN/IN TENSION

.0044 IN/IN COMPRESSION

Figure 5. Fuselage Design Criteria

COCURED RIB
MECHANICALLY CAPS/SHEARTIES
FASTENED

STIFFENERS EMBEDDED
INTO SKIN LAMINATE

Figure 6. Modular Wing Concept

44
 t,

LAY STIFFENERS IN TOOL _,

.)
TAPE LAY OUTER SKIN

Figure 7. Modular Wing Fabrication

The front and rear spars are fabricated by automatic tow placement. Alterna-
tives are filament winding or tape winding. The spars are designed as C-sections
and can be wound as pairs in the form of a rectangular box and cut into C-sections
after cure.

The ribs are fabricated as separate caps and webs. Prepreg plies are cut by
Gerber cutter, stacked, formed to shape and B-staged. The rib caps are then placed
in the cover fabrication tool along with the inner skin strips and the stiffeners
before the outer skin is laid directly on the curing tool. A caul plate is placed
over the assembly which is then bagged and cured.

The rib webs are compression molded using either thermosets or thermoplastics.

Final assembly is achieved by using mechanical fasteners.

CONCEPT #2 - RESIN TRANSFER MOLDED WING

The desisn concept for the resin transfer molded (RTM) wing is shown in Figure
8. The design calls for the wing box to be fabricated in two halves. Each half
consists of one complete cover and part of the inteEral front and rear spars and
with integral rib caps. Consequently, this design has no mechanical fasteners
penetrating the outer surfaces.

The wing box would require large woven/stitched preforms. Close stitching
would be required to debulk the preforms sufficiently to allow them to be assembled
in the RTM tool. Assembly of the final molded halves would be accomplished by
mechanical fasteners in the spar webs and by mechanical attachment of the separately
molded rib webs. Figure 9 shows the fabrication approach.

45
 INTEGRAL
SPAR CAP
INTEGRAL
SPAR CAP/WEB

STIFFENERS
STITCHED TO
COVERS

COCURED RIB
MECHANICAL
CAP/SHEAR TIE
ATFACHMENT

Figure 8. RTM Wing Concept

IN ASSEMBLE
COVERS AND RIBS
BOX

Figure 9. RTM Wing Fabrication

46
 The wing cover assemblies will weigh approximately 2500 pounds. The largest
RTM assemblies fabricated today weigh about 250 pounds. This design would thus
require considerable scale-up of current technology.

A major concern with this design is the large number of tool parts which would
be required. Stiffeners are usually normal to the skin, thus giving closed angles
which would require segmented tools between each adjacent pair of stiffeners. At
various locations along the wing span and probably chordwise, sets of vents would be
required for resin to escape and to release trapped air. These vents would be
closed progressively as the resin migrates outward from the injection ports. After
all vents are closed some pressure would be maintained via the injection port to
reduce the chance of entrapped air settling and causing voids.

CONCEPT #3 - ADVANCED TOW PLACEMENT WING

The design concept for the advanced tow placement (ATP) wing is shown in
Figure I0. This design calls for the wing box to be tow placed and cured on a
single mandrel. The large size of the wing box, however, made the handling of a
single mandrel a major logistical problem. Consequently, the design was modified to
fabricate the covers and spars separately. Rectangular tubes would be tape wound on
mandrels and cut into two channel sections to form the blade stiffeners. This

process allows plies to be picked up and dropped off to add localized reinforcements
and 0 degree plies in the stiffener webs.

Spars can be tow placed in pairs around mandrels. An alternative approach

would be to incorporate the spar caps in the covers and to tow place the webs in
groups. The fabrication approach is shown in Figure II.

The ribs would be press or diaphragm formed.

CONCEPT #4 - BRAIDED WING

The design concept for the braided wing is shown in Figure 12. This design
calls for a one-piece wing box. Both 2D and 3D braiding were considered. The
physical size of the wing box being considered does not fit within the confines of
any current or planned braider. Today's largest 2D braider would have difficulty
braiding 45-degree angles with 12k tow over a one-foot diameter mandrel. The size
of carrier for this type of braider is five feet in diameter. A machine capable of
braiding an L-1011 size wing box would require an enormous amount of floor and air
space. More importantly, the individual carriers could not dispense a high enough
quantity of fiber to make the process automatable. The carriers would require such
frequent replacement that the process is not feasible in the foreseeable future.

This is also partly true for 3D braiding. The limitations of closed section
tubular structures is less severe. Atlantic Research has developed an automated 3D
braider which utilizes 9216 fiber carriers. A fabrication approach is shown in
Figure 13.

Because of the problems in fabricating a complete wing box, this concept was
dropped from further consideration. The braiding process was retained as an option
for smaller assemblies as part of the modular wing box concept.

47
 COVERS WITH
INTEGRAL
SPAR CAPS

ATTACHMENT
COCUREDRIB
CAPS/SHEAR TIES

Fisure I0. ATP Wing Concept

ATP CHANNELS
ASSEMBLE CHANNELS TAPE LAY SPARS

I _ _-_ . __
ASSY

TAPE LAY OUTER SKIN DIAPHRAGM "_"

FiKure II. ATP Wing Fabrication

48
INTEGRALSPAR CAP

INTEGRAL SPAR
WEB/CAP

COBOND
RIB CAPS

INTEGRAL HAT
ENTIRELY STIFFENERS
BRAIDED WITH 0 ° FIBERS
STRUCTURE IN CAP

Figure 12. Braided Win8 Concept

Fisure 13. Braided Wing Fabrication

49
FUSELAGE CONCEPTS

Three fuselage concepts were selected for the trade studies. Concept #I is a
sandwich design incorporating braided triangular tubes in the sandwich. Concept #2
is a geodesic design based on an isogrid concept. Concept #3 is a hat stiffened
shell design.

CONCEPT #I - SANDWICH STIFFENED SHELL

The design concept for the sandwich shell is shown in Figure 14. This design
consists of a sandwich using braided triangular tubes as the core. The tubes are
oriented longitudinally. Periodically, there are flanged tubes as shown in Figure
15 which act as longerons.

The fabrication approach is to braid the tubes using dry fiber, then to
pultrude through a resin bath and B stage. A fly-away foam mandrel would be
required with this approach. The tubes can be fully cured and then assembled using
an adhesive. This would eliminate the need for foam mandrels but would create many
bond lines which would be difficult to inspect.

The inner skin is built up from C-sections which could also be pultruded. The
outer skin is formed by overwrapping with tow or tape. The fabrication plan for a
complete barrel section is shown in Figure 16.

CURVEDFRAME MEMBER

LONGERON WITH CUTAWAY

FOR FASTENER ACCESS

INNER PANEL
-__ INNER PAN
CHANNELSECTION
CHANNEL SECTION

BRAIDED TRIANGULAR
TUBES

FLANGED TUBES

OUTER SKIN

Figure 14. Sandwich Stiffened Shell Concept

50
 2"

1
3-D BRAIDED OR PULTRUDED
(TUBES 20" LONG)

Figure 15. Flanged Triangular Tube Section

PULTRUDE BRAIDED TRIANGLES

PULTRUDE INNER SKIN

Figure 16. Sandwich Fuselage Fabrication

51
CONCEPT #2 - GEODESIC FUSELAGE

The design concept for the geodesic fuselage is shown in Figure 17. This
design calls for an isogrid stiffened shell. The helical stiffeners are formed by
winding Filcoat material alternately in each direction. Filcoat is a patented
Lockheed designed material consisting of Gr/Ep tape coated with an equal thickness
of epoxy filled with glass micro-balloons called syntactic. At intersections the
syntactic is squeezed out. Figure 18 shows a schematic of an intersection. The
fibers in each direction are continuous and the intersections are the same height as
the stiffeners.

The hoop stiffeners are not continuous. They are pull-formed and cut to their
individual lengths. Intersection clips and overwraps are stitched dry fiber forms
containing mainly _45 degree and 90 degree plies to provide shear and flange bending
strength. These clips and overwraps can be combined to minimize parts and are resin
transfer molded and B-staged.

The skin is finally tape or tow wrapped over a mandrel. The fabrication
process is shown in Figure 19.

CONCEPT #3 - HAT STIFFENED SHELL FUSELAGE

The design concept for the hat stiffened shell fuselage is shown in Figure 20.
This design consists of pultruded hat stiffeners cocured to an advance tow placed
skin. The frames are designed to be resin infusion molded and the complete assembly
is cocured. An alternative fabrication method for the stiffeners is to braid prior
to pultrusion. The fabrication process is illustrated in Figure 21.

RESIN INFUSED
INTERSECTION CLIP
3-D WOVEN F!gER FORM

GEODESIC FUSELAGE
STIFFENER MANDREL TOOL
COVER

HOOP STIFFENER
SEGMENT

Figure 17. Geodesic Fuselage Concept

52
 LOW DENSITY SYNTACTIC
(DISCONTINUES THROUGH THE NODES

UNIDIRECTIONAL GR/EP
(CONTINUES THROUGH THE

SKIN

Figure 18. Geodesic Intersection

INTERSECTION CLIPS

.-,.-: :.--

STIFFENER OVERWRAP ATP HELICAL STIFFENERS

ATP FUSELAGE SKIN

Figure 19. Geodesic Fuselage Fabrication

53
 DESIGN FEATURES

• HAT STIFFENED TO INCREASE

SPACING BETWEEN STIFFENERS

• CO-CURED ASSY

• GOOD DAMAGE TOLERANCE

• RFI FRAMES

• PULTRUDED HAT STIFFENERS

Figure 20. Stiffened Fuselage Shell Concept

PULTRUDE STIFFENERS FRAMES __]_----- I_1

ASSEMBLE DETAILS
COCURE ASSEMBLY

Figure 21. Stiffened Shell Fuselage Fabrication

54
 The most effective way to fabricate these panels is as quarter panels.
Essentially, upper, lower and side panels would be required. The use of closed hat
stiffeners in the lower, or keel section, needs to be studied further because of
problems associated with entrapment of bilge fluids. This is not considered to be a
major problem as composites do not corrode. Drainage must, however, be provided to
prevent accumulation of fluids and bacteria growth as well as the additional weight.

TRADE STUDIES

Figure 22 shows a summary of the options which were considered during the trade
studies.

Blade, Jay and hat stiffeners were considered as options for wing skin
stiffening. The Jay stiffener was eliminated because it is more difficult to
fabricate than the blade and it did not show a significant enough weight saving to
justify a higher cost. Hat stiffeners posed several problems in fuel tanks. They
can trap fuel, they can provide leak paths and, being wide, they are difficult to
terminate outboard effectively. This led to the use of blade stiffeners in all of
the wing concepts. The blade configurations are, however, different. The blade
configurations for the three concepts which were carried to completion are shown in
Figure 23. The ATP stiffeners are built up from side by side channel sections. For
the modular wing the stiffeners are pultruded with tapered flanges so that they can
be buried in the skin. For the RTM wing the stiffeners are built up from woven
stitched fabric.

The fuselage concepts are unique in themselves, so the stiffener configuration

was not tradeable.

CONFIGURATION FABRICATION METHODS MATERIALS

CONCEPT
_.L _- _ ._ BRAID ATP RTM/RFI PULT T/S T/P
i

MODULAR
WING

RTM
WING •

ATP
WING

BRAIDED
WING

SANDWICH
FUSELAGE

GEODESIC
FUSELAGE •

STIFFENED SHELL • • • • •
FUSELAGE

Figure 22. Trade Study Options

55
 i I
ATP WING MODULAR WING

RTM WING

Fisure 23. Stiffener Comparison

The fabrication methods looked at for each concept are summarized in Figure 22.
The method selected for each component has already been discussed.

The trade study also looked at the possible use of thermoplastic materials
instead of thermosets. The high cost of thermoplastic materials today makes their
use in subsonic aircraft unlikely. The inherent toughness of the thermoplastic
materials has some advantages, but the toughened thermosets are much more cost
effective. Thermoplastics do look good for press formed ribs and even for frames.
The main disadvantage of mixing materials is that mechanical fasteners would be
required for assembly as thermosets and thermoplastics are very difficult to bond
together. Figure 24 shows typical thermoplastic material costs today. Figure 25
shows a comparison of current and projected material costs. If projected prices do
in fact become reality, then thermoplastics may be viable candidates for future
commercial subsonic transports.

CONCEPT EVALUATION

Each concept was evaluated for cost, weight, design technology advancement,
manufacturing technology advancement, producibility, damage tolerance, inspectabil-
ity, maintainability and repair. The ability of the concept to meet the program
goals was a major consideration. The scoring system used in the evaluation gave 40
points to cost, 30 points to weight and 30 points to all other factors. The cost
score is the cost goal divided by the concept cost multiplied by 40. The weight
score is the weight goal divided by the concept weight multiplied by 30. The other
factors' score is the total of all points other than cost and weight divided by the
maximum possible score multiplied by 30. This is summarized in Figure 26. The
individual scores for the other factors are shown in the appendix along with the
rationale.

56
 200

_,1

rr 100
-I
o
E3 • 0-99 LBS
[] I00-I000 LBS

[] OVER 1000 LBS

UNITAPE COMMINGLED TOWPREG

FABRIC

Figure 24. Thermoplastic Material Forms Cost Comparison

2OO

100
rr
[] TIP COMMINGLED
.J
o [] TIP UNITAPE
E3
[] BMI UNITAPE

[] EPOXY FABRIC
[] ,EPOXY UNITAPE

0
CURRENT PROJECTED
PRICES PRICES

Figure 25. Cost of Various Composites

57
 EACH CONCEPT WAS ALLOCATED A MINIMUM SCORE OF 100 POINTS BROKEN DOWN AS

• COST (40 POINTS)

COST SCORE = (COST GOAL/COST OF DESIGN) X 40

• WEIGHTS (30 POINTS)

WEIGHTS SCORE = (WEIGHT GOAL/WEIGHT OF DESIGN) X 30

• OTHER FACTORS (30 POINTS)

OTHER FACTORS SCORE = (SUM OF POINTS OF DESIGN / MAXIMUM POSSIBLE POINTS) X 30

Figure 26. Scoring System

WEIGHT TRADES

The weights for the wing concepts were based on the total weight of the wing
box structure per aircraft. Optimum sizing of the structure at outer wing station
151.1 was obtained. A spanwise variation was then used based on previous wing
studies. Additional weight was added to account for landing gear attach fittings,
engine mount fittings and access doors.

Weights for the fuselage concepts were based on sizing of the upper shell at
station 750. The sizing assumed maximum tension and shear or maximum compression
and shear. The sizing was then conservatively assumed to be constant at all circum-
ferential locations. Total weight between Fuselage Stations 235 and 983 was taken.

A comparison of wing box weights is shown in Figure 27 and the fuselage segment
weights in Figure 28. A summary of the weight trade study is shown in Figure 29.

COST TRADES

The cost trades were based on recurring costs only, although nonrecurring costs
were considered in the producibility trades. Recurring costs were based on a
production run of 300 ship sets at a rate of five per month. Labor rates are 1995
projected as agreed among the ACT program contractors at the Cost Workshops.
Material costs were assumed to be $40/ib. A sensitivity study on material cost will
be performed. It was assumed that there would be no purchase of facilities or
equipment. The fly to buy was dependent on fabrication method. The cost analysis
program used was ACCEM. It includes material burden, support labor, quality
control, learning curves and industrial engineering standards. A comparison of the
wing concept costs is shown in Figure 30 and a comparison of the fuselage costs is
shown in Figure 31. The cost trade study results are summarized in Figure 32.

58
[] COVERS
[] RIBS & BULKHEADS
[] SPARS

[] JOINTS,ETC.
*LBS/A/C

27,800*

BASELINE MODULAR RTM ATP

Figure 27. Wing Box Weights

[] SKIN/STRINGERS

[] MINOR FRAMES

*FS235-FS983 -- m
A A A
23.3% 14.9% 30.3%

9610 LBS

BASELINE SANDWICH GEODESIC HATSTIFFENED

Figure 28. Fuselage Shell Weights

59
 CONCEPT WEIGHT SCORE

BASELINE 27,800
(GOAL)-34% (17,792)

WING
MODULAR 19,831 26.91

RTM 21,976 24.29

ATP 18,953 28.16

BASELINE 9,610
(GOAL)-34% (6,150)

FUSELAGE SANDWICH 7,367 25.04

GEODESIC 8,175 22.57

HAT STIFFENED 6,700 27.54

Figure 29. Weight Trade Study Results

]_"_ASSEMBLY
COVERS
_SPARS
RIBS
I:FI:g_COVERS & SPARS
14%
77%

10°1,, 16%

18% 21%

21%_ 10% 24%

53%

52%

57%

ATP

13% SAVINGS 10% OVER BASELINE 24% SAVINGS

Figure 30. Wing Concept Cost Comparisons

60
 5% 20%
SKIN 14o,,_
FRAMES
STRINGERS
ASSEMBLY
58%
BONDED ASSEMBLY

9% 6% 26_,

19%
3%
31% __
40% 10% 20%

16%

CORRUGATED

38o/oOVER BASELINE 177% OVER BASELINE 30% SAVINGS

Figure 31. Fuselage Concept Cost Comparison

CONCEPT COST $ SCORE

BASELINE 2;636,425
(GOAL).25O/o (1,977,318)

WING
MODULAR 2,301,918 34.36

RTM 2,912,135 27.16

ATP 2,002,760 39.49

BASELINE 161,704
(GOAL)-25% (121,278)

i FUSELAGE SANDWICH 221,985 21.85

GEODESIC 448,918 10.81

HAT STIFFENED 112,962 42.95

Figure 32. Cost Trade Study Results

61
 Cost benefits and drivers for each concept are summarized in Figure 33.

DOWNSELECT

Based on the trade studies, one wing and one fuselage concept was selected from
the concepts shown in Figures 34 and 35. Figure 36 shows a summary of the rankings
of the concepts. The Advanced Tow Placement Wing and the Hat Stiffened Shell
Fuselage designs finished the clear winners. Both came close to the 25 percent cost
saving target and the 40 percent weight saving target and exceeded the 50 percent
reduction in parts count. The weight saving goal shown has been reduced to 34
percent to account for resizing. The 34 percent was an overall goal bearing in mind
the fact that the wing would be expected to have a larger savings from resizing than
would the fuselage which has volume constraints. The selected concepts are shown in
Figures 37 and 38.

The original program plan had called for continuing with a backup design for
both the wing and the fuselage through the end of Phase I. The primary concepts,
however, emerged as such clear winners that it was decided to put all the remaining
effort into developing and validating the primary concepts.

EXPECTATIONS FOR DEVELOPMENT

Both the selected concepts depend on minimization of mechanical fasteners and

the fabrication of large components. The development of these concepts depends on
the minimization of discontinuities, the development of analytical methods, the
demonstration of repeatable process and the use of in-process controls which will in

CONCEPT COST BENEFITS COST DRIVERS

MODULAR • MORE EFFICENT USE • TOOL LOADING

WING OF MATERIALS • MATERIAL COST
• ADDITIONAL FASTENERS
• INCREASED ASSEMBLY

RTM • STIFFENER SIMPLICITY • TOOL LOADING

WING • REDUCED PART COUNT • COMPLEX TOOLING
• MINIMAL MECHANICAL • MATERIAL COST
ASSEMBLY

ATP • USE OF ATP EQUIPMENT • HANDLING

WING • INTEGRAL SPAR CAPS • TOOL LOADING
• REDUCED ASSEMBLY

SANDWICH • USE OF AUTOMATED .HIGH PARTCOUNT

FUSELAGE FABRICATION PROCESSES •ASSEMBLY
.MATERIALCOST

GEODESIC • COMMONALITY OF DETAILS .HIGH PARTCOUNT

FUSELAGE .HANDPLACEMENTOF
DETAILS
.MATERIALCOST

STIFFENED SHELL • REDUCED NUMBER OF PARTS • COMPLEX TOOLING

FUSELAGE • COCURED FRAMES

Figure 33. Cost Benefits and Cost Drivers

62
 COVER SPAR RIB CAP

MODULAR
WING BOX
\

RTM
WING BOX

ATP "---I I

WING 130
1

Eigure 34. Wing Box Concepts

COVER FRAME

BRAIDED
TUBE

GEODESIC

STIFFENED
SHELL

Figure 35. Fuselage Concepts

63
 CONCEPT WEIGHT COST ILITIES TOTAL RANK

MODULAR 26.91 34.36 12.5 73.77 2

WING 27.16 20.5 71.95

RTM 24.29

ATP 28.16 39.49 18.0 85.65

SANDWICH 25.05 21.85 17.0 63.90

FUSELAGE GEODESIC 22.57 10.81 15.5 48.88

STIFFENED 27.54 42.95 17.5 87.99

SHELL

Figure 36. Downselect Summary

DESIGN FEATURES:

• ONE PIECE DESIGN - NO SPANWISE/

CHORDWISE JOINTS

• INTEGRAL CONTINUOUS BLADE

STIFFENERS

• UPPER/LOWER COVERS WITH

INTEGRAL SPAR CAPS

• NO FASTENERS/LEAK PATHS
THROUGH TI-IE COVERS

• INTEGRAL RIB CAPS WITH SHEAR

CLIPS

Figure 37. Selected Wing Concept

64
 DESIGN FEATURES:

• HAT STIFFENED TO INCREASE

SPACING BETWEEN STIFFENERS

• CO-CURED ASSY

• GOOD DAMAGE TOLERANCE

• RFI FRAMES

• PULTRUDED HAT STIFFENERS

Figure 38. Selected Fuselage Concept

effect build in the quality and minimize scrap and buy-off. The concepts involve a
moderate risk but can be approached in an incremental manner which will improve the
chances of success. Neither concept involves an all or nothing approach and altern-
ative paths are available if needed.

SUMMARY AND CONCLUSIONS

Advanced structural and material trade studies were carried out on four wing
and three fuselage concepts. The trade studies showed that the Advanced Tow
Placement Wing concept and the Hat Stiffened Shell Fuselage concept both showed
excellent potential for meeting the program goals. The weight savings are close
enough to the goal that there is a reasonable chance of meeting or exceeding this
goal with further development and refinement. The wing cost is within one percent
of the goal and the cost of the fuselage concept exceeds the goal. Efforts are now
underway to validate these designs by more detailed analyses and by fabrication and
test.

65
 APPENDIX

DESIGN
CONCEPT TECHNOLOGY RATIONALE
ADVANCEMENT

MODULAR CONVENTIONAL ASSEMBLY METHODS. LARGE COCURED COVER

WING 4 ASSEMBLIES REPRESENT A SLIGHT TECHNOLOGY ADVANCEMENT.

RTM CONVENTIONAL DESIGN CONFIGURATION WITH EXCEPTION OF COVER/SPAF

WING 5 INTEGRATION AND STITCHED PREFORM.

ATP CONVENTIONAL DESIGN CONFIGURATION. EMPHASIS ON LARGE COCURED

WING 5 ASSEMBLIES. MECI4ANICAL FASTENERS SIGNIFICATANTLY REDUCED.

SANDWICH UNIQUE DESIGN CONCEPT AMEANABLE TO AUTOMATED FABRICATION

FUSELAGE 8 TECHNIQUES. SIMPLIFIED FRAME TO COVER ATTACHMENT.

GEODESIC 6 HIGHLY EFFICIENT, DAMAGE TOLERANT DESIGN. DESIGN SUITABLE FOR

FUSELAGE AUTOMATED MANUFACTURING METHODS.

STIFFENED SHELL DESIGN CONCEPT IS CURRENT STATE OF THE ART. CONFIGURATION

FUSELAGE 4 ALLOWS FOR COCURING OF ALL STRUCTURAL ELEMENTS.

TECHNOLOGY
CONCEPT ADVANCEMENT RATIONALE
MANUFACTURING

DIFFICULT TO LAY UP RIB CAPS AND COCURING TO THE COVER

MODULAR 3
WING AND STIFFENERS. LI]-I'LE TECHNOLOGY ADVANCEMENT.

ONE-SHOT COMPLETE RTM WING HALF WOUED BE A TREMENDOUS

,RTM 9 ADVANCEMENT OF THE TECHNOLOGY. DESIGN ELIMINATES MANY
WING COMPONENTS AND FASTENERS.

ATP OF C-CHANNEL BLOCKS AND ASSEMBLY TO FORM COVERS

ATP 7 REPRESENTS AN ADVANCEMENT OVER CURRENT METHODS. CO-
WING CURING OF INTEGRAL RIB CAPS IS A SIGNIFICANT ADVANCEMENT.

SANDWICH SIGNIFICANT ADVANCEMENT IN MANDREL TECHNOLOGY. SOME

ADVANCEMENT IN COMPONET LOCATION ARENA WITH THE MANY
FUSELAGE 7 TUBES. A SIGNIFICANT ADVANCEMENT IF PULTRUDED.

GEODESIC ATP OF HELICAL STIFFENERS AND RFI OF INTERSECTION

FUSELAGE 7 CLIPS IS A SIGNIFICANT ADVANCEMENT. PULL FORMING OF
FRAMES IS AN EXTENSION OF TECHNOLOGY.

STIFFENED SHELL LITTLE ADVANCEMENT ASIDE FROM THE FACT THAT THE
3
FUSELAGE FRAMES ARE COCURED.

66
 APPENDIX

INSPECT-
CONCEPT
ABILITY RATIONALE

MODULAR SEPARATE COMPONENTS CAN BE INSPECTED, BUT ALSO WILL REQUIRE

WING EXTENSIVE POST PROCESS INSPECTION DUE TO COCONSOLIDATION
AND/OR BONDING

PREFORM MAY BE INSPECTED BEFORE MOLD FILLING. WIDE RANGE OF

RTM
IN-PROCESSMETHODS COULD BE USED FOR MONITORING THE MOLD FILL
WING
AND CURE, INCLUDING PROCESS MODELS.

ATP TOW QUALITY, SIZE AND PLACEMENT MUST BE MONITORED AT ALL TIMES,
6
WING WILL DEPEND ON MACHINE. PLACEMENT MONITORING NEEDS TO BE
DEVELOPED.

SANDWICH TUBES COULD BE INSPECTED IN-LINE, BUT POST PROCESS WILL BE VERY
FUSELAGE 5 DIFFICULT BETWEEN TUBES.

GEODESIC 2 VERY COMPLEX GEOMETRY. THE TRUSS INTERSECTIONS ARE

FUSELAGE UNINSPECTABLE.

STIFFENED SHELl HIGH SCORE BECAUSE COMPONENTS MAY BE INSPECTED BEFORE FINAL
FUSELAGE 8 CURE. IN-PROCESS INSPECTION OF PULTRUDED HATS AND RTM FRAMES
HAVE EASY GOEMETRY.

MAINTAIN-
CONCEPT RATIONALE
ABILITY

MODULAR 6 MODULAR CONSTRUCTION FACILITATES LESS COSTLY REPAIR

WiNG TECHNOLOGY. HEAVY STRUCTURAL DAMAGE IS UNREPAtRABLE AT
FIELD LEVEL.

RTM LEAK PATHS ARE ELIMINATED. CONSTRUCTION FACILITATES LESS COSTLY

WING 6 REPAIR TECHNOLOGY. HEAVY STRUCTURAL DAMAGE IS UNREPAIRABLE
AT FIELD LEVEL.

ATP LEAK PATHS ARE ELIMINATED. REPAIR AT FIELD LEVEL IS LESS COSTLY.
WING 7 HEAVY DAMAGE WILL INDUCE REMOVE AND REPLACEMENT OF ENTIRE
STRUCTURE.

SANDWICH COMPOSITE MATERIALS ELIMINATE MOST MAINTAINABILITY ISSUES.

FUSELAGE 8 REPAIR CAN BE EASILY DONE AT THE FIELD LEVEL

GEODESIC CREATES REPAIR PROBLEMS THAT CANNOT BE SATISFIED WITHOUT

FUSELAGE 4 MAJOR RECONSTRUCTION OF LARGE AREAS. REQIURES EXCESSIVE SPAR[
/REPAIR PARTS INVENTORY.

STIFFENED SHELL 6 REPAIR PROBLEMS IN TRANSFERING LOAD ACROSS DAMAGED AREA.

FUSELAGE

67
 APPENDIX

CONCEPT PRODUCIBILITY RATIONALE

MODULAR 3 RIB CAP CONFIGURATION DIFFICULT TO FABRICATE AND EXPENSIVE. NO

WING PROVISION FOR TOLERANCE FLOAT. LIMITED ACCESS FOR INTERAL FAST.

RTM 6 THIS CONCEPT IS HIGH RISK, BUT HAS HIGH PAY-OFF. ALSO, HAS SAME
WING ACCESS PROBLEMS AS THE MODULAR DESIGN.

ATP CONCEPT HAS LESS RISK THAN RTM DESIGN, AND ALSO LESS PAY-OFF.
WING 6 SAME ACCESS PROBLEMS AS OTHER WING CONCEPTS.

SANDWICH TOO MANY PIECES. THE LENGTH AND SMALL CROSS SECTION OF THE
TRIANGLES MAKE MANDREL REMOVAL DIFFICULT. INSPECTION OF BONDED
FUSELAGE 3 ASSEMBLY PRESENTS PROBLEMS.

GEODESIC TOO MANY PARTS. JUNCTION CLIPS VERY DIFFICULT TO FAB WITH
FUSELAGE 3 CONTINUOUS FIBER COMPOSITES. VERY COMPLEX TOOLING, MANY PARTS.

STIFFENED SHELL 8 ALL PROCESSES ARE AUTOMATED INCLUDING RTM, PULTRUSION AND
FUSELAGE FILAMENT WINDING.

DURABILITY /
CONCEPT DAMAGE RATIONALE
TOLERANCE

MODULAR 3 CONCERN IS THAT IMPACT DAMAGE WILL CAUSE STIFFENER TO PULL

WING AWAY FROMSKIN DRASTICALLY REDUCING MECHANICAL PROPERTIES.

THROUGH THE THICKNESS REINFORCEMENT SHOULD PREVENT STIFFENER

RTM 7
WING UNBOND AND MINIMIZE IMPACT DAMAGE. LOWER FIBER VOLUME IS STILL
A CONCERN, AS IT WOULD REDUCE STRUCTURAL INTEGRITY.
ATP
WING 5 THIS IS TYPICAL OF CURRENT STRUCTURES.

SANDWICH THINNESS OF FACINGS IS A DURABILITY CONCERN. IMPACT COULD CAUSE

FUSELAGE 3 SEPERATION OF TRIANGULAR TUBES OVER A LARGE REGION. THIS COULD
REDUCE RESIDUAL PROPERTIES.
THIS CONFIGURATION IS HIGHLY REDUNDANT AND SHOULD HAVE OUT-
GEODESIC
9 STANDING DURABILITY AND DAMAGE TOLERANCE. HOWEVER, THERE IS A
FUSELAGE HIGH RISK OF CRITICAL MANUFACTURING FLAWS IN THE DIAGONAL
CROSS-OVERS,
STIFFENED SHELL CONSIDERED SLIGHTLY BETTER THAN CURRENT STRUCTURES BECAUSE OF
FUSELAGE THE ELIMINATION OF FASTENERS AND HOLES.

68
 APPENDIX

TECHNOLOGY
CONCEPT ADVANCEMENT PRODUCIBILITY DAMAGETOL/ INSPECTABILITY MAINTAINABILITY/ TOTAL SCORE
DURABILITY REPAIR
DESIGN MFG

MODULAR 4 3 6 25 12.5
WING

RTM
WING 5 9 7 8 6 41 20.5

ATP
WING 5 7 6 5 6 7 36 18.0

SANDWICH
FUSELAGE 8 7 3 3 5 8 34 17.0

GEODESIC 6 7 3 9 2 4 31 15.5
FUSELAGE

STIFFENED SHELL 4 3 8 6 8 6 35 17.5

FUSELAGE

69
Page intentionally left blank
 COMPOSITES TECHNOLOGY FOR TRANSPORT
PRIMARY STRUCTURE

VICTOR CHEN, ARTHUR HAWLEY,

MAX KLOTZSCHE, ALAN MARKUS,
RAY PALMER
DOUGLAS AIRCRAFT COMPANY

The ACT contract activity being performed by the McDonnell Douglas Corporation

is divided into two separate activities: one effort by Douglas Aircraft in

Long Beach, California with a focus on Transport Primary Wing and Fuselage

Structure, and the other effort by McDonnell Aircraft in St. Louis, Missouri

with a focus on Advanced Combat Aircraft Center Wing-Fuselage Structure. This

presentation is on the Douglas Aircraft Transport Structure portion of the ACT

program called ICAPS - Innovative Composite Aircraft Primary Structure. The

McDonnell Aircraft portion of the ACT program will be presented by Mike Renieri

in a presentation scheduled for tomorrow.

71
 COMPOSITE APPLICATION GOALS

The Douglas Aircraft Company has a goal to have the wing technology readiness

in hand in early 1993. It is anticipated that both a military and a civil

application that require start of engineering development will occur during

that year. Plans to initiate specific development design and tests have been

identified.

The goal for a pressurized fuselage is to have the technology readiness in hand

in early 1994 so that the preliminary designs can be finalized and start of

engineering development can occur in early 1995. Douglas has targeted a

specific vehicle, the MD-XX, to incorporate this technology.

START ENGINEERING DEVELOPMENT OF A WING BOX

ON A SHORT-HAUL TRANSPORT IN 1993

START ENGINEERING DEVELOPMENT OF A

PRESSURIZED FUSELAGE IN 1995

72
 OBJECTIVE

There are several specific objectives in this program as well as the Douglas

funded effort. First and foremost is cost savings. In the ACT May 1989

kickoff review meeting at NASA Langley Research Center, we had one chart on key

barriers - one of which was "COST." We have attacked the major cost drivers as

well as almost all of the other cost areas.

The weight savings on individual parts can be as much as 50Z compared to metal,

but on an overall assembly for primary structure the specific objective is to

exceed 30Z.

For a production program the objective is to have Development (non-recurring)

and Production (recurring) costs to be equivalent to a metal system. This has

been demonstrated on a few secondary composite structures already.

To achieve the weight savings we have to have a high strength after impact

damage. Today to get that value we use toughened resin/high strength fibers

which cost much more than the currently used material systems in secondary

structures. Our objective is to use material systems that are less costly than

today's, yet end up with a component that has high strength after damage as well

as better tolerance to all types of impact.

COST SAVINGS OF 40% FROM STATE-OF-THE-ART

COMPOSITES

WEIGHT SAVINGS OF 30% OF METAL ON ASSEMBLED

STRUCTURE

DEVELOPMENT AND PRODUCTION COSTS EQUIVALENT

TO METAL SYSTEM

DAMAGE TOLERANCE EQUIVALENT TO TOUGHENED

RESIN/HIGH-STRENGTH FIBER MATERIAL SYSTEMS

ABILITY TO ANALYTICALLY PREDICT STRUCTURAL

BEHAVIOR AND FAILURE

73
 OVEI_J_LL FOCUS OF ICAPS

Evaluation in the mid-80's indicated that to make a cost breakthrough, a

deviation from the current trend of fabrication had to be made. Douglas

focused on the dry preform resin transfer concept. From early testing it was

evident that a considerable improvement in the damage tolerance could be

obtained, equivalent to the toughened resin high strength fiber prepregs just

reaching the marketplace. These new systems were also higher in cost. The

preliminary stitched preform concepts were sponsored by Douglas and were further

evaluated under NASA contract in the late 80's to broaden the understanding.

With the advent of the ACT program it was an opportunity to develop a more

comprehensive evaluation program. While Douglas funds were developing the wing

section using the new toughened resins, the NASA ICAPS program focused on the

dry stitched preform concept. The stitching parameter development will be

explained in the later presentation by Marvin Dow of NASA. All current

stitching work is being accomplished by our old (but modified) Douglas sewing

machine. A more versatile machine was needed to evaluate the speed-up of the

stitching concept. After numerous equipment developer visits, a general

specification was developed and the solicitation, proposal evaluation, and

source selection process was accomplished. Several new resins are being

evaluated that are more suitable for the RTM (pressure and/or vacuum) process.

Tools for each process are fabricated for the subcomponent and testing is in

progress. As test results are analyzed, the design/prediction methods are

further evaluated. As the larger components are made, the cost projection

analysis will be re-evaluated.

SIMPLE DESIGN
• ACCOMMODATES FABRICATION, INSPECTION, AND REPAIR

UNIQUE COMPOSITE MATERIAL

• STITCHED FABRIC PREFORMS AND RESIN TRANSFER MOLDING

PROCESSES AND TOOLING

• RTM AND ATP WITH POTENTIAL FOR USE ON LARGE
INTEGRATED STRUCTURES

DEVELOPMENT TESTING
• DESIGN VALIDATION AND PERFORMANCE PREDICTION DATA

STRUCTURAL MECHANICS

• ENGINEERING FOUNDATION FOR FUTURE DESIGNS,

PERFORMANCE PREDICTIONS, AND CERTIFICATION

74
 PRIOR TO ACT PROGFU%I_

The selected baseline transport configuration used for the advanced technology

related to subsonic transports in 1984 were the D3300 configuration for the

wing and the MD-100 for the fuselage. The specific features addressed on the

wing and fuselage would be utilized in a next generation new transport that

would utilize organic composite structures. The composite joints and cutouts

program under NASA contract used loads and design criteria from these baseline

aircraft. These development contracts were awarded after the ACEE program and

preceded the ACT program.

BASELINE STUDIES
• ADVANCED DESIGN HIGH-ASPECT-RATIO WING
• DOUGLAS DEVELOPMENT PROGRAM

BASELINE DESIGN
• LOW STRAIN CONCEPT
• BLADE-STIFFENED WING/J-STIFFENED FUSELAGE
• FEATURES
- SKIN, BLADE LAYUP PATTERN IDENTICAL
' REPAIRABLE
• DOUGLAS TOUGHENED RESIN WING BOX PROGRAM

DEVELOPMENT
• NASA JOINT AND CUTOUT CONTRACTS
• NASA PANEL FABRICATION CONTRACT
• DOUGLAS COMPONENT DEVELOPMENT

75
 WING BASELINE AIRCRAFT

The D3300 configuration was an intermediate range 200-passenger aircraft for

which detailed wing loading and configuration data was generated. Today the

Douglas Operating Plan refers to this aircraft as the MD-XX and is focusing on

very high bypass ducted fan engines,but anunducted counter rotating system could

still be a consideration.

D3304-2 CONFIGURATION

109 FT 7.5 IN.

_0ooooooooo
......o.....o....._ 0-_._
127 FT 10 IN.

76
 BASELINE WING TEST BOX RELATIONSHIP

The test units for both Douglas and ICAPS Program is identified on the baseline

wing plan form. The section is at the side of thebody tothe wing aerobreak,which

is approximately twelve feet. For the baseline wing configuration used in this

program the test units are full scale.

DIHEDRAL BREAK
TEST BOX
SWEEP BREAK

WING BREAK 3AL

AIRFOIL

HIGH ASPECT
RATI(

77
 WING DESIGN CRITERIA

Wing loads and criteria are summarized. Shear and bending loadings are a

maximum at the aerobreak station (outboard end of the test box) rather than at

the root. This is because, even though root shear and bending moments are

higher, this is offset by the much greater box dimensions at the root station.

The design fuel pressures derive from the 9g crash condition. Minimum flexural

and torsional stiffnesses are those required to meet dynamics and aileron

effectiveness requirements.

Nx LB/IN. TENSION 23,200

COVER PANELS-
MAXIMUM SPANWISE ULTIMATE

LOADING AVERAGED
COMPRESSION 23,600
ACROSS BOX

SPARS - Nxy LB/IN. ULTIMATE 3,700

MAXIMUM SHEAR FLOW [AVERAGED OVER SPAR DEPTH]

REQUIRED FLEXURAL
(El) 10 6 LB/IN 2 261,300
STIFFNESS (ROOT)

REQUIRED SHEAR
(GK) 106 LB/IN.2 137,500
STIFFNESS (ROOT)

LB/I N.2 INNER WING 36.2

MAXIMUM FUEL
PRESSURE ULTIMATE OUTER WING 46.7

78
 FUSEI_%GE BASELINE AIRCFJ_FT

The MD-100 configuration was a pre MD-II configuration on which advanced

structural configuration changes were evaluated. The NASA contract on fuselage

cutouts and joints for composite structures used loads from this preliminary

design. A considerable number of composite structural configurations were

evaluated and the "J" stiffened panel was selected.

MD-100 CONFIGURATION

169 FT

79
 FUSELAGE DESIGN CRITERIA

The fuselage loadings given represent maximum forward fuselage conditions that

occur at the front spar station. Hoop loading due to internal pressure

differential are based on the 2.0 ultimate factor condition with no other loads

present.

CONDITION CROWN SIDE KEEL

MAXIMUM LONGITUDINAL
N x LB/IN. 4,600 3,200 2,300
TENSION LOADING

MAXIMUM LONGITUDINAL
N x LB/IN. -1,700 -3,000 -2,100
COMPRESSION LOADING

MAXIMUM SHEAR FLOW 800 1,500 1,000

Nxy LB/IN.

MAXIMUM HOOP
TENSION LOADING 2,157 (AT 2p)
Ny LB/IN.
(PRESSURE = 9.1 PSI)
ALL LOADS ARE ULTIMATE

80
 EVOLUTION OF DESIGN STII_%INS

The baseline composite wing is not based on a high strain concept since this is

not an option because of stiffness and supportability constraints. A simple,

easily producible and repairable design based on 4,000 strains was shown in

an earlier NASA program to be weight-competitive on a damage tolerance basis

with high strain concepts. Further reductions in design strain have been

introduced _o meet specific stiffness and repair criteria. The wing is

designed to be repairable at any location by mechanically fastened patches.

HIGH STRAIN WING HAS HIGHER FABRICATION

COMPLEXITY AND REPAIR LIMITATION

SELECTED HARD SKIN DESIGN LIMITED TO 4,500_

STRAIN FOR DAMAGE TOLERANCE

STRAIN REDUCED IN PLACES TO 4,000_ TO MEET

STIFFNESS TARGETS

STRAIN FURTHER REDUCED TO 3,750_ TO ALLOW

BOLTED REPAIR PATCHES AT ANY LOCATION

81
 WING STIFFENER DESIGN

From the many studies and development tests performed, the stiffener selected

for the wing design being considered is a basic stitched preform 9-ply balanced

pattern arrangement from which stiffener and skin stacks are made. In the
cross section shown, the vertical blade is stitched, the skin stack is

stiffened, and then the blade flanges are stitched to the skin. It will be

possible to do most of this automatically after discussion with equipment

manufacturers.

...-- (0.432)
EIGHT 9-PLY SEGMENTS

l
STITCHED, 3/16-1N. SPACE

/--PRE-DENSIFIED

2.00
0-DEG SPACER

3/8 7/16 IN. "--'_1 I_-.- " (o.216)

L
IIIIIII IIIIIll
TWO STITCHES EVERY STEP

(ONE STITCH ON THE FIRST)

1/8-1N. SPACE, 0 ° ROWS
(0.324)
SIX 9-PLY SEGMENTS
STITCHED, 3/16-1N. SPACE

82
 FUSELAGE STIFFENER DESIGN

From the many studies and development tests performed prior to the ACT program,

the stiffener selected for the fuselage design is a "J" configuration. These

stiffeners will be dry stitched in the vertical stem and "J" cap, and then

stitched to the fuselage skin.

(0.120) --

0.50 I"--
I

(0, 45, 90, -45, 0)s 2

1.45

/
(0, 45, 90, -45, 0)s

(0.060)

3X R 0.09 -__ L
t
2.25 (0.072) SKIN

(0, 90, 45, 0, -45, 90)s

REINFORCED LONGERON

83
 GUIDELINE FOR REPAIRABLE LAMINATES

A generalized design guideline in use for sometime at DAC constrains the per-

centage of fibers in each of the four directions (0, 90, +45, and -45) to fall

between 12.5% and 37.5Z. The fiber directions must be thoroughly interspersed

to prevent the matrix from becoming the weak link in the laminate. These

guidelines were set to allow good bolt load transfer behavior and did not nec-

cessarily apply to thin lightly loaded laminates or to laminates where no

fasteners were required either in the original design or for repair. The chart

extends the range slightly and permits, for example, the wing cover panels to

increase the percentage of 0 ° fibers to 44.4%. This was desirable in meeting

stiffness requirements.

± 45 °
100 0

0 ° 90 °
0 20 40 60 80 100
PERCENTAGE OF 90 ° LAYERS

84
 COST I)I_IVEI_S

Since the early 1980's, Douglas has been focusing on the cost issue of

composite components in three specific areas: non-recurring, recurring and

ownership. During the 1970's, when we were making sizable components under the

NASA sponsored Aircraft Energy Efficiency program, the initial focus was weight

savings. Their weight savings goals were achieved, but at what price? When

cost estimates were made for production, it was clear that something different

had to be done in order to make the transition into production. For example,

on the DC-10 rudder program, by a few internal system changes, tool variations

the projected recurring costs could be reduced by 30%. The major drivers are

shown in the following charts.

NONRECURRING

RECURRING

OWNERSHIP

85
 COST DRIVERS -- NON--RECURRING

For non-recurring costs, an investigation was made to determine the percentage

range of the various items that make up the total cost to focus on, which were

the major contributors. The background data was from a number of programs and

a range of parts from small to full assemblies. Both commercial and military

components were included. Large percentage variables were identified in tool-

ing and capital equipment. Tooling related to the complexity of the part being

considered and the capital equipment variable depended on how the cost was

distributed to various programs or how many units were to be considered in the

spread. In the ICAPS effort, heavy emphasis will be focused on simplified

design to reduce tooling costs and define the fabrication process to not

require high cost equipment items.

CAPITAL

EQUIPMENT

FACILITIES

DESIGN

PROCESS

;ERTIFICATION

MATERIAL

CERTIFICATION

ANALYSIS

DEVELOPMENT

TESTS
TOOLING

86
 COST DRIVERS -- RECKING

For recurring costs, the data show that there is a wide percentage range for

the various elements from different data sources. It's clear that materials,

fabrication, assembly, and quality control are the major cost drivers. On very

small parts, the material cost is a high percentage in making the part. On a

composite component made up of several parts, the material cost percentage is

much lower. For ICAPS, the four major percentage areas are the ones on which

the major focus is being addressed with significant reductions anticipated with

the concept approach being taken. A majority of the other cost items will be

reduced to some extent. As larger elements and subcomponents are fabricated, a

specific set of percentages can be verified, and a cost relationship figure can

be generated and compared to the state-of-the-art fabrication of the same

component.

MATERIALS
FABRICATION

LABOR )

ODC

STORAGE

ODC
QUALITY

ENERGY
CONTROL

ASSEMBLY MArL

EQUIPMENT WASTEJSCRAP
LABOR
ODC
MAINTENANCE
ENVIRONMENT

87
 COST DRIVERS -- OWNERSHIP

Although we have some carbon epoxy components flying on the DC-10 since June of

1976 and have accumulated over 55,000 hours on the high time units, we do not

have a sufficient experience sample to draw specific conclusions at this time.

Where we look at a transport that will have 40 or more years of service it's

difficult to relate the inspection, maintenance, and repair history of only a

few years.

In D0C data used by airlines, the cost of ownership in todays environment is

approximately 50%. In ICAPS the major thrust is to develop the concept that

would offer a 50% reduction in the cost of ownership for the composite compon-

ents. As the percentage of composites increases, of the total operating weight

empty there would be a significant effect to the operator D0C relationship. In

the airlines D0C maintenance element, the structure represents half of this

value utilizing composites that virtually eliminate corrosion and have a sig-

fant reduction in fatigue, there would be a further cost savings impact to the

operator. On this figure we represent ownership as half and maintenance as the

other half with regard to structure, and until we have more time to work with

operators to develop cost impact trends, we will not connect the outside

element to an operator cost pie chart.

REPAIR

88
 SPECIFIC FOCUS OF ICAPS

This listing is what ICAPS is accomplishing to reduce the high cost drivers. A

number of key parameters with respect to stitching have been investigated to

optimize the process versus structural capability of the structure. Two new

machines are being procured: one that will handle large widths for several pro-

cesses and a computer controlled unit that will handle stiffener to skin

stitching. The fabrication investigation covers two types of tools for RTM

(resin transfer molding) in terms of vacuum or pressure impregnation. To com-

pare processes for the fuselage, the Hercules advanced automated tow placement

is being evaluated. New resins formulated for RTM are also being used. Of the

thousand specimens tested, the data are being used for stitching parameter

determination and structural mechanics methods improvement. As the larger com-

ponents are fabricated, the cost data will be used to update the predictions

that have been made to date. Each of these subjects is covered in further

detail.

STITCHING DEVELOPMENT

STITCHING MACHINE PROCUREMENT

FABRICATION PROCESS TOOLS

DEVELOPMENT AND RESINS

SUBCOMPONENT/COMPONENT TESTING

STRUCTURAL MECHANICS

COST DATA PROJECTIONS

89
 POTEI_'I_IAL FOR STITCHING/RN

Stitching has for a long time been considered to improve the through the thick-

ness properties of laminated structure. When used with prepreg there was

extensive fiber damage. The parts looked good, but the structural capabilities

were degraded. With the development of unidirectional cloth with straight

fibers, the probability of ply stacking and stitching was investigated.

Structural testing showed substantial increase in damage tolerance capability

with high residual strength approaching that of the toughened resins with high

strength fibers.

Evaluations show that cost reductions could occur in many areas. Some are

significant, others are not cost drivers but relate to facility requirements.

As the design concepts were formulated, additional benefits were identified;

i.e. near net shape stitched preforms, self-contained tool for RTM and cure,

etc. With large high speed stitching equipment, scale-up to large aircraft

structures is possible.

LARGE DAMAGE TOLERANCE INCREASE

LOW MATERIAL COST

LOW RECEIVING INSPECTION COST

LOW STORAGE COST

LITTLE AGING CONCERN

AUTOMATED NEAR NET SHAPE PREFORM

SELF-CONTAINED TOOL

LOW WORKPLACE ENVIRONMENTAL IMPACT

OPTIONAL FIBER PLY ORIENTATION

SCALE-UP CAPABILITY TO VERY LARGE STRUCTURE

90
 STITCHING DEVELOPMENT

This is a listing of the various parameters that were considered and evaluated,

and explained later as previously noted.

STITCH TYPE
• LOCK, CHAIN
STITCHING THREAD
• GLASS, KEVLAR
STITCHING PATI'ERN
• DENSITY AND DIRECTION
STITCHING TENSION
• NEAR NET SHAPE

COUPON/SPECIMEN TESTING

FABRICATED 60 PANELS CUT INTO APPROXIMATELY

1,000 SPECIMENS FOR NASA TESTING

91
 STITCHING DEVELOPMENT TESTS

More than half of the tests planned in ICAPS are for stitching development.

These are mostly done on quasi-isotropic laminates, which is close enough to

practically used laminates in composition. One stitching parameter, whether

stitch type, thread or pattern, is varied at a time. This slide shows the

various stitch patterns chosen. Each panel was stitched accordingly, impreg-

nated and cut into specimens. The goal of stitching development is to achieve

the best balance of improved damage tolerance properties with minimum loss of

basic tension, compression and stiffness properties.

PANEL SPACING STEP DIRECTION

!!_ 2130.s2130-5 1/4 IN• 1/8 IN. 0

13 IN. _ ........ o
1,.,.o
13 IN. 28 IN.
• I'--%,N. ,6 o 2130..6 3/16 IN. 3/16 IN. 0
........1--
I
'/8,N.--I 28 IN.__I

90
2130.1 I PANEL
J 2130-1
2130-3
SPACING
1/8 IN.
1/8 IN.
STEP
1/8
3/16
IN. DIRECTION
IN.
90
0
r/,,N--
i ,-- ,oJ 2130-7 3/16 IN. 3/16 IN. 90

2130-4 3/16 IN. 1/8 IN. 0 2130-8 118 IN. 1/6 IN. 0
28 IN.
,, 28 IN.

ii__t 213o-2 ii__, 2130-6

13 IN. II_
;; L 1/8 iN 45 0

_/_,N---
ii__ 90"'J ,-- ,o"J
28 IN. 2130-7

ii t 2130-3 ,.3/,_,N.
13 IN. ! !-- 45 0 13 IN. 45 0
i
, 28L3/,6,N.
'/_,N.--'i IN.
90____ 3/_,N--II-- ' ,o--_
28 IN. 28 IN.

i i.__, 2130., i! J _,3o-8

13 IN. ;; i'_,_' 45 0 13 IN. ; ; -- 45 9
! YsIN. i L'I/61N.

I -; ,-'/8,_. ,oJ

92
 ORIGINAL SINGLE NEEDLE
STITCHING MACHINE

All heavy density stitching development to this date has been on the manual
control single needle 5' long arm machine shown in this photograph. Adjustment
of stitch step and tension on the thread is available. The material to be
stitched is secured in a holding frame and an adjustable guide rail controls
the spacing of stitched rows. This machine is located at and operated by
Ketema, Anaheim, California (formerly Textile Products, Inc.).

93
 ORIGINAL MULTI-NEEDLE
STITCHING MACHINE

Light density stitching has been accomplished on this multi-needle machine at

Ketema. The machine is 60" wide with 1" needle spacing and maximum thickness
stitching of 10 layers of uniwoven fabric. The equipment does not have multi-
feed rolls so individual layers of material were cut, stacked, and then passed
through the machine.

94
 NEW STITCHING MACHINE

The stitching machine selection for this program is a Pathe machine that is
being modified to have up to twelve unidirectional roll feed to lightly stitch
the fuselage skins and a more unidirectional roll feed to lightly stitch the
wing preforms. The wing basic plies will be stacked and stitched together by
the basic pattern selected. This machine is in the process of being checked
out by stitching some of the subcomponent parts at the machine manufacturers.

PATHE
MULTI-NEEDLE STITCHING MACHINE

95
 BENEFITS OFFERED BY

NEW STITCHING EQUIPMENT

The DAC ICAPS program features the automated stitching process to fabricate dry

fiber near net shape stitched preforms. The development work completed to date

to prove the structural benefits of the process (damage tolerance) has all been

accomplished on a manual contour single needle long arm machine.

ICAPS will soon demonstrate the economic benefits of stitching that will become

available with the new automated stitching equipment now being built for this

project by Pathe.

This viewgraph shows

i. The description of the original machine, the computer control automated

single needle machine, and the mechanical control multi-needle machines

that are being purchased.

2. Identifies the assumed stitching speed for each machine. All machines are

capable of operating at much higher rates. These rates are the assumed

rates for a i/4" thick preform and are conservative in all cases.

3. The estimated stitching efficiencies from observations on the original

machine and from the experience of the automated machine builder.

4. Available penetration rate per minute is the assumed stitching rate times

the efficiency factor times the number of needles.

5. The time to stitch one square foot with penetration density of 64 per

square inch is the total penetrations per square inch (64x144 = 9216/ft 2)

divided by the penetrations per minute.

6. The time to stitch an 8'x12' uniform thickness panel is a multiple of step

5 above, times 96. Note that the current manual machine would require

24,576 minutes to stitch this skin where the new automated single needle

machine would require 1248 minutes. This is a 95% reduction in time.

The multi-needle machine with 128 needles would require only 55 minutes to

stitch this panel. This is a 95% reduction in time from the automated single

needle machine.

The multi-needle machine with all 256 needles will require considerable testing

before it can be considered for the application.

98
 FABRICATION PROCESS FOR

DEVELOPMENT AND RESINS

Two basic RTM processes and the automated tow placement process are being

developed and demonstrated in the Douglas ICAPS program.

RTM

Vacuum impregnation and vacuum film in fusion are being developed as a method

of fabrication for the Douglas wing effort. This method of RTM utilizes

either liquid or film resins depending on the resin processing characteristics

and final mechanical properties.

Pressure impregnation is being developed to fabricate composite fuselage skins

and longerons. This method of fabrication will rely entirely on liquid resin

systems and in-tool curing techniques.

A_TED TOW PLACEHI_rr

Finally, the automated tow placement process of Hercules is being demonstrated

under subcontract to fabricate the same fuselage skin and longerons components

being done under pressure RTM.

Completion of a subcomponent cost and performance evaluation (end of Phase A)

will depict which process or combination of processes will be used for

continued development in Phase B.

FABRICATION PROCESS AND TOOL EVALUATION

VACUUM IMPREGNATION

FILM INFUSION

PRESSURE IMPREGNATION

AUTOMATED TOW PLACEMENT

97
 VACUFUI_I IMPREGNATION OF STIFFENED P_EL

LIQUID RESIN INFUSION

The complete process requires only vacuum source and either self heated tools

or an oven.

Resin is mixed and degassed as required to remove trapped air. The tool is

heated to desired temperature and vacuum is drawn on the system. The tempera-

ture varies for each resin system, for optimum impregnation and ranges from

150°F to 275°F. When the heat has stabilized, the valve is opened and the

resin flows through the runners of a surface call plate, through holes in the

runners that connect to the dry preform. The resin passes through the skin, up

through the webs of the stiffeners, through holes in the upper pressure plate,

and into the upper bleeder cloth. The temperature and resin gel time are

coordinated so that the resin stops flowing and cures before it reaches the

vacuum outlet line. Temperature may be raised, after resin gel, to a higher

temperature to complete the cure cycle - all with only vacuum bag pressure.

BLEEDER CLOTH .---_ ,,_ TO VACUUM

TOP PRESSURE
PLATE
VACUUM BAG
DRY PREFORM SEAL

OOOOOOOOOOOOOOOOOOOOOOOOOO

EDGE DAM SEAL ELIMINATES-'-'-7__ .._ill

SIDEWISE RESIN FLOW _._

98
 WING ELE;MENT PREF0R.M

Douglas preforms are being developed with unidirectional graphite cloth being
stitched into the three-dimensional shapes representing structural wing
components. These structural shapes are then stitched directly to a stitched
wing skin to give a complete wing skin assembly preform. This method of
preforming has shown a g.reat cost reduction in both layup and assembly of
composite wing skins. Shown in this figure is a 3 blade stitched wing skin
preform.

99
 WING ELE;MENT TOOLING
VACUUM IMPREGNATION

The tooling f o r vacuum impregnation of three blade stiffened wing elements

consists of a set of machined aluminum blocks used to define the wing skin
control with a set of matching flat tooling plates used to sandwich the
preform/tooling blocks during impregnation. This entire assembly is then
vacuum bagged while the resin is infused and cured in the preform.

100
 RESIN FILM INFUSION OF

STIFFENED PANEL

The complete process requires only vacuum source and either self heated tool or

an oven.

Resin film of desired amount is placed on the surface of the tool. The dry

preform is placed on the resin, the upper tools located in place, and enclosed

in a vacuum bag. Heat is applied. The resin viscosity thins and the resin

flows upward through the skin, the webs of the stiffeners, through the outlet

holes in the upper pressure plate, and into the bleeder cloth where it gels.

The amount of resin, the temperature, and the gel time are coordinated so that

the resin will stop flowing and cure before it reaches the vacuum outlet line.

Temperature may be raised after resin gels to a higher temperature to complete

the cure cycle - all with only vacuum bag pressure.

A TO VACUUM
BLEEDER CLOTH ---_
TOP PRESSURE \ ////

PLATE "_ _ //// VACUUM BAG

, II/!
• ¢' ¢, _, •

II __T_Op_L._'//2f61'_P_T99__gL_mTgOLy_
_O_/_J.Z_ , ,,

/ SE_L
_.oEL,_,._T_S
_<--00UBLE.//

101
 NEW RESIN EVALUATION

In the continuing Douglas evaluation of new resin there were a considerable

number investigated that were in liquid form suitable for the ICAPS program.

From the past fifteen months, the selection was narrowed to three or four which

are in the ICAPS specimen/coupon evaluation process. The objective is to

narrow down to one or two for the component fabrication early next year.

DOW CHEMICAL CO. TACTIX 695

SHELL CHEMICAL CO. DRL 862

CIBA-GEIGY CO. MATRIMID 5292 AJB (BMI)

BRITISH PETROLEUM E-905L

(OPEN)

102
 RTM FUSELAGE ELE3KECNT TOOLING

The pressure impregnation process is being developed for composite primary

fuselage applications at Douglas Aircraft Company. This process is utilizing
dry stitched preforms in combination with matched metal tooling and a positive
displacement resin meter mis pump. The Douglas RTM fuselage element tool is
shown in this slide.

103
 DAC RTM MACHINE

All the pressure RTM development being done at Douglas Aircraft is being done
with a liquid control multi flow-4 machine. This machine can handle oneor two
component resins that require pre-heat temperatures of up to 175OF. This
machine will be used extensively in the RTM fuselage sub-component work to be
done at Douglas during the remainder of Phase A of the ICAPS program.

104
 PROCESS MODEL AND MONITOR

William & Mary College (W&MC) and Virginia Polytechnic Institute (VPI) are

combining forces to create a process model and process monitoring system.

W&MC, under the direction of Dr. David Kranbuehl, has a FDEMS system that will

record resin collation (as a preform is being impregnated along with resin

viscosity and resin cure condition). A resin kinetics model is being developed

to establish the relationship between DSC, viscosity, temperature, and time

versus FDEMS measurement.

WILLIAM AND MARY COLLEGE

DR. DAVID KRANBUEHL

PROCESS MONITOR

FDEMS SYSTEM
• RESIN LOCATION
• RESIN VISCOSITY

• RESIN CURE CONDITION

RESIN KINETIC MODEL

• DSC VERSUS VISCOSITY VERSUS TIME

VERSUS FDEMS RELATIONSHIPS

• CREATE RELATIONSHIP MODEL

• VERIFY MODEL PREDICTIONS

105
 C[[}{E MODEL,_N[I) MONITORING

VPI is conducting a porosity study of preforms and resin flow rates with

temperature/viscosity/gel time restraints. VPI, under the direction of Dr. A1

Loos, will use their data alone with the W&MC data to construct a process cure

model. The objective of this model is to assess a new design, predict if the

RTM system will work, and if so, specify a curing time/temperature cycle.

VIRGINIA POLYTECHNIC INSTITUTE (VPI)

DR. AL LOOS

POROSITY STUDY OF PREFORMS

RESIN FLOW RATE

TEMPERATURE/VISCOSITY/GEL

CREATE PROCESS PREDICTION CURE MODEL

• VPI AND WILLIAM AND MARY DATA

VERIFY MODEL PREDICTIONS

106
 ~ ~~

AUTOMATED F I B E R T O W PLACJZMENT

As part of the Douglas pressurized fuselage development, the automated fiber

tow placement process is being evaluated in conjunction with the Douglas RTM
fuselage effort. Hercules Inc. is fabricating element and subcomponent
fuselage panels with their most advanced fiber placement machine in this Phase
A effort. Shown below is a side view of the Hercules tow placement machine and
a photo of the element panel fabricated earlier this year.

107
 DEVELOPMENT TESTING

This is a summary of the wing and fuselage test articles being fabricated. The

majority of the elements (smaller sizes) are fabricated and in test. The sub-

components (larger sizes) are in the fabrication process.

WING (TEST ARTICLES - 15 TOTAL)

• STRINGER PULLOFF

• THREE-BLADE STIFFENED PANELS (24 IN. X 21 IN.)

• SIX-BLADE STIFFENED PANELS (42 IN. X 60 IN.)

FUSELAGE (TEST ARTICLES - 35 TOTAL)

• LONGERON PULLOFF

• THREE J-STIFFENED PANELS (15 IN. X 21 IN.)

• SIX J-STIFFENED AND TWO-FRAME PANELS (48 IN. X 60 IN.)

108
 EL_ENT/SUBCOMPONENTS

This illustration shows the sizes of the various coupons, elements, and sub-

components. The coupons were of thick and thin sizes to do development work

for both fuselage and wing structures. When the elements were designed, they

specifically focused on the J-stiffened panels for the fuselage application and

the thicker blade stiffened panels for the wing application. The subcomponents

are scale-up of the element test articles that were sized to fit the existing

testing fixtures.

.._ 2 FT .-----.-_ /

4FT

COMPOSITE FUSELAGE

RTM FUSELAGE
_5 IN._ I
ELEMENT I_ 4FT --

-- I'-- _--------- 2 FT ---'-"_"/

IN.

I
) IN.

=.!,!5
1.75
- IN.
1 --811-
RTM WING RTM WING
ELEMENT SUBCOMPONENT

109
 THREE BLADE WING PANEL

This photograph shows one of the three blade stiffened (element) wing panels
fabricated for compression development testing. Panels of this type as well as
coupons and test specimens using the current state-of-the-artresin and inter-
mediate strength fibers have shown that the dry stitch preform fabrication
concept structure yields strength after damage equivalent to the newer toughened
resins and high strength fibers. This can be accomplished at a much lower
material and fabrication cost.

110
 FUSELLAGE "3" STIFFENED PANEL
RTM

This photo represents a t h r e e "J" s t i f f e n e d f u s e l a g e element p a n e l f a b r i c a t e d

by t h e p r e s s u r e RTM p r o c e s s .

111
 FUSELAGE "3" STIFFENED PANEL
AUTOMATED TOW PLAC-ENT

This is a cross section photo of a three "J" stiffened fuselage element

panel fabricated with the automated two placement process under subcontract to
Hercules.
 STRUCTURAL MECHANICS

INVESTIGATION GOALS

As we introduce the innovative concept of Cost-Effective Resin Impregnated

Stitched Preforms (CRISP), we need to investigate many new dimensions of

composite structural mechanics. CRISP centers around two new approaches:

stitching dry preforms and resin transfer molding (RTM in the general sense,

including both vacuum and pressure processes). Both stitching and RTM alter

the material properties to the extent that we need to modify the conventional

laminate theory and composite design methodology. CRISP drastically improves

composite damage tolerance performance under compression. Therefore we need to

develop a new set of residual strength prediction tools for the design and

application of stitched composites. There are more than a thousand tests at

coupon, element, and subcomponent levels planned in ICAPS Phase A to validate

our structural mechanics development.

MODIFIED LAMINATE THEORY FOR STITCHED MATERIAL

DESIGN METHODOLOGY WITH STITCHED LAMINATES

RESIDUAL STRENGTH PREDICTION

SUPPORTED BY MORE THAN 1,000 TESTS

113
 MODIFIED lAMINATE THEORY FOR

STITCHED MATERIAL

Stitched laminates have greatly improved through-thickness strength, which

leads to higher compression after impact (CAI) strength, and in many cases,

higher effective compressive strength due to the suppression of brooming. We

are in the process of further quantifying these benefits of stitching. As

stitching helps CAI and compression performance, it alters other in-plane

properties, and our most important finding is that this property change is

non-uniform. Stitching disturbs the fibers perpendicular to stitch-line the

most, therefore causing more strength reduction in that direction. Further,

the property change is not uniform through the thickness either and the surface

plies are affected more. This causes more strength reduction in thin laminates

than in thick ones. In essence, stitched laminates are treated as if they were

hybrid materials -- the plies having their fibers parallel, 45 ° or perpen-

dicular to the stitching lines, are modeled as different materials. Their

strength and stiffness are added up using regular laminate theory.

± 45"

LAMINA PROPERTIES OF

MATERIAL 1: FIBER//STITCH

MATERIAL 2: FIBER/45 ° STITCH + //'XX

MATERIAL 3: FIBER _1_ STITCH ,7dA5/%
e_/ "/'_X F Y'/ 40\'5,

LAMINATE STACKING
/_'\ SEQUENCE INFORMATION/ \

USE STITCHING PARAMETER TO o,/ o',<,,

0 _20_40--60_80_100
SORT OUT MATERIAL PROPERTIES
PERCENTAGE OF 90" LAYERS
FOR EACH PLY, THEN
PREDICTED LAMINATE PROPERTIES
SUM UP STIFFNESS AND STRENGTH

114
 PREDICTION BY MODIFIED LJ_PIINATE TI-I]E_ORY

The predictions shown in this slide are derived from a phenomenological theory

modified for stitched laminates. Without any thickness correction factor, the

model tends to under predict compression strength. A study is ongoing to

determine if one extra factor is needed for the thickness effect. More tests

are planned to study improved uniwoven material and thickness/stacking sequence

combination effects.

120
107

lOO - 96.5 97.1

93
87
_×/ _ ....... 85 I
80 I
I
PREDICTED/ I I
J 66 _ ....
MEASURED
...... _ ......... 60*
(%) 60 52*
,z,z/× ...... ¢</2_2

4O

,-,:,
>:
20 - ):: ::
:.:/11/?

0 i

E(0°) El90 °) F t(0 °) Ft(90°) F c(0 o) Fc(90 °) E(0°) F t(0 °) Fc(0° )

* MLT TENDS TO ICAPS WING-SKIN, 0°-STITCH QUASI-ISOTROPIC, 0°-STITCH
UNDERPREDICT (44.4% MATERIAL 1, 44.4% MATERIAL 2, (25% MATERIAL 1, 50% MATERIAL 2,
COMPRESSIVE STRENGTH 11.1% MATERIAL 3) 25_0 MATERIAL 3)

115
 RESIDUAL STRENGTH

PREDICTION DEVELOPMENT

It is important to predict the ultimate strength (failure load) of damaged

parts. Compression after impact (CAI) is probably the most common means of

residual strength measure. However, large structures have redundant load paths

built in them and do not lose the same percentage of strength with the same

damage as the coupons. Also, most structures have stiffening members and

stiffener-skin separation can be the failure mechanism. To estimate the

ultimate strength of a stiffened part, this has to be accounted for in addition

to buckling, crippling, and pure compressive failure. Stiffener pull-off tests

will provide us with such strength data of stiffener-skin separation resistance

for various material/process parts. We will use the CAI coupon and stiffener

pull-off test results to establish impact damage modeling. Empirically matched

"reduced stiffness and strength" parameters will be reduced and fed into finite

element analyses. Element panel tests will be used to calibrate our prediction

model parameters. The refined model will be used to predict the ultimate

strength of damaged subcomponents.

tt_ , fLOAD SPLICE(S)

,--IMPACT
DAMAGE t, j r-CURVED DOUBLER

I"
,l_ I_ _
- !

BASE OF TEST FIXTURE

t LOAD INDUCTION FIXTURE j

]_ LOWER DOUBLER

CAI STIFFENER PULL-OFF DAMAGED STIFFENED

PANEL COMPRESSION TEST

ANALYSIS
• DAMAGE MODELING (REDUCED STIFFNESS AND STRENGTH)
• ULTIMATE STRENGTH PREDICTION ON DAMAGE PARTS
(FINITE ELEMENT ANALYSIS WITH PARAMETERS
OBTAINED FROM DAMAGE MODEL)

116
 OTHER DEVELOPMENT TESTING

Since the ACT-ICAPS program has been initiated there have been several specific

activities addressed to enhance the program results. Subcomponent fatigue

testing will be accomplished prior to the Phase B components to determine if

there are any issues that need to be addressed. Ply drop-off evaluation is

being conducted to determine if the wing panel stacking sequence is

acceptable. Validation of the spice joint is being made as well as thermal

cycling of the stitched preform concept. Many components have over the years

had varied compression failures related to the length between restraints called

"brooming failure." This has been documented in numerous technical papers. A

clarification of how stitching will affect these failure characteristics is

being made. Additional damage tolerance tests are being conducted to cover

additional impact force variables.

(DAC- NASA LANGLEY- UNIVERSITIES)

FATIGUE

PLY DROP-OFF CRITERIA

BOLTED JOINTS

THERMAL CYCLING

COMPRESSION FAILURE MODES

DAMAGE TOLERANCE

117
 COST MODELING

For the wing program there are three large scale wing box evaluations for

comparison of cost. The wing box size is 8 foot root chord, 5 foot tip chord

with a 12 foot section of span representing in size the side-of-body section of

a DC-9-30 wing. The three cost analyses are on a metal, conventional com-

posite, and ICAPS composite version.

CONVENTIONAL WING DESIGN - METAL

STATE-OF-THE-ART TOUGHENED RESIN WING - DOUGLAS

INNOVATIVE STITCHED DRY PREFORM WING DESIGN - ICAPS

118
 STATE--OF--THE--ART COMPOSITE

WING DESIGN

The Douglas composite wing program using toughened resin high strength fiber

will consist of five ribs, two spars, and stiffened lower skins. The spars are

typical spanwise "C" sections that will be mechanically fastened to the skins.

Ribs will be flat with cutouts and stiffener. The cover skin thickness will

taper by ply drip off and a blade stiffened cocured component. Rib shear clips

will be secondarily bonded and fastened to the skins.

The component to be fabricated will have hand layup in many areas, but the cost

model will search for and use the lowest cost fabrication concept (i.e. tape

layup, filament/tow placement, etc.).

.r,r qS_J_

INTERCOSTALCLIPS \ / _r_, > _, L, "-l, _-_-1

\\ / _ [-_'_-_' _ _-_'_ ._ RIBS (ACCESS HOLES NOT SHOWN)
STRINGERS _ _ _ _ [ _' j_,, /

/..-"

LOWER SKIN PANEL

FABRICATION PROCESSES: AUTOMATED TAPE LAYUP (ATP)

119
 ICAPS WING BOX ASSEMBLY

The ICAPS composite wing program will use dry stitched preform stiffened cover

skins with stitched rib shear clips. Once again the ribs will be flat and

nearly identical to the Douglas wing box ribs. The spars will be stiffened

flat webs. The ribs and spar web will be mechanically fastened to the cover

panels with shear fasteners with no tension fasteners through the cover skin.

Savings in assembly and number are realized with this approach as well as

future maintenance issues (i.e. fuel loads, etc.).

UPPER SKIN PANEL

ACCESS DOORS- _ INTEGRAL SPAR CAPS

LAT WEB

(FRONT SPAR) -f I_
FLAT WEB _

LOWER SKIN PANEL INTEGRAL SPAR CAPS,

120
 DESIGN DIFFF/_F/_CES

The significant difference between the state-of-the-art design and the ICAPS

design is simplification and avoidance of problem issues that exist today. On

the left there are mechanical fasteners joining the skin and spars that are in

tension. A problem area develops when there areless than perfectly matched

surfaces between the components. The ICAPS design avoids all fasteners through

the cover skin as well as minimizing the number of fasteners. The ribs and

spar webs are all stiffened flat members that are low cost minimum tool parts.

STATE-OF-THE-ART ICAPS

UNTAPERED STRINGER FLANGES TAPERED STRINGER FLANGES

C-SECTION SPAR INTEGRAL SPAR CAPS
SEPARATE RIB CLIPS INTEGRAL RIB CLIPS
MANY SKIN BOLTS STITCHING - FEW SKIN BOLTS

121
 WING BOX COST DATA PROJECTION

The initial cost/weight projections for the three wing structural concepts are

as shown. As additional data is generated from the elements, subcomponents and

the two large scale box components, the analyses will be updated. The elements

of cost data are listed as well as weight by component.

PERCENT PERCENT
COST WEIGHT

COST ELEMENTS
t
40% WEIGHT ELEMENTS

[_ OTHER
1.0
1
T
1.0
RIBS

[_ MATERIAL

B
REAR SPAR

MANUFACTURING FRONT SPAR

SUPPORT Z
I'_ UPPER SKIN
QUALITY
LOWER SKIN
ASSEMBLY

I--'1 FABRICATION

TOOLING

ITIT_ ENGINEERING LS

LS

IIIIIIIII
C W C W C W

METAL DOUGLAS ICAPS

COMPOSITE UNIT UNIT

122
 DEVELOPMENT _ VERIFICATION

SCHEDULE FOR iCAPS

We are 19 months into the ICAPS program and have developed a considerable

number of development tests focused at wing (thick laminate) and fuselage (thin

laminate). Over the next year, we will be making and testing the large sub-

component panels for both wing and RTM/tow placement fuselage panels. We will

be initiating detail design on the RTM wing box so that the full box can be

fabricated and structurally tested right after the Douglas wing box program.

The full fuselage barrel design will depend on the selection evaluation of the

RTM and tow placement test and cost data. The wing target dates are in agree-

ment with the Douglas and ICAPS goals and objectives,

DEVELOPMENT AND
VERIFICATION TASKS 1989 1990 1991 1992 1993 1994

RTM WING PANELS

RTM WING BOX

RTM FUSELAGE PANELS

ATP FUSELAGE PANELS

FULL BARREL SECTION

BARREL WITH CUTOUTS

(OPTION TO PHASE B)
I
I I
*COMPOSITE WING TECHNOLOGY READINESS END PHASE A END PHASE B

COMPLETED WORK _ DEVELOPMENT TASKS _ VERIFICATION TASKS

NOTE: THIS SCHEDULE IS AN EXAMPLE. START AND END DATES ARE NOT NECESSARILY ACCURATE. HOWEVER, IT IS THE BIG PICTURE SCHEDULE I RECOMMEND.

123
 PRODUCTION STITCHING

PREFORM PROCESS

This slide presents a pictorial step-by-step production process to make near

net shape stiffened wing skin like dry fiber preforms. The basic design uses a

9-ply repeat pattern layers of fabric.

The uni-woven fabric is purchased with major direction of reinforcement fibers

at 0 °, at 45 ° and at 90 ° . Nine rolls of this material are placedin a

feed system, passed through the multi-needle stitching machine and rolled and

stored for later use. Stitching is light density I" apart parallel rolls.

The next step is to unroll one 9 ply stitched pre-ply and cut the doubler

buildups required for the skin.

Six of the 9 ply stitched pre-ply rolls (54 total ply) are positioned on the

feed system and pulled to the multi-needle position of the machine. Doublers

are placed in position on the 54 ply area and then the entire skin and doublers

are stitched together with the desired high density stitch pattern.

The stiffeners are made by stitching 8 nine ply stitched pre-ply material in

areas of the web only for each stiffener. The stitched sections are then cut

to allow the web and flanges to be folder right and left ready for installation

on the stitched skin.

The stiffeners are properly located and the flanges are stitched to the skin

using an automated single needle machine.

The intercostals are cut from two stitched pre-ply layers, stitched in the web

area on a single needle machine and flanges are folded for stitching to the

skin. The intercostals are then located on the skin/stiffener detail and

stitched in the skin flange area.

The dry stitched preform is now out to near net size and is ready for the RTM

impregnation and cure process.

124
PRODUCTION STITCHING PREFORM PROCESS
CUT DOUBLERS FROM ONE
EIGHT ROLLS STITCHED PREPLY
LAYER - STITCHED PREPLY

._ STITCHED STIFFENERS
NINE-LAYER PREPLY
STITCHED __f_

WEB AREA

TCHED PREPLY FOLD OPEN TO MAKE

LADE STIFFENERS
HIGH-DENSITY STITCH __--------_J

SKIN AND DOUBLERS(/.__y

LOCATE STIFFENERS ON SKINS

NON-STITCH FLANGESTO SKIN

STITCHED WING SKIN

PREFORM READY FOR RTM
LOCATE INTERCOSTALS AND
STITCH FLANGES TO SKIN FOLD INTERCOSTALS

125
 S UIM_ML_RY

The goals and objectives of ICAPS are still the primary focus of the on-going
effort. There have been a considerable number of accomplishments that indicate

progress is being made in attacking the high cost drivers. The stitching

development has selected specific parameters. New resins are in evaluation for

the RTM process and Hercules has done an excellent job in providing the devel-

opment panels on tow placement for fuselage application. The equipment for the

RTM process monitoring has been ordered so that shortly we'll be able to go

from the laboratory effort to VPI/W&MC to the shop floor in fabrication of the

large subcomponent,

We have specific application targets and feel very strongly that we can accom-

plish the cost breakthrough we need to make it happen.

ACCOMPLISHMENTS

• WING AND FUSELAGE SUBCOMPONENT DESIGNS

• DESIGN GUIDELINES FOR STITCHED/RTM COMPOSITES

• STITCHING AND ANALYSIS DEVELOPMENT TESTS 80% COMPLETE
• WING PANEL TOOLING DESIGNED
• FUSELAGE PANEL TOOLING COMPLETED

• ATP FUSELAGE PANELS FABRICATED

• HIGH-CAPACITY STITCHING MACHINES PURCHASED

TARGET OF OPPORTUNITY EXISTS FOR APPLICATION TO TRANSPORT

STRUCTURE IN MID-1990

STITCHING DRY PREFORMS/RTM OFFERS BREAKTHROUGH

IN AFFORDABLE MANUFACTURING

ICAPS PROGRAM PROVIDES ENABLING TECHNOLOGY

126
 ADVANCED TECHNOLOGY COMMERCIAL FUSELAGE STRUCTUREI

L.B. Ilcewicz, P.J. Smith, T.H. Walker, and R.W. Johnson

Boeing Commercial Airplane Group

ABSTRACT

Boeing's program for Advanced Technology Composite Aircraft Structure (ATCAS) has focused on
the manufacturing and performance issues associated with a wide body commercial transport fuselage.
The primary goal of ATCAS is to demonstrate cost and weight savings over a 1995 aluminum
benchmark. A 31 foot section of fuselage directly behind the wing to body intersection was selected
for study purposes. This paper will summarize ATCAS contract plans and review progress to date.

The six year ATCAS program will study technical issues for crown, side, and keel areas of the
fuselage. All structural details in these areas will be included in design studies that incorporate a
design build team (DBT) approach. Manufacturing technologies will be developed for concepts
deemed by the DBT to have the greatest potential for cost and weight savings. Assembly issues for
large, stiff, quadrant panels will receive special attention. Supporting technologies and mechanical
tests will concentrate on the major issues identified for fuselage. These include damage tolerance,
pressure containment, splices, load redistribution, post-buckled structure, and durability/life.

Progress to date includes DBT selection of baseline fuselage concepts; cost and weight comparisons
for crown panel designs; initial panel fabrication for manufacturing and structural mechanics research;
and toughened material studies related to keel panels. Initial ATCAS studies have shown that NASA's
Advanced Composite Technology program goals for cost and weight savings are attainable for
composite fuselage.

INTRODUCTION

Technology advancements are needed to insure that the United States retains a majority share of the
world market in transport aircraft. Composites have shown the potential to achieve improved
performance (e.g., increased fatigue and corrosion resistance) and reduced weight relative to aluminum
aircraft structures. However, higher costs associated with past composite structures remain an
economic barrier to increased applications. A balance between cost and weight efficiency appears
possible by integrating composite design, manufacturing, material, and structures technologies.

The application of affordable composite technology to transport fuselage is addressed in Boeing's

program entitled Advanced Technology Composite Aircraft Structure (ATCAS). This program started
on May 12, 1989. Assuming that all phases an6 options are funded, ATCAS is scheduled to be
completed by the middle of 1995. The main ATCAS objective is to develop an integrated technology
and demonstrate a confidence level that permits cost- and weight-effective use of advanced composite
materials in future primary aircraft structures with the emphasis on pressurized fuselages.

An aft fuselage section directly behind the wing to body intersection is used for technology
development and verification purposes in ATCAS. This section of fuselage, referred to as section 46,
has many design details and associated technology issues that pose a staunch test of advancements in

1 This work was funded by Contract NAS 1-18889, under the direction of J.G. Davis and W.T. Freeman of
NASA Langley Research Center.

127
composite primary structures. The design envelope (i.e., size, loads, and configuration constraints)
chosen by ATCAS is based on preliminary data for section 46 of the 767-X (a current development
program for an airplane 80% the size of a 747). The 767-X has a 240 in. diameter aluminum fuselage.
The parallel efforts of 767-X and ATCAS provides an opportunity to completely evaluate cost and
weight advantages of composites in the 1995 timeframe.

Keel, side, and crown areas of section 46 are being considered in the ATCAS program. Some
manufacturing cost issues for this fuselage section are shown in Figure 1. Although many issues are
common to the entire section, each area has unique problems that must be solved in order to achieve
low costs. The complexity of design details such as window cutouts and section splices result in
significant cost centers. Therefore, they are included in ATCAS studies to insure that overall low-cost
composite technologies are adequately demonstrated. In order to save costs as compared to metals, the
relatively high price of composite materials must be countered by innovative composite fabrication
and assembly processes that minimize labor costs.

ADDITIONAL ISSUES FOR =,_,,,_ _u;,_ _lo_i,,, Low Cost Stiffening Elements
ENTIRE FUSELAG /
" / _ Frame to Stiffener Attachment
•• ASSEMBLY_ __-//
• PART COUNT / J_ _ /
• PROCESS LIMITATION .¢" " __.._ ..Complex Curvature Parts
• MATERIAL FLEXIBILITY _ ro.P2u_o_rr _
• DEFECT CONTROU /f/_L_.uwNI _ /
REQUIREMENTS /./_,J/'_ _ A_l_l_ i High Part Counts,
//.,Jt7"[l [..__ ",n_j__ Shimming, Skin Taper,
// kv_P'_,E-Que_ , ___ Complicated Clips
Joints and Splices " /" i<t.<-_ "w%e___u_
(Bonded vs Boliedl / / _ _TA'7_T_T--_. .,
• ' / / _._'_'E)_--f;" -- Multiple Attachments vs.
// _'O__/;J_'7 Integral Stringer/Skin

Z "_-_0_ Keel Beam - High Concentrated Load

Complex Cutouts / Redistribution Difficult
(IntegralFramingvs. Fastening) / for Composites

Complicated Floor Fixtures

Figure 1: Cost Drivers for a Composite Fuselage

Many of the important composite structural issues, shown in Figure 2, are also unique to individual
areas of the fuselage. Design of the crown is driven by tension loading. The side area is dominated by
shear and pressure load redistribution around door and window cutouts. Keel design is driven by
major load redistribution from the keel beam and combined loads dominated by axial compression.

128
 HailImpact
\ BiaxialTension(Damage
Tolerance)
Pressure
Pillowing__ _ / _'Oz_..
\_ / ""-_'_'.,_,,. Circumferential/
/ "o "_..... _-._O j..1 Longitudinal Joints
__ Top_.,_..-.. _ -_/(Durability/Life)
Combined
Stress Loads/
Conoen- _:__/ .
- Comb,ned .
Bend,ng/

trations _..r_OU.oF_,_T__//
_,_ __- Hoop Tension

t ucture / / "__ (Damage Tolerance)

Post-Impact Ground /
Strength Handling Major Load Redistribution
Equipment from Keel Beam
Damage

Figure 2: Structural Design Drivers for a Composite Fuselage

The remainder of this paper is broken into three main sections that overview the ATCAS program.
The first describes the design build team approach. Baseline fuselage concepts and associated
technical issues are described in the second section. Plans and progress to develop concepts and solve
technical issues are presented in the final section.

DESIGN BUILD TEAM APPROACH

The greatest potential for saving cost and weight occurs in early phases of product development. Once
the structural configuration is set, attempts to reduce weight or enhance manufacturing efficiency are
traded against added costs associated with design changes. Past Boeing studies, indicating strong
interactions between design details and manufacturing costs, led to a decision to consider assembled
structure early in the ATCAS design cycle. The approach used for this effort is based on a design
build team (DBT). The initial goal of the DBT is to identify composite design and manufacturing
concepts that have a strong potential for cost and weight savings as compared to 1995 metals
technology. An accurate estimate of the potential for cost and weight savings with a composite
concept is established prior to the commitment for solving major technical issues.

The ATCAS DBT approach was derived by team members representing manufacturing, structural
design, structural mechanics, materials, quality control, and cost analysis. As a result, each team
member was given a sense of ownership in the ATCAS program. Early efforts revealed that the
combined inputs from different disciplines was critical to identifying concepts with a potential for both
cost and weight savings. Initial activities also indicated that the majority of work performed in support
of a DBT occurs outside the group meetings used for coordination and review purposes.

Early developments by the ATCAS DBT prompted a need for an efficient method of studying
candidate fuselage design concepts and manufacturing processes. Initially, 30 candidate fuselage

129
panel concepts were produced by design personnel. The number of concepts was increased from 30 to
159 during subsequent brainstorming sessions with the full DBT. Schedules would not allow cost and
weight evaluations of all concepts. Instead, concepts were classified into the eight design families (see
Figure 3) having common manufacturing characteristics. This allowed a more viable DBT approach in
which a reduced number of concepts, representing chosen families, are evaluated for cost and weight
efficiency. After identifying the best family for a given application, the cost and weight relationships
of variables within that family are analyzed in greater detail.

'FAMILY A FAMILY B
FAMLBonded Stiffeners
Bonded Stiffeners and Frames
Mechanically Fastened

FAMILY D FAMILY E FAMILY F

Sandwich Corrugated Geodesic

FAMILY G FAMILY H
Integrally Stiffened Closed 360°

Figure 3. Representative Concepts From ATCAS Design Families.

Three steps are used by the ATCAS DBT to select, evaluate, and optimize fuselage concepts. The
starting point is selection of baseline concepts as those design and manufacturing ideas having an
apparent potential for cost and weight savings, combined with an acceptable risk. Technology issues
are also identified during this phase of design, helping to focus early efforts in manufacturing,
structural mechanics, and materials development. In the second step, referred to as global
optimization, cost and weight savings are evaluated by performing detailed studies for the baseline and
a limited number of alternative concepts. Global optimization effectively integrates manufacturing
data into the design evaluation process. The final step, called local optimization, attacks cost centers
and major technology barriers established during the first two design steps.

130
Figure 4 shows a flow diagram summarizing the global/local design optimization process. The cost
estimating procedure for this effort uses detailed designs and fabrication/assembly plans. This
facilitates cost and weight trades which consider enough details to select a cost-effective concept for
further study. The ATCAS program is considering new material forms and manufacturing processes.
Lack of sufficient data for these emerging technologies can reduce the accuracy of structural
performance and cost predictions made during global optimization. Any uncertainties are noted and
can influence the decisions made in this design step; however, the risk in selecting new technologies is
minimized during local optimization which allows sufficient time to perform manufacturing trials,
generate data bases, and complete more thorough analysis. Local design optimization addresses the
critical cost centers, technology issues, and structural performance details. The goal of this DBT step
is to minimize cost and weight, while insuring structural integrity.

Global Design Optimization Process

DesignBUiid Team I Cost/Performance Results

I Process
Selection _1
Design ConcepV : I

767-X

Detaiied Ea_0Uis

[ Local Design Optimization Process

ManUfaCtUdng Pian I Cost

767-X
Data Gene_ati6n :
(Aluminum) I _. , ..... I
==.._I COstanclWeightI _ | Nesu,s,rom_,ooaJ I
Estimation I _"_"_ DesignC°ncel_sSi_ I

J I Man ufac tu _r`

,ng Trials ! :I
X I _ X I Enhance Pr°cess Efficiency .....
I

_ IPotentially Low-Cost Maieriais I

Cost \ I

\ " I
v
Local Element (e.g.;Stillened
Panels, Frames) Design Oplimizer
I
I
Detailed Cost / I With Cost & Mechanics Constraint-,_
,na sis
o, L.
I optimizedDesign Jr"

Figure 4. Global/Local Cost and Weight Optimization

BASELINE CONCEPTS

Baseline concepts have been selected for all areas of the aft fuselage section under study in ATCAS.
The DBT selection rationale included several considerations. Critical manufacturing and performance
issues were considered in selecting a design family that is compatible with structural details such as
doors and the keel beam splice. Baseline manufacturing processes were selected to maintain cost
efficiency with design details such as skin thickness tapers and curved frame geometries. Since the

131
scheduled completion date for crown global optimization coincided with baseline concept selection for
the entire fuselage section, results from the former were used as an indicator of composite cost centers.
The more detailed crown studies also helped to identify promising fabrication processes for skin,
stiffener, and frame geometries. Finally, the only concepts considered were those judged to have a
strong potential for large fuselage subcomponent manufacturing and test demonstrations by 1995.

Quadrant Panel Definitions

A quadrant manufacturing approach was selected for baseline panels. The four quadrant segments
defined for the baseline fuselage section are shown in Figure 5. The quadrant concept is intended to
reduce manufacturing costs in two ways. First, fuselage section assembly costs are lower than those of
the metals benchmark due to a reduced number of longitudinal panel splices. Second, the quadrant
concept is compatible with an advanced tow placement batch method for processing skin panels. As
shown in Reference 1, crown trade studies projected this method of skin layup to be cost effective.

Figure 5. Exploded Circumferential View of Fuselage Quadrants. Note

That Each Quadrant Panel is 30 ft. Long.

Differing arc lengths were chosen for each quadrant to accommodate the load conditions and specific
design details for each area of the fuselage. Another goal was to select quadrants that minimize
material waste of the batch process for advanced tow placing skins. The crown and keel quadrants
were chosen as 90 ° and 34 ° segments, respectively. As shown in Figure 5, both crown and keel
quadrants were symmetric about a vertical centerline to promote compatable skin gages on the right
and left panel edges. The selection of a relatively small keel arc length was due to a cargo door
location. Both side quadrants were large 118 ° segments, and include the cargo door, passenger doors,
and window belt.

Figure 6 illustrates the number of manufacturing segments produced in a batch process for each
fuselage quadrant. Crown and keel batch processes yield four and ten panels, respectively. Each
crown or keel batch consists of uncured skin panels that are cut at the intersection of right and left
edges. The side panel skin layup process required that both right and left sides be tow placed in the
same batch. Each side panel skin design will have top and bottom edge layups that are compatible
with the corresponding edge on the opposite side. For example, right top and left top panel edges will
consist of the same thickness tailoring and layups that differ only in the sign of angle plies.

132
 4 Crown Skin Panels 10 Keel Skin Panels
(Limited Amount of Thickness Tailoring) (Large Amount of Thickness Tailoring)

2 Left and 2 Right Side Panels

(Thickness Tailoring from Top to Bottom and at Doors & Windows)

_8o8_to_ m 1

_. Larger Radius (160 in.)

i ' " / _ to Yield 4 Panels

of
ArcSufficient
Length

Figure 6. Cicumferential View of Tow Placed Manufacturing Segments.

Figure 7 shows the automated tow placement work.station for baseline skin panel manufacturing plans.
Crown and side panel mandrals are illustrated, with the latter having a larger radius to accommodate four
panels per wind. Note that both Figures 6 and 7 idealize.tow placement mandrels as circular cylinders.
Actual mandrel shape will accommodate design details. For example, a joggle at the edges of individual
quadrant panels will be needed to allow for the geometry of longitudinal splices and cutting waste.

SIDE SKIN WINDING TOOL

TOW PLACEMENT ROBOTS

SIDE PANEL LAYUP

AND CURE TOOLS

SIDE SKIN

PANEL LAYUP

AND CURE TOOL

CROWN SKIN WINDING TOOL

Figure 7. Automated Tow Placement Work Station.

133
Design Concept and Element Process Descriptions

A full factory flow from element fabrication through final assembly was envisioned to support baseline
concept selection. Individual element processes are summarized in the following discussion.
Mechanically fastened joints were chosen for splicing quadrants and section joining during final
assembly. Additional features of the manufacturing processes appear in Reference 2.

Detailed cost and weight estimates from global optimization substantiated the baseline choice
for the crown quadrant. Results from this effort are summarized later in this paper. A design concept
from Family C (skin/stringer/frame with bonded stringers and frames) was chosen as the crown
baseline. Figure 8 illustrates a representative area of the crown baseline quadrant. Frames are
mouse-holed to avoid a complex bonding detail with the hat shaped stringers.

TEXTILE PREFORM/RTM FRAMES CO-BONDED TO SKIN

CTLM/DRAPE FORMED STIFFENERS CO-CURED TO SKIN

ADVANCED TOW PLACED LAMINATE SKIN

Figure 8: Crown Baseline Design Concept and Fabrication Processes.

Skin panels constitute the bulk of crown quadrant weight. Computer automated advanced tow
placement was selected to layup the skins. The previous subsection gave some additional details on
manufacturing multiple skin panels in a batch process.

Additional crown baseline elements include hat stiffeners and J frames. A contoured tape lamination
machine (CTLM), followed by a drape forming process was selected to layup and shape the hat
stiffeners. Textile preforms in a J-shape and a resin transfer molding (RTM) process were chosen to
form curved frames. The frames will have sufficient thickness to account for stress concentrations at
the mouse-holes. Finally, the autoclave fabrication of full crown quadrant segments (= 192 in. by 374
in.) was envisioned as wet skin and stiffeners, cobonded with frames.

Windgw Belt (Side): A variation of design concepts from Family C (skin/stringer/frame with bonded
stringers and frames) was chosen as the side panel baseline. The variation includes door and window
design details. Figure 9 illustrates the window belt area of the side panel baseline. A
skin/stringer/frame design family was chosen to facilitate design details at doors and windows.

134
 TEXTILE PREFORM/RTM FRAMES
CO-BONDED TO SKIN

CTLM/DRAPE FORMED STIFFENERS

CO-CURED TO SKIN

ADVANCED TOW PLACED

TAPERED LAMINATE SKIN

TEXTILE PREFORM/RTM WINDOW

FRAME CO-BONDED TO SKIN

Figure 9: Side Baseline Design Concept and Fabrication Processes Near

the Window Belt.

Skins for both side quadrants will also be produced in a batch tow placement. The skin thickness of
side quadrants is close to minimum gage approaching the crown and relatively thick near the keel.
Increased skin gages also occur locally near doors and windows. Automated batch processing and
significant amounts of ply tailoring possible with the advanced tow placement method led to its
selection for the side quadrant.

The same baseline processes and material forms were chosen for side panel stiffener and frame
elements as were selected for the crown. The window and door frame design details were chosen to be
textile preforms fabricated in an RTM process. Autoclave fabrication of full side quadrant segments
(each side = 251 in. by 374 in.) was planned to include wet skin and stiffeners, cobonded with frames
(circumferential and window belt).

Keel;. A variation of design concepts from Family D (sandwich) was chosen as the keel baseline. The
variation includes a thick laminated plate to panelize the keel beam chords. The thick plate gradually
transitions into a sandwich panel as axial compression loads decrease away from the section splice.
The sandwich facesheets have tapered thickness in the transition zone. Figure 10 shows the heavily
loaded end of the keel panel baseline near the section splice.

This innovative panelized concept was chosen for the keel quadrant baseline for several reasons. First,
it avoids problems in fabricating and splicing two large composite keel chord members. Large fastener
hole diameters needed to mechanically splice discrete composite chord members would cause a
significant knockdown on the allowable strength. The panelized concept alleviates this problem to
some extent. The blended thick laminate/sandwich construction also yields a constant gage panel that
avoids problems in attaching frames and other elements to a skin/stringer design having skins with
considerable thickness taper.

135
 AFT END OF FUSELAGE SECTION

TEXTILEPREFORM/RTM FRAMESAND
FLOOR SUPPORTSTRUCTURE

SANDWICH

THICK LAMINATE
ADVANCED TOW PLACED TAPERED
LAMINATE/SANDWICH PANEL

TEXTILE PREFORM/RTM INTERCOSTALS FOR

STABILITY AND CARGO FLOOR SUPPORT

Figure 10: Keel Baseline Design Concept and Fabrication Processes.

Face skins for the keel panel will also be produced in a batch tow placement process. Again,
automated batch processing and significant amounts of ply tailoring possible with the advanced tow
placement method led to its selection for the keel quadrant. The sandwich core will be machined with
a constant taper to keep the keel panel at a constant gage as laminate plies are dropped.

Keel baseline elements include intercostals and full-depth frames, both of which will be stiffened.
These elements serve dual roles in overall panel stability and cargo floor support. Note that
intercostals are discontinuous at the frames. Baseline material forms and processes for both these
elements are textile preforms and RTM, respectively. Frames and intercostals will be mechanically
attached to shear tied blade elements on the fully cured keel panel. These blade elements will be
co-bonded to the keel panel at cure. Keel quadrant segments (= 72 in. by 374 in.) will each be small in
comparison to side and crown panels; however, the weight per unit area will be relatively high.

Technical Issues

The most critical manufacturing issues associated with baseline concepts selected for ATCAS relates
to final assembly. Crown and side panels have bonded stringers and frames, while the keel panel is
sandwich with bonded frames. This eliminates element assembly steps and part count, but high
bending stiffness of the baseline panels will limit attempts to deform them into shape prior to splicing.
Figure 11 illustrates some of the problems on a diagram of the full baseline crown panel. Variations in
locational tolerances of stringers and frames is critical to their alignment at circumferential and
longitudinal splices, respectively. Overall panel warpage may also be an issue due to the local
unsymmetric layups at bonded elements.

Another critical manufacturing issue is the quality control of quadrant panel fabrication processes.
Quadrant panel cost benefits assume that large panels will not be rejected due to manufacturing
defects. In order to avoid adding unnecessary costs to the manufacturing process, the panel fabrication
quality required for acceptable performance must be understood.

136
 ALIGNMENT OF FRAMES
AT LONGITUDINAL SPLICES

/ OVERALL PANEL WARPAGE AFFECTS

ALIGNMENT OF STIFFENERS BOTH SPLICES
AT CIRCUMFERENTIAL SPLICES

Figure 11: Assembly issues for the crown quadrant panel.

Crown: The fuselage crown area is the simplest quadrant in terms of design detail and manufacturing
complexity. A smaller number of technical issues are also associated with the crown. Two critical
manufacturing issues were discussed at the start of this subsection. Tension damage tolerance is the
most critical structural issue for crown panels. Analysis and tests are needed to evaluate the effects of
dynamic pressure release on tensile residual strengths following a through-penetration impact event
(e.g., engine blade penetration). Other major issues for the crown baseline which wilt be addressed in
ATCAS are listed in Reference 1.

Window Belt (Side): The fuselage side quadrants have considerably more design detail and
manufacturing complexity than the crown. Many of the crown technical issues are also critical to side
panel locations above the window belt. The two critical manufacturing issues that were discussed at
the start of this subsection are of greatest concern for the side quadrants due to their size. The lower
portion of side quadrants (i.e., approaching the keel) have considerable combined load interactions.
The combined effects of axial compression, inplane shear, and hoop tension on damage tolerance
need to be understood. Other major issues for the side baseline concepts which will be addressed in
ATCAS are described in Table 1.

Keel: The aft fuselage keel quadrant poses one of the most difficult design challenges for applications
of composites to transport fuselage (i.e., load redistribution of very high compression loads entering
the fuselage shell at the keel beam attachment). It also has considerable design detail and
manufacturing complexity related to the cargo bay floor. The critical manufacturing issues that were
discussed at the start of this subsection are of particular concern for the baseline keel quadrant due to
its high stiffness. Other technical issues listed for the crown and side quadrants are also critical to the
keel. These include side issues number 5, 6, and 9 from Table 1 for circumferential frames. Note that
loads affecting durability of keel frame to panel adhesive bonds differ from those in the crown. The 2
ft. deep baseline keel frames and intercostals also add manufacturing complexity to the textile
preform and RTM fabrication processes. Other technical issues for the keel are given in Table 2.

137
 _o
o
0_'0_ ,
_._
"___
"b • _ o O't_
._
_N
_ _& _m
¢_ r_ o. __, _
o
_-_ _ e'_
,N
o _'_'_
e:m _
_'_
_d
e:m !
.I
m
li.0_
_= [..
.1"_ _..__.__
_0 = m=l _E
_.'_
.=__
._'_
_=
t'q
138
 m
t_.
m
.m
[-.
e4
_ --= [,.
_H'J'u'J
.__::_
.E ,_ -= =.='-
.--
-,
III ¢
._.__
t_ t'.-
139
 ATCAS SIX YEAR PLAN & PROGRESS TO DATE

Crown, keel, and side areas of the aft fuselage section will be studied in Phases A and B of the ATCAS
program. These two phases span a five year time period ending in 1994. An optional addition to
Phase B for fabricating a large pressurized panel test fixture and performing combined load tests is
scheduled to occur from 1993 to 1995.

Figure 12 shows a timeline of the six main areas of development and verification tasks within the
ATCAS program. Comprehensive schedules were created in each of these areas to integrate design,
manufacturing, structures, materials, and test tasks. Crown, keel, and window belt areas will initially
be studied separately because each has unique manufacturing and structural issues to be resolved.
Efforts are spread such that the most difficult problems receive greater attention. For example, Figure
12 shows that 2.5 years are dedicated to crown panels, while keel studies will last a full 5 years.

I,JUVII_ IU_,,_I I II_l It ill lU

Verification Tasks 1989 1990 1991 1992 1993 1994 1995

Crown Panels
I
t t I
L i I

l l l
Keel Panels

Window Belt Panels

Frames & Attachment

Details (e.g., Splices)

Full Barrel Section

,ii, ..................
',

I
',

i
I.................
!i, 1

===============================================================================================
i _ ::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
I J .... -.+.., ..., .,.1,1,1,1,,, 1
i = i End
' : I ' Phase
B
Start i
Large Pressurized Phase B I I
Panel Tests
En_ [_!__!_!___!_!_______!___!_!_____ii_i_____i!__
(Option 1 to Phase B) Phas, A I I

Completed Work Developmental Tasks Verification Tasks

Figure 12. Development & Verification Timeline for ATCAS

Development and verification tasks for frames and attachment details support each quadrant. The final
design optimization efforts with these elements and splices are left until the "Full Barrel Section" effort
at the end of Phase B. This will promote assembly cost savings by striving for part commonality
throughout all regions of the study section.

The three step DBT approach described earlier for achieving ATCAS cost and weight saving goals will
be applied to each quadrant. Note that design activities for side quadrants are associated with the
window belt timeline. The first step, baseline concept selection, has been completed for each
quadrant. Global evaluation has been accomplished for crown panels and is currently underway for
the keel quadrant. This second design step is scheduled to be completed for all quadrants by the end of
Phase A, substantiating concepts chosen for verification tests with detailed cost and weight saving
estimates. Local optimization (Step #3) has just begun for the crown and will be completed by early
1991. The time needed for local optimization varies for each quadrant depending on the engineering
efforts required to overcome the associated technology barriers. For example, local optimization for
the keel lasts more than 2.5 years.

140
There is considerable overlap in technology issues that need to be addressed for each quadrant.
Fabrication, ana!lysis, and testing effo_rts were scheduled to group common tasks. Figure 13, which is a
modification of the ATCAS quadrant diagram (i.e., Figure 5), shows that many of the side quadrant
technical issues will be addressed by crown and keel work tasks. The side panel manufacturing and
test efforts will include windows, but not doors. Keel panel manufacturing and test studies will
include compressive load redistribution approaching the forward end of the section but not the actual
keel beam splice. Although doors and the keel beam splice are not included in ATCAS test and
manufacturing hardware, both are considered in the design cost and weight estimates.
Crown
.. I_ Crown ,J Top ol _ CtoMI

Side

Keel

Figure 13. Schedules for Tracking Fabrication, Analysis, and Test Tasks

Industrial subcontractors and other Boeing groups play important roles in ATCAS. Early experiences
suggest that the integration of diverse expertise from groups outside the immediate ATCAS team
(Boeing Commercial Airplanes) is crucial to the development and verification of advanced
technologies. Figure 14 shows groups that have supported ATCAS to date.

Boeing Ae¢ospaoe & EleeUonlos

Engineering Consultant
(W, Avery, R. Snyder, D. Wright)
(j. McCany)
Nondestructive Inspection
Impact Analysis & Test Integrated Teohnoioglee trio.
Fuselage Struclur al
(B. Coxon)
Textile Preform/RTM Processing
Design & Analysis
Fracture Coupon Testing
Stiffened Panel Impact Experiments

[S. Johnston, Ft. Lundquist)

Computational Structural
Me(::_anics Problem

Herouiel Inoorporated
Grant, G. Walker, R. Nuisrner,
D. Cairns. Y. Tokita)

Boeing I-leleoopeers Group

Panel Fabrication Using
Advanced Tow Placement

iCI Flberl_e Braided Composite Technology

(S. Clark, R. Hoeltho) for Fuselage Applications

Tow and Tape Materials Bolted and Bonded Joints

i (C. Gunther, P. Grant, M Fedro)
Textile Preform Manufac=udng

Figure 14. Other Boeing and Industrial Groups Supporting ATCAS

141
As shown in Figure 15, several university subcontracts have also been issued to support ATCAS. The
approach taken to infuse universities in ATCAS has been beneficial to both parties. The
Boeing/NASA ATCAS goal of advanced technology development complements the universities'
engineering goal for improved education. Each university subcontract was defined to start with
fundamental research and end with applications directly supporting work on a technical issue in
ATCAS. The engineering tools developed at universities include analysis methods, test techniques,
material characterization, and manufacturing developments. In return, the students education is
supplemented with an understanding of fuselage technical issues and the opportunity to apply their
research to a "real world" engineering problem. Secondary benefits from university infusion have
included a cost-effective use of ATCAS funds, and budget for graduate student programs.

of Washintq_ i
M. Tut_e & Z. Zab_nsky l K. Un
Long-Term Solvent Resistance MullJ-Parameter Design I Impact Damage Tolerance •
of Materials
J. Seferis
for Keel Panel Cost Optimization Models for I Analysis Models for •
and Bonded Frame Applications
Crown Panel Appflcations

on State University
University of Iowa
T. Kennedy R. Lakes
Damage Resistant
Core Materials for
Keel Panel Applications

Drexel University
F. Ko
W. Bascom Braided Composite Technology for
Frame Applications
Toughened Matrix
Universityof
Failure Utah
Mechanisms for i J. Awerbuch & A. Wang
Keel Panel Applications
Durability/Life Methods for Keel Panel
and Bonded Frame Applications

Figure 15. Integration of University Subcontracts in ATCAS

Crown Panels

Global Optimization: Crown panel efforts started in the fall of 1989. Most of the work performed to
date supported the global evaluation of design and manufacturing concepts. Two designs from each of
three families were developed for global optimization of the crown quadrant. Each design was sized
considering multiple load cases, damage tolerance, and assembly design details. The three families
were B, C, and D (see Figure 3 for family definitions). A detailed fabrication and assembly plan was
developed for each design. These were used to estimate weight, material costs, and labor rates. Both
recurring and nonrecurring (minus capital equipment) costs were estimated assuming specified
groundrules (e.g., 300 shipsets at a rate of 5 per month). A synopsis of the results from global crown
studies appears on the following paragraphs. A more complete account of this effort appears in
Reference 1.

142
The two designs and manufacturing plans for each family varied material types, manufacturing
processes, and structural details. This helped to establish a range of cost and weight variation for each
family. Design trades within a family also yielded data on cost centers and variable interactions
crucial to local optimization studies.

The majority of weight for all designs was in the skin where tension damage tolerance was found to
drive skin thickness and layup of most of the crown panel area. Hail impact requirements also
controlled the skin gage for the aft end of some panels. Stringer thicknesses were driven by reversed
load stability requirements. Both skin and stringer gages near panel edges were controlled by joint
bearing requirements.

Figure 16 shows the final results from global optimization. All composite crown designs studied were
found to be cost and weight competitive relative to the metallic benchmark. Relative weights for
composite designs ranged from 49% to 80% of the metal. The estimated relative manufacturing costs
for composite concepts ranged from 99% to 139% of the metal. When considering an economically
acceptable cost increase per unit weight savings (i.e., concepts to the left of the slanted solid line in
Figure 16), all composite designs showed an advantage over the metal benchmark.

Aluminum costs were dominated by labor, while material and labor costs for composite designs were
close to equal. Composite designs were competitive with the metal benchmark primarily because
reduced assembly labor offset lower aluminum material costs. For example, a composite crown panel
had 10,000 less fasteners than the metal counterpart. This relates to large composite quadrant panels
with bonded elements for Family C.

140 ..................... '........... _ ............ o_ ..................... _...................... _..........

c2 i _.
120 :: i'_ ............... !....
....................
i....................... ! ...................
!.....
X
100 ............................................................ ip--_.-- _ ....................................... _ ...... ._
¢.
¢.D Family [] B _$_'_ i _ J
80
Fam_y C ...... , ...................... _............... -_ ..... : ...................... ; ..........
o_
60
Fa_y B
carl
O Glob. i_Pt. Family B
¢._)
40 Glob. Opt. Family C

Glob. OApt.Family D
20
Aluminum (767-X)

0 I i
0 20 40 60 80 100

Weight, % 767-X

Figure 16. A Comparison of Results from Global Optimization Studies

and the Metallic Benchmark.

Figure 16 also shows globally optimized family designs derived by mixing and matching the best
features of concepts studied (e.g., cost-efficient materials and processes). The most promising family
was selected based on results from this exercise and an evaluation of the potential for further cost and
weight optimization. Globally optimized Families B and C were both found to be nearly equivalent in

143
cost and weight. Family D yielded a lower cost with some weight penalty (i.e., 94% the cost and 64%
the weight of metal). The weight increase for Family D was not found to justify the cost savings. This
is shown in Figure 16 where concepts to the left of the dashed line represent an acceptable cost
increase per unit weight savings. Family C was chosen over Family B, despite greater manufacturing
risk, due to good local optimization potential and bonded frames which may help damage containment.

Local Optimization: Local optimization of crown panels is currently underway. This task will further
refine the design by attacking cost centers identified in global optimization. The most significant cost
centers for crown panels were found to be skin, stringer, and frame fabrication; panel bonding (i.e.,
element subassembly, bagging and cure); and fasteners required for quadrant assembly and body join.
The relative percentage of material and labor components of cost varied for individual fabrication and
assembly steps. A rough estimate of the local optimization potential for composite crown panels
indicated cost savings up to 25% (relative to the metallic benchmark) within the range of acceptable
weight penalties.

The use of a high performance fiber to reduce skin weight was not found to be cost effective in global
optimization of crown panels. This relates to the trade between a higher material purchase price and
the costs saved from added performance capability. Attempts to reduce the composite material cost
center for crown panels will include studies with graphite/fiberglass hybrids. The high tensile strength
and low cost of fiberglass makes it a candidate for crown applications. The economically acceptable
increase in weight per unit cost savings will be considered in this evaluation.

The relatively low aluminum material costs generally allow an effective trade of added weight for
reduced labor. High composite material costs are expected to lead to relationships between cost and
weight that differ from those of metals. A software design tool that includes cost and mechanics
constraints is being developed by the University of Washington to support the optimization of crown
panels (see Reference 7). Figure 17 shows this optimization tool schematically. Critical design
variables are expected to include material type, stringer geometry, and laminate thickness.

Global Optimization: Cost and Weight Results Develop constraints

150 • Cost
O
• Damage tolerance
• Manufacturing
100 A
• Load conditions
Cost. %
of 767-X
DFan_ B
50
O Fa,._iyc If constraint function (eg., cost,
A Fmn,_,D
"YX"AJummum('t_'.X)
violated, add weight, etc.)
i 1 t I I I l I I material Minimize objective
2O 40 60 80 100
Weig_L % of 767.X

No violated Check constraints

constraints
• Damage tolerance • Cost
Best possible • Buckling • Materials
crown panel design • Minimum gauge

Figure 17. Local Crown Panel Optimization Tool

144
Several other factors will be considered to reduce cost in local crown optimization. These include the
manufacturing approach to panel subassembly and the use of low-cost composite fasteners. Efficient
processes for stringer and frame fabrication are also sought. Considerable costs were saved during
global studies by envisioning batch processes for these elements; however, their costs were still high
relative to the metallic counterparts. One potential cost savings may be in the use of tow placement
rather than CTLM to layup stringer charges. This relates to projections that the future costs of prepreg
tow will be less than that of tape. A continuous RTM process involving textile preforms may also be
suitable for low-cost stiffening elements.

Analysis, Fabrication. and Test: Tension damage tolerance is the most critical performance issue for
crown panels. Several damage tolerance analysis methods have been developed and implemented as
mechanics constraints in the local optimization software shown schematically in Figure 17. These
procedures, which utilize a characteristic dimension failure criteria, include the effects of pressure,
panel curvature, bonded stiffening elements, and material anisotropy. Figure 18 plots typical results
from residual strength analysis showing an interaction between pressure, panel curvature, and local
bending stiffness. Pressure is shown to decrease residual strength due to local bending at the notch tip
that increases the stress concentration. A sandwich panel construction alleviates some of the effects of
pressure by increasing local bending stiffness. For this reason, a sandwich panel design was selected
as a backup to the crown baseline.

80
AS4/3501-6
_ ,¢/ Laminate: 4-0's/8-45's/2-90's
70
Honeycomb skin: 2-0's/4-45's/1-90
Honeycomb core thickness: 1/2"
60 Shell Radius: 122"

(/) 50

t-
40
O1
e" Flat Panel /
L_ 30 k w J _ Curved Honeycomb Panel
t/)
20

10

Curved Panel
0 i i i i i

0 1 2 3 4 5 6

Crack Half-Length (in)

Figure 18. Predicted Tensile Strengths of Damaged Panels Subjected to

Inplane and Pressure Loads

In addition to the simplified analysis described above for predicting tensile residual strength, a more
complete progressive damage model is being developed at Stanford University. The Stanford models
will be used during final test verification to evaluate the growth and accumulation of local damage that
occurs prior to catastrophic failure. To date, progressive damage models have been developed to
account for fiber breaks, matrix cracking, nonlinear shear behavior, and fiber/matrix splitting. Future
enhancements will incorporate the effects of nonlinear longitudinal moduli, delamination, variable
notch geometry, combined load conditions, and panel element geometries.

145
Advanced tow placement and the batch quadrant processing concept were shown to be efficient in
detailed cost studies for crown skins. Tow placement layup costs were found to be on the order of
20% of the skin material costs. Since skin constituted the largest portion of panel weight, an efficient
layup process helped to lower total composite crown costs. Advanced tow placement is ideally suited
for the graphite/fiberglass hybrid concept being considered for reducing crown material costs.
Tow-placed intraply hybrid panels can be fabricated without affecting process efficiency. Figure 19
shows one of the four graphite/fiberglass hybrid panels fabricated in ATCAS to date. Additional
discussion of advanced tow placement manufacturing trials completed in ATCAS appear in the next
section on keel panels.

Figure 19. Tow Placed AS4/S-Glass Hybrid

Specimens cut from the hybrid panels will be used to evaluate notched tensile residual strength. The
base material type in these studies is AS4/3501-6. Hybrid material and process variables under
consideration include hybridizing fiber type (S-glass and TlOOO), percent hybridization (25% and
50%), and repeating unit tow width (0.35 and 1.1 in.). The screening test variables include notch type
(hole, slit, and impact induced through-penetration) and notch size (0.25,0.875, and 2.5 in.).
Plans and schedules in ATCAS are coordinated to yield test panels from process verification studies.
Fabrication of crown panel demonstration hardware is scheduled for the first half of 1991. This effort
will start with some tooling trials and end with the fabrication of several panels, the largest of which is
8 ft by 10 ft. Current plans will emphasize the baseline design (Family C, with bonded stringers and
frames). If problems occur due to the complex bonding operation or insufficient tensile damage
tolerance an alternate plan, involving sandwich panels with bonded frames, will be implemented.
Verification tests using stiffened panels (three to five stringer) that represent portions of the optimized
crown design will occur during 1991. These will include (1) uniaxial panel stability tests for reversed
load conditions (with and without critical impact damage), (2) impact trials to validate that minimum
gage/hail requirements are satisfied, and (3) tensile damage tolerance tests (pressure, uniaxial, and
biaxial) for various through-penetration damage scenarios.
Keel Panels

Aside from baseline concept selection, little keel design work has been completed to date. Initial keel
work tasks have included analysis, fabrication, and tests to address some of the critical issues listed
earlier. Specifically, issues number 1, 6, and 7 from Table 2 have been studied during the first 16
months of ATCAS. Results from these studies will support final material selection and detailed
design. This subsection will summarize work on each of the three issues, followed by a description of
the large keel hardware tests planned towards the end of ATCAS.

Designed Experiment for Fuselage Panel Impact Resistance: All quadrants of the fuselage have
technical issues related to impact damage resistance. A test program was defined early in ATCAS to
evaluate the effects of thirty two combinations of material, laminate, structural, and extrinsic variables
on impact damage resistance. The experiment was designed to ascertain main effects with a minimum
number of tests. Some of the details of this experimental design are described in Reference 8. The
desire to integrate ATCAS tasks (e.g., manufacturing trials and test hardware) led to a more general
plan for panel fabrication and usage than would have been required for impact screening alone. Figure
20 shows some of the manufacturing and performance issues addressed by this initial fuselage panel
fabrication effort. Large panels were fabricated to study manufacturing scaleup issues such as
low-cost tooling concepts in the presence of ply dropoffs.

Manufacturing/Process Technology

Process Low Cost Advanced Elements &

Panels for
Feasibility Cost Data Tooling Fiber Assembly
Concepts Placement Trials

Damage Impact Stiffness

Damage and
Tolerance Resistance Stability

Performance Issues

Figure 20. Multifunctionality of the Initial Fuselage Panels Fabricated

for ATCAS

Sixteen 110 in long stiffened panels were fabricated to support the tasks shown in Figure 20. Each
panel had thick (=0.18 in.) and thin (=0.089 in.) ends, with a central transition zone where 12 plies were
dropped over a 10 in. length. Note that different portions of these panels were used to study baseline
crown, keel, and side quadrant issues. The thick end provided data for the lower portion of side
quadrants (i.e., below the windows). The thin end was used for the crown and the upper portion of
side quadrants (i.e., above the windows). Finally, laminates in the transition region of the panel were
used to fabricate sandwich panels that supported the keel baseline. All activities related to the
fabrication of the sixteen stiffened panels were tracked on the keel schedule of ATCAS monthly
reports, despite the relationships with other quadrants.

147
Eight of the sixteen panels were fabricated at Hercules Inc. using advanced tow placement. Figure 21
shows a photograph of one of the tow placed panels. All other panels were processed using tape
material forms at the Boeing Company. Drape forming and low-cost rubber tools were used to
facilitate fabrication of the hat stiffening element geometry at both companies.

Figure 21. Tow Placed Hat Panel

As discussed earlier, the most critical assembly issue related to cured panel warpage and locational
tolerances for bonded elements. The approach used in ATCAS to solve this problem involves three
steps. First, manufacturing demonstration hardware will be fabricated with design details. Second,
cured panel imperfections will be experimentally measured. Figure 22 shows warpage measurements
for one of the hat stiffened fuselage panels. Finally, a suitable analysis method will be developed to
predict the warpage and perform sensitivity studies to select design variables that minimize potential
assembly problems. Early developments supporting this final step included an ABACUS finite
element model that predicted the shape and overall deflection shown in Figure 22 to within 5%.

Figure 22. Panel Warpage Measurements Using a Moire' Technique

148
Toughened Material Models: Composite materials with compliant, resin-rich, interlaminar layers
(RIL) tend to have improved interlaminar shear toughness and impact damage resistance. Both appear
to be advantages for keel applications. Considerable efforts have been spent in ATCAS to develop
analysis tools and generate data bases for toughened materials with RIL (see References 8-12 for more
complete documentation).

Toughened materials studies have included (1) material property tests, (2) characterization of failure
mechanisms, and (3) analysis method development. Some relationships with the RIL microstructure
were found in each area. Mode II delamination toughness was directly related to RIL thickness;
however, no such correlation was found to exist for mode I toughness (Ref. 11). The tendency for RIL
to suppress delamination during an impact event has been found to lead to fiber failure as an alternate
failure mechanism (Refs. 10 and 11). The presence of RIL was found to alter matrix crack resistance,
reducing the in situ transverse strengthening effect observed with traditional composite laminates (Ref.
12). Not all analysis methods developed for materials with RIL required an accurate model of the
microstructure. It was crucial to simulate the RIL for matrix crack predictions. However, accurate
predictions of compressive strength after impact (CAI) were obtained without modeling the RIL.

Impact damage is expected to be a critical issue, particularly in the lower side and keel quadrants.
Models have been developed and verified in ATCAS to predict CAI for different composite materials
and laminate stacking sequences (Ref. 8-11). Laminate layups and specimen geometries used for
material screening received special attention in ATCAS. Models were used to interpret results from
experiments. For example, the effects of finite specimen width on CAI were eliminated with the help
of analysis. Parametric analysis studies were used to identify key variables affecting CAI, resulting in
recommendations for improved material screening procedures (Ref. 9). Figure 23 shows why it is
crucial to screen materials over a range of impact energies and correct for specimen width effects.

110
......... Material A
\ (HighUndamagedCompressive
Strength/LowToughness)
l

9O tl
Material B
Il
(High Toughness/Low Undamaged Compressive Strength)

7O ", Finite Width Corrected Results for Low Energy Impact

t/)
x.,

<
O
5O

3O

Finite Width Corrected Results for High Energy Impact

10 I I I I I I

0 1.0 2.0 3.0

Damage Diameter, in

Figure 23. Theoretical Damage Tolerance Curves for Two Materials

149
Studies are underway in ATCAS to determine optimum fiber/resin combinations and resin contents
(both intra and inter-laminar, see Ref. 11) from the standpoint of both cost and performance.
Conceivably, lower cost fibers and higher resin contents may acheive the compression performance
advantages of a RIL microstructure with reduced costs. The compressive failure strain was found to
increase for toughened laminates having higher resin contents (Ref. 11). This attribute may be even
greater when considering laminates with significant amount of ply dropoffs, characteristic of the keel
baseline. Figure 24 shows that the ATCAS analysis methods developed for toughened materials are
general enough to account for variations in resin content.

6O
T800H/3900-2
4"Wide Specimen
5O

145'0"45'901 4S

,¢ 40
_c"
C_
¢.-
_ 30

_ 20
LL

10

0
0 ' 0.4' 'o'8. ' 1.2' ' 1.6' ' 2" ' 2.4' '2'8.
Damage Diameter, in

Figure 24. CAI Analysis and Experiments for a Toughened Material

With Increased Resin Content (Vf= 0.49)

As shown in Figure 15, the development of toughened material models involves considerable
interaction with universities. Two groups at the University of Washington are involved in the separate
tasks of developing (1) methods to screen solvent resistance, and (2) general impact damage analysis
models. The University of Utah is using microscopy to study relationships between toughened
material microstructure and impact damage accumulation. Drexel University is involved with analysis
and test characterization of the long-term durability of toughened materials, including environment and
real time effects. Results from these studies will be presented in future ATCAS documentation.

Advanced C0r¢ Dcv¢10pment: As discussed earlier (see issue number 7 in Table 2), damage resistant
core materials are needed to facilitate the keel baseline concept. Considerable progress has been
achieved in this area with the University of Iowa subcontract. Figure 25 shows a new class of
sandwich core concepts which have potential advantages in fuselage applications. Advantages of the
re-entrant cell structure which results in a negative Poisson ratio include high fracture toughness and
indentation resistance for foam materials. One manufacturing advantage of the negative Poisson ratio
is the ability to bend re-entrant foams and honeycombs into cylindrical shapes characteristic of

150
fuselage panels (note that traditional honeycombs tend to deform into a saddle shape under bending
loads). Future work with this new material form will include scaleup of a suitable low-cost process;
development of ultrasonic inspection procedures; and analysis/test studies on impact damage
resistance, environmental sensitivity, and durability.

Foam With Re-entrant Cell Structure

Honeycomb With Re-entrant Cell Structure

Figure 25. Advanced Sandwich Cores: Foam and Honeycomb

Large Panel Fabrication and Test Plans: A substantial portion of ATCAS funds addresses keel issues
with manufacturing demonstration hardware and subcomponent panel tests. Several tests involving
small panels (e.g., CAI with three stringer panels) are also planned. As with the crown, process
demonstration studies yield panels for large verification tests. Combined compression/shear tests will
be performed with curved panels (=3.5 ft by 5 ft) representative of the lower side quadrant designs. A
large curved panel (=5 ft by 8 ft), representing the aft keel design, will be tested for damage tolerance
in different combinations of compression and shear. Repairs will also be performed and verified with
this panel. Finally, another large curved panel (--5 ft by 8.5 ft) will be tested to validate the
compressive load redistribution capability of the forward keel design. This panel will also be
damaged, tested, repaired, and then tested to failure.

Window Belt Panels

Very little work has been performed in support of the window belt. A baseline concept was selected
and the associated technical issues identified. Global optimization of the side quadrants will occur
between March and October of 1991. Local optimization of the window belt lasts one year, ending in
the fourth quarter of 1992.

As shown in Table 1, several issues (3,4,5,6, and 7) relate to RTM processing of the window frame
detail. Manufacturing and supporting technology tasks have been initiated to address these issues. In
particular, a request for proposals to fabricate RTM/textile window frame modules were sent to
candidate manufacturers. Initial tasks will include coupon tests to determine basic material properties
and the effects of defects for the window frame material form. Flat and curved window frame modules
will be fabricated in the 1992 to 1993 timeframe to bond to window belt manufacturing hardware and
test subcomponents. Analysis methods will be developed in an Oregon State University subcontract to

151
predict the overall performance of window belt panels. The effects of local design details (e.g., frame
rib geometry) will be considered in this effort.

Panel and subcomponent tests will be used to verify the window belt design concepts. Included in the
larger subcomponents is a curved panel (=3.5 ft by 3.5 ft) with a central window cutout. This panel
will be tested in hoop tension to evaluate the effect of cutout geometry and cutout-doubler
configuration. Picture-frame shear tests (=3.5 ft by 3.5 ft) will be conducted to verify the capability of
the most promising window-belt concepts and the accuracy of predictions (load-redistribution and
stability). The optimized window-belt configuration, including longitudinal stringers and simulated
frames, will be tested under compression/shear loading to provide final verification of the design
concept and analytical predictions. This panel (=5 ft by 8 ft) will also address damage tolerance
issues, including impact damage and loss of elements.

Frames and Attachment Details

Design development of frames and splices is integrally linked to other ATCAS efforts. Global
optimization of these details will be accomplished with each panel quadrant. Local optimization will
be performed in the "Full Barrel Section" effort, near the end of Phase B.

To date, the detail development has focussed on the crown quadrant (Ref. 1). These studies indicated
that (1)curved frames have much higher weight-normalized costs than simpler geometries (e.g., skins);
(2) braid/RTM processes are the most attractive for frame geometries; (3) recurring fastener costs can
be significant, and (4) eccentricities at panel splices are a major fabrication issue. Significant efforts
have been accomplished in addressing braided-composites variables and fastener costs.

Plans are in place to address other frame and splice issues. Mechanical joints data will be generated to
support the splice design of new material forms selected for use in ATCAS (e.g., graphite/fiberglass
hybrids and textile materials). Adhesives suitable for the frame-to-skin bond will also be evaluated.
This assessment will include residual strength and durability tests. The latter will include sustained
loads characteristic of cabin pressure. Robust splice plates that allow some locational tolerance
mismatch between stringers or frames at panel joints will be designed and tested.

Braided Composites: Test plans have been developed to evaluate and compare properties of braided
composites formed using two processes: (a) RTMing a thermoset matrix, and (b) consolidation of
preforms with commingled graphite fibers and thermoplastic strands. Both two- and three-
dimensional braids are being considered in this study. Data will be obtained for unnotched tension;
open-hole tension, compression, and inplane shear; CAI; transverse tension and shear; bearing; and
crippling. Analysis methods are being developed in parallel, to predict stiffness and strength of
braided parts with known fiber architectures. Reference 13 gives additional details on progress to date.

Advanced Fasteners: By working with vendors, major fastener cost centers and potentially low-cost
fasteners and/or installation methods have been identified. To date, spin-forming of
graphite/thermoplastic rivets appears most attractive. Trial installations of the selected concepts and
mechanical tests will be conducted to evaluate cost and strength performance.

Full Barrel Section

The primary objective of work associated with the "Full Barrel Section" timeline shown in Figure 12 is
to complete work tasks which provide the degree of technology verification required to commit to
development of a full scale barrel section (i.e., a potential Phase C option to ATCAS). The
manufacturing steps for each quadrant of the aft fuselage will be optimized to minimize costs of a full
barrel section. One of the main design activities will be to evaluate potential cost advantages of a 360 °

152
concept from Design Family H. Local design optimization and verification tests will also be
performed with frames, attachment details, and splices in attempts to minimize the number of parts.
Any of the long-term research efforts with new material forms will also be completed by the end of
ATCAS, allowing them to be included during full barrel design studies.

Large Pressurized Panel Tests

The optional addition to Phase B of ATCAS is scheduled to address dynamic aspects of pressure
release following a through-penetration damage event. Additional studies of combined loads
characteristic of fuselage bending will be coupled with the pressure release problem. Both analysis
and test will be used to obtain solutions to these problems, resulting in additional verification on the
damage tolerance of composite fuselage design concepts.

Three main analysis steps appear necessary. The first step involves a fluid flow analysis of pressure
released through an opening in a curved panel. The second step is a mechanics problem to determine
the stress state acting on the damaged composite structure due to a combination of dynamic pressure
release and existing load state. Finally, verified failure criteria need to be developed which account for
the combined effect of in-plane and out-of-plane load conditions. Some of the analysis development
and coupon test verification for this effort will be performed under university subcontract by the
Massachusetts Institute of Technology.

The full scale test fixture, illustrated in Figure 26, will be designed and fabricated to permit testing of
large pressurized fuselage panels (=6.5 ft by 10 ft). Flight conditions will be simulated by attaching a
loading frame to a bulkhead. The loads will be reacted through another bulkhead which is attached to
a strongback. The interface between the composite test panels and aluminum test hardware will be
designed such that panel failure will not damage the test fixture.
122-In radius metal
_lel res_ion
_Ik_sd

S_eel

line

te_ fixture

Figure 26. Pressurized Fuselage Damage Containment Tests

Four large panel tests are planned: (1) crown panel, (2) keel panel, (3) crown-to-side longitudinal
splice, and (4) keel-to-side longitudinal splice. Each panel will be tested for multiple load conditions
and damage scenarios prior to the large penetrating damage event.

153
 SUMMARY

Keel, side, and crown areas of an aft fuselage section directly behind the wing-to-body intersection of
a wide body aircraft are used for composite technology development and verification purposes in
ATCAS. Boeing Commercial Airplanes Group is the prime contractor for the ATCAS program, with
significant contributions also coming from other Boeing Groups, industrial subcontractors, and
universities. Initial ATCAS efforts plan to develop composite technology that meet NASA Advanced
Composite Technology goals for cost and weight savings, relative to 1995 metals technology. During
the final years of the ATCAS program, work tasks are planned to provide the degree of technology
verification required to commit to development of a full scale barrel section.

A design build team approach has been established to minimize manufacturing costs and structural
weight. Three steps are used by the ATCAS DBT to select, evaluate, and optimize fuselage concepts.
First, the DBT selects baseline concepts that appear to have a strong potential for cost and weight
savings. This helps to focus early efforts in solving major technical issues. Second, cost and weight
savings potential of the baseline and several alternate concepts are evaluated using detailed designs
and manufacturing plans. This step has been termed global design optimization because it identifies
cost centers associated with fabrication and assembly of the entire structure. Finally, the DBT
attempts to reduce significant cost centers associated with the chosen concept. This third step has been
termed local optimization because it addresses individual manufacturing steps and subcomponent
design within the global cost constraints identified for the entire structure.

Four large quadrant panels (e.g., crown, 2 sides, and keel) were selected to minimize the number of
panel splices and to take advantage of a low-cost batch process for fuselage skin layup. A skin panel
concept with bonded stringers and frames, was selected for the crown and side quadrant baseline
design concepts. Selection of this concept was justified by global design optimization studies for the
crown, which indicated cost and weight savings potential. An innovative thick laminate/sandwich
concept was selected for the keel panel. This choice was made to avoid anticipated problems with
composites in fabricating and splicing a more traditional keel panel design. The major manufacturing
and performance issues associated with baseline concepts were identified. Detailed cost estimates for
side and keel baseline concepts will be performed in the following year.

All composite crown designs considered in global optimization studies indicated economic advantages
over the metal benchmark in cost and weight trades. The most efficient composite crown designs were
found to have 50% the weight and equivalent manufacturing costs. Composite concepts were
competitive with the metal benchmark primarily because reduced assembly labor offset lower
aluminum material costs. Cost savings on the order of 25% were projected for composite crown
panels by attacking cost centers (e.g., using tow placed fiberglass/graphite hybrids to reduce material
costs) in local optimization studies.

Considerable work was completed in addressing technical issues. A crown design tool that combines
manufacturing cost and structural performance constraints with an optimization algorithm is near
completion. Panels (e.g., 110 in. stiffened panels and intraply hybrid skin panels) were fabricated to
serve multi-purposes in evaluating technical issues such as process feasibility, low-cost tooling
concepts, impact damage resistance, and tension damage tolerance. Analysis methods were developed
and experimentally verified for toughened composites in support of keel applications. These included
post-impact compressive strength prediction. Data bases were obtained for manufacturing and
performance evaluation of new material forms, including braided composites and sandwich core.
Twenty technical papers, including seven in this proceedings, were published to document ATCAS
work completed to date.

154
 REFERENCES

1.) Walker, T.H., et al, "Cost Studies for Commercial Fuselage Crown Designs," First NASA
Advanced Composites Technology Conference, NASA CP-3104, Part 1, 1991, pp. 339-356.

2.) Willden, K., et al, "Process and Assembly Plans for Low-Cost Commercial Fuselage Structures,
First NASA Advanced Composites Technology Conference," NASA CP-3104, Part 2, 1991,
pp. 831-842.

3.) Smith, P.J., Thomson, L.W., and Wilson, R.D., "Development of Pressure Containment and
Damage Tolerance Technology for Composite Fuselage Structures in Large Transport Aircraft,"
NASA CR 3996, Contract NAS1-17740, August 1986.

4.) Horton, R., Whitehead, R., et al, "Damage Tolerance of Composites, Final Report,"
AFWAL-TR-87-3030, Vol. 3, May 1988.

5.) Rhodes, M.D., "Impact Fracture of Composite Sandwich Structures," AIAA/ASME/SAE 16th
Structures, Structural Dynamics, and Materials Conf., May, 1975.

6.) Friis, E.A., Lakes, R.S., and Park, J.B., "Negative Poisson's Ratio Polymeric and Metallic
Materials," J. of Materials Science, 23, 1988, pp. 4406-4414.

7.) Zabinsky, Z., Tuttle, M., et al, "Multi-Parameter Optimization Tool for Low-Cost Commercial
Fuselage Crown Designs." First NASA Advanced Composites Technology Conference, NASA
CP-3104, Part 2, 1991, pp. 737-748.

8.) Best, E., et al, "Developments in Impact Damage Modeling for Laminated Composite Structures,"
First NASA Advanced Composites Technology Conference, NASA CP- 3104, Part 2, 1991, '
pp. 721-736.

9.) Ilcewicz, L.B., Dost, E.F., and Coggeshall, R.L., "A Model for Compression After Impact Strength
Evaluation," Prec. 21st International SAMPE Technical Conf., 1989, pp. 130-140.

10.) Dost, E.F., et al, "The Effects of Stacking Sequence On Impact Damage Resistance and
Residual Strength for Quasi-lsotropic Laminates," Presented at 3rd Symposium on Composite
Materials: Fatigue and Fracture, Nov. 6-7, Buena Vista, Fla., ASTM, 1989.

11 .) Grande, D., Ilcewicz, L., Avery, W., and Bascom, W., "Effects of Intra- and Inter-laminar Resin
Content on the Mechanical Properties of Toughened Composite Materials," First NASA
Advanced Composites Technology Conference, NASA CP- 3104, Part 2, 1991, pp. 455-476.

12.) Ilcewicz, L.B., et al, "Matrix Cracking in Composite Laminates With Resin-Rich Interlaminar
Layers," Presented at 3rd Symposium on Composite Materials: Fatigue and Fracture, Nov. 6-7,
Buena Vista, Fla., ASTM, 1989.

13.) Fedro, M.J., Gunther, C.K., and Ko, F.K., "Mechanical and Analytical Screening of Braided
Composites for Transport Fuselage Applications," First NASA Advanced Composites
Technology Conference, NASA CP- 3104, Part 2, pp. 677-704.

155
 DESIGN, ANALYSIS AND FABRICATION OF THE TECHNOLOGY

INTEGRATION BOX BEAM

C. F. Griffin and L. E. Meade

Lockheed Aeronautical Systems Company

SUMMARY

Numerous design concepts, materials, and manufacturing methods were

investigated analytically and empirically for the covers and spars of a transport
wing box. This information was applied to the design, analysis, and fabrication of
a full-scale section of a transport wing box.

A blade-stiffened design was selected for the upper and lower covers of the
box. These covers have been constructed using three styles of AS4/974 prepreg
fabrics. The front and rear T-stiffened channel spars were filament wound using
AS4/1806 towpreg. Covers, ribs, and spars were assembled using mechanical
fasteners. When they are completed later this year, the tests on the technology
integration box beam will demonstrate the structural integrity of an advanced
composite wing design which is 25 percent lighter than the metal baseline.

INTRODUCTION

Current applications of composite materials to transport aircraft structure,

most of which are stiffness critical secondary structural components, have
demonstrated weight saving from 20 to 30 percent. The greatest impact on aircraft
performance and cost will be made when these materials are used for fabrication of
primary wing and fuselage structures that are 30 to 40 percent lighter than their
metal counterparts and have a reduced acquisition cost. Achievement of this goal
requires the integration of innovative design concepts, improved composite
materials, and low cost manufacturing methods.

In 1984, the Lockheed Aeronautical Systems Company began a program to develop

engineering and manufacturing technology for advanced composite wing structures on
large transport aircraft. The program was sponsored by the National Aeronautics and
Space Administration (NASA) under contracts NASI-17699 and NASI-18888 and Lockheed
Aeronautical Systems Company independent research and development funds.

The selected baseline component is the center wing structural box of an

advanced version of the C-130 aircraft. The existing structural box, shown in
Figure I, is a two-spar multi-rib design, 440 inches long, 80 inches wide, and 35
inches deep at the crown. A preliminary design of a composite wing box was
completed as were many design development tests. A full-scale section of the wing
box was designed in detail, analyzed, and fabricated. This paper will summarize the
major technical achievements of the box beam program.

It should be noted that some concessions as listed herein were made to reduce

the cost of the program; the conclusions drawn thus far in this program are valid
and achievable.

157
 FiKure I. Technology Integration Box Beam

- Surface contour was eliminated

- Same tooling used for upper and lower covers
- L/E and T/E attachments were omitted
- Ribs were made of aluminum

- Cut-outs in the spars were omitted

- Stiffeners on spars were made of aluminum
- Constant sections were used
- Hand lay-up of hat-stiffeners, doublers, and C-channels was used
- Additional autoclave cycles were used for doublers
- Both spars were filament wound at once on common mandrel
- Aluminum access door hole doubler was bolted
- Spar/cover joint doublers were secondary bonded

BOX BEAM DESIGN AND ANALYSIS

GEOMETRY

The technology integration box beam, see Figure I, represents a highly loaded
full-scale section of the C-130 center wing box. The test section of the box is 150
inches long, 50 inches wide, and 28 inches deep. The test section contains a large
access hole in the upper cover, wing box-to-fuselage mainframe joints, and center
wing-to-outer wing joints.

158
 DESIGN LOADS AND CRITERIA

The design loads for the box beam are based on the baseline aircraft

requirements. Maximum ultimate loads are 26,000 ib/inch compression in the upper
covers and 24,000 ib/inch tension in the lower covers. Ultimate spar web shear flow
is 4,500 ib/inch. These loads are combined with the appropriate pressure loads due
to beam bending curvature and fuel. The stiffness requirements for the wing were
established to meet the commercial flutter requirements specified in FAR Part 25.
Stated briefly, at any wing station the composite wing bending stiffness and
torsional stiffness could not be less than 50 percent of the baseline wing, and the
ratio of the bending to torsional stiffness must be greater than one but not more
than four.

Structural requirements for damage tolerance considered civil as well as

military criteria. Thus, the criteria used for this program requires the structure
to have ultimate strength capability with the presence of barely visible impact
damage anywhere within the structure. Barely visible impact damage is either the
kinetic energy required to cause a 0.i inch deep dent or a kinetic energy of i00
ft-lb with a 1.0 inch diameter hemispherical impactor, whichever is least.

COVER DESIGN

Design trade studies and structural tests were conducted on various

configurations for the wing covers. Figure 2 describes several of the designs and
presents results of compression tests on impact damaged panels. Based on these
investigations and manufacturing cost estimates, the blade stiffened design was
selected for the box beam.

BLADE STIFFENED

J- STIFFENED

61/33/16 FIBERGLASS FIBERGLASS

OVERWRAP WRAP
BLADE CAP ./
75/25/0 -.. .,.'--_ FASTENERS AT
SKIN / PAN '"-_, t"=_
,-_. RIB LOCA T IO N S
42/50/81 17/-6/17

COMPRESSION (9
/ ' kX, SKIN
DESIGN MATERIAL FAILURE LOAD
(IN.) O'PAO
OPINSK,N 17/66/15
AREAL WEIGHT

BLADE IM7/8551-7E 940,000

(TAPE)
T-STIFFENED
BLADE AS4/2220 590,000
FIBERGLASS
(TAPE)
OVERWRAP - ,__i---_ BLADE

T AS4/1806 810,000 .'4l 72/25/3

(TAPE)

AS4/1806 800,000
7- -I ....,
SKIN FASTENERS AT
(FABRIC) RIB LOCATIONS
15/70/15

(9 100FT-LB IMPACT

Figure 2. Wing Cover Designs

159
 The lower cover design, shown in Figure 3, consists of back-to-back channels
laid up on a skin laminate to form a blade stiffened panel. Note that the flanges
of the channels contain additional 0 degree plies compared to the web thus resulting
in a blade containing 67 percent 0 degree plies, 29 percent plus/minus 45 degree
plies, and 4 percent 90 degree plies. The blades, which are spaced at 5 inches, are
tapered in height to account for the increased axial loading from the outboard joint
to the wing centerline. A constant thickness laminate containing 27 percent 0
degree plies, 64 percent plus/minus 45 degree plies, and 9 percent 90 degree plies
makes up the lower skin.

The configuration of the upper cover, shown in Figure 4, is similar to the

lower cover with the exception that the blades are slightly taller. Also, the
central bay of the upper cover is reinforced by a hat stiffener which is terminated
at each rib location. The cut-out in the upper cover is reinforced by an
aluminum plate which is mechanically fastened to the cover laminate.

The chordwise splice of the composite covers to the aluminum load introduction
box is accomplished with the double shear joint illustrated in Figure 5. Note, the
bending stiffness continuity of the cover is maintained by inserting the aluminum
splice Ts between the composite blades.

SPAR DESIGN

As with the covers, trade studies and subcomponent tests were conducted on
various spar designs. Figure 6 shows the results of tests on stiffened shear panels
manufactured using several different materials and methods. The results of these
studies when combined with manufacturing cost estimates led to the selection of the
T-stiffened channel configuration shown in Figure 7. The spar webs and caps are of
constant thickness with the exception of the doublers located at the mainframe
attachment and the spar splice locations. This spar was designed to be filament
wound using AS4/1806 towpreg with unidirectional, bidirectional, and bias fabrics
used for the spar cap inserts, and doublers. The stiffeners were made of aluminum
for economy and were bolted and bonded to the spar web.

RIBS AND BOX ASSEMBLY

For the box beam, a J-stiffened skin configuration constructed of aluminum was
selected for all of the ribs. As shown in Figure 8, a T-shaped shear tie connects
the rib web and rib cap to the cover. All ribs will be mechanically fastened to the
spar webs and covers. Also the spar caps are mechanically fastened to the covers
using a double row of fasteners.

STRUCTURAL ANALYSIS

A detailed structural analysis was completed on the box beam using the methods
shown in Figure 9. A three-dimensional finite element model was constructed and
used to obtain internal loads for sixteen loads cases. Detailed two-dimensional
models were used to analyze the cover chordwise joint, cover cut-out area, and the
mainframe to spar joint. The compression stability of the covers was predicted
using the PASC0 code obtained from NASA. Several Lockheed computer programs were
used to obtain local stresses and strains using the internal loads obtained from the
NASTRAN models.

160
 150.00

AS4/974 FABRICS

• UNIDIRECTIONAL
• BIDIRECTIONAL
• _+45°BIAS
I
50.0

SPAR CAP DOUBLER

/ PLAN VIEW _CHORDWISE

JOINT DOUBLER
/ CONSTANT TAPER FROM

WS 18.00 TO WS 70.00

2.4B @ WS 70.00 / 2.78 REF @ WS 18.00

I
WS
WS WS WS u__ WS
75.00 03o
WS 0.00 18.00 52.00 _ d 75.00
18.00 r_

.605
_'J I_ 5.00 TYP

I IF[_l IL..._ F]BI:RGLASS

rll 3 PLY

- 50.00 ,

Figure 3. Lower Cover Box Beam Design

FIAT STIFFENER SPAR CAP DOUBLER

, \ ,
i \. i
k \ I

\ CHORDWISE;JOINT DOUBLER AL. DOUBLER PLATE _OVER GR/EP DOUBLER

150 IN .-

PLAN VIEW

CONSTANT TAPER FROM

3.38 @ WS 70.00
WS 18.00 TO WS 70.00

/ C3.68 REF @ WS 18".00

I

WS
75.00
'1 WS
18.00
I
WS
0.00
I
WS
I

18.00
WS
I

52.00
WS
I' WS
75.00

WS 70.00
70.00
5.00TYP'--'_ _ _I'---15.00--" I

i
_ R H /_L Jl]LL__
I. 50.00 .I

Figure 4. Upper Cover Box Beam Design

161
 WS75 SHEAR TIE
(ALUMINUM)

SPLICE PLATE
(ALUMINUM)

COMPOSITE
COVER

SPLICE TEE
(ALUMINUM)

Figure 5. Cover Chordwise Splice

PT

Pc
SHEAR (_ _TEST FIXTURE
MATERIAL MANUFACTURING FAILURE LOAD I
METHOD AREAL-WEIGHT (N.)

AS4/1806 HAND LAY-UP 410,900

FABRIC

AS4/1806 FILAMENT WOUND 483,000

TOWPREG

AS4/APC-2 ROLLFORMED AND AUTO- 412,500 PT

TAPE CLAVECONSOLIDATED
RIB FITTINGS
AS4/1806 HAND LAY-UP 355,600 MECHANICALLY
FABRIC (POST BUCKLED DESIGN) ATTACHED

(_ IOOFT-LB IMPACT
STIFFENER
PRE-MADE
:°
-
OR
PULTRUDED OR
ALUMINUM

DIRECTLY
IJE STRUCTURE
TO COVER
ATTACHES !i _""_" _"

Figure 6. Wing Spar Designs

162
 MATERIALS
AS4/1806 TOW PREG

AS4/1806 FABRIC
+44144112

_ ,_o.,ws_ _ I _;_''

.225_,.1_-/

I I_._Jl_>/j_ AT MAINFRAME AND _- 2 0--'1

__PLICE LOCATIONS

Figure 7. Spar Assembly

MATERIAL=ALUMINUM

WEB _

POST

RIB CAP , i//

_J/_ FITTING

_i" It/! ,_._ii_ _:_ _SHEAR TIE

Figure 8. WS75 Rib

163
 ,NT,
,LEMENTANAL
°A=,,ANSS ;

CUT-OUT
REINFORCEMENT

FiEure 9. Structural Analysis Methods

Fisure 10 presents the typical design allowables obtained for the AS4/1806 and
AS4/974 fabric prepreg materials. These allowables were computed based on laminated
tests, and in the case of the impacted laminate allowables, stiffened panel tests.
Note that the allowable strain is plotted as a function of the percentase of
plus/minus 45 degree plies within the laminate minus the percentage of 0 degree
plies. This value is called the AML for angle minus longitudinal plies. For
example, a quasi-isotropic laminate has an AML value of 25. The blade stiffener on
the cover has an AML of -38 and the cover skin a value of 37.

Margins of safety were computed for numerous locations on the covers and spars
using the applied strains and the design allowable strains. Several minimum margins
of safety are presented in Figure II. Both the upper cover and spar webs have a 0
margin of safety for the impact damaged condition. The lower cover and the spar cap
are critical for bearing/bypass and net tension, respectively.

A preliminary design of a C-130 center wing box was completed using the design
concepts and materials described for this technology integration box beam. Weights
analysis indicated an overall savings of 25.4 percent compared to the metal
baseline. The predicted spar weight savings was 35 percent and the cover was 28
percent. These weight savings could be improved if a higher modulus fiber such as
IM7 were used in conjunction with the latest generation of touEhened epoxies.

SPAR FABRICATION

The spars were fabricated by filament winding AS4/1806 towpreg onto a mandrel
that when trimmed apart lengthwise produced both the front and rear spars at the

164
 STRAIN

(IN/IN)
PLY LEVEL ELASTIC CONSTANTS COMPRESSION DESIGN
ALLOWABLE
m 0.008

PROPERTY UNIDIRECTIONAL BIDIRECTIONAL

FABRIC FABRIC

17.70 9.70
0° TENSILE MODULUS (MSl)

1 47 880
90 ° TENSILE MODULUS (MSI)
--0004

0° COMPRESSION MODULUS (MSI) 1700 9.70

DESIGN ALLOWABLE
180" F. W OPEN HOLE
0 ° INPLANE SHEAR MODULUS (MSI) 0,62 ' 0.62
-- 0.002

0 ° POISSONS RATIO 030 0.05

I I I I I I
.6o -40 -2O 0 2o 40 6o

% :t: 45 = - %0 °

STRAIN STRAIN
(IN/IN) FILLED HOLE' COMPRESSION DAMAGE (IN/IN)
TESION DESIGN
ALLOWABLE 0.25 IN. DIA. TOLERANCE ALLOWABLE

-- 0.008 -65 = F. DRY -- -- 0008

- -- -- 0.006

OPEN HOLE
0.25 IN. DIA.
-- 0.004 -65 ° F. DRY

" 0004

0.002 IMPACTED DESIGN -- -- 0.002

ALLOW 180 ° F. WET

I I I I I 1 I I I 1 I I
-60 -40 -20 20 4O 60 -60 -40 -20 20 40 60

% + 45 = . %O + % ::t:45 ° - %0 °

Figure I0. Design Allowables

® " _ M.S. MODE

0.00 CAI

LOWER COVER 0.01 BRG/BYP

" UPPER COVER 0.04 BRG/BYP

li. ® ® sP.RWE. 0.00 CAI

k_ (_ SPAR CAP 0.05 NET TENSION

Figure II. Minimum Margins of Safety

165
 same time. Figure 12 shows the spars as filament wound and cured on the mandrel
prior to removal. Figure 13 shows the two spars after separation. The aluminum
stiffeners were fabricated, located on the spars, and drilled for fastener
installation. The stiffeners were then phosphoric-acid anodized for bonding,
adhesive coated with Hysol #9339 glass-microballoon filled adhesive and bolted and
bonded in place. Figure 14 shows a spar with the stiffeners installed.

COVER FABRICATION

The covers were fabricated by separately laying-up C-channels, 64 tow stuffers

and the skin laminate and then assembling the lay-ups for cure as shown in Figure
15. The C-channels were alternately laid-up on 'hard' mandrels and 'soft' mandrels
to achieve both positive location of each blade stiffener and full fluid bag
pressure during cure.

Note: 'Hard' mandrels were hollow aluminum mandrels formed into closed boxes
by welding two Ls together. 'Soft' mandrels were U-shaped silicone
rubber mandrels reinforced with included graphite fabric placed directly
into the C-channel lay-ups to apply fluid pressure to the laminate while
maintaining some stiffness for dimensional and shape control.

After laying up the skin laminate directly on the steel cover tool plate the
center hard mandrel with its C-channel in place was positioned and pinned in place.
(NOTE: Each hard mandrel, being aluminum to reduce worker handling weight, has a
large difference in thermal expansion from the steel base; therefore, tooling pin
holes were solid on one end and slotted on the other.) Next, a towpreg stuffer made
of 64 tows was installed in the radius of the C-channel-to-skin joint. Then a soft
mandrel C-channel layup was installed, another stuffer, another hard mandrel C-
channel, etc., until the lower cover layup assembly was complete. Figures 16, 17,
18, and 19 show sequentially this layup process. A layer of FM 300 0.030 psf
adhesive film is used between each layup interface. Thermocouples were installed,
the tool corners padded, breather material applied, and the tool covered with the
curing blanket which was sealed to the tooling base. The cover was then cured in
the autoclave with 85 psi at 350"F for two hours. As seen in Figure 20, taken
after unbagging and removal from the tool, the cover exhibited a curve which is due
to differential shrinkage of resin at each blade stiffener location. This curve,
however, is easily removed with moderate force at time of assembly with spars and
ribs.

The cover has approximately four inches of trim excess on one end and I0 inches
of excess on the opposite end which contains NDI standard flaws. This end will
become the NDI standard for these panels after trimming to net size.

Additional material has to be added in a separate layup and cure cycle to serve
as doublers for the load introduction box joint at each end of the covers. At the
same time, spar cap doublers, separately laid up and cured, will be bonded on using
FM 300 0.030 psf adhesive film. The hat stiffeners are to be installed on the cover
with fasteners and micro-balloon filled paste adhesive, as were the stiffeners on
the spars, in a bolted/bonded joint. After trimming the cover to final size, a
3-ply fiberglass layer overwrap will be installed with a vacuum cure at 250°F on
the upstanding leg of each stiffener for damage tolerance protection as shown in
Figure 2.

166
 SPAR ASSEMBLY

MATERIAL
GRlEP (AS4118061

LOCAL REINFORCING
AT MAINFRAME A N D
SPLICE LOCATIONS

Figure 12. Spar Assembly

F i g u r e 13. Two S p a r s

167
 Figure 14. Spar Assembly

/ IC' \

\ STUFFERS \SKIN

Figure 1 5 . Skin Panel Layup Scheme

168
F i g u r e 16. S o l i d v e r s u s S l o t t e d Mandrel T o o l i n g H o l e s

169
 Figure 1 7 . S o f t Mandrel t o be I n s e r t e d

Figure 1 8 . Panel Ready f o r Bleeder

170
 L

Figure 1 9 . Bagged Lower Cover i n Autoclave

Figure 2 0 . Curve i n Panel Due t o Blade Resin Shrinkage

171
 RIB FABRICATION

The ribs being made of aluminum use standard aircraft assembly methods with
mechanical fasteners as seen in Figure 21. An auto-fastener machine was used to
install most of the rib assembly fasteners. Fastener locations were placed directly
onto the part by using a mylar reproduction of the blueprint as an overlay.

BOX ASSEMBLY

The box assembly sequence will utilize assembly of the spar subassemblies with
the rib subassemblies on a surface table using tooling knees initially for alignment
and positioning. Once the spar-rib subassembly is assembled and squared, the covers
will be joined to it and drilled using a Cybotec brand robot which will allow
drilling of the cover-to-box holes without reaming. Although the box will not be
sealed for fuel tightness, fasteners will be wet installed with corrosion inhibiting
sealant where necessary to prevent corrosion between metal and composite surfaces.
The box assembly includes an aluminum load introduction box extension on each end
for testing. The test fixture is being fabricated under independent funding and
will interface with the technology integration box beam via the aluminum box
extension. Figure 22 details the assembly sequence of the box beam.

COST ANALYSIS

A detailed cost analysis compared a composite center wing structural box with
an advanced aluminum version of the C-130 center wing box. The cost analysis
evaluated the final technology integration box beam design, which was extrapolated
to a full sized 80 by 440 inch wing box. The results demonstrate a potential 5
percent labor and material cost savings for a composite wing box compared to a new
state-of-the art metallic design. Cost benefits are achievable in the current
composite design concept through a reduction of labor costs; innovative design
concepts result in less time required for fabrication and assembly operations.
Also, automated manufacturing processes such as filament winding and pultruding have
the potential to reduce costs. Estimated costs of the composite wing box, project
recurring costs that will be incurred during a typical full-scale production program
producing 200 ship sets of wing boxes. Figure 23 illustrates the cost breakdown for
each major component as well as final assembly. Costs are distributed for both the
advanced aluminum and composite wing box, illustrating relative costs of covers,
ribs, spars, and assembly. Recurring production costs of the composite box are 95
percent of the baseline as a result of fewer parts in composite subassemblies and
automated fabrication processes. Aluminum costs are based on actual C-130 cost
history. Composite material costs are based on material vendors projections for
material at high quantities. Fabrication costs, where possible, are generated from
actual composite production experience. Where cost tracking data is not available,
Value Engineering cost estimating methodology is used.

172
 _l--_F-1---F_ _

/7

Figure 21. Rib Assembly

STABILIZER CLIPS (68)

DOUBTER

UPPERAssyCOVER/_)i / _/>
RIB POST
181

R,a^CCESSPOOHS
LOWER
_O_E%__
ASSY 2<.

II

<,_ _

REAR SPAR __

ASS_ ill ..__i _,,

RIB CHAP, I NELS_ _ FINAL ASSEMBLY
_1¢i --
j"
RIB SHEAR TIES
(4)

FRONT DRAG ANGLE WEBS

SPAR //. _/_
(2l

ASSY/ _ FUEL CLIPS (32)

CORNER FITTINGS
(8}

Figure 22. Task 5 - Box Beam Assembly Breakdown

173
 SPARS
RIBS
1.20 -- [[_ ASSEMBLY
COVERS
1.00
1.00 -- 0.95

0.80 --

0.60 --

0.40 --

0.20 --

0.00
ALUMINUM COMPOSITE

Figure 23. Wing Box Relative Cost Comparison Cost Breakdown

The covers comprise 49 percent of the total cost and account for 64 percent of
the total weight of the wing box (reference Figure 24). A meaningful cost reduction
could most easily be achieved through reducing costs associated with the covers.
Assembly costs also have a considerable impact on costs, accounting for 25 percent
of the total. Cost reductions for the spars and ribs will not result in an
appreciable cost savings since their impact on total costs is only I0 percent and 16
percent, respectively.

Figure 25 further illustrates the total cost breakdown, showing material and
labor costs separately. This provides a more specific means with which to target
and assess cost drivers. Material costs for the covers stand out, accounting for 35
percent of the cost. Figure 26 shows a breakdown of material costs, demonstrating
the significance of the cost of the covers as a percentage of total material costs.
Approximately 74 percent of the total material costs is in the covers as shown.
Material costs are based on projected costs estimated by vendors for graphite/epoxy
prepreg at I0,000 pounds per year quantities. The plus/minus and minus/plus 45
degree knitted fabric is priced almost 87 percent higher than the uni-directional
and 0/90 degree fabric. There is a potential for reduction in the former as the
vendors have limited experience producing this material; consequently, their
estimate may be conservative. A reduction in this price could have a significant
effect on cover material costs as the plus/minus 45 and minus/plus 45 degree fabric
accounts for 47 percent of the cover material costs.

174
 25% ASSEMBLY

49% COVERS

10% RIBS

16% SPARS

Figure 24. Total Cost Breakdown

5% ASSEMBLY MATERIAL
13% COVERS LABOR_

,_-- 21°/o ASSEMBLY LABOR

35% COVERS MATERIAL -

"-- 8 % RIBS LABOR
6% SPARS MATERIAL

10% SPARS LABOR

Figure 25. Labor Cost Breakdown

175
 11% ASSEMBLY

74% COVERS

Figure 26. Material Cost Breakdown

Assembly costs are 25 percent of the total cost, 20 percent of which is

assembly labor and 5 percent material. Assembly labor costs account for 39 percent
of the total labor cost (reference Figure 27); assembly is labor intensive due to
the necessity of a two-step drilling procedure required for graphite/epoxy compo-
sites to obtain acceptable holes. Improved composite drilling techniques/equipment
or a reduction in the overall part/number of fasteners (see Figure 28) could signi-
ficantly reduce the total costs. Material costs, at 5 percent of the total, are not
a driver even though titanium fasteners are required.

Spar costs account for 16 percent of the total cost and probably represent an
optimized design; the cost is based on a filament wound spar which significantly
reduces the number of parts and fasteners, compared to a new metallic spar design.
Further cost reduction is unlikely and would have a negligible effect on the total
cost.

The ribs are only I0 percent of the total costs, 8 percent of that for labor.
It is assumed that the skins and channels would be hand laid-up. Material costs are
based on the same assumptions as the covers, and may be reduced; however, only 4
percent of the total material costs (reference Figure 26) is for the ribs. This is
not an ideal target for emphasis on cost reductions.

176
 26% COVERS

39% ASSEMBLY

20% SPARS

15% RIBS

Figure 27. Labor Cost Breakdown

CONSTRUCTION TYPE PARTS FASTENERS

ALUMINUM
"C130 CENTER WING" 160 3500
"FRONT SPAR"

COMPOSITE 40 200
SPAR

Figure 28. Part/Fastener Comparison

177
 CONCLUDING REMARKS

The concurrent engineering approach used in this project has resulted in a wing
box design which has a 25 percent weight saving and a 5 percent cost saving compared
to the baseline advanced metal wing box. Incorporation of improved materials and
the evaluation of alternatives to the bias fabrics could lead to further reductions
in weight and acquisition costs. Spars were successfully filament wound, back-to-
back on a common mandrel. Box covers were also successfully co-cured. These
successful fabrication demonstrations point up still more lower-cost fabrication
methods that could be incorporated in the future.

178
 DESIGN AND MANUFACTURING CONCEPTS
FOR THERMOPLASTIC STRUCTURES

Michael P. Renieri, Steven J. Burpo,

and Lance M. Roundy

McDonnell Aircraft Company

St. Louis, Missouri

SUMMARY

Results to date on the application of two manufacturing techniques, fiber

placement and single diaphragm/coconsolidation, to produce cost-effective,
thermoplastic composite (TPC), primary fuselage structure are presented.
Applications relative to fuselage upper cover structure indicate potential cost
savings relative to conventional approaches. Progress is also presented on
efforts concerned with other design details which take advantage of
thermoplastic composites such as fastenerless stiffener/frame attachments. In

addition, results are presented on the development and verification testing of

a composite lug analysis program which incorporates through-the-thickness
effects.

INTRODUCTION

A major obstacle to widespread use of high performance composites in primary

aircraft structures is the high cost of manufacture and assembly. Under NASA's
ACT program, McDonnell Aircraft Company is investigating cost-effective,
innovative techniques for the fabrication and joining of primary airframe
structure using thermoplastic composite materials. MCAIR is teamed with
Douglas Aircraft Company (DAC) under the ACT initiative in a program entitled
Innovative Composite Aircraft Primary Structures (ICAPS).

Primary effort on the MCAIR portion of the ICAPS program has concentrated on
developments relative to an advanced fighter fuselage section which are
applicable to commercial vehicle structure. These include the application of
two innovative manufacturing techniques, fiber placement and single diaphragm/
coconsolidation to fabricate fuselage cover panels.

In addition to panel fabrication, elemental specimens are being fabricated and

tested to address key design issues associated with the fuselage section such
as pull-off strength of fastenerless frame attachment concepts and the
performance of thick composite lugs. In support of the lug evaluation, an
analytical program has been developed incorporating through- the-thickness
effects.

GENERIC FUSELAGE SECTION

The advanced aircraft system selected for the fighter development effort was
the Model 4629 ASTOVL design developed by MCAIR under the NASA-Ames sponsored
U.S./U.K. ASTOVL Technology Development program. Based on representative

179
 fuselage cross sections of the Model 4629 aircraft, a generic center fuselage
structure, Figure i, was developed as the primary structure demonstration
component. While the fuselage structure contains design features particular to
advanced ASTOVL aircraft, cost effective fabrication techniques and innovative
design concepts developed in this program demonstrate technology related to all
emerging aircraft systems.

The generic center fuselage structure contains many challenging structural

components:
o Upper Cover
o Tank Floor
o Carry-Thru Bulkhead
o Closure Bulkhead
o Keel Webs
o Frames
o Inlet Ducts
o Longerons

An upper fuel cell cover subcomponent was selected for design/analysis,

fabrication, and structural testing for Phase A. The cover structure ties the
upper longerons and bulkheads of the generic fuselage section togetheE and is a
primary load carrying component for flight induced structural and fuel cell
loading. The cover must be capable of a 255°F (II0°C) operating temperature,
have a limited number of fasteners on the outer moldline (OML), and resist

hydrodynamic ram loading.

MATERIAL AND PROCESS SELECTIONS

Material

The material chosen was based on temperature requirements, solvent resistance,

component design, manufacturing approach, and processing ease. Due to the
255°F design requirement, the baseline thermoplastic material selected was
Imperial Chemical Industries' ITX (intermediate temperature crystalline), which
has service capability to 300°F. The fiber selected was an intermediate
modulus fiber produced by Hercules, IMT. ITX has processing characteristics
similar to ICI's APC-2 (PEEK) system which is a mature resin that MCAIR has
worked with extensively. In addition, AS4/APC-2 was selected for early forming
studies due to availability and to verify analytical predictions for thick

composite lugs.

Processes

Manufacturing processes were selected using a concurrent engineering approach.

Processes were rated based on innovativeness, cost, risk, supportability,
survivability, and weight. Two manufacturing techniques, fiber placement and
single diaphragm/coconsolidation, were selected to fabricate subscale fuselage
cover panels.

180
Fiber placement (FP), one of the more promising methods of fabrication, is the
in situ consolidation of individual material layers using pressure and heat at
the point of contact. This procedure eliminates the autoclave requirement and
automates the material deposition process reducing significant cost elements in
a typical composite production environment.

Secondly, MCAIR is evaluating a method that uses only a single upper diaphragm
to form a skin and hat structure in one step. This process, single diaphragm/
coconsolidation (SDCC), simultaneously consolidates the hat stiffened inner
skin plies with the outer skin plies while coconsolidating the two yielding a
high quality interface and reducing the number of process steps from three to
one.

PRODUCIBILITY ANALYSIS

A producibility analysis was performed to determine the cost-effectiveness of

the selected processes versus alternate approaches, Reference I. Three
material systems and five design options were considered for the cover. TPC
materials are considered in three approaches: SDCC, fiber placement, and
manual lay-up. The other two options include a manual lay-up of toughened
bismaleimide (BMI) and a titanium superplastic formed diffusion bonded (SPF/DB)
design.

The three approaches for thermoplastic composites include (I) SDCC in which a
pressure box is employed to consolidate the outer skin while at the same time
forming and consolidating the inner stiffened pan, (2) fiber placement, using a
tow placement process over preformed hat stiffeners recessed into a fiber
placement tool, and (3) a thermoplastic composite manual lay-up approach with
autoclave consolidated unidirectional and comingled material forms. A
traditional manual lay-up process was considered for the toughened BMI
thermoset composite (TSC) design utilizing rubber mandrels and female tooling
to produce a co-cured structure. In addition, the TSC design included
stitching of the stiffeners to increase stiffeners-to-skin interface strength.
A titanium superplastic formed/diffusion bonded design was the metal option.
Diffusion bonding allows the economical creation of high performance hat
stiffened skins without fasteners.

The analysis explored the impact of component complexity on producibility and

cost. Two levels of complexity were considered. The fuel cell cover under
development for this program is a single curved component. A producibility
analysis of this generic cover was established to serve as a baseline. A
complex, doubly curved version of the cover was also considered since OML
fighter skins are typically complex surfaces.

High processing temperatures for thermoplastic composites (750°F, 385°C) impose

two major fabrication constraints: (I) flexible rubber mandrels (for hat
stiffener tooling) cannot be used since they are unable to survive the
processing temperatures and (2) high temperature tooling is required instead of
aluminum tooling. The influence of these constraints for both recurring and
non-recurring costs was considered. Each fabrication approach listed above
was evaluated in order to identify the best technique for both levels of
complexity.

181
 Recurring component costs were generated by summing material and labor costs
for each step of process plans for each fabrication approach. Labor costs were
burdened to include equipment/facilities costs. Non-recurring costs took into
account tooling expenses, including any duplicate tooling required to produce
the theoretical rate of 85 ship sets per year (600 aircraft total). Cost
comparisons for this study were normalized; the least expensive simply curved
approach is set equal to one with the cost of other options appropriately
ratioed.

The cost study results for both complex and simply curved components showed
that the SDCC approach was most cost-effective for the cover due to flat ply
collation and short cycle times, Figure 2. TPC fiber placement was the next
most cost-effective approach due to automated processing of the skin. Although
TPC's are difficult to manually lay-up, this process is less expensive than TSC
manual lay-up due in part to stitching requirements for TSC in order to
increase hat pull off strength. Titanium SPF/DB and TSC were close in
recurring cost due to the labor intensive operations required for these
approaches. As expected, the recurring cost of fabricating complex structure
was consistently higher than simple structure.

Non-recurring costs (tooling) for the five fabrication approaches showed that
duplicate tooling requirements for TSC and TPC manual lay-up increase their
respective tooling costs to a level comparable to the other fabrication
approaches, Figure 3. Even with duplicate tooling, non-recurring costs for
simply curved TSC and TPC are the least expensive options. Five-axis machining
requirements for tooling on complex curved manual lay-up TSC and TPC
approximately doubles their respective non-recurring costs. Although press
forming and fiber placement tooling costs are identical for simply curved
applications, a substantial cost increase is incurred in press forming versus
fiber placement costs for complex curvature. This increase is attributed to
difficult machining requirements (five-axis) for not only the press forming
tool but also for the associated pressure box. High temperature matched metal
steel tools must be supplied for the titanium SPF/DB approach resulting in the
highest tooling costs of any approach.

SUBCOMPONENT DESIGN/ANALYSIS AND STRUCTURAL TESTS

Design loads for the upper cover subcomponent, Figure 4, were developed from
maneuvering flight conditions consisting of a 9g symmetric steady-state pull-up
(SSPU) for down bending, a -3g steady state pushdown (SSPD) for up bending, and
7.2g rolling pull-out (RPO) for combined vertical and lateral loads. All
flight conditions are at sea level and 0.95 Mach. A 22.0 psi (ultimate) fuel
pressurization load condition is also included.

Design loads were obtained at a fuselage station in the forward-center fuselage

section at the most forward wing shear attach location. A hat stiffened skin
of single curvature was sized for these loads. Various laminates were
evaluated for static strength and panel stability. A finite element model of
the cover was used to examine effects of combined loads and determine static
deflections. Panel stability was determined using SS8 Anisotropic Curved Panel
Analysis Program, Reference 2.

182
Two designs were developed for the upper cover. The first design, Figure 5,
contains discrete hat stiffeners with a constant thickness skin. The second
design, Figure 6, contains a constant thickness Inner Mold Line (IML) pan with
a buildup under the stiffener in the OML skin. The first design will be
utilized in the FP manufacturing process. The second design will be fabricated
using the SDCC.

The laminate stacking sequences for the cover subcomponents represent the
minimum necessary to sustain all flight load conditions, buckling constraints
and fabrication requirements. Skin buckling occurs at 120% of design limit
load, a requirement common in new fighter aircraft designs with composite mold
line skins.

Planned structural testing of the subcomponents consists of compression static

and fatigue loading. Static tests will be used to compare results of
structural tests to analytical predictions, while fatigue tests will determine
panel resistance to delamination modes of failure due to repeated loads.

ELEMENT DESIGN/ANALYSIS AND STRUCTURAL TESTS

Two structural areas of particular interest in the fuselage structure were

selected for elemental evaluation. These are thick composite lugs and
stiffener-to-skin joints.

Lug Elements

A method for predicting the static response of thick, highly loaded, composite
lugs has been developed. Composite lugs provide a mechanism for the transfer
of concentrated loads from one structural member to another. The most notable

examples are lugs that transmit wing loads into bulkheads such as those pres-
ent on the generic fuselage structures. The geometry of these lugs can vary
substantially for different applications, and they may be required to carry
in-plane as well as out-of-plane loads. In addition, effects such as pin
bending may result in complex stress states through the thickness of a lug,
even when it is subject to only in-plane loads.

The wide range of variables associated with this problem necessitates the use
of an analytical method that is very flexible, both in the range of geometries
and the types of load conditions that it is capable of analyzing. The complex
through-the-thickness stress distributions that may develop require that the
method also be capable of predicting three-dimensional stress fields. The
finite element method is one such approach and was the method used in this
development.

Laminated composite aircraft components are typically analyzed using plate

elements that are based on classical laminated plate theory. These elements
are simple to use since the geometry of the element can be defined in two
dimensions, and they account for in-plane and out-of-plane loads. They are,
however, restricted to the analysis of thin plates, which limits their
usefulness for the analysis of thick lugs. The thin plate derivation permits

183
 only linear variations in the in-plane stresses through the thickness of the
lug, and out-of-plane stresses are assumed to be negligible.

At the other extreme, a thick lug can be modeled with three-dimensional solid
finite elements. Using the approach, each ply can be modeled (one or more
plies through the thickness of each element). These elements allow complete
generality in defining the lug geometry and loads and are based on assumed
three-dimensional displacement fields. Although models generated with these
elements provide accurate results, their use is cumbersome and they require
substantial computing resources. They are, therefore, not recommended for the
type of parametric study that would be required to optimize a lug design.

The approach taken in this program was a compromise between the two methods
described above. A subparametric laminated solid element based on cubic

displacement functions was used. This element was developed in Reference 3 to

study through-the-thickness stress fields that develop during low velocity
impact events.

The geometry of this element is defined by four nodes in the X-Y plane, and the
stacking sequence of the laminate under consideration, Figure 7. There are
twenty-four degrees of freedom at each node. These degrees of freedom are the
translations at the upper and lower surfaces of the laminate in each of the
three coordinate directions and the derivatives of these translations with
respect to each coordinate.

Using conventional finite element methods, the stiffness matrix for the element
is generated by integrating the strain energy density over the volume of the
element. The effects of stacking sequence are included by performing this
integration numerically over the volume of each discrete ply and summing the
results. This approach also allows the average strain in each ply and at each
interface to be calculated once the translations and derivatives of the
translations have been determined.

To account for pin bending effects, both the lug and the pin are modeled. The
generation of these models is relatively simple since the geometry is defined
in only two dimensions. The lug/pin contact is modeled by coupling lug and pin
displacements in the radial direction. Since the extent of the contact area is
not known a priori, the problem is solved iteratively. An initial contact area
is assumed, loads applied, and displacements and reactions forces in the lug
and pin calculated. The forces that develop between coupled points on the lug
and pin are then investigated. If the force at a given point is compressive,
the lug and pin remain in contact at that point; if tensile, the two separate.
The model then analyzes the new contact area, and the contact forces are again
investigated. This process continues until the contact area stops changing.

The converged displacement field is then used to perform a laminate analysis on

an element by element basis. For each element, the average strain in a given
ply is calculated by integrating the strain field over the volume of that ply
contained in the element, and the average strain at an interface is calculated by
integrating over the interface area. These strains are then used to calculate
average ply and interface stresses within the element. Either stress or strain

184
may then be used in an appropriate failure criteria to evaluate the integrity
of the element.

Advantages of this approach over existing finite element models are I) model
geometry is defined in only two dimensions, 2) three dimensional stress and
strain fields are used, 3) the effects of the laminate stacking sequence are
included, 4) both in-plane and out-of-plane loads are included, and 5) ply and
interface stresses and strains are calculated. Although existing methods have
some of these advantages, no other approach has all of them.

In order to verify the analytical code developed, three lug lay-ups and two pin
diameters were chosen (Figure 8). Each lug has a different
through-the-thickness stiffness distribution, but all have the same average
in-plane stiffness. The lug with the smaller pin diameter was sized to fail in
bearing, and the lug with the larger pin diameter was sized to fail in shear.

Based on sensitivity study results, four elements through the thickness were
used for strength prediction of lugs. A comparison of bearing strain
distributions at 40 kips predicted by models consisting of one, two or four
elements through-the-thickness is shown in Figure 9.

The lug specimens were tested as shown in Figure I0. Test results and
associated predictions for the lugs are summarized in Figure II. Failure
prediction in the critical elements was based on using a modified Hashin
criteria. The 1.75 inch diameter holes exhibited a tensile fiber failure at

the net section while the 1.0 inch diameter showed permanent yielding around
the hole prior to shear bearing failure. Good correlation between test and
prediction was obtained. The initial bearing failure load was determined by
using axial strain data from the rosettes located 0.5 inches away from the edge
of the hole. The load versus strain curve in Figure 12 shows that axial strain
decrease associated with material failure ahead of the pin. Typical bearing
and net section failure modes can be seen in Figures 13 and 14 for the 1.0" and
1.75" diameter specimens, respectively.

Frame Elements

The Y-section frame elements, Figure 15, will be fabricated and tested against
a T-section of contemporary design. Both sections will be coconsolidated to a
typical skin laminate during fabrication.

The Y-section was chosen for its potential formability in a diaphragm forming
process and for its lower peel stress components. The effect of changing the
angle of the Y-section is also being investigated. Two-dimensional finite
element models of the T- and Y-sections were created for the purpose of
defining the boundary conditions to be used in testing.

The fastenerless moldline Y-frame attachment element design is structurally

simple, allowing it to be diaphragm formed and coconsolidated. Transverse
tension strength for semicrystalline thermoplastic composites has been found to
be appreciably greater than comparable thermosets. This property is expected
to enhance the peel strength and survivability of the frame element and will be
verified by our planned tests.

185
 Test loads for the Y-sections will be introduced through an internal mandrel
(Figure 16). This method alleviates possible failures other than those desired
and allows for later design of several frame attach possibilities. Possible
frame attachments include: amorphous bonding, resistance joining, adhesive
bonding, coconsolidation, and mechanical fastening.

SUBCOMPONENT TOOLING AND MANUFACTURING

Tooling concepts for each cover design/manufacturing concept is discussed.

Internal geometry of the hat stiffeners was designed to be similar to allow
mandrel tools to be interchangeable.

The fiber placement manufacturing process consists of placing inverted roll

formed hat stiffeners into an aluminum fiber placement tool. Aluminum mandrels
will be placed in the hat stiffeners to prevent skin deflection during the
fiber placement process. Tooling conceptual design is illustrated in Figure
17. Retainers are utilized to hold the stiffeners and mandrels in place during
processing. The hat is positioned in the tool as shown in Figure 18. The hat
flange is offset slightly above the aluminum tool to allow for adhesive and the
first ply. Also, a heat blanket will be embedded into the aluminum tool that
supports the preformed thermoplastic composites stiffeners. The blanket will
provide a greater degree of temperature control where required. Following
fiber placement of the skin, the panel will be trimmed and retainers removed.
The mandrels will then be removed and the part prepared for nondestructive
testing.

The SDCC concept is unique in that there is but one diaphragm, and the IML pan
and OML skin are consolidated and coconsolidated during the diaphragm forming
process. The SDCC tooling concept is il'lustrated in Figure 19.

The greatest risk in diaphragm forming over hat mandrels is the chance for
bridging. To minimize this risk, the manufacturing and tooling team members
utilized lessons learned criteria to optimize hat height, cap width, and skin
thickness. Hat spacing was maximized to increase ply surface area between
mandrels. The increased surface area will increase the force exerted to form

the ply pack and prevent bridging. In addition, the mandrel will be fabricated
with a slight radius (Figure 20). The gap between the mandrel, OML skin, and
IML pan will be filled with a predetermined amount of unidirectional tow. This
fillet area has the highest probability for bridging; however, with the
unidirectional fillet, the pressure will be equally distributed to facilitate a
quality consolidation.

A vacuum ring and a neat film layer will aid in ply pack location. The IML ply
pack will be contained between the aluminum diaphragm and a layer of neat film.
A vacuum ring surrounds the IML ply pack and vacuum draws the aluminum
diaphragm to the upper surface of the IML ply pack and the neat film to the
lower surface of the IML ply pack. The IML ply pack is then positioned
correctly above the tool prior to application of heat and pressure. The neat
film is coconsolidated between the IML and OML ply packs during the press
operation. This will permit accurate location of the IML ply pack and aid in
prevention of wrinkles.

186
Concepts for the subcomponent tool include machined steel weldment, cast bulk
ceramic, machined aluminum, and a metal arc sprayed tool. A metal arc sprayed
tool which can accommodate integral heating, faster cycle times, and low tool
cost for production-type environments shows high potential.

ELEMENT TOOLING AND MANUFACTURING

Lug Elements

Tooling for the lug specimens consisted of simple project plates with steel
dams positioned to allow for expansion during consolidation. The lug
geometries and lay-ups that will be used were previously described in Figure 8.

The lugs were fabricated from AS-4/PEEK unidirectional tape. Eighteen 30" x
16" sublaminate panels of four different 30 ply lay-ups were consolidated in a
hydraulic press. Six sublaminates were then stacked to form the three
different 180 ply stacking sequences. The three stacking sequences were
co-consolidated in the autoclave. The lug specimens were water jet cut from
the panels and the holes reamed to final dimensions. Excellent consolidation
was achieved in all lug specimens as evidenced by ultrasonic and
photomicrographic inspections.

SDCC and Y-Frame Elements

An SDCC element verification tool (Figure 21) was developed which can
incorporate either two hat mandrels or a single triangular mandrel to fabricate
the Y-section frame elements. The hat dimension, spacing, height and width
simulate the subcomponent design. The hat stiffener mandrels, located by pins,
float on the unconsolidated skin. The inner skin is then formed over the

mandrels in a press operation. The aluminum tools are readily extracted

following forming.

To focus on the critical process variables and not on geometric complexities,

element forming trials on the verification tool commenced with the Y-frame
elements and will then proceed to the 2-hat section element. Scale-up is a
very significant concern and is being considered in all the fabrication
development activities.

Forming trials on the single "Y" configuration used one diaphragm to

consolidate the upper ply pack with the lower plies. Initially full pressure
(150 psi) was applied after the melt temperature of the ITX was reached but was
maintained for only 5 minutes at which point the diaphragm ruptured. In spite
of the short hold time the pressure was sufficient to fully consolidate the
flat areas of the part and to form the material over the mandrel. The upper
ply pack conformed to the mandrel surface very nicely, but the diaphragm rupture
caused the outer ply to lift and bridge across the mandrel/skin intersection.
The other plies remained in the formed condition, nesting closely to the
mandrel, and showed excellent definition at the interface between stiffener web
and lower skin.

187
 Forming was next done below melt temperature because of anticipated problems
where the two packs met each other beyond the stiffener area. For the next
trial full melt temperature was achieved before pressurizing. Other changes to
the setup included lengthening the mandrels to rest next to the ramp surfaces
and widening the upper ply pack so it extended out to the ramp surfaces in all
directions. This change required notches to be cut along the edge of the ply
pack to prevent buckling and rupture of the diaphragm. Kapton tape was used to
cover the notches for additional protection.

The next forming runs were performed with the noted changes and the diaphragm
survived well up through 150 psi. Since the plies were above melt temperature,
good consolidation was achieved between the upper and lower packs. The rupture
occurred along the edge of the mandrel in a notch location that allowed the
film to over elongate and burst. A large percent of the plies remained formed
to the mandrel surface along its base. Only one ply lifted and bridged away
from the radius area of the formed plies (Figure 22). The inside of the
stiffener shape revealed very good contact between the plies being formed and
the base of the mandrel even with loss of the diaphragm. Photomicrographs of
the area show that the upper plies dragged the lower plies in toward the
mandrel and formed wrinkles in the lower skin.

In an attempt to alleviate dragging of the base ply pack, the upper ply of the
base pack was extended to run under the ramp areas of the tool. This change
would maintain pressure on the top ply to allow slippage of the two ply packs
without wrinkling. Also, a fiber glass cloth (picture frame), was placed
around the ramp areas and over the mandrel to cover any areas that could
potentially allow the diaphragm to rupture. During this run the diaphragm
ruptured in a gap between the ramp and forming box causing incomplete forming
of the element. However, less ply slippage was noted.

Due to the frequency of rupture of the UPILEX diaphragms, an aluminum (SUPRAL)

diaphragm was selected for further trials. The aluminum diaphragm offers
greater elongation capabilities not only at processing temperatures but also
at temperatures below the melt temperature of the PEEK resin.

During the first run with an aluminum diaphragm, the pressure was applied at
550°F, (below the melt temperature of the thermoplastic resin). Applying the
pressure at this low temperature allowed the lower plies to slip prior to a
viscosity change of the resin. During this fabrication attempt, the top ply of
the lower ply pack was extended beneath the forming ramps in an attempt to
"lock" the ply in place, thus avoiding wrinkles. After applying pressure (120
psi) at 550°F, the temperature was increased to 750°F and held for 30 minutes.

The result was a stiffened panel with good surface quality but with bridging in
the radius. NDT results revealed a porosity free part in the flat areas. How-
ever, photomicrographs revealed the lower ply pack wrinkled. Since the upper ply
of the lower ply pack wrinkled and the ends were contained beneath the forming
ramps, the ply obviously split between the fibers of this outer 45 ° ply.

Following review of the results of the run, two changes to the manufacturing
process were identified to alleviate the wrinkling problem in the next run.
The next attempt will incorporate a neat resin film between the two ply packs

188
to serve as a lubricant. This will reduce frictional forces to allow the two
contacting plies to slip past each other. Another potential solution is to
change the two contacting plies from 45 degrees to 90 degree orientations.
This will increase the strength in the direction of slippage and reduce
friction.

Blade Frame Elements

Using two aluminum block details a blade panel was hand laminated by bending
and edge tacking each of seven plies with a soldering iron (Figure 23). The
fillet was filled with thin strips (.30" to .90") of ITX unidirectional tape
using a sharp cone tip on the soldering iron, Figure 24. A flat skin was
preconsolidated and a strip grit blasted across the center where the blade
attached (Figure 25). The two angles with fillet in place were inverted onto
this skin, Figure 26, and vacuumed bagged to a project plate. There was a
released UPILEX film between the angle plies (web) and the aluminum details.
Upon consolidation (Figure 27) this configuration did not show acceptable c-scan
results. The web area had many depressions in it that appeared to be oriented
along the second ply down from the surface, i.e., normal to the surface ply
fiber direction.

Outgassing from the release coated UPILEX and the lack of ears on the vacuum
bag at the base (which may have prevented sliding of the blocks) were
identified as probable causes for the poor consolidation. As such, a second
blade was fabricated with no UPILEX on the tool details and with extensive ear

folds in the vacuum bag. Nondestructive inspection of the second blade

revealed porosity in the radius areas. Although the part quality was improved
over the first blade, it was not the level desired. After a careful review of
the part it was evident one of the tool details had slightly rotated during
consolidation.

The consolidation tools are presently being modified to permit a positive

control of the details. A trimetric view of the modification is shown in
Figure 28. Keyways will be milled into the ends of the web details and fit to
keys in the end plates. This modification will maintain the movement of the
detail in the direction desired, thus maintaining constant and equal pressure
across the part surfaces. In addition, an upper slotted plate will maintain
minimum differential vertical displacement between the tooling blocks for the
back-to-back L-sections which comprise the T-section.

189
 CONCLUSIONS

Based on work to date, the following conclusions have been made:

Selection of the upper cover allows for the demonstration of two

promising cost-effective manufacturing approaches, fiber placement and
single diaphragm/co-consolidation, which have applications in a majority
of the remaining generic fuselage section.

The producibility analysis indicated that the selected manufacturing

approaches show potential for low cost fabrication of the upper cover.
The accuracy of the analysis will be verified during subcomponent
fabrication.

The use of an aluminum diaphragm during SDCC verification trials

prevented diaphragm rupture due to increased durability and elongation
properties compared to available polymeric films.

During SDCC pressure can be applied below resin melt temperature without
fear of diaphragm rupture since aluminum diaphragms provide sufficient
elongation at those temperatures.

Based on blade fabrication attempts, it is anticipated that control of

the tooling movement through selected keyways will guarantee cap and web
thicknesses and supply adequate pressure in the fillet area.

o Quality, thick (I in.) panels (for lug elements) can be successfully

fabricated.

o Abrasive waterjet cutting can be used to efficiently machine thick

panels.

o A lug analysis program capable of investigating through-the-thickness

effects showed good correlation to experimental results.

REFERENCES

l.
Hoffman, P.L.R. and Gibler, J.A., "Design for Manufacturing, Producibility
Issues for Thermoplastic Composites," presented at SAMPE Spring Meeting,
Anaheim, CA., April 4, 1990.

.
Wilkins, D.J., "Anisotropic Curved Panel Analysis," Report FZM-5567,
Air Force Contract No. F33615-69-C-1494, May 1973.

.
Goering, J., "Initial Impact Damage of Composites," Material Science
Corporation Report MSC TFR 1801/1207, March 1987.

190
 URE BULKHEAD

KEEL

INLET DUCT

FLOOR
CARRY-THRU BULKHEAD

FRAME
LONGERONS

Figure I Generic Fuselage Section

191
 2

(/)
O
o

[] Simple Curvature
O [] Complex Curvature
cc
"lO

E
O
z

0
Hand
SDCC FP Hand SPF-DB
Lay-up Lay-up
TP TS

Figure 2 Recurring Cost Comparisons

(/)

0
o
t-

O
• Simple Curvature
t_ [] Complex Curvature
t-
O
z
"0
(D

0
z

0
SDCC FP Hand SPF-DB Hand
Lay-up Lay-up
TP TS

Figure 3 Non-Recurring Cost Comparisons

192
 Phase A Subcomponent

Ultimate Loads

Condition Nx (lb/in) Ny (lb/in) Nxy (lb/in) P (psi)

SSPU, 0.95 MACH, SL, 9.0 g 2500 +100 0 1.0

SSPD, 0.95 MACH, SL, -3.0 g -800 +I00 5.0

RPO, 0.95 MACH, SL, 7.2 g 2000 +100 +500 5.0

Fuel System Over
Pressure Malfunction 0 22.0

Figure 4 Subcomponent Designed for Actual Flight Loads

193
 6.50 Typ

[_45,,o,o,,o,±45] ! ,L
(.Md_mensions shown are in Inches.)

[ ], k_,.=__J
Figure 5 SDCC Subcomponent Configuration

IA _0 Typ

26.00

[+45190101901+45] -_
(AI dimensionsshownare in inches.)

Figure 6 Fiber Placement Subcomponent Configuration

194
Y

Interface n- 1

Plyn--

z
> X

Z-O

4_

: Inteff*ace 2
Interface I
P; 2

Figure 7 Subparametric Laminated Solid Element

I. Hole to_orance _o

2. Denotes percentages of O's, + 45"s and 90"s tespecdvely.

3. Ply lay'ups for the sublamlnato d.g sequences
sta_ing _qu,nco= are .,I as
_ follows:
* l "_ ) If/
(s_4o/lo): (%14..%
,,_ 1.4.%/o
zj4s,_.4s,%)s

(echo/to): [oz/4smz/-4.._/4._
/4s%/-4-_%_o1=
1.4._ lois _ /
_ _ Notes: _ j_,.4sz,So14sloI.,_R
90"= mspecdvely.
(2o/7o/lo): [4s2_4sz,so/4..% s

_, 10.0

Sublamlnete Stacklng Dletdbutlon 2,'I

Lug Speclmene Ouantlty Each Hole Dlemotor 1
Sublamlnete 1 Sublamln-ta 2 Sublamlnate 3

Static 1 1.00 (50/40/10) (50/40/10) (50/40/10)

Static 2 1.00 (50/40/10) (30/60/1 O) (60/30/10)
Static 3 1.00 (sor3o/lo) (so/4o/1
o) (20/70/10)
Static 4 1.75 (5u4uto) (50/40/10) (50/4 O/1O)
Static 5 135 (50/40/10) (3o/6ulo) (60/30/10)
Static 6 1.75 (6u3ulo) (50/40/10) (20/70/10)

Figure 8 Lug Element Tests to Verify Analytical Methodology

195
 SYMMETRY
PLANE

c
7
.e
ONE EL.
W

.-c
2 000-
3;
I

.
m
I
X
4-ool -

-0.02
0 20 40 60 80

Ply No. (Thickness)

Figure 9 Convergence Study Indicates Four Elements Through-the-Thickness

a r e Necessary

Figure 10 Lug Element Under Test

 i_TT
_,G Specimen Quantity Hole Dia. Pred. Failure Load Test Results
Each (inches) (kips) (kips)

#1 4 1.0 64.2 60.1 (76.7)

#2 3 1.0 65.9 62.3 (74.3)

#3 4" 1.0 57.3 62.2 (74.7)

#4 4 1.75 66.9 69.3

#5 4 1.75 66.5 68.7

**
#6 4 1.75 61.5 66.8

* Initial Bearing Failure Load (Final Failure Load)

** Failure Load

Figure II Lug Test Results Showed Good Agreement to Predictions

6000
/_ LUG # 1 - 3

4000

_ 2000

[..,
o
r_

-2000 "'_'%'_'_'_._.,. st. #1

""%: f':'_"%:; :'2 "':;':':'_"'_'

-4000
' I ' I ' I '
0 20 40 60 80

LOAD(kips)

Figure 12 Strain Data Indicates Point of Bearing Failure

197
 F i g u r e 13 Typical Lug Bearing Failure

Figure 14 Typical L u g Net Section Failure

198
 60.0
°. I_ 0.12 RTyp

| . |

l
,-
L 1.50 --_

7.00 Ref
!
60 ° Y-Frame Specimen

1.50

O.125 R Typl

0.36---_

L. 8,ooRof_1
Blade Specimen

45.0°_ O.12 R Typ

t j

,- 7.00 Ref

45 ° Y-Frame Specimen

Figure 15 Frame Element Test Specimens Will Determine Viability

of Fastenerless Moldline Designs

199
 L( )AD

Blade
%

I \ I

\
SNn

Static
Framo
Configuration
RTA E'i'W"

Blade 3 3

Y(a = 45) 3 3

Y(a = 60) 3 3

255 ° F

LOAD

Loading Block
X
\
Y-section
Steel Plates /

• , *..L..L_.
I I I
I
/
Skin
Load Adapter

Figure 16 Pull-off Test Method Will Ensure Failure In Critical Regions

200
 Thru

Figure 17 Fiber Placement Tooling Concept

__._ Adhesive
1st Ply /--- Aluminum

Mandrel Bonded
i to Tool

_Heat
Blanket
..................................... I

, F,,,,.__ Insulation

Figure 18 Hat Stiffener Embedded In Fiber Placement Tool

201
 n S Upper Platten

Ply Pack
rl I I D--_-_Vacuum Ring
Neet Film

Detail A Aluminum Mandrel---_ _-- Vacuum Box

i-Ply Bu_d--Up Under Mandrel _// _

'_
Lower Platte_ _
I Aluminum Lower Tool

Figure 19 SDCC Tooling Concept

UnkF='ectional ?
TOW Fillet.--/

Figure 20 Unidirection Tow Used In Fillet Area to Assure Part Quality

202
 Ramps
Y-Mandrel for Fastenerless
,Rae ?!ate
Frame Eiement Fabrication
Box

Mandrels for Subcomponent

Manufacturing Verification Segmented Locator Plate

Figure 21 SDCC Element Tool

Figure 22 Initial Y-Frame Element Experienced Bridging in Radius

203
 Figure 23 Blade L-Section Plies Hand Laid O n Tooling Blocks

Figure 24 Tow Material Was Placed In Fillet Area

204
Figure 25 Base Plies Consolidated, Separated and Grit Blasted Prior
To Assembly

Figure 26 Assembly P r i o r To Autoclave Consolidation

205
 Figure 27 Intra Blade Elements Contained Porosity In Radius Due
T o Insufficient Pressure

Figure 28 Modifications To Blade Element Tooling T o Ensure P r o p e r P r e s s u r e

I n Radius

206
 STRUCTURAL EVALUATION OF CURVED STIFFENED COMPOSITE PANELS
FABRICATED USING A THERM-X sm PROCESS

Christos Kassapoglou Albert J. DiNicola Jack C. Chou

Structures Research Section
United Technologies Sikorsky Aircraft
Stratford CT

Jerry W. Deaton
NASA Langley Research Center
Hampton, VA

INTRODUCTION

The use of composites in aircraft structures is often limited by material and

manufacturing costs which, for some designs and applications, are prohibitively
high. To increase the frequency of application of composites in primary
airframe components alternative manufacturing processes are sought that reduce
cost and/or enhance structural efficiency.

One alternative process involves the use of THERM-X sm as the pressure transfer
medium during autoclave curing. THERM-X sm, a silicon- based flowable polymer
which behaves like a liquid under autoclave presssure, transmits
quasi-hydrostatic pressure to all contacting surfaces of the part to be cured.
Once the autoclave pressure is relieved, THERM-X sm reverts back to the powdery
solid state and can be reused many times.

The THERM-XSm process to be evaluated is depicted in Figure 1 and consists of

(a) enclosing the tool and part to be cured by a set of frames that create a
box, (b) pouring THERM-X sm powder onto the part and filling the box, and (c)
placing a vacuum bag over the box assembly. In this program, a separating
non-porous film (Teflon) was placed between the part to be cured and THERM-X sm
powder to avoid any contamination.

The use of THERM-X sm has two significant advantages over conventional

manufacturing procedures. First, it eliminates complicated hard tooling since
it guarantees uniform pressure transfer and thus good compaction at complex
structural details (such as frame-stiffener intersections and corners).
Second, it greatly simplifies vacuum bagging, since once the part to be cured
is covered by THERM-X sm powder, the vacuum bag need only conform to a
relatively flat shape reducing significantly the number of pleats required.

A program is on-going at Sikorsky Aircraft to evaluate the structural

performance of complex composite fuselage structures made with this THERM-X sm
process and to quantify the impact of THERM-X sm on manufacturing labor hours
and cost. The program involves fuselage panel optimization analysis, a
building block test program where structural details representative of the
full-scale article are analyzed and tested, and static and fatigue
test/analysis of the full-scale test articles. The main results of this program
are reported in this paper.

2O7
 Surrounding
box THERM-X sm

ZJo. 7;o_..<_. . >,-O-o. h o .o. r. o,_ ,,,"

• o 01 "_ .D e ee o

Skin Blade
stiffener

sm

Figure 1. Illustration of THERM-X Process

DESIGN SELECTION

An airframe construction representative of both helicopter and fixed wing

structure was selected in order to demonstrate the general applicability of the
results of this program. Several structural members were evaluated by
estimating the impact of using THERM-X sm versus conventional manufacturing.
The most common detail with the largest cost savings due to THERM-X sm
processing, a curved panel with cocured frames and stiffeners, representing
helicopter tailcone and fixed wing fuselage panels, was chosen (see Figure 2).
The selection procedure is described in reference I.

A simple method was developed to optimize the stiffened panel. In this process
the skin thickness, frame, and stiffener spacing, and frame and stiffener area
and moment of inertia were treated as variables and the weight and cost were
minimized subject to loading constraints. The loading constraints were the
following: (I) Applied loads were shear and compression (the latter along the
stiffeners), (2) Panel failure occurred at a predetermined ultimate load, (3)
Buckling of each bay and the panel as a whole occurred at a preselected load
combination (fixed postbuckling factor), (4) No material used would be below
minimum gage.

Panel failure of the postbuckled panel was determined as first-ply-failure of

any of the structural members under the applied loads. A Tsai-Hill stress
interaction criterion was used in conjunction with a maximum stress failure
criterion. The latter was used to give an idea of the failure mode since, for
first ply failure, a transverse tension failure of zero degree plies is very
conservative. No such failure mode was noted.

208
 _ L _I_'_ _

Nxy
Parameters
• Geometry Nx---_
_ I ds
' -q_-- Nx _
- Spacing --_ b
- Cross section
- Thickness
• Materials
J,
a -_ Nxy

Figure 2. Stiffened Panel Usage in Aircraft Structures

The cost was a combination of material cost ($50/ib for prepreg material) and
labor hours ($30/hr). Previous Sikorsky experience was used to estimate labor
hours required and the effect of increasing stiffener and frame spacing on the
total number of manufacturing labor hours. It was estimated that for each
additional stiffener or frame, the manufacturing labor hours for the entire
panel would increase by 13% for conventional manufacturing and 8% for THERM-X sm
processing. The 5% difference is due to the reduced bagging complexity of
THERM-X sm processing especially around intersecting members such as frames and
stiffeners. For each panel configuration then, the cost was the sum of the raw
material cost and the cost to manufacture that particular configuration. The
latter comprised of manufacturing cost for the skin and the cost to fabricate
the frames and stiffeners which took into account the 5% cost difference (per
added frame or stiffener) between the two manufacturing approaches. For
simplicity in the calculations, the panel was assumed flat and square with 30
inch sides.

The iterative optimization and sizing scheme was applied to various materials
and loading configurations. This process is shown schematically in Figure 3.
The cross-sectional area to spacing ratios for the stiffeners and frames (As/ds
and Af/df respectively) are treated as independent parameters. These ratios
are independently selected and the steps outlined in Figure 3 followed until
convergence is reached and the panel weight is minimized. Then, another set of
As/ds and Af/df values is selected and the procedure is repeated. The pair of

209
 I,ssume,s,_s
I
v

I AssumeAf/df, t [

m_

angle
Assume alpha
postbuckling I

I M,ncos,-_s,d,
! Increase t or
t stiffener/frame web
thickness

Optimize number of
I Bay buckling _ ds
halfwaves m_n
V
I Minimum weight _ t/hf I

Overall panel buckling ---IP'hs/hf I

Optimize no of halfwaves m,n I
t
I Calculate
t, hs, bs, all
hf,parameters:
bf, ds, df [

t
._ Find m,n for
number min buckling
of halfwaves load
same? I
[

t_
Compare current with previous
parameters. Same?

Y
Y
Failure check; failure? J

N
Calculate angle alpha
same as before?

_Y

I Min gage
below min check;
gage? I Y

Figure 3. Stiffened Panel Optimization Procedure

210
The As/ds and Af/df value finally selected is the one providing minimum weight for
all design variations.

A comparison of THERM-X sm processing versus conventional manufacturing for

typical helicopter tailcone and fixed wing fuselage loads is shown in Figures 4
and 5. For the helicopter tailcone configuration the ultimate loading design
requirement was 250 ib/inch in compression and 250 Ib/inch in shear (selected
based on S-76 tailcone ultimate design loading). For the fixed wing
configuration the ultimate loading was 2500 ib/inch in compression and 1250
ib/inch in shear corresponding to typical fuselage loads [2].

The (normalized) cost to manufacture stiffened panels for typical helicopter

tailcones as a function of stiffener spacing ds is shown in Figure 4. The
frame spacing df is determined by the optimization procedure (cost minimization
equation) to be very nearly equal to 3.2 ds. Two cost curves are shown, one
for standard manufacturing and one for THERM-X sm processing. For each geometry
configuration, the cost is calculated as the raw material cost plus labor hours
to manufacture based on previous Sikorsky Aircraft experience. It is important
to note that the minimum cost configuration involves few frames and stiffeners
(ds = 5 in.) of large area and moment of inertia and with thick skin, thus
corresponding to a relatively high weight. For the loading considered here,
the minimum weight configuration would be a minimum gage configuration (with
minimum gage thickness for frame and stiffener webs) corresponding to a
stiffener spacing less than 1 inch. Thus, the lightest configuration is labor
intensive because it involves many stiffeners.

A tradeoff between weight and cost can then be established. At small stiffener
spacings the panel weight is low but the manufacturing cost is high. At high
stiffener spacings the cost is low but the panel weight is high. An
equilibrium between the two driving quantities (weight and cost) can be found
by considering the premium in dollars per pound (termed value of improved
performance in Figure 4) the customer is willing to pay to reduce the
structural weight by one lb. For example, for UH-60 (BLACKHAWK) helicopters,
that value is $750/ib.

In this context, since the minimum weight (still meeting the loading
requirements) is that corresponding to a minimum gage design, any other
acceptable configuration will have a potential weight penalty equal to the
difference in weight from the minimum gage configuration. By multiplying this
weight difference by the weight premium dollar value (termed here value of
improved performance), an upward sloping curve (with increasing ds) is obtained
that shows the weight penalty for each configuration translated to dollars.
Various curves corresponding to different values of improved performance are
shown in Figure 4.

The points of intersection of the weight penalty curves with the cost curves
define optimum points, each corresponding to a different selection of
manufacturing process and dollar value of improved performance. Any
configuration away from the intersection points implies that for the
particular manufacturing method selected, either the weight or the cost of the
panel can be reduced and still meet the load requirements and the selected
value of improved performance.

211
 Value of improved
Normalized cost Dimensions: 30" x 30" performance
L_ Loading: Nx:-250Ib/in
Nxy=2501b/in
1000 $/Ib

750 $/Ib
Optimum point
\] for std. panel

500 $/Ib

\ (standard
\ manufacturing)
__ 300 $/Ib
Optimum point for "_" :ost"
THERM-XSM (THERM-X sM tooling)
panel

Current production
Min. gage
cutot configuration
\
0 1
Stiffener spacing ds (in)

Figure 4. Determination of Optimum GeometrE for Stiffened Panels in

Helicopter Structures

212
 STIFFENED PANEL OPTIMIZATION
Weight Normalized Dimensions: 30" x 30"
(Ib) cost Loading: Nx= -2,500 Ib/in; Nxy = 1,250 Ib/in

Frame spacing • 30"

Stiffeners fail
under Region of acceptable
14- 2.8 m

geometry
,,_compression _

t \ /
12- 2.4
t
I \
/
I \
/
I /
I \ /
10- 2.0
/
Cost standard
Weight /
manufacturing \ \ (both methods) /
\ \ /
/
1.6
J o_°_l°mmm°m_° I='= _m°ma_°°mma_ml_lllmasml mR_
/
Cost THERM-X sm %%% . /
processed ........_.'_ /
_=_ °°'=

1.2
eoeeo_--Im_Imu _ _ _
/
/
/
/
0.8
/
/
/
/
0.4 m
/
m

Min gage /
cutoff /
/
/
0 I I I I [ I
0 1.0 2.0 3.0 4.0 5.0 6.0 7.0
Stiffener spacing (in)

j I I I 1 I I I
0 3.6 7.2 12.0 16.2 21.5 26.0 30.1

Frame spacing (in)

Figure 5. Determination of Optimum Geometry for Stiffened Panels in
Fixed Wing Structures

213
 The configuration selected for the full-scale article corresponds to a
stiffener spacing of 6.5 inches. This corresponds to a value of improved
performance of $300/ib which was felt to be more representative of commercial
fixed wing transport. At that spacing the THERM-X sm process results in panels
approximately 10% less expensive than conventionally manufactured panels. At
smaller stiffener spacings (for higher values of improved performance) the
savings can be as high as 22% (ds= 3 inches). It should be noted that these
savings do not include savings in tooling. The THERM-X sm process requires
relatively simple tooling even for complex parts with cocured frames and
stiffeners.

The effect of applied loading can be seen if Figure 4 is compared to Figure 5.

In Figure 5, the optimization process was applied to panels with loadings
representative of fixed wing transport fuselages. In this case, the minimum
gage configuration is not attainable since the loading is high. The limiting
factor at low stiffener spacings is the compression failure strength of the hat
stiffeners assumed in this case to be 36000 psi which corresponds to the first
ply failure load of a predominantly 45 degree stiffener web layup. This is
shown by the left vertical curve at ds=3.5 inches.

Two cost curves are shown in Figure 5 much like the ones in Figure 4. In this
case however, the weight change between the minimum weight configuration (at ds
= 3.5 in.) and any other acceptable configuration is so small that the value of
improved performance should be higher than $10000/ib for the weight penalty
curves to intersect the cost curves. For that reason, the weight penalty
curves are not included. Instead, to show the tradeoff between cost and
weight, the minimum weight of the panel for each value of stiffener spacing is
shown as an upward sloping curve. The corresponding frame spacings are also
shown in Figure 5.

Figure 5 suggests that the minimum weight configuration corresponds to ds=3.5

inches while the minimum cost configuration corresponds to ds=3.75 inches for
THERM-X sm processed panels and 4.75 inches for conventionally manufactured
panels. The user will then have to make a choice on which configuration to
select favoring either a minimum weight or a minimum cost design.

The conclusions can change significantly if the compression strength of the hat
stiffeners is increased. That would move the left cutoff line to the left
increasing the number of acceptable configurations. This would also increase
the savings of the THERM-X sm processed panels from approximately 6% (at ds=5
inches for example) to over 15% (at ds=3.0 inches).

As a final comment on the optimization study, the current approach does predict
that the commonly used configuration of ds=6 inches and dr=20 inches in
aircraft structures is one of the acceptable configurations but corresponds to
a higher weight configuration than can be attained with lower ds values. It
should be borne in mind that the current process assumes a flat panel and
neglects the stiffening effect afforded by curved panels. This effect would
yield optimum configurations with ds values higher than currently predicted,
thus closer to the commonly used value of ds=6 inches.

The configuration selected based on the optimization process (ds=6.5 inches,

df=20 inches, Nx=-250 ib/in, and Nxy=250 ib/in at ultimate) was evaluated with
a detailed test and analysis program that is described in the next section.

214
 BUILDING BLOCK TEST PROGRAM

A representative sketch of the curved and stiffened composite panel under

consideration is shown in Figure 6. In order to provide a basis for
verification of predicting failure modes and the panel's ultimate load, the six
building block tests also shown in the figure were performed prior to
full-scale testing. Detailed results of these building block tests were
reported previously [I]. Comparable, and often enhanced, strength and
stiffness values were noted for THERM-XSm processed test specimens versus those
conventionally manufactured, and excellent laminate quality control was
attainable with substantially less effort. A summary of the data generated
during the tests, which will verify analysis and assist in predicting ultimate
load for the full-scale test, is presented in Figure 7.

Compression after impact tests were performed for the two damage regimes
envisioned for the composite fuselage panel: low speed-high mass impact and
high speed-low mass impact. The former is representative of "tool drop" style
impact damage and the latter is characteristic of in-flight impact damage.
Shear after impact tests were performed under low speed conditions only [i]
since the trends of high versus low speed impact established for the
compression specimens are expected to apply in this instance.

Skin/stiffener_L -
separation
#
F] Skin
/LJ tearing 1_3 Compression after

°s,;,;n% _., _ L_J impact

_'_X / I --_ h r r
< _ _]_ Sea affe
1
_" /k +_ impact

I IIII I
il IIII I
/
Frame/stiffener
30'_ 0''
intersection

Y Hat top:
F T F
(_+45/0/_+45)

Skin: (-+ 45)3 (fabric)

-El F T F Hat stiffeners: 11_ " t webs: (_+452)

Frames: 2.0._(+45/O2/_+45) s

,% 4" l FT
(+45/O 2/_+45)
Hat bottom: (+45/0)
F T F

Material F: C3K/5225 Fabric (0.0075", 0.015" thick)

T: C12K/5225 Tape (0.012" thick)

Figure 6. Building Block Approach and Full-Scale Article Configuration

215
 Building Block Number
Configuration Results
Evaluation Specimens

Skin tearing 8 Tensile modulus: 10 MSI

Tensile strength: 99 KSI

Stiffener crippling 6 Crippling strength:

14,628 PSI

t
Skin/stiffener 5 Pull-off strength:
separation 90.5 Ib/in

Compression 10 Residual strenght vs

after impact impact energy and
indentation at
point of impact for high
and low speed impact
(See fig. 8-10)
Shear after impact 5

Frame-stiffener 3 Failure load at the

intersection intersection of the frame
and Hat stiffener :
614 Ib/in

Fiaure 7. Summary of Bu Iding Block Approach Results

216
A summary of the average normalized compression after impact (CAI) strength
values is presented in Figure 8 for both velocity regimes and fabrication
procedures. Conventional manufacturing appears to provide moderately superior
CAI strength at both 600 and 1200 in-lb/in impact energies (15% and 3.3%
respectively). The slight advantage afforded by conventional manufacturing was
noted for both impact velocities. The underlying reason for the strength
discrepancy is currently being investigated. The only significant difference
between the two manufacturing methods, whlchmay account for the residual
strength discrepancy, is that the cure pressure for THERM-X sm processing is
twice that used during conventional manufacture (I00 psi versus 50 psi). More
tests are needed to quantify these differences with statistical significance.

Internal damage resulting from impact as measured by ultrasonic C-scan is shown

in Figure 9 for both fabrication procedures and velocity regimes. Consistent
with other literature citations [for example references 3 and 4], the high
speed impact event produced greater levels of internal damage than low speed
impact for a fixed energy level. For 600 in-lb/in of impact energy, THERM-X sm
processed specimens exhibited greater internal damage area (by 21%) whereas at
1200 in-lb/in of energy both fabrication procedures yielded similar amounts of
internal damage.

There is evidence that indentation at the point of impact is an effective means

to correlate a measureable quantity to resultant post-impact strength [I]. A
comparison of average indentation for conventionally manufactured and THERM-X sm
processed specimens is shown in Figure I0. Although indentation data for high
speed impact of THERM-X sm panels were not available, the consistency of the
data at hand indicates, irrespective of either velocity regime or manufacturing
procedure, indentation may be a reliable way to predict resultant strength
after impact.

1.0 1 m
Th-X Hi Speed
Th-X Lo Speed
Conv. Hi Speed
Conv. Lo Speed
0.6 Trend (Th-X)
0 .,,.
Trend (Conv)

0.4 -

0.2 -I

, ,, , I ,, ,, I,, , , I, ,, ,I ,, ,,I ,, ,, I, ,, ,
0.0
0 200 400 600 800 1000 1200 1400

Impact Energy (in-lb/in)

Figure 8. Compression after Impact Strength as a Function of Impact

Energy

217
 8.0

J
< 6.0 J
[] J

J
z[]
_._
a "_ 4.0
C J • [] Th-X Hi Speed
0 / / • Th-XLoSpeed
I [] Cony. Hi Speed
° 2.0 o Conv. LoSpeed
-- Trend Hi Speed
' -- Trend Lo Speed
0.0 • , , , I , , , , I, , , , I , , . , I', , , , I , , , , I , , , '1

0 200 400 600 800 1000 1200 1400

Impact Energy (in-lb/in)

Figure 9. Variation of C-Scan Indicated Damage Area with Impact
Energy

0.040
O
[]
Q. [] Th-X Lo Speed
E
m Conv. Hi Speed
N,.,.,.

O 0.030 [] Conv. Lo Speed

.B Trend
O
13.
@ 0.020
C
O []
=m

0.010
"O
N

0.000
0 200 400 600 800 1000 1200 1400
Impact Energy (in-lb/in)

Figure 10. Indentation at Point of Impact as a Function of Impact

Energy

218
Flat Frame/Stiffener Intersection Specimen

The frame-stiffener intersection specimen is used to provide experimental

evaluation of failure mechanisms present in the full-scale article at the
intersecting corners of frames and stiffeners, a link between flat and curved
specimens, and to support corresponding analysis predictions.

The specimen is shown in Figure Ii. It represents two bays of the full-scale
curved stiffened panel. A closeup of a typical frame/stiffener intersection
corner is shown in Figure 12. Excellent consolidation and radius definition is
evident. Part quality around the shear tie which consists of the outer plies
of the frame web cocured on the hat stiffener webs is also very high with
accurate placement and contour definition. The layup of the skin, frames and
stiffeners is identical to that of the full scale article as shown in Figure
6. Aluminum doublers 0.5 in. thick and 3.0 in. wide were used for load
introduction fastened on three-ply graphite/epoxy doublers that were cocured
with the specimen. The aluminum doublers formed a picture frame fixture for
testing the specimens in shear.

The finite element model used is shown in Figure 13. One specimen end is
loaded in tension (along a diagonal) and the opposite end is fixed. MSC
NASTRAN SOL 66 geometric nonlinear solution was used to determine the buckling
load and post buckling behavior of the panel. The model consists of 606 grid
points, 576 CQUAD elements, and 3601 degrees of freedom.

Comparison of Test Results to Finite Element Predictions

The strain gage data obtained from the frame-stiffener intersection specimens
was compensated for gage transverse sensitivity and percent reinforcement
(resulting from gage bagging and adhesive material) following procedures
recommended by the gage manufacturer (Micro-Measurements Division, Measurement
Group Inc., Raleigh NC) and reference 5.

For the type of gages used (CEA-03-063UR-350) the transverse gage sensitivity
is insignificant (only 1% change to the apparent strain). The percent
reinforcement effect however, for the materials and layups used, ranges from
0.6% to 15.8% (depending on the gage installation such as back-to-back or
single face, laminate thickness, and open face versus encapsulated gage
configuration). The results reported below have this correction included
wherever it is considered significant (more than 5%).

The strain gage locations (total of 18 rosettes) were chosen to give a detailed
strain distribution throughout the specimen and in particular at skin bays and
near the frame-stlffener intersections. Finite element predicted surface
strains are compared to test results at various panel locations and load levels
in Figures 14 through 17. The locations are (I) Hat Stiffener Center (Figure
14), (2) Frame-Stlffener Intersection Corner (Figure 15), (3) Bay Quarter Point
(Figure 16), and (4) Bay Center (Figure 17). At low applied loads (except for
the frame-stiffener intersection location) and high loads close to the failure
load (in all cases), the finite element predictions are in very good agreement
with the experimental results. At intermediate loads the correlation ranges
from poor (bay center and frame stiffener intersection corner) to excellent
(hat stiffener center and bay quarter point).

219
 F i g u r e 11. F l a t F r a m e / S t i f f e n e r I n t e r s e c t i o n Specimen ( S t i f f e n e d
Side Overview)

F i g u r e 12. Frame-Stiffener Intersection Detail

220
The shadow moire method was used to monitor the out-of-plane displacements of
the panel during the test. The first moire fringe pattern appeared at an
applied load of 2600 ibs and is shown in Figure 18. The postbuckling mode
shape just before the panel failure load of 20000 ibs is shown in Figure 19.
The analysis of the photographs of the moire fringe pattern follows standard
procedures outlined in the literature [6].

Typical experimental and analytical results for the out-of-plane displacement

along the panel skin bay at the applied load of 16000 ibs is shown in Figure
20. The moire measured amplitude correlates well with the finite element
prediction. However, the wavelength of the deflection mode shape is less than
the finite element prediction. The discrepancy between finite elements and
moire pattern data is attributed to local eccentricities of the specimen and
resulting differences in load transfer.

The failure prediction for these specimens was obtained by determining the most
highly loaded element in the finite element model and using the forces and
moments on that element as input in a first ply failure criterion. That
element coincided with the location where a crack initiated (near the bay
corner) during testing. Using mean material allowables the failure prediction
using a stress interaction criterion [7] is 26000 Ibs of applied load. The
corresponding B-Basis prediction is 22950 ibs. The test failure load (average
of two specimens) is 21000 ibs (614 ibs/in). The failure predictions are based
on allowables for conventionally manufactured parts and are off by 9 to 24%
(B-Basis versus mean allowable predictions). There are two reasons for the
discrepancy: (I) Loading of the first test specimen was stopped when the first

SKIN

lee
- 18.76"
\

Figure 13. Finite Element Mesh for Frame/Stiffener Intersection

Specimen

221
 I ÷ FEM BAG SIDE
0 TEST BAG SIDE
7 -
n FEM TOOL SIDE
ATEST TOOL SIDE
- I _ ROSETTE
i
E

F-
5
-
/ F/% \ -
4
C C
L 0

3
0
Z

2
£

_J

0
T
I
-1 t I I I I t I I I I I I I I I I I I I I I

0 2 ¢ 6 8 10 12 14 16 18 __20
(T_u_=nd=)
APPUED LOAD (Ibs)

Fi gure 14. Comparison of Finite Element Predictions to Test Results

at Hat-Stiffener Center (Shear Strains - Intersection
Specimen)
+ FEM 90 ° (perpendicular to stiffener)
0.9
0 TEST 90 °
018
n FEM 0 °
07
z_ TEST O °
0.6
ROSETTE
05

04.

c 0,3

.=_-_ o.2
o_ 0.1
bo
-- _ 0
C
0 --131 boo o OoO00 _ -°--------0 _------ o _ ---__ _
cF-
,-v --0.2

--0,4,

--0.5

--0._

-0.7

--0.8

--0.9

--1 I I I ! I I I I I I ! I I I i I I | I I

0 2 4. 6 8 10 12 14 16 18 ."20
(TFwmu_nd_)
App,_dLooa,Ob,)
Figure 15. Comparison of Finite Element Predictions to Test Results
at Frame/Stiffener Intersection Corner

222
 + FEM BAG SIDE
7- <>TEST BAG SIDE _>
[]FEM TOOL SIDE
A TEST TOOL S IDE J+
_=- I _ ROSETTE _J
C

5-

4
C
"6 o_

0
(-

2-

I I I I 1 I I I I I I I ! ! I I I ! I

0 2 4 6 8 10 12 14 16 18
(Tl',ou_nds)
.._i,_ Lood Oh,)
Figure 16. Comparison of Finite Element Predictions to Test Results
at Bay Quarter Point (Shear Strains - Intersection
Soecimen)
8
+ FEM BAG SIDE
OTEST BAG SIDE
7 _ [] FEM TOOL SIDE
| ATEST TOOL SIDE O" _, Ojrl

ROSETTE
A

5-
.c_ o
0---.
L

.._ o
C :l 4 - 0 ° 4) _._,,_o

b
tu 2
t-
G'i

A
• --"-_A A ,.,

"" & "" A h, '&

-1 I i i i i i i i t i i i J I I i i _ t i

0 2 4. 6 8 10 12 14 16 18 20
[Thousand,,)
_op,_ LoodOb_)
Figure 17. Comparison of Finite Element Predictions to Test Results
at Bay Center (Shear Strains - Intersection Specimen)

223
 Figure 18. First Shadow Moire Fringes on Frame/Stiffener
Intersection Specimen (2600 1 b s )

Figure 19. Shadow Moire Fringes Near Failure of Frame/Stiffener

Intersection Specimen (20000 lbs)

224
 0.15

0 0
0.1
¢.-
U []
c
[]
oo5
,_1
r, o
i/i
0
Z

IM
--0.05
6.
0
p-

0 --0.1
O El3

• FEM
--0,15

|
o M01R{

! I I I I I I |
i
| I I I ! | I I I |

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

_KIN I_Y IJ3NGITUDINAL DIST.,inch:

Figure 20. Comparison of 0ut-of-Plane Deflection Shape at Bay Center

Along Stiffener Axis
cracks developed (at 20000 lbs) in order to see where failure started. The
load capability may have been significantly higher as is indicated by the
failure load for the second specimen (22000 ibs). Thus, the average test
failure load of 21000 ibs may be conservative. (2) Based on post-test
examination, final failure was determined not to result from corner cracking
(that was noted in the specimen) but rather from high local strains in the
vicinity of the root of the hat stiffener (near the frame/stiffener
intersection) due to the buckled shape. This is verified by the shadow moire
fringes (Figure 19), which do not cross the centerline of the specimen but stop
where the stiffener webs meet the skin. The fringes, which indicate
out-of-plane deflection, tend to come close together in the vicinity of the
stiffener. This implies a large displacement gradient is present in this area
and the associated high bending moments precipitated final failure. The
existence and location of this high strain area was confirmed by the finite
element analysis.

FULL-SCALEARTICLE

The full-scale article (shown in Figure 6) was manufactured by laying up the

skin in an aluminum tool that was surrounded by an aluminum frame similar to
that shown in Figure I. The hat stiffeners were laid up around teflon mandrels
and placed in position on the skin. The frames were laid up on a specially
made tool and then lowered into place in the assembly. Two aluminum cross
members, one for each frame, were used to keep the frames in place during
curing, in a manner similar to the tool shown in Figure I. The outer web plies
of the frames opened up to accommodate the hat stiffeners going through and

225
 Figure 21. Full-scale Test Setup

served as shear ties between frames and stiffeners. The top skin ply was
placed l a s t , covering all stiffener and frame flanges (embedded flange concept).
The full-scale panel was tested in shear using a picture frame fixture (Figure
2 1 ) . Axial load was applied to diagonally opposite ends of the fixture in
order to introduce the desired shear loads through the structure. The finite
element modelling procedures were the same as for the frame-stiffener
intersection specimens. The model had 2530 nodes and 7590 degrees of freedom
with 2 4 2 4 CQUAD4 and 136 CBEAM elements.

A comparison of strain gage data near a frame/stiffener intersection (corner of

outer bay) to finite element predictions is shown in Figure 2 2 . The shear
strain at a point inside one of the bays (quarter of the distance between the
two hat stiffeners) is shown in Figure 2 3 . The axial strain along the frame
axis at the center of one of the outer bays is shown in Figure 2 4 . In all
cases, test and finite element analysis are in good agreement up to 10000 to
12000 lbs of applied load (postbuckling factor of about 3 ) . The differences at
higher loads are due to local failures that occurred (manifesting themselves
with loud noises and sharp increases in the deflection gage measurements) and
redistributed the load. These local failures were not modelled by the finite
element model. The deflection pattern over the whole panel was monitored by
shadow moire. In addition, a deflection gage positioned at the center of one
of the two center bays was used to measure deflection locally. The
out-of-plane deflection at that location is compared to the finite element
predictions in Figure 2 5 . For the same load, the deflections predicted by
finite elements are 20-30% less than test results through panel failure.

Figure 25 shows that the bay buckling load predicted by finite elements is in
excellent agreement with the test result of 4000 lbs ( 9 4 lbs/in). A more

226
 z
2.5
z

___o 2-

1 4-
"1-
"t-
.< + +
n, 1.5-

+/+ _FEM BAG SIDE

+TEST BAG SIDE
_ROSETTE
o.5-
0 [ "_ "Jr" | T I |
I I I I I I 1

0 4 8 12 16 20 24
(Thou_nds)
._pUED LOAD (LeS)
Fi gure 22° Comparison of Finite Element Predictions to Test Results
at Hat-Stiffener Center (Shear Strains . Full Scale Panel)

detailed finite element analysis (up to buckling) with the same mesh but with
smaller load increments showed the predicted bifurcation load to be 102 ib/in
or 8% higher than the bifurcation load indicated by the deflection gage during
test. It was a snap-through buckling where the skin, up to that point
deflecting in the direction of the panel curvature, reversed direction with a
jump in deflection of more than an order of magnitude. The average failure
load of 23500 lbs (554 lbs/in) gives a postbuckling factor (failure load to
buckling load ratio) of 5.9.

The failure progression as observed during test and inferred from examination
of failed specimens was as follows: (I) Upon loading to 18000-19000 ibs, fiber
cracking and matrix splitting were observed at one of the loaded corners of the
specimen (see Figure 26 point 0). This is consistent with the starting crack
observed during testing of the flat frame-stiffener intersection specimen and
is at approximately the same location. Like the flat frame/stiffener
intersection specimen, the crack on the full-scale article stopped after
growing to a size of 3-4 inches. It is believed this crack initiated as a
result of the test fixture pinching the lower corner of the specimen and served
as a stress relief mechanism. The conjecture that this crack did not cause
final failure was verified by the fact that each test specimen failed at
significantly higher load, 4000-5000 Ibs after the corner crack developed.

At 23000 to 24000 ibs of applied load, new fiber and matrix cracks were noted
to initiate as a result of large displacement gradients caused by the buckled
shape at the frame and hat stiffener intersection (see point F in Figure 26).
These cracks, accompanied by some delamination, progressed from the

227
 []

4-
4 _

z n

Z._

o_

---b
P
ffl
2 -

0
/ + 4-
+ +
4-
-I-
+

4-

+ + 4- B FEM BAG SIDE

+ TEST BAG SIDE
ROSETTE

o I_m I 1 I ! I | I I I I

0 4 8 12 16 20 24
(Thoumr_s)
APPUED LOAD (LBS)

Figure 23. Comparison of Finite Element Predictions to Test Results

at Bay Quarter Point (Shear Strains - Full Scale Panel)
4

4-

3.5 I 4-

_9 2.5

Z_
e}

oF

vO !.5
2] 4-
+

F- I
U_ o FEM BAG SIDE
÷ TEST BAG SIDE
0.5 ROSETTE

-43.5 I I I I I I - - -| ....... | - " _f I-...... | ....

0 4 8 12 16 _ 24
(T_-_nd_)
APPUED LQAD (LBS)

Figure 24. Comparison of Finite Element Predictions to Test Results

at Bay Center (Shear along Frame Axis - Full Scale Panel)

228
 20000.0
..Q

c;
O
-.I

10000.0
(3
ILl
_1
0-
O.

FEM

I , I , I , I , I , I
0.0 ==
-0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6

OUT-OF-PLANE DEFLECTION, Z INCHES

Figure 25. Load versus Deflection Plot at Bay Center (Finite Element
Prediction and Test Measurement)

229
 intersection along the edge of one of the hat stiffeners, propagated across the
hat stiffener, and continued to final failure as marked by separation of the
specimen into two large sections. The resulting fracture pattern is als0 shown
in Figure 26. It should be noted that the flanges of tb.e frames and stiffeners
were embedded in the skin and no stiffener separation from the skin was noted.
This is different from the common failure of these panels when the flanges are
bonded directly on the skin (see for example reference 8) and shows that this
configuration is effective in altering and delaying stiffener-to-skin failures
in such panels.

Failure of the full-scale panels is predicted using the results of the flat
frame-stiffener intersection specimens which showed very similar failure mode.
As is shown in Figure 27, the shear strains at the bay center for the two
specimens are very close to each other up to a load of 12000 ibs. At that
point, the full-scale panel diverges probably due to a change in the mode shape
that essentially reversed the buckling pattern. It is believed that the
strains at the location where final failure started for both flat and curved
specimens are similar and thus the loads (in ibs/in of shear) at which internal
strains reach the material allowables should be the same for both types of
specimen. The failure load for the flat specimen then should be a reasonable
approximation to the full-scale article failure load. As already mentioned,
the failure load for the full-scale article was 554 ibs/in which is 11% lower
than the value of 614 ibs/in that the flat specimen failure would predict.

SUMMARY AND CONCLUSIONS

A program is on-going to evaluate the structural performance of composite

fuselage structure fabricated using an autoclave THERM-X sm process. A building
block approach is used to isolate failure modes and quantify load paths and
failure loads for the full-scale article which is a curved skin panel with
cocured hat stiffeners and frames with a tee-shaped cross section. Tests at
the coupon and element level have shown that the THERM-X sm processed parts have
comparable stiffness, strength, and failure modes. The only instant where the
THERM-X sm processed parts showed moderate inferiority (up to 15%) to
conventionally manufactured parts was in high speed-low impactor mass
compression after impact test.

Finiteelement predictions of deflection shape and strains in the flat element

level tests are in good agreement with test results except at some locations in
the panel where at intermediate loads the test results suggest a difference in
the postbuckled deflection shape.

The curved stiffened panel performed very well, exceeding the design ultimate
load of 250 lbs/in by over a factor of 2 and failing at a postbuckling factor
of 5.9. The finite element predictions are in good agreement with the test
results up to a postbuckllng factor of 3. At higher loads the test results
suggest that local failures and changes in mode shapes took place that were not
accounted for by the finite element analysis. The failure mode involved cracks
starting at a frame stiffener intersection but no flange separations were noted
(due to the use of embedded flanges) nor any shear tie failures. This suggests
that failure is driven by in-plane skin failure in the vicinity of frame/
stiffener intersections where stress concentrations will be present. More work
is needed to better quantify this failure mode.

230
 9

-1

-2
0 4 8 12 16 20
(Thous onds)
APPLIED LOAD. ( l b s )

Fiqure 27. Shear Strain Comparison for Intersection and Full-scale

Tests (Bay Center)

231
 Failure predictions for the curved panel can be obtained using the failure load
from the flat element level test which had a similar failure mode. That

prediction is 11% higher than the curved panel failure load.

The fact that the finite element analysis assuming the THERM-XSm processed
parts had the same properties as conventionally Llnufactured parts showed good
agreement with test results up to the point where local failures and mode
changes not accounted for in the finite element analysis became significant,
suggests that undamaged THERM-X sm processed parts are structurally equivalent
to conventionally manufactured parts.

THERM-X sm processing enables manufacture of high quality complex parts with

corners and tight radii with minimum tooling and at low cost.

ACKNOWLEDGMENT

This work was performed under NASA contract NASI-18799. Jerry W. Deaton is the
technical monitor. The authors wish to thank S. Garbo, G. Schneider, and A.
Dobyns for their comments and suggestions regarding failure of the stiffened
panels.

REFERENCES

I. Kassapoglou, C.; DiNicola, A.J.; and Chou, J.C.: Structural Evaluation of

Composite Fuselage Structure Fabricated Using a THERM-X sm Process. Proceedings
of 46th American Helicopter Society Forum, Washington DC, 1990, pp. 671-682.

2. Boeing Structural Test Components Presentation, Workshop on Common

Structural Test Components for ACT, Newport News, May 1990.

3. Cantwell, W.J.; and Morton, J.: The Influence of Varying Projectile Mass on
the Impact Response of CFRP. Composite Structures, vol 13, 1989, pp. 101-114.

4. Cantwell, W.J.; and Morton, J.: An Assessment of the Residual Strength of

an Impact Damaged Carbon Fibre Reinforced Epoxy. Composite Structures, vol 14,
1990, pp. 303-317.

5. Error Due to Transverse Sensitivity in Strain Gages, M-M Technical Note

TN-509, Micro Measurements Division, Measurement Group, Inc, Raleigh, NC,1982.

6. Dally, J.W.; and Riley, W.F.: Experimental Stress Analysis, McGraw Hill,
Inc, 1978, chapters 12.5 and 12.6.

7. Tsai, S.W.: Strength Characteristics of Composite Materials. NASA OR-224,

April 1965.

8. Rouse, M.: Postbuckling and Failure Characteristics of Stiffened

Graphite-Epoxy Shear Webs. 28th SDM Conference, Monterey CA, April 1987, AIAA
paper 87-0733.

232
Noise Transmission Properties and Control Strategies for Composite Structures

Richard J. Silcox, Todd B. Beyer, and Harold C. Lester

NASA Langley Research Center

Abstract

A study of several component technologies required to apply active control techniques to

reduce interior noise in composite aircraft structures is described. The mechanisms of noise
transmission in an all composite, large-scale, fuselage model are studied in an experimental
program and found similar to mechanisms found in conventional aircraft construction. Two
primary conditions of structural acoustic response are found to account for the dominant
interior acoustic response. A preliminary study of active noise control in cylinders used
piezoceramic actuators as force inputs for a simple aluminum fuselage model. These
actuators provided effective control for the same two conditions of noise transmission found in
the composite fuselage structure. The use of piezoceramic actuators to apply force inputs
overcomes the weight and structural requirements of conventional shaker actuators. Finally,
in order to accurately simulate these types of actuators in a cylindrical shell, two analytical
models are investigated that apply either in-plane forces or bending moments along the
boundaries of a finite patch. It is shown that the bending model may not be as effective as the
force model for exciting the low order azimuthal modes that typically dominate the structural
acoustic response in these systems. This result will affect the arrangement and distribution of
actuators required for effective active control systems.

Introduction

The expanded use of composite materials for primary aircraft structures is evidenced by the
Boeing 360 and tiltrotor programs and new business aircraft. In commercial vehicles of this
type as well as aircraft with conventional construction, the acoustic environment is an
important element for passenger acceptance. Therefore, the understanding and control of the
vibration and acoustic transmission properties of composite structures is an important element
which will promote the widespread use of composites in large scale commercial aircraft. It is
expected that the integrated nature of the composite skin with the frame, along with the
decreased weight compared with conventional construction, will provide for less structural
damping and potentially higher overall acoustic and vibration levels.

Although finite element models have now been developed for composite layered media, most
of this work has been directed towards determining the static properties of structures. Little
work has been done towards developing finite element methods for built-up structures
responding to high frequency dynamic inputs. To date, finite element methods have not even
found widespread use for conventional structures at frequencies applicable to noise
annoyance. This is due to the complexity and cost of developing and running the required
numerical models. Additionally, these numerical approaches provide little insight into the

233
structural/acoustic coupling mechanisms controlling the noise transmission process. A
simpler approach that effectively models the gross shell motions and acoustic response is a
more effective first step.

A number of simplified modal models have been developed that provide a basic
understanding of the noise transmission mechanisms of aircraft structures.l,2, 3 These
techniques have successfully modelled the low frequency mechanisms of acoustic-structural
coupling determined from experimental evaluations of simple fuselage models.4, 5 These
works have shown that the acoustic response of interior fuselage spaces is dominated by a
limited number of low order spatial acoustic modes that are generally excited by off resonant
response of the fuselage structure. These simple models do not take into account the
anisotropic material characteristics of composite structures. However, it has been shown that
the individual resonant structural and acoustical behavior of composite structures is similar to
that of conventional isotropic structures at low frequencies. 6 Therefore, these analytical
modal models may be applied to composite structures with the qualification that the structural
modelling may be less than realistic.

Passive noise control techniques have traditionally relied on stiffened structures or structural
and acoustic damping to increase the transmission loss. The increased weight associated
with these techniques can effectively offset the weight savings due to composite structures.
Active control technology, however, has emerged as a realistic alternative for efficient control
of the interior noise in propeller driven passenger aircraft. Flight testsT, 8 and ground tests9, lo
of active control systems have shown that a 10 to 15 dB attenuation of the global interior noise
is attainable with either acoustic or vibration control sources. Experience with distributions of
acoustic control sources has shown that large numbers of sources are required in order to
provide control over a wide range of conditions and engine harmonics. The use of force
inputs has provided nearly equally effective control with significantly fewer actuators. This
reduction in number of actuators, however, requires more intelligent placement in order to
couple efficiently into the range and order of structural acoustic modes encountered over the
aircraft operational range.

Another aspect relating to active control is the type of actuator and sensor elements used to
implement the active control system. In reference 9, 10 Ib conventional shakers acting against
their own inertial mass were used to generate the necessary control forces. This approach
has obvious weight and force limitations on real aircraft. For this reason, piezoceramic
elements are currently being evaluated for this application as they input strain energy directly
into a localized region of the structure and their weight and cost are negligible.

The purpose of this paper is threefold. First, to illustrate the dominant mechanisms of noise
transmission for a large-scale composite fuselage model. Second, to examine an approach
for active noise control using piezoceramic actuators as demonstrated on an aluminum
cylinder. And third, to present and examine two analytical models of piezoceramic actuators
to apply either in-plane forces or bending moments to the structure. This work is part of a
larger ongoing effort at Langley Research Center to apply active controls concept to
ameliorate the noise and vibration environment of aerospace vehicles.

234
 Experimental Configuration

In this section, two experimental efforts are outlined. The first describes a large scale
composite fuselage structure excited by exterior acoustic sources. The shell and interior
cavity responses are presented and discussed. The second describes a preliminary
experiment in which piezoceramic actuators bonded to the side wall of an aluminum fuselage
model were used to effectively control the interior noise due to a simple propeller model. This
second experiment has led to a current interior noise control effort using similar transducers
on the full scale composite shell described above.

Fig. 1.- Composite cylinder in Fig. 2.- Cross section of

.anechoic room. composite cylinder.

Composite Cylinder Noise Transmission Tests

Figure 1 shows the test cylinder mounted on a stand in the large anechoic chamber of the
Acoustics Research Laboratory at NASA Langley Research Center in Hampton, Virginia. This
test configuration provided an environment for a comprehensive mapping of the exterior and
interior sound fields in a free field environment.

The aircraft fuselage model used in the current study is a filament wound, stiffened cylinder
1.68 m in diameter and 3.66 m in length. The composite material of the cylinder shell consists
of carbon fibers embedded in an epoxy resin. The ply sequence of the cylinder skin is
T45/T32/90/~32/~45 for a total thickness of 1.7 mm. The cylinder is stiffened (longitudinally)
by 22 evenly spaced composite hat-section stringers. The stringers pass through 10
composite J-section ring frames spaced 0.381 m apart. The ring frames and stringers are tied
together with a clip and all elements of the fuselage are rivet-bonded together. A schematic of
the cylinder cross section with detail of the stringer-frame geometry is shown in Figure 2. A
12.7 mm thick plywood floor is installed 0.544 m above the bottom of the cylinder. The
supporting beams and posts for the floor are made from aluminum extrusions. An aluminum
clip ties the floor to the shell at discrete locations. The plywood floor is bolted to the aluminum
supporting beams. Rubber gaskets and silicon rubber sealant fill the gaps between the floor

235
edge and the cylinder structure in order to acoustically isolate the spaces above and below
the floor. Additional details of the composite cylinder may be found in reference 11.

The cylinder endcaps were constructed from three layers of 32 mm thick particle board with a
3.2 mm wide groove cut out for the end of the cylinder to rest in. The endcaps are sufficiently
massive so that any airborne sound transmission through them is negligible compared to the
sound transmission through the cylinder sidewall.

The acoustic source used to excite the system was a point source located at 8=90°, x=0.333t
and 0.2a from the shell outer surface. The coordinate orientation is shown in Figure 2 with
positive x into the paper. Here I is the length of the cylinder, 3.66 m and a is the radius. The
source was assumed to approximate a propeller noise with known temporal and spatial
characteristics. The shell response was measured with an azimuthal array of 22 mini-
accelerometers equispaced around a circumference at x=0.333t, the source plane. The
interior pressure field was measured using six 12.7 mm condenser microphones mounted
along a radius. These microphones could be traversed both azimuthally and axially such that
a complete mapping of the interior acoustic field was obtained. The reader is referred to
reference 12 for additional details.

Piezoceramic
control
transducer
1 / Microphones

/
Lead
vinyl
\
\
Removable
11 Exterior
source
damping floor

Fig. 3.- Photograph of experiment rig and Fig. 4.- Schematic of aluminum
piezoceramic actuator. cylinder test apparatus.

Interior Noise Control using Piezoceramics

Figure 3 is a photograph and Figure 4 is a schematic of the test arrangement of the second
experiment consisting of an aluminum cylinder, 0.508 m in diameter, 1.245 m long, and 1.63
mm thick. The floor was 0.381 m wide and consisted of thin aluminum skin attached to a
lattice structure. The cavity below the floor was filled with acoustic foam in order to inhibit
acoustic resonances in this space which might complicate the system response. Propeller
noise was simulated by a 60 W horn driver attached to a tapered horn whose outlet was

236
positioned 76 mm from the exterior of the cylinder. All tests were performed at single pure
tone frequencies. The interior pressure field was measured by three 12.7 mm microphones
mounted on a movable traverse. The results will be presented as sound pressure level
contour plots located in the source plane. Additionally, the structural response was measured
by an array of 24 uniformly spaced mini-accelerometers attached to the cylinder in the source
plane. In order to simulate a free-field environment, the experiments were performed in an
anechoic chamber at NASA Langley Research Center.

To generate control inputs, two bimorph piezoceramic actuators (see inset of Figure 3) of
dimensions 50.8 x 12.7 x 0.51 mm were bonded to the exterior of the cylinder in the source
plane at -90° and 45 ° (90 ° corresponds to the acoustic source location). A bimorph actuator
has two co-located piezoceramic elements driven 180 ° out-of-phase in order to produce
surface bending. A reference signal was used to drive the acoustic noise source and the
same signal was passed through a two channel manually operated phase shifter. The control
signals were then amplified, passed through transformers with a voltage gain of 7:1, and
connected to each bimorph element. The experimental procedure was as follows: The
exterior noise source was driven at the desired frequency and level. The amplitude and
phase of each control signal (for some tests only one channel was used) were adjusted to
minimize the interior sound levels at one and/or two error microphones located at fixed
positions in the interior cavity. In practice, a computer based adaptive controller could be
used to perform this function. However, the manual system used here was more convenient
for the purpose of these tests. Once the error signals were minimized, the interior field was
mapped using the traversing microphone array. The control signal(s) were then turned off
and the interior noise of the primary field mapped.

Piezoceramic Actuator Models

In order to take effective advantage of any type of control actuator, reasonable analytical
models must be developed and studied. Previous actuator models for plate and beam
applications have assumed that the piezoelectric material occupies a small region of the total
plate.13 With shells, however, the piezoelectric material has been assumed to occupy an
entire layer of a multilayered shell. 14 That is, simple piezoelectric actuator models for shell
applications have not yet been developed. In the present work, two plate type actuator
models are coupled to a finite length, isotropic cylinder model developed in reference 2. The
displacement response of the cylinder and the interior acoustic pressure are expressed as
modal expansions in the characteristic functions of each physical system. For more details on
this model, the reader is referred to reference 2 which defines the response of a simple
fuselage model due to adjacent acoustic point sources (simulating a propeller) and to a point
force exciting the structure directly. In order to evaluate the effect of the piezoceramic
transducer elements, extensions to reference 2 are currently being evaluated.

Two piezoceramic patch models have been formulated for active vibration/noise control of
cylinders. These models, which are described below, represent adaptations of piezoceramic
models previously developed for flat plate applications, 13 where in-plane and bending (out-
of-plane) deformations are uncoupled. Hence, these piezoceramic models are most likely
valid only for patches whose dimensions are small relative to the radius of curvature.

237
The bending model simulates the effect of two out-of-phase piezoceramic patches attached
on opposite sides of the cylinder wall. By driving the two piezoceramic patches out-of-phase,
a normal stress distribution is produced which varies linearly through the thickness of the
shell. This approximates a state of pure bending about the middle surface of the shell.
However, some extensional deformation is produced since the out-of-plane and in-plane
deformations in a shell are coupled due to curvature effects. The essence of the modeling
involves replacing the piezoceramic patches by a uniform, line moment applied along the
perimeter of the patch area. The amplitude of the moment is proportional to the piezoceramic
supply voltage.

An in-plane piezoceramic model was also developed in which the adjacent patches are
driven in-phase creating primarily an extensional deformation of the shell's middle surface.
Again, because of coupling (due to curvature) some out-of-plane (bending) deformation of the
cylinder is produced. It is this out-of-plane motion which couples with the interior acoustic
space. With the in-plane piezoceramic model, the patch is replaced by a uniform, in-plane
line force distribution applied along the patch perimeter. In this case, the force amplitude is
proportional to the piezoceramic supply voltage.

Results

Results are presented in three sections. In the first section, representative shell and cavity
acoustic responses will be used to illustrate the noise transmission characteristics of the
composite fuselage model. The second series of results will illustrate the use of piezoceramic
transducers as actuators on a simple aluminum fuselage model. Finally, predictions using the
previously described analytical models for piezoceramic-patches will show the differences in
the wave space mode spectra response of the shell model.

Composite Cylinder Response

0 Figure 5 compares the spectra recorded by an

interior microphone (at r=0.924a, e=-84 °,
136 216 -- Microphone
I .... Accelerometer x=0.333t) and an accelerometer on the
exterior shell (at e=-82 ° and x=0.333t) of the

g
-10 I
-20 composite fuselage model. These frequency
response functions are measured with respect
to a fixed microphone mounted at the source
_ -40 :, ,,w:l ,o face. Both levels are then normalized to their
I
respective individual peak levels. The exterior
-50
I acoustic source used pseudo-random noise to
excite a single acoustic monopole mounted
-60
_: I 1 I I
I
0.2a from the shell wall at e=-90 ° as discussed
100 200 300 400 500 600 700 801 9O0 in reference 12. These frequency response
Frequency, Hz functions indicate a strong modal response
across the spectrum particularly for the interior
Fig. 5.- Typical interior pressure and pressure. The interior microphone response
shell acceleration spectra retains the sharp peaks out to the highest

238
frequency of 900 Hz. The accelerometer response begins to smooth out above 500 Hz
indicating that the modal density is increasing such that individual modes no longer dominate
the shell response. Note that the peaks of the vibration and acoustic responses do not in
general coincide at many frequencies. At those frequencies at which they do correspond, a
resonant shell response drives an off resonant acoustic response. There are however many
strong acoustic responses that correspond to relatively weak shell vibration responses.
These frequencies correspond to co-incidence frequencies where the spatial wavelength of
an off resonant shell mode is nearly equal to that of a strongly coupled resonant acoustic
cavity mode. Often, the shell mode that drives the acoustic response is not the dominant
response in the shell. However, it is the one that most strongly couples to the acoustic cavity
response. As will be shown, the acoustic cavity response is typically dominated by spatial
distributions that correspond to individual modes of the interior cavity geometry.

Two cases representing typical coupling mechanisms will be examined in detail. These are
noted in the frequency response functions of figure 5 as the two frequencies of 136 Hz and
216 Hz. At 136 Hz, the shell has a strong resonant response and forces an off resonant
response of the interior cavity. At 216 Hz, the opposite is true, an off resonant shell mode
couples strongly into a resonant cavity mode.

Figure 6 shows a plot of the shell vibration distribution plotted at 18 time increments over 1
cycle of vibration. This data is expanded from the 22 accelerometer measurements and
illustrates the motion of the shell at 136 Hz. A strong modal response characterizes the upper
cylinder response and corresponds with a structural mode defined in reference 6. The floor
structure forces nodes at e=+110 ° and the under floor response is reduced due to the
00
.-: :._::_i_._.:.:...._ !:_:!:._._.:._..::
.:cii:i:i_:i_iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii_iiiiii:ii:_:::
............,..-............,,....,...........................

..... ::iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii!
i_iiiiiiiiii!i!i!iiii!_i!i!iiiii!iiiiiiiiiiiiiiiiiiiii!ili!..

::_:_:_:_:_:_:_:_:_:_:_:i:_:i:i:i:i:i:i:i:i:i:i:i:i:i:i:i:i:i:i:i:i:i:i:i:i:i:i:i:i:i:i:_:i:_:_:i::::::::::::::::::::::::::::::::::::::::::::::::::::::_
________.........______._..._..._`_....`_._......`_..._._._`.._..._..._._`_..__.___..._._
........................
:_:_:+:_:_:_:_:_:+:_:_:`:`:`:_:_:_:`:`:_:_:`:_:`:`:_:_:_:_:_:_:_:`:_:_:_:_:`:`:`:`:`:_:_:_:+:_:_:+:_:_:_:
:::::::::::::::::::::::::::::::::::::::::::
:!$i:i:i::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

;;i!!!!!i_i!;!i!!_:_:_':=_'
a) x = 0.333J _:
-108 ° 108 °

39 SPL 74
+

b) e : 0.0 °
Fig. 6.- Shell vibration distribution for Fig, 7.- Interior cavity pressure distribution
exterior monopole excitation of 136 Hz. due to monopole excitation at 136 Hz.

239
stiffening effect of the floor and its supporting structure. This strong resonant response of the
structure forces a relatively uniform interior response as illustrated by the pressure
distributions of figure 7. Here, the pressure contours are plotted for the source cross section
at x=0.333/in figure 7a and an axial/radial distribution at e=0.0 ° in figure 7b. Because the
excitation frequency does not correspond with any resonant cavity mode, the result is a
combination of forced off-resonant responses. The overall interior cavity response is relatively
uniform with the variations that do exist correlating with the shell vibration distribution.

.................................. ;:_:::!::::!:-::.,:

• ..... %_1iiii_i:iiif::ii:_ii:::iiiiiiii_i_:::ii_i_i_iiiiiiiiiii_:::::?:?:::iiiiii_iiii_
:_::.,

..::_:::i:_:_:i:_:_:_:_:_:i:_:i:_:_:_:i:i:_:!:i:i::::.:..
........ :::":4i:"":__iiii'_
.,..,.,._._.::_:i:i:i:i:i:i:i:i:i:_:_:_:_:_:_:_:_:iii
_ _:'_ ........

.)) :::::::::::::::::::::::::::::::::::::::::::::::

a) x = 0.3331 :'_':':'_!!_!_!iiiii''!!ii_!_i_
-108 ° 108 °

39 SPL 74

:i_iiii_iiiii_i_i_i_iiiiiiii_iiiii_iiiiiiiliiiii. ' .::iiiiiiiiililili!iiiiiiiiiiiiii!iiiiiiiiiiiii!i!!!iiiiiiiii!iiiiiiii. ' . .:_::ilili!iiiiiiiiiiii_i_iii!i!iiiiiiill

b) 0 =0.0 °
Fig. 8.- Shell vibration distribution for Fig. 9.- Interior cavity pressure distribution
exterior monopole excitation at 216 Hz. due to monopole excitation at 216 Hz.

The second coupled response condition is shown in figures 8 and 9. The shell response
distribution is illustrated in figure 8 and is shown in correct relative scale to the accelerometer
response at 136 Hz in figure 6. At this frequency, the shell is not responding in a resonant
condition as evidenced by the magnitude of the response relative to figure 6 as well as the
accelerometer spectra of figure 5. The shell vibration distribution does not correspond to any
of the resonant modes of reference 6 and is thus inferred to be a combination of off-resonant
modal responses. The interior acoustic response of figure 9 is, however, a strong resonant
response. The microphone spectra of figure 5 illustrates a strong peak at 216 Hz and the
cavity pressure distribution of figure 9 corresponds to a resonant acoustic mode of reference
6. The strong nodal lines in the source cross section of figure 9a illustrate the dominate
behavior of this acoustic mode. The exterior source is in this cross section at e=-90 ° adjacent
to the peak interior acoustic response. The axial/radial pressure contour shown in figure 9b is
for e=0.0 °. This variation displays a cos(3_x/t) modal response. For this case, the acoustic
pressure spatial distribution is not well correlated to the shell vibration distribution. This is due
to the filtering effect of the structural acoustic coupling mechanism. Most of the shell vibration

240
modes couple poorly to the interior acoustic space. Only those modes that couple efficiently
must be considered in any active control scheme. This simplifies the control task and makes it
necessary to control only a limited number of interior cavity modes.

Active Control with Piezoceramic Actuators

Although active control results for a composite cylinder have not yet been obtained, an
aluminum cylinder with the floor installed was excited at two frequencies that were
characterized by a structural and an acoustic resonance. At 240 Hz, the structure is on a shell
resonance and the acoustic field is being forced in an off-resonance response. This is similar
to the condition for the composite cylinder at 136 Hz (figures 6 and 7). The interior acoustic
pressure contour in the source plane cross section due to only the primary acoustic source is
shown in figure 10a.

Using a single actuator located at e=-90 ° as indicated by the controller in figure 10b, the
control input was adjusted to minimize the interior acoustic field as measured by the error
microphone at e=-90 ° and r=0.925a. The resulting controlled acoustic field is shown in the

0°
source

(a) Primary "60_ O°

"120° _2.150o_......._L_..._ 150o O°

180 °
Controller

Acceleration level (dB)

Primary source
..... Controlled
(b) Controlled

Fig. 10.- Interior sound pressure level Fig. 11.- Shell radial acceleration distri-
for two source conditions at 240 Hz. bution for two source conditions
at 240 Hz.

241
contour plot of figure 10b. The peak acoustic levels are reduced on the order of 10 dB. This
was also found to be the case for the out of source plane response with a consistent global
reduction of about 10 dB.

Normalized shell acceleration levels in the source plane corresponding to the above primary
and controlled conditions are shown in figure 11. The vibration response for the primary
source alone is predominantly a distorted cos(2e) mode. The reduced response for the shell
under the floor is thought to be due to foam damping packed under the floor. Compared to
the vibration distribution for the controlled case, the single piezoceramic actuator is seen to
reduce the vibration level by over 10 dB over most of the circumference. Only at a few
isolated angles has the vibration level not been reduced. The residual response appears to
be of a much higher azimuthal order. It appears that the controller has forced a significant
reduction in the level of the dominant structural mode which is on resonance. This has
produced a corresponding reduction in the forced acoustic response in the shell interior
space.

The 687 Hz frequency of the second case corresponds to a cavity acoustic resonance as was
the case for the composite cylinder in figure 9. The pressure contour arising from the primary
acoustic excitation is shown in figure 12a. This antisymmetric mode has a strong nodal line
aligned with the vertical radius and out-of-phase antinodes on either side as shown in this

0°

(a) Primary
,i o o
,oo
troller -120°____12.150
° 150 ° 0°

Controller 180 °

Acceleration level (dB)

Primary source
..... Controlled
(b) Controlled

Fig. 12.- Interior sound pressure level Fig. 13.- Shell radial acceleration distri-
for two source conditions at 687 Hz. bution for two source conditions
at 687 Hz.

242
source cross-section contour plot. For this case, control was exercised using both
piezoceramic actuators placed at e=-90 and e=+45 degrees as noted in the contour plot for
the controlled case of figure 12b. Peak reductions of 10 dB were obtained, although with
somewhat less overall reductions. Again, however, reductions were obtained throughout the
entire volume.

The normalized shell radial acceleration distribution corresponding to the previous primary
and controlled cases is shown in figure 13. In this case, the dominant shell vibration motion
does not couple effectively into the interior cavity space. One of several modes comprising
the shell vibration however does couple effectively into the resonant cavity mode response.
The effect of the controller is to couple effectively into this mode such that the overall interior
pressure is substantially reduced. However, from figure 13, it is seen that the vibration levels
in the shell have increased significantly. This increased motion of the shell does not
effectively couple into the interior acoustics. This phenomenon may be taken as modal
spillover, an effect of the limited number of control actuators and sensors allowing extraneous
modes to be excited in the shell because they are not sensed by the error sensors. In cases
such as this, distributed arrays of sources and sensors (on the shell) will be needed in order to
control the overall shell response.

Piezoceramic Actuator Models

In this section, cylinder displacement modal response spectra are presented for each of the
two piezoactuator models discussed previously. The results are for an aluminum cylinder
having the same radius, length and thickness as the composite cylinder shown in figure 1.
These results, though preliminary in nature, have implications in terms of the number and
distributions of actuators as well as the input power to achieve control. This work is ongoing
and illustrates the type of actuators currently under consideration.

The figures that follow are plots of the wave-space mode amplitude excited at 136 Hz for the
radial displacement in a cylindrical shell due to different actuator models. The radial motion of
the shell wall is the motion that couples with the interior acoustic pressure and is described by
a sin(m=x/t) cos ne variation. The in-plane motions of the shell only excite an acoustic
response by coupling into the radial shell motion. The magnitude is plotted in gray scale
normalized to the maximum level for each actuator vs axial mode order m on the horizontal
axis and azimuthal order n on the vertical axis. Because the model actuator is geometrically
located at the axial center of the cylinder (x=0.5t), no even axial modes may be excited. This
is the reason that all modes with m=0,2,4,.., are identically zero in the results to follow.

Figure 14a illustrates the modal distribution produced by the bending piezoceramic actuator
model. In this model, the actuator does not couple effectively into lower order azimuthal
modes. The modes most effectively excited are low order axial modes of azimuthal order
greater than 10. This is expected from the theory which shows the primary mode of response
for low azimuthal order is an in-plane response rather than the out-of-plane radial response.
Preliminary analysis indicates that the bending piezoceramic model couples most effectively
with the higher-order axial and circumferential modes. Hence, the bending piezoceramic
model may not be effective at low frequencies where the vibro-acoustic environment is
dominated by the low-order cylinder modes. For this reason, an in-plane piezoceramic model
was also developed.

243
 :.:.:.:.:.:
:+:.:.:.:.
:.:.:.:.:.:.,

lg _ Iiiiiiiiii!iiil
Iiiiiiiiiiiiili_i_i_i_i_i_ :::::::::::::-
::::::::::::::
:+:.:.:.:+ ii!i!i!i!_i!l
lii!iii!i!ii_l
liiiiiiii!iii!l
Eiiiiii!iiii?l
:iiiii:iiiiiii
i:i:i:i:i:i:i_

13

!il ............

.:.:.:.:.:,:.:

i
1
11

1-1 lO

IM
o
- llJll
IIIIII
0 1 Z 0 4 15 O 9' 8 g 10 1 1 11 1_ 14

(a) bending (b) in-plane force

0 --6 --12 --18
Mode Level, dB

Fig. 14.- Modal amplitude distribution for two source models, f - 136 Hz.

Figure 14b shows the modal distribution excited by the in-plane piezoceramic model. The
overall levels of the modes are down by a factor of 100 or greater. This is due to the shell
being much stiffer in its in-plane response than in its radial or out-of-plane response. The low
order azimuthal modes are responding in this case because the in-plane excitation is directly
forcing the dominant in-plane response. These results indicate that the in-plane model will
couple more readily with the lower order cylinder modes and therefore has the potential of
providing better vibration and/or noise reductions at the lower frequencies. However, this is at
the cost of reduced excitation efficiency, i.e. more control power is required.

In the above cases, the 63.5mmx38mm actuator subtends an angle of only 4.3 °. Initial
parametric studies show the bending model can excite lower order azimuthal modes by
increasing the dimension of the actuator relative to the circumference of the shell. It is
important to be able to excite these low order modes in order to couple effectively into the
modes of the primary source that are creating the interior noise. In general, these actuators
have been shown to provide effective excitation and by using distributions of tailored
actuators, it is expected that effective control may be exercised.

244
 Concluding Remarks

This paper has described an ongoing effort to apply active noise control concepts to affect the
noise transmission of composite fuselage models. Results were presented for a built-up
composite structure excited in vibration by an external acoustic source representing a
propeller. An analysis of the shell response and interior acoustic cavity responses indicate
that the noise transmission mechanisms are similar to that of metal construction and are two-
fold. First, a resonant shell mode may force an off-resonant acoustic response. For this case,
the control system may be designed to best advantage to control the dominant shell response.
It may be expected that a minimization of either the shell response or the interior acoustic
response will give equally good results. Second, an off-resonant shell mode excites a
resonant acoustic mode. In this case, the shell modes most efficiently coupling into the
acoustic response must be controlled. In general, the interior acoustic response must be
minimized rather than the shell response. Minimizing the vibration response may even give
rise to higher interior noise levels. Care must also be exercised in selecting and distributing
the actuator elements. Finally, it has been shown that piezoceramic elements are effective
transducers with many advantages over conventional force actuators. Additional work is
required to model these actuator elements, but progress is being made in understanding the
coupling mechanisms involved.

Bibliography

. Lester, H. C.; and Fuller, C. R.: Active Control of Propeller Induced Noise Fields Inside a
Flexible Cylinder. AIAA Journal, Vol. 28, No. 8, pp. 1374-1380, August 1990.

. Silcox, R. J.; and Lester, H. C.: Propeller Modelling Effects on Interior Noise in
Cylindrical Cavities with Applications to Active Noise Control. AIAA Paper No. 89-1123,
April 1989.

. Bullmore, A. J.; Nelson, P. A.; and Elliott, S.J.: Theoretical Studies of the Active Control
of Propeller-Induced Cabin Noises. Journal of Sound and Vibration. Vol. 140, No. 2, pp.
191-217, 1990.

. Silcox, R. J.; Fuller, C. R.; and Lester, H. C.: Mechanisms of Active Control in Cylindrical
Fuselage Structures. AIAA Journal, Vol. 28, No. 8, pp. 1397-1404, August 1990.

o
Fuller, C. R.; and Jones, J. D.: Experiments on Reduction of Propeller Induced Interior
Noise by Active Control of Cylinder Vibration. Journal of Sound and Vibration, Vol.112,
No. 2, pp. 289-395, 1987.

.
Grosveld, F. W.; and Beyer, T. B.: Modal Characteristics of a Stiffened Composite
Cylinder with Open and Closed End Conditions. AIAA paper no. 86-1908, presented at
10th AIAA Aeroacoustics Conference, Seattle, WA, July 9-11,1 986

. Elliott, S. J.; Nelson, P. A.; Stothers, I. M.; and Boucher, C.C.: In-Flight Experiments on
the Active Control of Propeller-Induced Cabin Noise. Journal of Sound and Vibration.
Vol. 140, No. 2, pp. 219-238, 1990.

245
 o Dorling, C. M.; Eatwell, G. P.; Hutchins, S. M.; Ross, C. F.; and Sutcliffe, S.G.C.: A
Demonstration of Active Noise Reduction in an Aircraft Cabin. Journal of Sound and
Vibration, Vol. 128, No. 2, pp. 358-360, 1989.

.
Simpson, M.; Luong, T.; Fuller, C. R.; and Jones, J. D.: Full Scale Demonstration of
Cabin Noise Reduction Using Active Vibration Control. AIAA paper no. 89-1074, April
1989.

10. Simpson, M. A.; Luong, T. M.; Swinbanks, M. A.; Russell, M. A.; and Leventhall, H.G.: Full
Scale Demonstration Tests of Cabin Noise Reduction using Active Noise Control.
Proceedings of InterNoise 89, Newport Beach, CA. December 1989, pp. 459-462.

11. Jackson, A. C.; Balena, F. J.; LaBarge, W. L.; Pie, G.; Pitman, W. A.; and Wittlin, G.:
Transport Composite Fuselage Technologymlmpact Dynamics and Acoustic
Transmission. NASA CR 4035, December 1986.

12. Beyer T. B.; and Silcox, R.J.: Noise Transmission Characteristics of a Large Scale
Composite Fuselage Model. Presented at AIAA 13th Aeroacoustics Conference,
Tallahassee, FL, AIAA paper no. 90-3965, October 22-24, 1990.

13. Dimitriadis, E. K.; and Fuller, C. R.: Piezoelectric Actuators for Distributed Noise and
Vibration Excitation of Thin Plates. Proceedings of the ASME 8th Biennal Conference of
Failure Prevention and Reliability, pp. 223-233, Montreal, Canada, 1989.

14. Jia, J.; and Rogers, C. A.: Formulation of a Laminated Shell Theory Incorporating
Embedded Distributed Actuators. ASME AD-vol. 15 Adaptive Structures, Book No.
H00533, 1989.

246
 Long Discontinuous Fiber Composite Structure
Forming and Structural Mechanics 1.

R. B. Pipes, M. H. Santare, B. J. O'Toole, A. J. Beaussart, D. C. DeHeer

Department of Mechanical Engineering and the Center for Composite Materials University of Delaware
R. K. Okine

E. I. du Pont de Nemours and Co., Inc.

Abstract

Cost effective composite structure has motivated the investigation of several new approaches to
develop composite structure from innovative material forms. Among the promising new approaches is
the conversion of planar sheet to components of complex curvature through sheet forming or stretch
forming. In both cases the potential for material stretch in the fiber direction appears to offer a clear
advantage in formability over continuous fiber systems. In the present study the authors have
established a framework which allows the simulation of the anisotropic mechanisms of deformation of
long discontinuous fiber laminates wherein the matrix phase is a viscous fluid.
The initial study focuses upon the establishment of micromechanics models for prediction of the
effective anisotropic viscosities of the oriented fiber assembly in a viscous matrix. Next, the developed
constitutive relation is employed through an analogy with incompressible elasticity to exercise the finite-
element technique for determination of local fiber orientation and laminate thickness after forming.
Results are presented for the stretch bending of a curved beam from an arbitrary composite laminate and
the bulging of a clamped sheet.
Structural analyses are conducted to determine the effect of microstructure on the performance of
curved beams manufactured from long discontinuous fiber composites. For the purposes of this study,
several curved beams with ideal and non-ideal microstructures are compared for response under pure
bending. Material parameters are determined from a separate microstructural analysis.

Micromechanics qf Forming

Models for prediction of the effective anisotropic viscosities of an oriented assembly of

discontinuous fibers (Figure 1) suspended in a power-law, viscous fluid with finite yield stress have

been developed as summarized in Table 1. The present theory for the axial elongational viscosity is

compared to a relation for non-dilute suspensions developed by Batchelor [1] in Figure 2. The

Batchelor predictions are shown to be restricted to relatively small fiber volume fractions while the

present theory was shown to be valid for volume fractions above 0.196. The models for both shearing

viscosities are shown to satisfy the bonds at zero and maximum fiber volume fraction.

Results presented in Table 1 show the influence of power-law exponent upon each viscosity

term as a function of strain rate. For finite yield stress of the matrix fluid, the effective yield stress of

the assembly was determined for the two shearing and two elongational modes of deformation. Only in

1. NASA Contract NASl-18758 Program Monitor: Dawn Jegley

247
the case of longitudinal elongational mode did the effective yield stress differ from the matrix yield
stress.

Fiber overlap length and geometry have been evaluated for their influence upon the effective

elongational viscosity of an oriented fiber array. The results indicate that the developed relations are

equivalent for both symmetric and asymmetric overlap geometries as shown in Figure 3. For variation

in fiber length and overlap length within the assembly, an integral representation has developed and

normalized frequency functions for fiber length and overlap length of zeroth and first order have been

evaluated assuming independence of the two frequency distributions. Effective viscosity calculated for

typical fiber length data shown in Figure 4 reveal that small percentages in long fibers can have a

disproportionately large influence upon results as shown in Figure 5.

Two laminate theories have been developed for the prediction of the effective anisotropic

viscosities of oriented discontinuous fiber assemblies suspended in a Newtonian fluid. The first theory
is an analogy with classical laminate theory for elastic systems wherein the laminae are constrained to

exhibit identical states of strain rate. Results presented in Table 2 for the pseudo laminate theory reveal
that the highly anisotropic nature of the oriented fiber lamina is transferred to the laminate behavior.

Indeed, laminates were constructed wherein the effective elongational viscosities are several orders of

magnitude greater than the effective in-plane shearing viscosity, while other laminates were shown to

possess effective in-plane shearing viscosity which greatly exceeded both effective elongational

viscosities. The second theory allows individual lamina to be unconstrained by one another. In all the

cases presented the values of effective viscosities of the laminates were greater for the constrained
condition than for the unconstrained condition as shown in Table 3.

The influence of fiber orientation on elongational viscosity 1"1'11(0) for the highly anisotropic

lamina is shown in Figure 6 as a function of fiber aspect ratio (I/D). These results show that the

viscosity of the lamina diminishes by three orders of magnitude for 0=1 ° and I.dD = 10,000. In Figure 7

the expected value of viscosity <rll 1> for disoriented fibers is shown. Here the probability of fibers at

orientation, 0 is equal between +01 and -01. Again note that the apparent viscosity is a strong function

of fiber misorientation.

Macr0mechani¢, of Forming

Consider the equivalent anisotropic constitutive relationship for the oriented fiber assembly
described above in a state of plane stress:

248
 /1 1 "
"_ 0

1111 2r!11

_. 1 1 0
. 221
= 21111 1122
(1)

_12J
1
0 0
_. 2TI12_

The (3 x 3) matrix in equation 1 can be inverted although the material is incompressible. The reciprocal
stress/strain rate relation is

/i11
li °0
_/=

12,,]
12 C22

0 2C66
I.8221

_.E12,,]
(2)

where

411112 4 11111122

Cll = 4 1"111 - 1122 C22= 41111 -1122

(3)

C66 = 1"112

In the present work we have followed the approach used in finite dement analysis for

viscoplastic forming of metallic material. Two differences between the case of a viscoplastic metal and a

reinforced thermoplastic must be noted. First, the constitutive equation is linear for the composite

material considered here, while a non-linear model is followed for metal forming to account for the high

strain-rate dependence. Each step of the solution scheme is solved in one iteration for the composite

material. The linear model is obviously a simplification of the problem. Average values of the

anisotropic viscosities must be chosen in the range of values typical for the forming processes

considered. The second difference is due to the anisotropy of the composite material. The flow

249
formulation and the finite-element discretization are perfectly valid for an anisotropic material.

However, the fiber orientation change must be taken into account after each stage (time step) of the
forming simulation, which corresponds to a small deformation. One can easily imagine that the fiber

orientation in a sheet will not keep its initial direction, but will vary locally if the process involves large

deformations. The solution scheme for the simulation of the sheet forming of a viscous, anisotropic
material is as follows:

• Input: initial geometry and material parameters.

• Simulation:

1. Solve for the velocity field "V" using the analogy to classical elasticity theory.
2. Update the geometry by "VAt" where "At" is an appropriate time step.

3. Update the local fiber orientation and the sheet thickness.

4. Change the boundary conditions if new points come into contact with the mold
surface.

5. Repeat the process until pre-determined deformation is reached or until contact

with mold is complete.

• Output at each step: components of stress and strain rate, updated geometry, thickness and
fiber orientation distribution.

Implementation of the Plane Stress Formulation

One of the advantages of this method is the possibility of using existing finite element codes

developed for elasticity to solve viscous creeping flow problems. Consider the case of the off axis

tensile test shown in Figure 8 a. Results for deformed shape after 20 time steps are shown in Figure 8b

and corresponding fiber orientation in Figure 8 c. Consider the next the bending of a flat sheet of

unidirectional material (Figure 9). The resulting deformed shape and fiber orientation are shown in

Figure 9 b. Finally consider the bulging of a flat plate of 45 ° fiber orientation subjected to a lateral

pressure shown in Figure 10. The results offer 20 time steps and after 59 steps are shown in Figures 10
b and 10 c, respectively.

Forming Experiments

Curved C-channel structures have been manufactured from a planar, long discontinuous fiber

laminate through a diaphragm forming process. The chosen diaphragm material is a superplastic

aluminium alloy 2004. The long, discontinuous fiber laminates consist of a PEKK thermoplastic resin

and PAN-based fibers, AS-4, that are approximately 2.2 inches in length.

To insure successful manufacture of the C-channel structure, two molding techniques were

employed simultaneously. First, PEI films were placed between the composite and the diaphragm

25O
material to act as a lubricating layer. For added flexibility, the PEI film can be used in joining

thermoplastics by means of resistance welding. The second technique was to reduce the composite

preform area, thus resulting in less excess material after molding. The size of the composite preform
was found to influence the final shape and success of the structure.

The composite was isothermally formed at 3900 C and was pressurized at 5 psi/min. (on

average) with 5 minute dwells at 20 psi and 150 psi. The maximum pressure of 150 psi was maintained

during cooling. Vacuum pressure between the diaphragms was held throughout the process.
The curved C-channel structures were manufactured from 8 ply, unidirectional laminates.

Future studies will be performed using various stacking sequences and thicknesses to determine these

parameters' effects on microstructure. The resulting components are shown in Figure 11.

Stru¢toral Micromechanics

The elastic properties of advanced composite materials depend strongly upon the orientations of

the reinforcing components utilized. To better understand the structural performance of discontinuous

fiber components the microstructural characteristics of formed parts must be determined. Toward this

end, a study was conducted to evaluate two orientation parameters; relative fiber alignment and average
fiber orientation.

Ideally, a ply of unformed fibrous composite sheet is unidirectional. However, processing

limitations create sheets of the material that are not truly unidirectional. There will be some degree of

fiber misalignment in a typical sample of unprocessed material.

During the manufacture of parts with discontinuous fiber material, some microstructural changes

occur. The formation of curved components involves the forming of the originally flat material preform

over a contoured surface. The fibers in the some portions of the material may tend to align themselves

while fibers in other regions may become less aligned during the forming process. The local elastic

properties are altered accordingly, producing a material with mechanical characteristics different than
those of unformed sheets.

In order to determine the relationship between fiber alignment and mechanical properties, a

computer program developed at the University of Delaware's Center for Composite Materials was

employed [2]. The Sheet Molding Composites (SMC) program calculates the engineering constants of

composite materials from basic input data.

The SMC program utilizes an aggregate model approach that reduces the composite material to a

collection of typical reinforcing fiber regions and orientationally averages the properties of these regions.

The relative alignment of these regions is described by a value known as Herman's orientation parameter
fp. Completely randomly oriented fibers produce a Herman's orientation of 0, while a value of 1 results

from perfectly aligned fibers. In addition to the effective aspect ratio and Herman's orientation, the

251
program requires the mechanical properties of the fibers and matrix as input. These characteristics

include elastic modulus, shear modulus, and Poisson's ratio. Fiber volume fraction of the composite
material is also needed.

The parameter fp is calculated in the following manner. The equation given below defines

Herman's orientation as a function of a trigonometric average

fp = 2 <cos 2 ¢> - 1 (4)

where

<cos 2 ¢_> = j n(¢) cos 2 ¢ dq_

(5)

and _Ois the orientation of the individual regions. Here n(¢0 is the distribution function that describes the

relative number of regions oriented at an angle ¢ with respect to the average fiber orientation. This

function can be calculated for a particular sample with a histogram. N is assumed to be the total number
of fibers counted and N(q_i) is the number of fibers falling within _i + Aq_ of the average fiber direction.

The value <cos 2 _> can then be found from the following relation:

N
1
<c°s2 ¢> =N X N(¢i) c°s2 ¢i. (6)
l= 1

According to the supplier, the fiber volume fraction of the material samples studied was 0.60. In

addition, the fiber length for these samples was given to be approximately 2 in. Using an average fiber
diameter of 295 gin, an effective aspect ratio of 6780 was calculated.

An example of the estimation of Herman's orientation is summarized here. A micrograph of

unprocessed material was analyzed and the histogram depicted in Figure 12 was obtained. A value of

0.997 was subsequently calculated for the Herman's orientation. Two samples of unprocessed sheet

were analyzed, with an average Herman's orientation of 0.994. In addition, five samples of material

formed into a spherical shell were examined. An average orientation parameter of 0.995 was found.

The calculation of mechanical properties by the SMC program utilized the material parameters of PEKK

resin and AS-4 carbon fibers. The relationship between relative fiber alignment and several material

252
properties is shown in Figure 13. The results show that relative fiber orientation has little effect on
elastic properties in these unidirectional discontinuous fiber composites before and after processing.

In parts manufactured using a diaphragm forming process, the average fiber axes are not

perfectly aligned with structural axes. Consequently the elastic properties of a given part vary
depending upon local fiber orientation. It is necessary to determine structural engineering constants

from the material properties of the composite and the average reinforcing fiber orientation. The most

direct method of performing the transformation makes use of the compliance matrix of the material. The

transformation matrix for the properties in the plane of the material is depicted below:

$22| m_n2 _42 2m2n 2 m_2 /ISyy] (7)

= m n - m n //Sxy/
-] ( m4 n4 2m2n 2 m_n22 Sxx
/Sial
lS66-1 4m2n 2 4m2n 2 m 4 + n 24
_ 8m2n (m 2- n2) 2 )/Sss_ 1

Where the 1, 2 subscripts refer to the structural axes (the tangential and radial directions in the case of

the curved beam) and the x,y subscripts refer to the material axes (x-being the local average fiber
direction). Also m = cos 0t, n = sin 0t, and et is the angle between the average fiber axis x and the

structural axis 1.

The structural engineering constants may then be determined by the following relations:

1 1
E2 -
E 1 - $11 $22

(8)
1 S12
G12- $66 _12--$22

A typical curved part was manufactured in a diaphragm forming process as described in the

previous section. A microstructural analysis was performed to determine the average fiber orientation at

a number of points in the curved beam. Equation (8) was then used to calculate the the local material

properties relative to the structural axes for use in a structural performance analysis.

Structural Macromechanics

The following analysis compares the relative performance of curved beams with several different
microstructures based on a maximum stress failure criterion. The microstructures are chosen to be

253
representative of different manufacturing procedures. Efficiency of the structure is determined as the
ratio of failure load to weight. No attempt was made to include the cost of manufacture or material.

Material heterogeneity is most prevalent in the web section; therefore analysis is conducted for a

rectangular cross-section curved beam which represents the web of an I-beam, channel beam, T-beam,

etc. Analysis shows that the principal stresses are in the radial and tangential directions for a beam

loaded in pure bending and the maximum tangential stress can be several times greater than the

maximum radial stress depending on the geometry. This indicates that the efficiency of a beam can be

improved by orienting the majority of the fibers in the maximum load direction. Several beams have

been produced by diaphragm sheet forming. Preliminary results show that the fiber orientation along
the web of these beams is not quite tangential so that the material properties, relative to the beam

coordinates, vary along the length of the beam. Figure 14 compares schematically the microstructures

of unidirectional and quasi-isotropic curved beams made from stretching a straight beam and diaphragm

forming processes with those made from a process which entails cutting a curved pattern out of a flat

panel. Both an elasticity analysis and a finite element analysis are conducted to determine the effect of

material heterogeneity on the stresses in the beam.

Elasticity Analysis
The analysis is conducted for a curved beam with heterogeneous material properties loaded in

pure bending. Figure 15 shows the geometry of a beam loaded in pure bending where a and b are the

inside and outside radii, r and q are the radial and tangential coordinates, M is the applied moment, R is

the average beam radius, R =(a +b)/2, t is the beam depth and d is the distance from the beams

centroidal axis. An important geometric parameter is the average radius to depth ratio, R/t.

The analytical solution is based on an elasticity approach similar to that found in Lekhnitskii [3]

with the only difference being that the constitutive relation allows for material property variation in both

the radial and tangential directions. The analysis is conducted by expressing the radial, tangential and

shear stresses in terms of a stress potential, F. The details of this analysis can be found in [3] but the
constitutive relation used here is

eij = Okj,
where (9)

aik = _k cos(k 0) r -n

and aij are the elements of the compliance matrix, 0qj are the homogeneous material property base

values of aij, r is the radius, and n and k are the degrees of heterogeneity in the radial and tangential

directions respectively. The sign of n determines whether the beam is stiffer on the inside or the outside

254
radius; a negative value of n represents a beam which is stiffer on the inside, and a positive value of n

represents one which is stiffer on the outside. Tangential heterogeneity can be represented by adjusting
the value of k to match the frequency of the fiber waviness.

Finite Element Analysis

A linear finite element model is used to verify the results of the elasticity analysis. The model

consists of 300 four node quadratic elements where the properties are input for each element to handle

the material heterogeneity. The pure bending boundary conditions are applied by distributing the

appropriate normal loads at the nodes along the straight edges. The results are within 8 % of the

analytical results from the elasticity analysis.

Results

Stresses in the curved beams are affected by the geometry of the beam as well as the material

heterogeneity. Figure 16 shows the relative shape of several beams with different radius to depth ratios,
R/t. An R/t ratio of 25 or greater is typical of the curved beams used as primary structures in an aircraft

fuselage. Figure 17 shows the effect of radial heterogeneity on the maximum radial stress for several
different R/t values. The degree of radial heterogeneity, n, is varied from -2 to 2, where an n value of -2

corresponds to a 20% drop in stiffness from the inside radius to the outside and an n value of +2

corresponds to a 20% increase in stiffness. Figure 17 shows that for a radius to depth ratio of 1.0, the
maximum radial stress is affected greatly by the value of n, but for an R/t value greater than 10.0, the

degree of radial heterogeneity has no effect on the maximum radial stress. Figure 18 shows similar
results for the effect of n on the maximum tangential stress. The degree of tangential heterogeneity, k,

was found to have no effect on the stresses in the beam, but does have an effect on the deflection as well

as the failure of the beam.

Maximum Stress Failure Analysis

The maximum stress failure criterion is used to compare the performance of beams made with

the microstructures shown in Figure 14. The maximum tensile, compressive and shear stresses are

determined for the beam and are compared to the corresponding ultimate loads of the material. The

maximum moment to failure is determined by increasing the applied moment until one of the stresses

exceeds the corresponding ultimate load. The moment to failure is determined for a beam with an inside
radius of 3.5 inches, an outside radius of 4.5 inches and an included angle of 90 ° .

Several different unidirectional microstructures are compared; the first has tangentially oriented

fibers, the second is the curved beam cut from a straight unidirectional panel, and the third has curved

fibers, but not the same curvature as the beam. Data for this microstructure was taken from a part

255
 produced by diaphragm forming. A beam with patchwork tangentially oriented fibers is also compared
to the unidirectional microstructures. Two quasi-isotropic microstmctures are also compared, one with

axisymmetric fiber orientation as obtained from stretching a straight beam and another which represents

a curved pattern cut from a quasi-isotropic panel. The analysis is conducted for an 8-ply lay-up so that
the weight of all microstructures is the same. Table 4 compares the moment to failure of the different
microstructures as well as the maximum deflection for a unit load. Unidirectional beams with non-

tangentially oriented fibers result in considerable reduction of the moment to failure and a considerable

increase of the maximum deflection. The two quasi-isotropic microstructures have virtually the same
moment to failure and maximum deflection.

Conclusions

This paper briefly outlines a series of analyses that can be used to predict structural performance

from basic processing information and material preform design for long discontinuous fiber

thermoplastics. This information is important to the material manufacturer and the process designer as

well as the structural designer. Although much needs to be added, these tools represent the basis for a

truly integrated design methodology for composite structures.

References

o
G. K. Batchelor, "The Stress Generated in a Non-dilute Suspension of Elongated Particles by

Pure Straining Motion," Journal of Fluid Me_h_lni_, Vol. 46, part 4, 1971, pp. 813-829.

.
R. L. McCullough, G. J. Jarzebski, S. H. McGee, "Constitutive Relationships for Sheet Molding

Materials," The Role of the Polymeric Matrix in the Processing and Structural Properties of

Composite Mst¢ri_l_," Plenum Press, New York, N. Y., (1983).

3. S.G. Lekhnitskii, Theory_ of Elasticity of an Anisotropic Body, Mir Publishers, Moscow, 1981

256
 Table 1. Summary of Property Predictions

Term Newtonian Fluid Power-Law Fluid

B m

rl::/rl --i (L/D) 2 2-mf

2 !_l__---f _

-- -m

2_ m 2-_"-f
T112/T1

_ 1-_/__

r123/rl [1 "_]-m_231

1122/1"I 4[ 1__7- l-m_2m-1

1"123/1"112

T111/1"122 2-3f "_ 1/2(L/D)2 2" (m+2)f T m/2(L/D)m+l

257
 Table 2. Apparent Laminate Viscosities, Laminate Theory

laminate f'l'l 1/'q '22/T1 {'1'12/1"1 X'i2 _"21

1.04x106 1.04x106 4.48 1.52x10 -5 1.52x10 -5

[0/90],

[0/+45] s 6.96x105 2.32x105 3.48x105 1.0 3.33x10 -1

[0/+45/90] s 6.96x105 6.96x105 2.61x105 3.33x10 -1 3.33x10 q

[+-45]s 1.79x101 1.79x101 5.22x 10 5 1.0 1.0

f = 0.6 I_/D =1000

TI11]'r I = 2.09x106 r122_ = 3-18x101 TII2/T1 = 4.48

].tll/rl = 2.09x106 _12t't'i= 1.59x101 _a/rl= 3.13x101

_66/r1= 4.48

258
 Table 3. Constrained (LPT) and Unconstrained (UCT) Laminate Theory Results

Lami fl' /n
LPT UCT LPT UCT LPT UCT

[0/90], 1.04x106 1.04x106 1.04x106 1.04x106 4.48 4.48

[0/+45] s 6.96x105 6.96x105 2.32x105 2.10xlO 1 3.48x10 -5 2.26x101

[0/_+45/9O], 6.96x105 5.22x105 6.96x105 5.22x105 2.61x105 1.81x101

[_+45]s 1.79x101 1.57x101 1.79x101 1.57x101 5.22x105 3.17x101

Table 4. Microstructural Comparison of Beams Loaded in Pure Bending

Fiber Moment Maximum

Orientation to Failure Deflection

Unidirectional 1.00 1.00

Tangential

Unidirectional 0.39 1.14

Cut-out

Unidirectional 0.46 1.09

Non-tangential

Unidirectional
0.70 1.36
Patchwork

Quasi-isotropic
axisymmetric 1.00 1.00

Cut-out
Quasi-isotropic I 0.97 1.00

259
 3Z

L X

Figure 1. Oriented Fiber Assembly

35000 I I

30000
SQUAREARRAY
WD=1000 ]

25000
¢-q
c-,l
m=l .0
20000

15000

10000

m=0.5
5O0O
m=0.3

0 J

0 0.2 0.4 _-- 0.6 0.8

Figure 2. Ratio of Axial and Transverse Elongational Viscosities

260
Symmetric Antisymmetric
Figure 3. Overlap Length Geometry

261
 40 - 41%

AS4/PEKK LDF
30 L AVG.= 2.13 in.

2O 23%
tO

G)
18%

10

6%
J 5% [4 % 3% I
00 1 2 3 4 5 6 7

Fiber Length (Inches)

Figure 4. Fiber Length Distribution

m
I
20 - 22%

18%
e-.
o 15 m m
a_

--1 15%
14%
=1
L-

c-
o 10
O 11%

e,
o
I_
5
n

1%
0
0 1 2 3 4 5 6 7

Fiber Length (Inches)

Figure 5. Fiber Length Contribution to Effective Viscosity

262
 10o

10-2
UD=100

L/D=1000

L/D=10000

10-8
UD=100000

10-10

0 10 20 30 40 50 60 70 80 90
e

Figure 6. Influence of Fiber Orientation on Elongational Viscosity

100
f=.6

SQUARE ARRAY
10-1

L/D=100

.._ 10-2
UD=1000

V_" 10-3-

L/D=10000

10 -4

L/D=100000

10-5

0 10 20 30 40 50 60 70 80 90
0
1

Figure 7. Effect of Orientation Distribution on Elongational Viscosity

263
 4 3

1 2

_e= 45 deg.
_X

Figure 8. Off Axis Tension Test a) Problem Geometry b) Deformation After 20 Time Steps c) Fiber
Orientation After 20 Time Steps

264
 Y

°-°°_.,

Figure 9. Bending of a Sheet of Unidirectional Material a) Problem Geometry b) Deformed Geometry

Figure 10. Bulging of a Clamped Sheet a) Problem Geometry b) Deformation After 20 Time Steps c)
Deformation After 59 Time Steps

265
 Figure 11. Diaphragm Formed Curved Beams

1.3 5.3 3.3 1.3 -0.7 -2.1 -4.1 -6.1 -8.7 -10.7

phi (degrees)

Figure 12. Fiber Orientation Histogram of Processed Material

266
 1.84e+7

1.82e+7

Q..
v 1.80e+7
X
U..I

1.78e+7

1.76e+7 , . , . , . , .
0.980 0.984 0.988 0.992 0.996 1.000

fp

0.32

0.30

0.28
X

0.26

0.24

0.22
0.980 0.984 0.988 0.992 0.996 1.000

fp

Figure 13. Material Properties vs. Relative Fiber Alignment

267
 Diaphragm Formed Patchwork

Axisymmewic Cut-Out
Uni-directional

Axisymmetric Cut-Out
Quasi-isotropic

Figure 14. Comparison of Beam Microstructures

f Y

Figure 15. Pure Bending Load Case

268
 Figure 16. Comparison of Several Different Beam Geometries

0.5

0.4
----m----

/-'-
_>
R1olj
R/t=1.5
R/t=2.0
R/t=2.5
"_--'-0
/]
_r
0.3
6M
t2

0.0
-2 -1 0 1 2

Degree Of Radial
Heterogeneity (n)

Figure 17. Maximum Radial Stress vs. Degree of Radial Heterogeneity

269
 5

R/t=1.0
4 R/t=1.5
-- R/t=2.0

(_t <_ R/t=2.5

R/t=4.0
6M
R/t=10
t2

0 ! I

-2 -1 0 1 2

Degree Of Radial
Heterogeneity (n)

Figure 18. Maximum Tangential Stress vs. Degree of Radial Heterogeneity

27O
 RESINTRANSFER MOLDING
for
ADVANCED COMPOSITE
PRIMARY AIRCRAFT STRUCTURES

Alan Markus and Ray Palmer

Douglas Aircraft Company

INTRODUCTION

Resin Transfer Molding (RTM) has been identified by Douglas Aircraft Company (DAC) and
industry to be one of the promising processes being developed today which can break the cost
barrier of implementing composite primary structures into a commercial aircraft production
environment. The following paper will discuss the RTM process developments and scale-up
plans Douglas Aircraft will be conducting under the NASA ACT Contract.

271
 PROCESS I)ESCRIPTIONS

Resin Transfer Molding at Douglas Aircraft consists of two distinct methods of resin
impregnation: 1.) Vacuum infusion and 2.) Pressure injection.

Figure 1 below describes the two forms of vacuum resin infusion being developed at Douglas.
First is resin film infusion, in which a resin (film) is placed on the tool, the preform is placed
over the resin, the entire assembly is vacuum bagged and oven cured according to the proper
cure cycle for that resin system. The second form of infusion is liquid resin infusion. In this
method, instead of placing the resin film on the tool, a porous distribution manifold is placed
on tile tool to allow the liquid resin to infuse the preform. As with the film infusion method,
the entire assembly is vacuum bagged at 30 in mercury and cured at 350 degrees in an oven.
(Note: tlard tooling can and often is used in place of vacuum bag.)

PROCESS DEVELOPMENT

VACUUM IMPREGNATION

UTILIZES RESIN FILM WHICH IS SOLID

AT ROOM TEMPERATURE

BREATHER _ ff BAG VACUUM

1//
RESIN
F,,.______ ! tJJ__
TOOL--'A 1

UTILIZES LIQUID RESIN AT ROOM TEMPERATURE

BREATHER_ ,_/-'-- BAG VACUUM

DRYPREFORM _ .............. '_' _'1 ///

RESI.
MA.IFOLD
_ !t //'
TOOL--'t /// I

TO LIQUID

RESIN

Figure 1

272
 PROCESS I)ESCRIPTIONS

The second form of Resin Transfer Molding is pressure injection. As described in Figure 2
below, this process transfers liquid resins through a mix chamber into a compacted stitched
preform contained within a matched metal tool. The tooling for this process is usually self-
heated and requires some type of press or restraining fixture.

In the Douglas RTM development, the resin injection port is located at the center of the part
allowing resin to flow directly into the preform, while venting occurs at locations deemed
necessary by preform geometry (resin pressure of 40 psi is desirable). In the case of flat
panels, venting usually occurs at each corner.

This type of process is currently being implemented into many automobile prototype manu-
facturing facilities.

PRESSURE IMPREGNATION

LIQUID RESIN I
RESIN INLET

" 511
J LIQUID HARDENER I

MATCHED METAL TOOL /

RESIN OUTLET _[_

RESIN OUTLET

Figure 2

273
 PROCESS OBJE(TI'IVES

The objectives of Douglas' RTM development is to I.) Exploit the benefits of RTM to max-
imum potential, 2.) Become cost competitive. Below is a list of the primary benefits RTM
offers as a process (Figure 3).

RTM/STITCHING BENEFITS

LOW-COST MATERIAL SYSTEM (WITH PREMIUM DAMAGE

TOLERANCE)

REDUCED ASSEMBLY COSTS (DUE TO STITCHING)

SHORTER PROCESS CYCLES

BUILT-IN QUALITY (REDUCED QUALITY INSPECTION COSTS)

GREATER CONTROL OF DIMENSIONAL ACCURACY

REDUCED FABRICATION COSTS (CUTTING, LAY-UP,

PROCESSING)

INHERENT STRUCTURAL QUALITY

Figure 3

274
 PROCESS OB.IECTIVES

If all benefits of the RTM process are realized, significant advances can be made in the
application of composite primary structure to commercial aircraft. Those benefits of most
importance in leading to application are primarily cost related. Figure 4 below compares the
Douglas funded 8' x 12' wing box fabricated with conventional materials and methods versus
the NASA funded 8' x 12' wing box to be made by RTM. If full advantage of the stitching
process can be realized, (stitching rib clips to skin) a cost savings of 48% is projected over the
conventional box being fabricated by Douglas.

During wing box fabrication, detailed data will be compiled on material and labor costs. This
data will provide a direct comparison between conventional and advanced composites fabri-
cation. Obviously, a major issue is the cost of composite wing structure versus the cost of
conventional aluminum construction. Attempts are underway to define aluminum costs at the
wing box level. Another approach will involve wing cost models to project composite wing
box costs to full scale wing structures.

RECURRING COSTS FOR 8 X 12 TEST BOX

$
1.0

I
MAJOR ASSEMBLY

CUPS/
SUBASSEMBLY
BONDING

C_ 0.6
ILl 48%
N RIBS/SPARS

,.J
,<
MAJOR ASSEMBLY
n-
STIFFENERS
O RIBS/SPARS

I MAJOR A_SEMBLY |
Z
CUPS/ RIBS,_PARS

SUBASSEMBLY

BONDING

SKINS SKINS
SKINS/ STIFFENERS
STIFFENERS CLIPS

MATERIALS

0.0 MATERIALS MATERIALS

CONVENTIONAL BASELINE MANUFACTURING

COMPOSITE STITCH/RTM GOAL
CONSTRUCTION PROCESS

Figure 4

275
 MANUFACTURING DEVELOPMENT
for
COMPOSITE PRIMARY WING STRUCTURES

276
 WING SUB-SCALE TOOLING I)EV'_EI.OPMENT
Effect of Preform

In developing a manufacturing strategy for the Douglas wing effort, it became clear that
vacuum infusion with heavy density stitched preforms was a natural for this structure. The
heavy density stitching gives the damage tolerance needed as well as the compaction necessary
to utilize the vacuum impregnation minimum pressure process. Below in Figure 5, com-
paction studies run at Douglas indicate that the stitched preform gives 92% of what full
compaction would be for 60% fiber volume. The real benefit of this is that the tooling for the
structure can be greatly simplified. In fact, tooling is basically used for part volume definition
only since no compaction or movement is required.

WING ELEMENT TOOLING DEVELOPMENT

EFFECT OF HEAVY DENSITY STITCHING

ON TOOLING DEVELOPMENT

90 ..... 60oo/:0
FIBER VOLUME
HEAVY % FIBER VOLUME
COMPACTION
PRESSURE
REQUIRED
(PSI) I
I

0
0.35C "_0.324

THICKNESS
(IN)

Figure 5

277
 WING SUB-SCALE TOOI.ING DEVEI.OPMENT
Tool/Process Definition

Douglas tooling for three stringer element wing panels is currently set up for both film resin
infusion and liquid resin infusion. In Figure 6, on the left is a schematic of the film infusion
tooling. As with the flat panel tool, the resin fihn is first placed on the tool, the preform and
mandrels are then placed on the resin, and the entire assembly is bagged and cured according
to the proper resin cure cycle. (Note: The current method of processing with 3501-6 film
resins requires the use of an autoclave to insure proper wet-out.) In Figure .6(right), the liquid
vacuum infusion method is illustrated. In this case, the resin film is replaced with a porous
distribution manifold that allows the liquid resin to uniformly permeate the preform. Once
resin infusion is complete, the entire assembly is allowed to cure under vacuum bag pressure
and oven heat only. (It should be noted that this tooling currently uses rubber expansion for
compaction between stiffeners. In future tooling, this will not be necessary. Also, rigided
periphery tooling can be used in place of vacuum bag to hold tolerances of preform.)

WING ELEMENT TOOLING DEVELOPMENT

KEY FEATURES

• VACUUM RESIN INFUSION

• POTENTIAL FOR NON-AUTOCLAVE CURING

• PROCESS WELL KNOWN

_A
TO°RR:2
C'OTN--
: TOVAC-- BLEEDER CLOTH _ TO VACUUM

-\ \/// o. PREEO::"
TOP PRESSURE "_ _SL

• _ 4 4 •

I ! _,\.SEp._ / ,,I" _ I
L- St*LA.O EUMI.ATE"'_M-oous_R// L_ RESIN ,_ eeeeeee _ "'_;'"_ .... e ....... :
SIDEWISE RESIN FLOW SPACER FILM
TOoLER._ -_f

LIQUID RESIN

TO MANIFOLD
PLATE

FILM INFUSION METHOD LIQUID INFUSION METHOD

Figure 6

278
 WJNG SUB-SCAI,I! TOOI,IN(; J>J3VEIAOI’MEN?
Douglas Wing Vacuum Infusion 7’001

Below in Figure 7 are photos of the preforms and tooling for wing element work. On the
upper left is a heavy density stitched 3 hladc wing elenient preform ready for resin infusion.
Below to the right is thc matched metal aluminum tooling assemhled on the preform beFore
vacuum bagging the assembly.

Figure 7

279
 WING SCALE-UP TOOI.ING DEVELOPMENT

In scale-up of the vacuum infusion process, Douglas will select between liquid and film resin
infusion. To date, the infusion process performs better with liquid resins rather than the film
resins, ltowever from a tooling scale-up perspective, the film resins are preferable. To utilize
a film resin, it appears that a viscosity profile should be maintained well below 250 centipoise
at 250 degrees F or below for between one and two hours.

In using such a film resin, the vacuum infusion tooling and process have a significant advan-
tage in that both are currently well defined for scale-up. The primary development necessary
is in selecting tool materials that satisfy classical tooling concerns (i.e.. coefficient of thermal
expansion, thermal mass/weight and cost.) The three Douglas concepts evaluated are shown
in Figure 8. In evaluating each concept, expansion control for scale-up and cost were driving
factors. The tooling concept selected by Douglas is the aluminum/graphite combination.
While this aluminum/graphite concept is approximately 14% more costly than the all alu-
minum tool, the benefits in tooling tolerance control and scale-up outweigh the small cost
penalties.

TOOLING CONCEPTS EVALUATED

VACUUM IMPREGNATION OF WING

SCALE-UP ISSUES (SUBCOMPONENT)
O Oo Oo O

0 O0 O0 0 0 0

0 0 0 0 0
0 0 0 0 0 0 GRAPHITE
TOOL

ALUMINIUM-.-..-"
I MANDRELS
ALL ALUMINUM ALL GRAPHITE

• WEIGHT • NO LATERAL COMPACTION

• TOLERANCE CONTROL • COST
• COEFFICIENT OF THERMAL • HEAT-UP
EXPANSION
• POSSIBLE BULK FACTOR
o °o °o o o o PROBLEM
o ° 0 ° 0
o o o Oo
GRAPHITE
SIDE RAILS"_
•.. GRAPHITE
TOP PLATE
GRAPHITE I
TOOL PLATE

ALUMINUM MANDRELS

ALUMINUM/GRAPHITE COMBINATION

Figure 8

280
Shown to the right in Figure 9 is the tooling concept that will be used to fabricate 4' x 6' wing
skin panels. This concept is a combination of graphite tooling plates with aluminum man-
drels. This was chosen over an all aluminum tool bccause the graphite/epoxy tooling plate is
better able t o control tolerancefi, maintain resin bleed holes in proper location, and allow for
possible integration of stitched rib to skin clips in future development. (Note: Resin bleed
holes will consist of' steel bushings pottcd into the graphite tooling plate to maintain hole
dimensions and prevent excessive wear.) This aluminum/graphite tool once assembled is then
placed in a restraining fixture as shown below on the l e k A compaction pressure of 20-30
psi can be applied with a pressure bladder restraining fixture to squceze cxcess resin out and
insure a quality surface finish.

WING SUBCOMPONENT TOOLING CONCEPT

AND PROCESS DEFINITION

HOLES FOR GRAPHITE

BLEEDING OFF
BLOCK7
EXCESS RESIN 7

BOLT

STITCHED
PREFORM RESERVOIR BLOCKS

Figure 9

281
 MANUFACTURING DEVELOPMENT
for
COMPOSITE PRESSURIZED FUSELAGE STRUCTURES

282
 FIJSEIAGE SUB-SCAI,E TOOLING T)IZVEI,OPMI:N?'S
Effect of Preform

In developing the pressure injection RI'M process for thin fuselage structures, two critical
points to tool design need to be considered u p front. First is matched metal tooling needed
for compaction pressures required to yield 60";" fiber volume or the final desired thickness.
(Compaction is needed due to the fact this preform is not heavily stitched.) In Figure 10, thc
chart on the left shows that approximately 48 psi is required to give the appropriate com-
paction pressure. This pressure becomes important when designing for tool rigidity and when
pressurizing the tool with resin during fabrication. Another piece of information found on the
chart is bulk factor. A 0.01 inch bulk factor is inherent in the prerorm. While this bulk seems
small, in scale-up (Le.. multiple longerons) it does become additive causing tolerance problems
for tooling. To circumvent these problems, the bulk factor must bc accounted for in tooling.
Figure IO to the right illustrates both types of matched metal tooling being looked at by
Douglas. First is a clamp/fit fixture used to eliminate bulk factor problem before assembly
into a matched metal tool frame. The second tooling concept is a sidewall compaction
matched metal tool that allows bulk factor and compacts after the tool is closed.

FUSELAGE ELEMENT TOOLING DEVELOPMENT

EFFECT OF PREFORM ON TOOLING DESIGN

BULK FACTOR

48

COMPACTION
PRESSURE
(Psi)

0.0815 0.072
PREFORM THICKNESS FINAL
THICKNESS (IN.) THICKNESS

Figure 10

283
 FUSELAGE SUB-SCALE TOOIANG I)I_VEI_OPMI{NT
Effect of Preform

Another reason matched metal tooling is required for fuselage structure has to do with flexi-
bility of the preform. Since the thickness of each "J" section longeron is only 0.072 inches, the
preform will not maintain its required shape and orientation. In Figure 11 (left), preform
tlexibility is illustrated. Figure 11 to the right, shows the matched metal tool required to shape
the longerons properly.

FUSELAGE ELEMENT TOOLING DEVELOPMENT

EFFECT OF PREFORM ON TOOLING DESIGN

L I_
I - 7 IN.
I
= I = 7 IN.
I_
= 1-3.5 IN. =
I

1 'Ill 1.45 IN.

TOLERANCE G141.06

Figure 11

284
 FIJSEI.AGF. SUB-SCAI.E TOOLING I)EVEI.OPMENTS
Effect of Tooling Tolerances

In progressing from flat panels to stiffened structures, various unforeseen problems were
encountered. While it was already known that resin injection pressures for thin preform were
extremely high (125 psi), the effect of tooling tolerance had not yet been discovered. Figure
12 below, indicates that the resin pressure required to permeate a preform appears to follow
some exponential form for very thin laminates. In examining the graph, note that as the
thickness of the preform decreases, significant increase in the amount of pressure required to
permeate the preform occurs. This result, translated to tool design, says that if tooling man-
drel tolerances (for matched metal tools) are not extremely close, the resin flow profile
expected may not occur. The resin will follow the path of least resistance to the exit vent, thus
resulting in a part that is not fully impregnated. (This has only been found to present a
problem in thin structures.)

FUSELAGE ELEMENT TOOLING DEVELOPMENT

EFFECT OF PREFORM AND TOOL TOLERANCES

ON PROCESSING THIN STRUCTURES

,2, • PRESSURE GRADIENTS

T...s,T,o._/J.//I CREATED FROM TOOL
TIME TOLERANCE DEVIATIONS
PRESSURE
HAVE TREMENDOUS
(PSI)
IMPREGNATION AREA _i IMPACT ON FLOW
60 30 MIN CONTROL AND
PREDICTION

10 MIN

0.25 0.1 0.072 0.042

PREFORM THICKNESS
* 3 BY 3 FOOT PREFORM SIZE

Figure 12

285
 FUSELAGE SUB-SCALE TOOLING DEVELOPMENT
Douglas Fuselage RTM Tool

In developing the tooling for the sub-scale fuselage structure, strict attention was given to tool
tolerances. All mating surfaces in the horizontal plane of the tool were blanchard ground to
+/- .005 inch to aid in regulating resin injection pressures. This tool is constructed of alu-
minum, with self-heated calrod heaters and is clamped between I-beams to apply compaction
pressure. The fabrication process is simply to inject resin into the skin at the center ofthe tool
until resin appears at the corners, then inject the stiffeners until resin appears at the opposite
vent at the other end of the tool. Resin pressure is maintained at 40 psi while the part is
cured. See Figure 13 below.

FUSELAGE ELEMENT TOOLING DEVELOPMENT

ELEMENT TOOLING CONCEPT

PANEL

INJECTION PORT _ ,,FPUSH'OUTPINS

..s,. -4.,oo000 o ooooo
j

STIFFENER
RES,N ._.LO.INU..ANDREL
ASSE.BL.
_,
BLEED "1 I&ooooooooooooooooo ,LI,I

I___=_INJECTION

I • k • V _ SKIN RESIN BLEED

STIFFENER R_

INJECTION

Figure 13

286
 FUSE LAG E SUB- S (1A 1, E 'I'OO I, I N (; 1) 13VE LO P MENT

Figure 14 below shows the Douglas fuselage tool in various stages of fabrication. The upper
left photo is a view of the clamp/fit tooling used to inscrt the preform and mandrel assembly
into the matched metal self-heated containment tool shown on the upper right. Below is the
entire tool assembly with heaters ready for final clamping from the pressure bladder
restraining fixture.

FUSELAGE ELEMENT TOOLING DEVELOPMENT

ELEMENT TOOLING

MANDREL ASSEMBLY TOOUMAN DREL INTEGRATION

TOOL ASSEMBLY

Figure 14

287
 FUSELAGE SCAI.E-UP TOOl.ING DEVEI.OPMENT

Major drivers in tooling scale-up development for fuselage as related to cost include tolerance
control, tool heating and press/restraining fixture. In addressing these issues, numerous tool
concepts were evaluated at Douglas. The best two are shown below in Figure 15. On the
right of Figure 15 is a scaled-up version of the sub-scale fuselage tooling with minor modifi-
cations. These include use of flexible caul in place of a hard upper tool (to ease tool
assembly), use of hot air convection heat system in place of calrod heaters (due to cost), and
use of a dedicated restraining press with stainless steel inflatable pressure bladders (due to cost
of large press). Below to the left is an alternate concept consisting of a one piece
carbon/epoxy tool which is being considered to replace the mandrel concept if matched tool
tolerances continue to be a problem in tile resin injection cycle. (Note: A careful
cost/performance study of the RTM composite fuselage panels will be done with duplicate
panels being made by the Automated Tow placement process at llercules. Data from this
study will determine the manufacturing approach for Phase B fuselage development.)

PRESSURE IMPREGNATION OF FUSELAGE

TOOLING CONCEPTS

B INFLATABLE B INFLATABLE
STAINLESS STEEL AIR STAINLESS STEEL AIR
BLADDER BLADDER
RESTRAINING FIXTURE_ RESTRAINING FIXTURE_

\
, .;_ GRAPHITE/EPOXY _11 , • ._-" SIDES) HOLD
GRAPHITE/EPOXY _..l_g_g,JJ t] REINFORCED A'R PAD II _ ,.-i-/i,_a-_JJ,,[_ MANDRELS IN PLACE

TOOL i " _=g===_l. il

ll I RUBBER CAUL
I_
SHEET'-'dlPt [_ll _'
_ _I
II I
_BOLTEDTO
SKIN. ./C MACHINED
TOOL)

17 NI Iii f , NI M'N<'RE<s
J I
[ BRAKE
,ORMED
6061-T6 SKIN PLATE I L
I "--DRAKE,OR"
MOEN HEATER CHANNELS _ TOOLING BASE 118.0-1NCH RADIUS / TOOLING BASE 6061-T6118.0-1NcHSKINRADIusPLATE
/
MOEN HEATER CHANNELS '_j

GRAPHITE TOOL CONCEPT ALUMINUM TOOL CONCEPT

Figure 15

288
 P R 0D 1JC T I ON IJ '1'I 1, I Z A T I 0N NE ED S
Tooling Scale-up to Production

In developing the tooling for the fuselage, it is clear that the OMI, tool surface and tool base
are from a design perspective common for any tool concept Douglas may have. For this
reason, an existing fuselage tool built under previous NASA contract will be modified to fit
the RTM process being developed. Shown in Figure 16 is the size fuselage to be manufac-
tured in Phase B and the existing tools to be used for the fuselage. This lower tool (similar
to the tool base in Figure 15) will be used with the best concept for the upper tool portion
and modified to include as much automation and heavy tool handling capability as possible.

Primary focus to date on fuselage has been the RTM fabrication development for fuselage
quater panel skins with "J" stirener longerons. Therefore, to date, manufacturing develop-
ment of other fuselage elemcnts (clips, framcs, floor beams, etc..) has not been addressed in
detail.

PHASE B FUSELAGE DEVELOPMENT

FUSELAGE BARREL
4
201 N . b ,-FRAME
I

Figure 16

289
 PRODUCTION UTl1,IZA'I'ION NEEDS
In Line Proccss Monitoring

Process monitoring must ultimately become an integral part of thc RTM process if it is to go
to production. For this reason in line process monitoring with process modeling is part of a
continuous Douglas funded improvement program and part of the NASA ACT program.

Figure 17 below illustrates the work being done under NASA contract with William and Mary
College and Virginia Polytechnic and State IJniversity (VPJ). I n Figure 17 (above left) are the
sensors being used by William and Mary College to gcticratc kinetic relationships for various
RTM resins. nelow left are kinetic relationships for the Ilow Tactix 138 resin. Figure 17 to
the right represents permeability studies on stitchcd fabrics being done at VPI to begin mod-
eling the RTM process. Shown in this chart arc permeahilities for 54 ply wing skin lay-ups
and 12 ply fuselage lay-ups.

PROCESS MONITOR / MODEL DEVELOPMENT

THROUGH THE THICKNESS PERMEABILITY COMPARISON
LOG (WdP2) VERSUS POROSITY
01

0.05

0 02

0 01

0.005

Kcidf"2
0.002

0.001

0.0005

EQUATION PARAMETERS 105' 121' 149" 1 7

0.0002
K. OW697 001689 0 1177 0767
K,(l-ImIN N 2 590 1.335 2.340 1.345 0.0001
RESIDUAL O.WO03 OWW2 0.00551 0.005 0 0.2 0.4 0.6 08 1

K, OW2 0018 0.002 0.143 POROSITY

Kn*(l-a)" M 4.478 0.031 -2.252 -3.71
2 N 1.286 1.382 -0 885 o.gn
RESIDUAL 0.000 0.000 0.002 0.003
K. 0662 0.027 4487 46.259 71890 AND 72090 - 54 PLY STITCHED
K, 0.0663 0.029 447 46 11
(MCDONNELL DOUGLAS)
(K,+K+r")(l-m)" M OW4 0.329 0.011 0.002
N -1.M 4.761 0.939 0.997

,
DumT = K (1 - 01" N AVO = 1 902

TEMPERATURE = 105' 14p

Kt 0.00566 002033 0.09967 033540

Figure 17
 PRODUCTION JTILIZATION NEEDS
Future Factory

In looking forward, a futuristic factory must be envisioned that is equipped to handle the
uniqueness of RTM fabrication concepts. At Douglas, we envision a close tie with major
resin suppliers that will equip us with the capability of handling large quantities of liquid and
film resins. In Figure 18 below, the resin storage is located within the building foundation for
space conservation and system delivery purposes. Douglas also envisions 1.) the use of
standard automated handling equipment to move preforms from automated stitching
machines to the tools, 2.) state-of-the-art resin film laying equipment to calendar out resin on
the tools, and 3.) hydraulically controlled tools with in linc process control to supply complete
process control from a central control point.

PRODUCTION NEEDS

ADVANCED FACTORY CONCEPT

KEY FEATURES

0 IN-GROUND RESIN STORAGE

0 HYDRAULIC-CO NTR0LLED
VACUUM-TIGHT TOOL CAVITIES

0 COMPLETE AUTOMATED CONTROL OF

STITCHING AND CURE OPERATIONS

Figure 18

29 1
 CONCI.USIONS

1. Thru-thickness vacuum infusion process well suited for large wing structures

2. Projected cost savings of 48% for RTM composite wing over conventional technology
composite wing

3. Resins currently acceptable for vacuum infusion not meeting 3501-6 mechanical property
standards

4. Resin requirements:

a. Film form resin desirable

b. I.ess than 100 centipoise resin viscosity

c. l.ess than 250 degrees F resin infusion temperature

d. Resin infusion time of 1 hour or more

e. Resin capable or oven cure under vacuum bag desirable

f. 3501-6 composite laminate properties desirable

5. Scale-up tooling costs for pressure RTM is cause for concern

6. Tooling and preform integration critical to success or pressure RTM

7. Fuselage cost/performance comparisons of Automated Tow Placement versus pressure

RTM are key to design/manufacture or fi]ll scale l)hase B fuselage barrels

8. Advanced RTM factory will require multi-industry participation to succeed

292
 CONSOLIDATION OF GRAPHITE/THERMOPLASTIC TEXTILE PREFORMS

FOR PRIMARY AIRCRAFT STRUCTURE

J. Suarez and J. Mahon

Grumman Aircraft Systems
Bethpage, New York

Abstract

The use of innovative cost effective material forms and processes is being considered for fabrication of
future primary aircraft structures. Processes that have been identified as meeting these goals are textile
preforms that use resin transfer molding (RTM) and consolidation forming. The Novel Composites for
Wing and Fuselage Applications (NCWFA) program has as its objective the integration of innovative
design concepts with cost effective fabrication processes to develop damage-tolerant structures that can
perform at a design ultimate strain level of 6000 micro-inch/inch. In this on-going effort, design trade
studies were conducted to arrive at advanced wing designs that integrate new material forms with
innovative structural concepts and cost effective fabrication methods. The focus has been on minimizing
part count (mechanical fasteners, clips, number of stiffeners, etc.), by using cost effective textile
reinforcement concepts that provide improved damage tolerance and out-of-plane load capability, low-cost
resin transfer molding processing, and thermoplastic forming concepts. The fabrication of representative
Y spars by consolidation methods will be described. The Y spars were fabricated using AS4 (6K)/PEEK
150g commingled angle interlock 0/90-degree woven preforms with +45-degree commingled plies stitched
using high strength Toray carbon thread and processed by autoclave consolidation.

293
 INTRODUCTION

The Novel Composites for Wing and Fuselage Applications (NCWFA) Program is performed by the
Grumman Aircraft Systems Division of Grumman Corporation and its subcontractors, Textile
Technologies, Inc., and Compositek Corporation, under the sponsorship of NASA Langley Research
Center, Hampton, Virginia 23665-5225. Mr. H. Benson Dexter is the NASA/LARC Contracting Officer
Technical Representative.

Background

The full weight savings and life cycle cost savings potential of state-of-the-art
composites have not been achieved because of design strain levels limited by
the materials LOW
• Damage tolerance
• Fracture toughness
• Notch strength
• Out-of-plane strength

and HIGH
• Material cost
• Manufacturing cost

We need
• Novel material forms
• Design strain levels of 6,000 + micro in./in.
• Cost effective manufacturing concepts

294
 PROGRAM OBJECTIVES

The overall objective of the NCWFA program is to integrate innovative design concepts with cost
effective fabrication processes to develop damage-tolerant primary structures that can perform at a design
ultimate strain level of 6000 micro-inch/inch.

1) Develop optimum wing desigr_, concepts with high performance fiber architecture to achieve
• Improved damage tolerance and durability
• High notch strength
• Increased out-of-plane load capability

2) Develop integrally stiffened fuselage bulkhead concepts that

• Eliminate skin/stiffener separation failure modes
• Minimize fabrication cost

3) Explore textile processes to achieve affordable integral skin/stiffener structures by automated

• Weaving
• Knitting
• Stitching

4) Explore RTM and consolidation/forming hybrid Gr/TP fiber forms for cost-effective fabrication of
primary wing and fuselage components

5) Conduct tests to validate structural performance and correlate test results with analytical
predictions.

295
 PROGRAM TASKS

The program is divided into five major tasks: Task 1 - Wing Design Concepts; Task 2 - Fuselage
Bulkhead Design Concepts; Task 3 - Wing Spar-Rib Intersection Design Concepts; Task 4 - Design and
Fabricate a Generic High Strain Wing Torque Box Component; and Task 5 - Structural Test of the Wing
Torque Box Component. The work reported here was accomplished in Task 1.

Task I - Wing Design Concepts

Design Trade Studies:

• Baseline, VSTOL 698-420
• Generate advanced wing designs using innovative structural concepts
(target weight savings of 50%)
- Multi-directional weavings (fiber form architecture)
- Stitching
- Softening strips
- Compliant laminates
- Integral cover to substructure
- Crack arrestment strips
- Y, blade and hat stiffeners
- Spread 0 ° covers
- Discrete cap covers

Newer/emerging materials:
- Fibers (E>40 msi, F>600 ksi)
- Commingled yarns
- Powder-based resin matrices
- Toughened TS and TP resin matrices

And cost effective processes (target cost savings of 25%):

- RTM
- Autocomp
- Consolidation forming

296
 BASELINE AIRCRAFT

The baseline aircraft selected for this program is a subsonic patrol VSTOL aircraft, Grumman design
698-420. This design is a high-wing, T-tail, turn-tilting nacelle configuration which combines both
power-plant and control vanes immersed in the fan stream.

BASELINE SUBSONIC PATROL V/STOL SEA

CONTROL CONFIGURATION

f,

I" 44'-0" _I
1__._16..o.. II
--
,_60-,-

16'-10"

_--- 16"-0" ------_ ,- 298

i" 42'-0" •;

297
 WING STRUCTURAL CONFIGURATION

The structural configuration for the wing is shown below. The wing has a span of 44 ft and a fold
span of 16 ft and is sized to allow installation of the conformal radar. The thickness ratio is 14% at the
root and 12% at the tip with a maximum depth of 14.4 in. at the centerline. Fuel is carried in the wing box
from fold-joint to fold-joint. Roll control in conventional flight is provided by spoilers mounted on the
rear beam.

Five-Spar Wing Box Configuration

BL 145.00

BL 258.00 /
_ BL 24.00
I / BE 200.00 _ _ , _BL 0.00 FRONT SPAR
W" / BL 60 00 _ , I / FOLD AXIS (FSRL)
FS 553.20 .L I_'/ "_- . _ 1t FS 581.00 " / /

WING/FUS ATTACH RIB CL RIB

298
 WING BASELINE CONFIGURATION

Consistent with the structural arrangement, design requirements, and advanced composite wing design
technology, a baseline wing configuration was established from previous design efforts on the High Strain
Wing Program. The upper and lower covers are Gr/Ep laminates, working to a design ultimate strain level
of approximately 6000 micro-in./in., and include G1/Ep for softening strips and crack arrestment strips
(for damage tolerance). The substructure consists of a front, rear, and three intermediate spars. The spar
webs are flat angle-stiffened Gr/Ep laminates, with the intermediate spars integrally co-cured and stitched
(with kevlar) to the lower cover. Gr/Ep sinewave ribs were used, except at the wingfold and tip where
titanium and Gr/Ep plain panels were used, respectively.

-r" gt.NP SOCTINII_I

"!" GL/llP IOF11NIlU ITRI@ In'RIP. a"

TYPICAL FOR AL L SPAR /

CAPS 0* Q*q/IEP T _ .312 _ jGRIIIP • lie
it4 " ILINO FAJ_rIN|fI
I_I¢IOTI S,lll//, OtjIEp CRACK "IlG ,OQ'r- WSU.L01C

\\ i
FIN(.

I_t*4,1l
I//,,n.,."
I _ / I
AAIIISTMINT

/
/
\. I I I " -- ' .-- '_

ITIF_ENER °fl/CP |TIE FENICI SPAR

¢OCUREO WITH TVP FRONlr AND HEAR
SPAR TYP ALl*

-""" ""MI [J
RIVt.AJll STITCH

_l._P
/
SOfTI[mN4 STRIP
,\,=, GN/IP T

_G1
• .141

!I
IN KIV_ _TTTCN..IS IqTCH
,,,
FOil I.|AOING IDOl ATTAC_ _ttP _r * 3_I I_ L IkI liAI[AI[I_Q S'TRII/iQTH _ SI[CTION C-C

TYPICALIC_I
ATTACH_FOllTIIIAILINq.
I_ 1034/1_/t_111 J LO0 TVff

I41tOl( F Jt.STf RIR.ENO_

OF i'TIFFINI[R ONLY
AJET1,Bt It L

299
 STRUCTURAL CONCEPTS

In achieving the objectives of the NCWFA Program, composite design concepts incorporated into the
baseline wing have been studied. These can be classified into two categories: multi-spar and multi-rib.
The structural arrangement considered spar/stiffener orientation, spar/stiffener spacing, and rib pitch. The
structural geometry was varied to achieve a least weight/cost cross section of detail structural elements.

Structural Concepts Trade Studies,

Wing Torque Box Structures

Torque box covers

- Flat monolithic (unbuckled)
- Discrete cap (unbuckled and buckled) - Multi-spar
- Lumped spar cap (buckled)
- Isolated discrete cap (buckled)

- Hat stiffened (unbuckled and buckled) -7

- Tee stiffened (unbuckled and buckled) [__ Multi-rib
- Wye stiffened (unbuckled and buckled) |
- Blade stiffened (unbuckled and buckled)_J

Spar webs
- Shear resistant
- Stiffened web (unbuckled and buckled)
- Sinewave Shear Resistant

Spar caps • Rib webs and caps

- Angle - Similar to spars
- Tee
- Wye

300
Structural Concepts
Structural Sizes/Weight Derivation

• Minimum weight design of each concept considered derived to compare

concepts on a common basis

° Structural sizing/weight analysis performed using multi-station-wise basis

Appropriate analytical methods used to determine least weight geometry &

thickness to satisfy strength/stability and stiffness for each
component/concept considered

• Analytical methods yield geometry/thickness that represent upper limit to

efficient use of structural material from which theoretical weights are derived

• Non-optimum factors (NOFs) applied to theoretical weights to project realistic

assembly weights

• NOFs derived from previous composite programs/studies account for weight

penalties associated with
- Load introduction
- Build-ups around cut-outs
- Fasteners
- Ply transition and smoothing
- Variation in laminate thickness and density

301
 MULTI-SPAR CONCEPT

Our multi-spar concept was derived from previous studies, which indicated a five spar configuration to
be the most efficient from both a weight and cost standpoint. A total of 15 fibs was used in our
arrangement.

Two types of wing cover configurations were evaluated which have the potential of successfully
increasing the working strain to levels at least 50% higher than those of the baseline. The two types
evaluated were plain panel-spread and discrete cap. The first type, plain panel-spread, is essentially a
monolithic skin of approximately constant thickness at any chordwise cut. In addition, the laminate
consists of the same family of lamina orientations (0 °, 90 °, +450) at any point. The second type, discrete
cap, utilizes a skin of two distinct laminate orientations. Between spars, the skin panel consists of a high
strain to failure laminate of 90 ° and +45 ° layers. The absence of 0 ° layers, in this panel has two additional
advantages: first, for a given thickness it will possess a higher resistance to buckling loads; and second,
the laminate's EA (extensional stiffness) is very low as compared to the total section, resulting in a lesser
axial load applied to the unsupported segment of skin. At each spar 0 ° layers are added to the panel
laminate resulting in a local pad. The 0 ° layers provide the axial filament control to the laminate and carry
the preponderance of axial load. Located over the spar, the high loads are rigidly supported, minimizing
any instability problems.

Structural Concepts
Upper Cover Configurations
Advantages Disadvantages

Spread zero upper cover (with "Y" • Least cover fab cost • Max rows of fasteners - assy
intermediate spars) cost impact
• Very efficient comp cover
when combined with Y spar • High strain concentration -
fasteners through laminate
• Easily repaired
with high % of 0-deg plies
• Good battle damage tolerance
'%// /Upper
'q_ covertly
I due to multiple load paths • Damage
material
tolerance
dependent
totally

• Excessive SIS shear carrying

material

Spread zero upper cover with • Minimizes number of spars • High strain concentrations -
integral 'T' stiffner • Reduced assembly costs fasteners through laminate
I I with high % of 0-deg plies
• Good battle damage tolerance
• Damage tolerance totally
due to multiple load paths
material dependent
FW--t
Difficult to repair

Discrete cap upper cover • Most efficient comp cover • Increased laminate tailoring
design for multi-spar • High strain concentration
configuration factor - fasteners through
=_ I l I Min number of substructure laminate with high % 0-deg
T Pa L T Pad attachments plies
nel_2_i_ n '"]'_
Excellent damage tolerance • Difficult to repair
due to multiple load paths &
compliant laminates

302
 Structural Concepts
Lower Cover Configurations
Advantages Disadvantages

Spread zero lower cover • Least cover fab cost • Max rows of fasteners - assy
- Tee & angle spar support • Very efficient comp cover cost impact
when combined with Y spar • High strain concentration -
• Easily repaired fasteners through laminate
with high % of O-deg plies
• Good battle damage tolerance
ir ' to multiple load paths • Damage tolerance totally
material dependent
• Excessive SIS shear carrying
material

Lower cover with integral "1" • Minimizes number of spars • High strain concentrations -
stiffener fasteners through laminate
• Reduced assembly costs
with high percent of O-deg
i_vv_q • Good battle damage tolerance plies
due to multiple load paths
• Damage tolerance totally
i i ' material dependent
• Difficult to repair

Discrete cap lower cover • Most effficient comp cover • Increased laminate tailoring
design for multi-spar • High strain concentration
2 in. configuration factor-fasteners through
• Min number of substructure laminate with high % O-deg
attachments plies
Pane, Jtl_l_ __
• Excellent damage tolerance • Difficult to repair
T ..- __} '
due to multiple load paths &
Wmax
compliant laminates

Spar Design Concepts

I , Fastener/fastener I , , Fastener/stitching I

/-r- Fr-
I

I
I

=V = I

Kevlar
s
B
I
stitching
r
I
f
I
I
I
/
I

I I ._;-_
Y spar Sine wave spar Channel spar Y spar-stitching Y spar J spar-stitching

303
 MULTI-RIB CONCEPTS

In general, subsonic patrol aircraft wings are relatively thick where strength/stability conditions
govern. Consequently it is of interest to investigate cover configurations in which load-carrying material
in the cross section is redistributed so as to achieve increased flexural stiffness over that of the relatively
uniform thickness cross-section multi-spar cover designs. This concept leads to the consideration of
longitudinal stiffening systems as an integral part of the cover itself. The integral hat and open J stiffener
configurations were considered in the study and are illustrated below in combination with the skin. The
designs were sized for all combinations of stiffener pitch equal to 4, 5, 6 and 7 in. and fib spacing equal to
15, 20, 25 and 30 in. Because of relatively high structural efficiency and potential ease of manufacture,
stiffeners parallel to the front spar were selected as the preferred stiffener orientation.

Cover Concepts, Multi-Rib

0 ° -- Gt/Ep SKIN CAP 90 °, -*45 = Gt/Ep SKIN

\ / / ,,

90 °, ±45 ° STIFFENER
CO-CURED WITH SKIN

0° - Gr/Ep STIFFENER CAP

INTEGRAL HAT-STIFFENED

0° - Gt/Ep COVER CAP 90 °, ±45 = Gr/Ep SKIN

90 °, :!:45 _' STIFFENER

COCURED WITH SKIN

0° - Gt/Ep STIFFENER
CAP

INTEGRAL Jo STIFFENER

304
 MATERIAL-ORIENTED DESIGNS WEIGHT SUMMARY

For the baseline wing and the various material-oriented high strain wing concepts, theoretical cover
and substructure weights were derived analytically on a multi-station basis utilizing t's generated for
the various forms of construction. To account for such weight items as penalties associated with load
introduction, attachments, cutouts, variations in laminate thickness and density, the theoretical weights are
multiplied by an empirically determined "non-optimum factor", thereby fielding a realistic assembly
weight.

A weight breakdown of the multi-spar design, for both material systems (IM7/8551-7A and G40-
80(0584 Gr/Ep), and multi-rib design (IM7/Tactix 123 Gr/Ep) is shown below. The multi-rib design
concept consists of four hat stiffeners outboard of the fold, seven inboard and twenty-five ribs at a 25-inch
pitch.

The weight savings generated by these concepts show significant improvement over the baseline. The
multi-spar design, using G40-800/F584, provided a savings of 543 lb, or 42% of the metal torque box
weight of 1233 lb. The multi-rib design, using IM7/Tactix 123 Gr/Ep, yielded a 541 lb savings or 41% of
the metal torque box weight.

Weight Breakdown for Selected Wing Designs (Material-Oriented)

BASEUNE WING MUL_PAR MULTI-SPAR MULTI-SPAR (CONCEPT C)

COMPONENT TORQUE BOX (LBS) IM7/8551-7A (LBS) G40-800/F584 (LBS) HAT-STIFFENED (LSS)

UPPER COVER 293.80 258.30 246.20 266.60

• BASIC INTERSPAR COVER 240.50 223.90 211.80 266.60

• SOFTENING STRIP PENALTY 2.60
(00 GL/EP)
• DAMAGE TOLERANCE PENALTY 10.60
(0 ° GL/EP)
• DAMAGE TOLERANCE PENALTY 0.00
(+45 ° GIJEP)
• SPAR CAPS (INC WRINKLING 40.10 34.40 34.40
PENALTY)

LOWER COVER 224.10 192.40 182.70 185.70

• BASIC INTERSPAR COVER 187.20 175.30 165,60 185.70

• SOFTENING STRIP PENALTY 2.20
(0" GL/EP)
• DAMAGE TOLERANCE PENALTY 8.00
(0 ° GL/EP)
• DAMAGE TOLERANCE PENALTY 1.50
(+45" GIJEP)
• SPAR CAPS (INC WRINKLING 24.90 17,10 17.10
PENALTY)
88.60 73.80 73.80 98.80
FRONT SPAR (SYNCORE STIFF) 36.00 32.90 32.90 40,10
REAR SPAR (SYNCORE STIFF) 110.00 101.80 101.80
INTERMEDIATE SPARS (SYNCORE
STIFF)
52.20 52.20 52.20 96.60
RIBS
8O6.20 711.40 689.60 687.80
TOTAL TORQUE BOX 94.8O 116.60 118.40
WEIGHT SAVINGS OVER BASELINE
% SAVINGS 11.80 14.50 14.70

305
 COMBINED MATERIAL/CONFIGURATION CONCEPTS

After completing the material-oriented design concepts, our efforts were directed toward developing
combined material/configuration concepts. This involved the use of Y-spars and Y-stiffeners to support
the covers. For the upper cover, the following concepts were studied:
1) Spread 0 ° supported by 5 or 6 Y-spars
2) Isolated discrete cap supported by 5 or 6 Y-spars
3) Isolated discrete cap supported by integrally cured Y-stiffeners

The basic philosophy in using intermediate Y-spars is that they reduce panel widths and required
thickness on the upper cover. Although an increase in weight is expected for the intermediate spars, the
weight savings produced by the upper cover will adequately compensate for it, and yield an overall weight
savings.

The same loading conditions that sized our previous spar concepts were used to size the Y-spar
configuration. For all Y-spar designs, the angle was set at 120 ° to provide equilibrium and balance. The
distance between the legs of the Y-spar at the attachment to the upper cover depends on the spar spacing.
To obtain the maximum benefit from the Y-spar configuration, the fastener spacing is half the spar
spacing.

G40-800/F584 tape was used for the upper and lower covers and 3-D IM7/Tactix 123 and 3-D
commingled AS4/PEEK 150G weave was utilized to size the Y-spars. For the multi-spar concepts, only
the intermediate spars were designed with the Y configuration. The SynCore-stiffened design was used
for the front/rear spars.

The corresponding lower cover design configurations were as follows:

1) Spread 0 ° supported by 5 or 6 Y-spars
2) Isolated discrete cap supported by 5 or 6 Y-spars
3) Spread 0 ° supported by blade stiffeners.

306
 MATERIAL- AND CONFIGURATION-ORIENTED DESIGNS

Weight Summary - Similar to previous design concepts, the theoretical weights, derived for the
material/configuration concepts, were multiplied by an empirically determined non-optimum factor, to
yield a realistic assembly weight. The table below shows a weight breakdown of the multi-spar designs
(spread 0 ° and discrete cap) and multi-rib design. The weight savings generated by these concepts show
significant improvement over the baseline. The multi-rib design, using G40-800/F584 with Y stiffeners,
provided the greatest savings (573 lbs. or 46% of the metal torque box weight of 1233 lbs.)

Weight Breakdown for Selected Wing Designs (Material & Configuration Oriented)

BASEUNE WING MULTI-SPAR MULTI-SPAR MULTI-SPAR

COMPONENT TORQUE BOX (SPREAD 0°) (DISCRETE CAP) (Y-STIFFENED)
(LBS) (LBS) (LBS) (LBS)

UPPER COVER 293.80 182.80 191.50 215.50

• BASIC INTERSPAR COVER 240.50 169.00 177.70 215.50

• SOFTENING STRIP PENALTY 2.60
(0° GL/EP)
• DAMAGE TOLERANCE PENALTY 10.60
(0° GL/EP)
• DAMAGE TOLERANCE PENALTY 0.00
(-+45° GL/EP)
• SPAR CAPS (INC WRINKLING PENALTY) 40.10 13.80 13.80

LOWER COVER 224.10 207.00 186.60 2O8.9O

• BASIC INTERSPAR COVER 187.20 200.20 179.80 2O8.9O

• SOFTENING STRIP PENALTY 2.20
(0 ° GL/EP)
• DAMAGE TOLERANCE PENALTY 8.00
(0 ° GL/EP)
• DAMAGE TOLERANCE PENALTY 1.50
(-+45° GWEP)
• SPAR CAPS (INC WRINKLING PENALTY) 24.90 6.80 6.80

FRONT SPAR (SYNCORE STIFFENED) 88,60 73.80 73.80 98.80

REAR SPAR (SYNCORE STIFFENED) 36.00 32.90 32.90 40.10
INTERMEDIATE SPARS (Y-SPARS) 110.00 159.00 159.00

RIBS 52.20 52.20 52.20 96.60

TOTAL TORQUE BOX 806.20 707.70 696.00 659.90

WEIGHT SAVINGS OVER BASELINE 98.50 110.20 146.30
% SAVINGS 12.20 13.70 18.10

307
 CONCEPT EVALUATION

Each design concept was rated in terms of the following parameters: weight, risk, manufacturing and
production costs, durability/damage tolerance, repairability, inspectability and operation and support costs.

With the concept rating forms, along with layouts, engineers from different disciplines were able to
rate the various novel composites wing concepts. These disciplines included Advanced Materials and
Manufacturing, Tooling, Design/Structural Analysis, Quality Control and Reliability/Maintainability. The
results are incorporated in the following table, for the multi-spar and multi-rib concepts. The Total Score
column represents the total of each discipline's score for that parameter. The Average Score column
represents the Total Score divided by the No. of R (rating disciplines), which is four in all cases. For
example, for the first parameter, Weight, Concept I received scores of 100, 125, 125, 100, from the four
disciplines, which resulted in a Total Score of 450 and an Average Score of 113 (450/4). The sum of the
Average Scores then represents the rating for that particular concept.

Concept Evaluation, Multi-Spar Concepts

CONCEPT I CONCEPT II CONCEPT IV CONCEPT V

IM716551-7A G40-800/F584 G40-80OlF584 G40-800/F584

PARAMETERS WEIGHTING SYNCORE-STIFFENED SYNCORE-STIFFENED SPREAD 0*/Y-SPARS DISCRETE CAP/Y-SPARS

FACTOR TOTAL NO. OF AVG TOTAL NO. OF AVG TOTAL NO. OF AVG TOTAL NO. OF AVG
SCORE R SCORE SCORE R SCORE SCORE R SCORE SCORE R SCORE

WEIGHT 25 450 4 113 7O0 4 175 575 4 144 675 4 169

RISK 10 3O0 4 75 3O0 4 75 240 4 6O 240 4 60

MFG ROT & E AND 18 552 4 138 528 4 132 552 4 138 552 4 138
PRODUCTION COSTS

DURABILITY AND 14 378 4 95 364 4 91 378 4 95 406 4 102

SURVIVABILITY

REPAIRABILITY 15 435 4 109 435 4 109 405 4 101 420 4 105

INSPECTION/ACCESS 10 290 4 73 290 4 73 220 4 55 220 4 55

OPS & SUPT COSTS 8 224 4 56 224 4 56 236 4 59 236 4 59

SUM OF AVG. SCORES 659 711 652 688

R90-4125-007

308
 CONCEPT SELECTION

The relative closeness of the ratings for Concepts II, V and VI, and subjective nature of the evaluation
make them virtually equivalent. However, continued effort will be directed towards Concepts V (Multi-
Spar: Y-Spar) and VI (Multi-Rib: Y-Stiffened), since they represent the latter stages of development and
the most potential to attain the program's goals.

Concept Evaluation, Multi-Rib Concepts

CONCEPT III CONCEPT Vl

IM7/TACTIX 123 IM7/TACTIX 123

HAT-STIFFENED Y-STIFFENED
PARAMETERS WEIGHTING
FACTOR TOTAL NO. OF AVG TOTAL NO. OF AVG
SCORE R SCORE SCORE R SCORE

WEIGHT 25 7OO 4 175 925 4 231

RISK 10 250 4 63 250 4 63

MFG RDT & E AND 18 456 4 114 480 4 120

PRODUCTION COSTS

DURABILITY AND 14 406 4 102 364 4 91

SURVIVABILITY

REPAIRABILITY 15 405 4 101 390 4 98

INSPECTION/ACCESS 10 260 4 65 250 4 63

OPS & SUPT COSTS 8 245 4 61 212 4 53

SUM OF AVG. SCORES 681 719

309
 DESIGN, ANALYSIS, FABRICATION AND TEST OF Y-SPARS

• Y-spar representative of intermediate wing spar segment in size, complexity and load carrying
capability (shear flow of 1,015 lb/in, in 5-spar wing configuration)
• Three 40-in. Y-spar preforms woven by Textile Technologies, Inc. on NASA Jacquard loom
using angle interlock fiber architecture
- Commingled AS4(6K)/PEEK 150g Tows
- 0/90-degree weave and +45-deg fabric stitched with Fiberglass/Toray H.S. thread
• Demonstrate structural integrity of Y-spars
- Two Y-spars destructively sectioned
• Mechanical & physical properties tests
• Standard for ultrasonic NDI
- One Y-spar tested in four-point beam bending

Commingled AS4/PEEK 150G

Y-Spar Configuration
D19B8220-13

I. 25
i 40.00 1

T > .102

I 0" 1.45" 190" I

I I OX I 50Y. I 40X I
T_ .135 (9.74) n'_

MAT ORIENTATION

I .12 R

T _ .102
r L//TYP

IF-_1.25

Q "Y" SPAR
_! [2.50)

_ "Y"SPAR
(_ "Y"SPAR

310
 MANUFACTURING EFFORT OVERVIEW

Commingled AS4/PPS 00/90 ° I-Beams

- Design and fabrication of woven commingled AS4/PPS 3-D I-Beam preforms
- Consolidation/forming of I-Beam preforms
- NDI and dimensional analysis of I-Beams

Commingled AS4/PEEK 150g Y-spars

- Design and fabrication of woven commingled AS4/PEEK 150g Y-spar preforms
- Consolidation/forming of Y-spar preforms
- NDI and dimensional analysis of Y-spars
- Structural test of Y-spar

311
 COMMINGLED AS4/PPS 0o/90 ° I-BEAMS

Textile manufacturers at present cannot weave 3-D preforms with 0 °, 90 ° and +45 ° fiber reinforcement
orientations. Preforms with fibers oriented 0°/90 ° to each other are woven to provide 3-D carcasses. Plies
of fabric are then stitched to these carcasses in 45°/135 ° orientations to provide preforms which when
consolidated will provide structural parts.

A series of woven commingled AS4(3K)/PPS 0°/90 ° I-beam carcasses were consolidated to provide
processing data for these intermediate preform configurations. Emphasis was placed on the consolidation
characteristics of these material forms and resultant percent fiber volume values. It was realized that the
structural properties of the consolidated 0°/90 ° I-beams were minimal (without 45°/135 ° reinforcement);
therefore testing for physical properties was emphasized in these processing studies.

Textile Technologies, Inc. (TI'I) was requested to weave three I-beam preforms using commingled
AS4/PPS fiber. The 3-D preforms were to be fabricated using commingled tows with a 60/40 graphite
fiber to PEEK 150g filament volume ratio. The 00/90 ° preforms were to be configured to accommodate the
following target dimensions after consolidation: length - 40 in., cap width - 5.5 in., and web height - 4.0
in. The thickness of the web was to be 0.072 + 0.006 in. The web flanges were to be 0.60 in. wide and
0.036 + 0.003 in. thick.

Woven Commingled AS4/PPS 00/90 ° I-Beam

312
 DESIGN METHODOLOGY

"Iq7 reviewed configuration and material requirements plus cost concerns for the proposed commingled
AS4/PPS I-beam. These customer requirements were then translated into an architecture for a woven 3-D
preform which upon consolidation would yield a structure meeting design specifications. Preparation of a
preform architecture is supported by the use of the interactive computer program Framework. This
computer program permits the textile designer to rapidly determine the effect of fiber mix on the final
thickness and fiber volume in given areas of the woven structure.

The loom is set up to accommodate the weaving requirements of the preform architecture. (A Jacquard
loom was used to fabricate the commingled AS4/PPS preforms.) A series of trial weaving runs are then
performed. Following each operation the appearance, dimensions and weight of the preform are
evaluated. Based on these determinations the preform architecture is refined as necessary to produce an
acceptable part. Since 3-D weaving is not an exact science this design/manufacturing iteration may have to
be performed a number of times.

REVIEW
CUSTOMER TRANSLATE
CONCERNS PART CREATE
-I_ PARAMETERS TO -_. DRAWING-IN
• PART REPRESENTATIVE SEQUENCE
DIMENSIONS WOVEN
• MATERIALS ARCHITECTURE
• COST

PATTERN
DESIGN
CHAIN
I
PREPARE
PATTERN
CHAIN

ILOOMSET-O
I-I
• REWIND
• CREEL
• DRAWING-IN
WEAVING
START-UP
PASS,F^,,
I - I'DRAW'N 'N
REDESIGN

WEAVING I

313
 WOVEN PREFORM ARCHITECTURE

The architecture of a given preform shows the relative orientation of the fibers used to build that fabric
part. In the case of the commingled AS4/PPS I-beam preforms, three fiber types (by function) were used:
fill, stuffers and warp weavers. The stuffers are continuous, non-crimped fibers, installed in the same
direction as the interlaced (with the fill fibers) warp weavers.

The architecture for the commingled AS4/PPS 00/90 ° I-beam describes fiber configurations in the web
and caps of the I-beam. The stepped area in the I-beam cap is also shown.

Architecture of Woven Commingled

AS4/PPS 0°/90 ° I-Beam

TOWS TH K

O FILL 4 .040"

-- STUFFERS 8 .064" (EXCEPTIN WEB SECTION)

WARP 4 .040"
N WEAVERS

MATERIAL: AS4 (3K)/PPS COMMINGLED

314
 FRAMEWORK
Consolidated Thickness Spread Sheet

The Framework is an interactive computer program that predicts the thickness of a consolidated
preform based on the physical characteristics of the fibers used by the weaver in the fill and warp
directions of the preform. These inputs are identified by an asterisk in the Framework shown for the web
of the I-beam preform. They are end. count (fiber ends per in.), denier (grams per 9000 meter) and fiber
density. In the case of the I-beam web, a thickness of 0.060 in. and a percent fiber volume of 62.8
percent were predicted for the consolidated part.

Consolidated Thickness Spread Sheet

Woven Commingled 3-D AS4/PPS I-Beam

HYBRID YARN FABRIC FIBER RESIN TOTAL

I-BEAM WEB

WARP * END COUNT (ENDS/IN) 128 128

• MANUFACTURER HERCULES BASF
° PRODUCT CODE AS-4,3K PPS
• DENIER (GR/M) 1963 900 2863
YIELD (YDS/LB) 2276 4965 1561
• DENSITY (GR/CC) 1.73 1.34 1.58
AREAL WEIGHT (GFUSQ M) 1099.1 503.9 1603.1 47.24
THICKNESS (MILS) 25.0 14.8 39.8 ^OZ/SQ YD ^
VOLUME FRACTION (%) 41.9 24.8 66.7
WEIGHT FRACTION (%) 45.7 21.0 66.7

FILL * END COUNT (ENDS/IN) 64 64

t
MANUFACTURER HERCULES BASF
t
PRODUCT CODE AS-43K PPS
*
DENIER (GR/M) 1963 900 2863
YIELD (YDS/LB) 2276 4965 1561
DENSITY (GR/CC) 1.73 1.34 1.58
AREAL WEIGHT (GPJSQ M) 549.6 252.0 801.5 23.62
THICKNESS (MILS) 12.5 7.4 19.9 ^OZ/SQ YD ^

VOLUME FRACTION (%) 20.9 12.4 33.3

WEIGHT FRACTION (%) 22.9 10.5 33.3

TOTAL AREAL WEIGHT (GR/SQ M) 1648.7 755.9 2404.6 70.85

FABRIC THICKNESS (MILS) 37.5 22.2 59.7 ^OZ/SQ YD ^
VOLUME FRACTION (%) 62.8 37.2 100
WEIGHT FRACTION (%) 68.6 31.4 100

DENSITY (GPJCC) 1.58

315
 JACQUARD WEAVING MACHINE

This machine was purchased by ?TI under NASA Contract NAS 1-18385. The machine features a
mechanism which permits individual control of every yarn weaving across the width of a given fabric or
preform. With a Jacquard head, weaving possibilities become infinite. It is this capability that permits the
weaving of 3-D preforms.

The NASA machine has a 2500 warp end capability.

316
 JACQUARD CARDS

Punched cardboard cards convey the fabric design or preform architecture to the Jacquard loom. The
cards operate much like the rolls of perforated paper which control the operation of player pianos. The
cards determine whether or not warp yarns are raised or lowered to accommodate the insertion of a fill
yam.

317
 WOVEN COMMINGLED PREFORM

Three 40 in. (length) woven commingled AS4PPS Oo/900 preforms were received from TI’I. The
visual appearances of the preforms were satisfactory; no floaters were detected in either the cap or web of
the preform and the parts were tightly woven. (A floater is a warp weaver that has relatively long
distances between successive interlacing fill yarns.)

The average thickness dimensions of the preforms were as follows:

Web: 0.627 in.
OuterCap: 0.1251 in.
Inner Cap: 0.297 in.

A -
B
I I
D

SECTIONS: A B C D

DIMENSIONS: 2.12 .6265 4.0 5.5

THICKNESS: 0.251‘ 0.297” 0.170‘ (SEEA 8 B)

318
 CONSOLIDATION/FORMING MANDRELS

It was decided to consolidate/form the commingled AS4/PPS I-beam preforms using matched
monolithic graphite mandrels. The male tools would insure achievement of the required web height and
thickness plus optimum forming/consolidation of the steps on the inner surfaces of the I-beam caps. Flat
0.125 in. thick aluminum caul plates were selected for use in consolidating the I-beam caps against the
underlying monolithic graphite mandrels.

I-Beam Consolidation/Forming Tools

(Monolithic Graphite)
4.00
REF

3.947
.026 TYP---_
REF

.09R
_, .09R REF .60 TYP

t
2.52
REF
,T± 3.5

40.0 _'-- NET TRIM OF PART REF

Q 2DETAIL
REQ'D

TOLERANCES NOT TO SCALE PRELIMINARY SKETCH

I-BEAM, WOVEN
.X = + .060 PER TI'I, INC. SKETCH
.XX =_+.030 DATED 19 JULY 1989
.XXX = + .005

319
 MONOLITHIC GRAPHITE TOOLING

Monolithic graphite was chosen as the material of construction for the tooling mandrels. The
coefficient of thermal expansion (CTE) for this material ranges from 1.5 to 2.0 x 10-6 in./in./OF. This
value compares favorably with the commonly used CTE range for GrEp composite laminates, 1.0 to 1.5 x
10-6 in./in.PF. (The CTE value for consolidated woven 3-D commingled AS4/PPS Oo/900structure is
presently not available.) Corresponding CTE ranges for competitive tooling materials such as aluminum,
steel and nickel alloy are 13.0-13.6, 6.1-6.7, and 6.6-7.4 x 10-6 in./in./'F, respectively. The thermal
conductivity of monolithic graphite is greater than that of steel and nickel tool alloys (68 vs 26 and 34
BTU-ft/ft2-hr-OF, respectively) but less than that of aluminum alloy (68 vs 117 BTU-ft/f$-hr- OF).
Aluminum tool alloys, however, are not useful at temperatures above 700'F whereas monolithic graphite
tools can be reliably utilized at temperatures up to 1000OF.

Monolithic graphite tools also have exceptional resistance to thermal shock at high temperatures due to
the material's low CTE, high thermal conductivity, high tensile strength and relatively low modulus of
elasticity.

Finally, monolithic graphite is very machinable and tools with precise tolerances (~0.003in.) and
smooth surfaces ( 4 0 rms) may be cost effectively manufactured.
 AUTOCLAVE VACUUM BAG ARRANGEMENT

The commingled AS4/PPS I-beams were consolidated using an autoclave/vacuum bag procedure. The
workpiece (1-beam preform installed between matched graphite mandrels) was installed on a steel base
plate. The workpiece was vacuum bagged to the baseplate using Kapton film material. A weighted steel
frame, in combination with medium and high temperature sealant materials, was used to provide a vacuum
tight bag over the workpiece.

The commingled AS4/PPS preforms were consolidated for 30 minutes at 630°F and 200 psi.

Woven Commingled AS4/PPS 00/90 ° I-Beam

Autoclave Vacuum Bag Arrangement
C GRAPHITE MOLD FORM
(MONOCOTE E-148B COATED/
FREKOTE 44 COATED)

(FREKOTE 44 COATED)
MONOLITHIC tiM CAUL PLT.
GRAPHITE (FREKOTE 44 COATED)
MOLD
FORM KAPTON TAPE

8 PLIES, 181 CLOTH

MONOLITHIC
GRAPHITE PLIES, 181 CLOTH
MOLD
FORM 4 PLIES, 181 CLOTH

TAPE

BASE PLATE (V-22 FLAT MOLD FORM) ]

(D19B1746-701 MOF)

VACUUM PORT, 2 PLCS., WITH PERFORATED ALUMINUM

PLATE OVER PORT
J
SASH CHAIN, CONNECTING 2 VACUUM PORTS IN LOOP, 2 ROWS --

HIGH-TEMP. SEALANT TAPE, GENERAL SEALANTS (GREEN)

MEDIUM-TEMP. SEALANT TAPE, SCHNEE-MOREHEAD 5150 (BEIGE)

HIGH-TEMP. SEALANT TAPE, GENERAL SEALANTS (GREEN)

.O02-THICK KAPTON FILM, PLEATED WHERE REQUIRED

_-- STEEL FRAME WITH WEIGHTS, ALL AROUND

321
 CONSOLIDATED I-BEAMS

The autoclave/vacuum bag consolidation procedure for I-beam - 1 was unsatisfactory. The graphite
mandrels hung up and did not properly close on the web of the preform. As a result, this I-beam did not
pass ultrasonic or dimensional inspection. The layup procedure for the workpiece and the vacuum
bagging arrangement WE refinedto ensure optimum closure of the forming dies. I-beam preforms -2 and
-3 were subsequently consolidated to provide I-beams that were acceptable with respect to ultrasonic and
dimensional NDI. The trim drop-off from each of the I-beams was used to provide coupons for percent
fiber volume analysis and edge photomicrographic inspection. Edge photomicrographic inspection
indicated that I-beams -2 and -3 were free of porosity. The average percent fiber volume for these
structures was approximately 57 percent.

Ultrasonic Inspection: No Defects

Edge Photomicrographic Inspection: OK
Percent Fiber Volume: 57%
Final Web Thickness: 60% That of Preform

322
 TARGET, PREFORM AND FINAL PART DIMENSIONS

Target, as received preform and final consolidated dimensions are presented for each of the three
I-beams. Emphasis is placed on final web thickness since a specific target value was given to TFI, i.e.,
0.072 + 0.006 in. As expected, the web thickness of I-beam -1 was unacceptable, 0.115 in. The web
thicknesses of I-beams -2 and -3, however, were acceptable, i.e., 0.065 and 0.074 in., respectively.

Average percent consolidation values for the web and cap areas of the preforms are presented for
I-beams -2 and -3. (Percent consolidation equals preform thickness - consolidated thickness divided by
preform thickness multiplied by 100.) The average percent consolidation value for the webs was 56.5
percent.

The 56.5 percent consolidation value was used in the shimming operation for consolidation/forming of
the AS4/PEEK 150g Y-spar preforms.

Woven Commingled AS4/PPS I-Beam

Target, Preform & Final Part Dimensions

CONSOLI DATED
PREFORM I-BEAMS PERCENT
T TARGET 1 2 3 CONS.

G ON.) (IN.) (IN.) (IN.) (IN.) (%)

0.251 0.112 0.112 0.102 59.4
B ::-0.120 0.297 0.144 0.140 0.144 51.5
C O.066 0.170 0.115 0.065 0.074 56.5

!c
-0.078
0.6 0.6 0.6 0.6
A B E >2.0 2.2 2.2 2.2 2.2
F >5.0 5.6 15.6 5.6 5.6
G 4.O 4.0 14.0 4.0 4.0

323
 COMMINGLED AS4/PEEK 150G Y-SPARS

Three 40 in. (length) commingled AS4/PEEK 150g Y-spars are being fabricated by the autoclave
consolidation of woven/stitched preforms. Cap details will be mechanically attached to the Y-spars to
provide elements suitable for destructive testing.

Y-spar 00/900 carcasses were woven by "vrI using their Jacquard loom. Sewing Machine Exchange,
Inc. (SMX), a qq'I subcontractor, stitched woven commingled AS4/PEEK 150g fabric in 45°/135 °
orientations to the carcasses to provide woven/stitched 3-D Y-spar preforms.

The Y-spar preforms are being sequentially consolidated by Grumman using vacuum bag processing
in an autoclave. Grumman is presently evaluating non-autoclave consolidation procedures which have the
potential to be more cost effective than the baseline process. These innovative processes include
thermoforming and the Therm-X process. Since the primary goal of our program is to provide reliable
engineering data and since a limited number of Y-spars are to be fabricated, it was decided to use the low
risk autoclave process. Furthermore, the autoclave process had already been demonstrated by the
successful consolidation of the commingled AS4/PPS I-beam preforms.

The Y-spar preforms are being sequentially consolidated using the following approach. The first
Y-spar will be inspected using ultrasonic NDI and dimensional analysis. The Y-spar will be destructively
sectioned and reinspected using dimensional and micrographic analysis. Design/Manufacturing/Quality
Control personnel will review these data and refine the consolidation procedure accordingly. The second
Y-spar will be consolidated and the above review process will be repeated. The third Y-spar will be
consolidated using the refined process and then destructively tested using four-point beam bending.

324
 COMMINGLED AS4/PEEK 150G Y-SPAR CONFIGURATION

'ITI provided three woven/stitched commingled AS4/PEEK 150g Y-spar preforms which upon
con solidation would meet modified configuration requirements of D 19B 8220-13. The percent fiber
volume of the consolidated preforms was to be 60 percent.

Commingled AS4/PEEK 150G

Modified Y-Spar Configuration

F
40. O0

1.25

. l ORIENTATZO[_ I
I O" I _4_" I 90" /

/
_ (5.90) --1
• 7._ ] 6_._,_ I 29_ ]_
T > .asJ

I OR]:ENTATIQN I
0" 1-_4_ • 190" I -.%

I _o7_ I _s._ I42.9_ 10.70 I 1.95

' T > .21s

MAT ORIENTATION
I ORIENTATION
. 2R
l 9" I ±4_" I 90"
'T _ lsl TYP

_- I .25

(2.50)

325
 ARCHITECTURE OF PREFORMS

The architecture of the woven commingled AS4/PEEK 150g 00/90 ° preforms is presented below. The
preform webs consist of 76.59 percent fill yarns, 19.14 percent warp stuffers and 4.25 percent through
the thickness warp weavers. The preform flanges consist of 75.00 percent fill yams, 18.75 percent warp
stuffers and 6.25 percent through the thickness warp weavers.

The critical dimension for the preforms was the web height as measured from the radius of the 90-
degree flange to the radius of the Y-flange, i.e., 10.70 in. This dimension was made important by the
decision to use male mandrels for consolidation of the preforms. Web and flange thicknesses were to be
such that upon preform consolidation to a 60 percent fiber volume, modified D 19B8220-13 target
dimensions would be achieved. A 55 percent consolidation value was expected.

Architecture of Woven Commingled

AS4/PEEK 150G 00/90 ° Preform
WEB SECTION

-_ __1-_10.70 _
NN
YARNS/ % FIBER BY
WEB SECTION INCH WEIGHT FLANGE SECTION
O FILLING 72 76.59
WARP 18 19.16

THROUGH-THE THICKNESS 4 4.25

FLANGE SECTION

O FILLING 36 75.00
WARP 9 18.75 MATERIAL TYPE: AS4/PEEK 150G
COMMINGLED YARN
THROUGH-THE THICKNESS 3 6.25

326
 FRAMEWORK
Consolidated Thickness Spread Sheet - Web

The Framework print-out for the web of the commingled AS4/PEEK 150g 00/90 ° Y-spar preforms is
shown below. The 27 ends/in, end count for the warp fibers was composed of 22 warp stuffers and 5
through-the-thickness weavers. The denier value for the AS4 and PEEK 150g fibers making up the tows
used in the warp and fill directions were 3927 and 1800, respectively.

Target thickness and percent fiber volume values for the consolidated web were 0.071 in. and 61.0
percent, respectively.

Woven Commingled 3-D AS4/PEEK150 G 0°/90 ° Y-Spar Web Preform

HYBRID YARN FABRIC FIBER RESIN TOTAL

WARP * END COUNT (ENDS/IN) 27 27

• MANUFACTURER BASF BASF
• PRODUCT CODE AS-46K PEEK
" DENIER (GR/M) 3927 1800 5727
YIELD (YDS/LB) 1138 2482 780
• DENSITY (GR/CC) 1.80 1.29 1.60
AREAL WEIGHT (GR/SQ M) 463.8 212.6 676.4 19.93
THICKNESS (MILS) 10.1 6.5 16.6 ^OZ/SQ YD ^
VOLUME FRACTION (%) 14.3 9.2 23.5
WEIGHT FRACTION (%) 16.1 7.4 23.5

FILL * END COUNT (ENDS/IN) 88 88

MANUFACTURER BASF BASF
PRODUCT CODE AS-4 PEEK
DENIER (GR/M) 3927 1800 5727
YIELD (YDS/LB) 1138 2482 780
DENSITY (GR/CC) 1.80 1.29 1.60
AREAL WEIGHT (GR/SQ M) 1511.7 692.9 2204.6 64.96
THICKNESS (MILS) 33.1 21.1 54.2 *OZ/SQ YD ^
VOLUME FRACTION (%) 46.7 29.9 76.5
WEIGHT FRACTION (%) 52.5 24.1 76.5

TOTAL AREAL WEIGHT (GR/SQ M) 1975.5 905.5 2881.0 84.89

FABRIC THICKNESS (MILS) 43.2 27.6 70.8 ^OZ/SQ YD ^
VOLUME FRACTION (%) 61.0 39.0 100
WEIGHT FRACTION (%) 68.6 31.4 100
DENSITY (GR/CC) 1.60

327
 FRAMEWORK
Consolidated Thickness Spread Sheet - Flange

The Framework print-out for the flange of the commingled AS4/PEEK 150g 00/90 ° Y-spar preforms is
shown below. The 15 ends/in, end count for the warp fibers was composed of warp stuffers and through-
the-thickness weavers. The denier value for the AS4 and PEEK 150g fibers making up the tows used was
again 3927 and 1800, respectively.

Target thickness and percent fiber volume values for the consolidated 00/90 ° Y-spar web carcass were
0.036 in. and 61.0 percent, respectively.

Woven Commingled 3-D AS4/PEEK150 G 0°/90 ° Y-Spar Flange Preform

HYBRID YARN FABRIC FIBER RESIN TOTAL

WARP " END COUNT (ENDS/IN) 15 15

* MANUFACTURER BASF BASF
* PRODUCT CODE AS-4 6K PEEK
* DENIER (GR/M) 3927 1800 5727
YIELD (YDS/LB) 1138 2482 780
° DENSITY (GR/CC) 1.80 1.29 1.60
AREAL WEIGHT (GR/SQ M) 257.7 118.1 375.8 11.07
THICKNESS (MILS) 5.6 3.6 9.2 ^OZ/SQ YD ^
VOLUME FRACTION (%) 15,5 9.9 25.4
WEIGHT FRACTION (%) 17.4 8.0 25.4

FILL * END COUNT (ENDS/IN) 44 44

°
MANUFACTURER BASF BASF
PRODUCT CODE AS-4 PEEK
o
DENIER (GR/M) 3927 1800 5727
YIELD (YDS/LB) 1138 2482 780
DENSITY (GR/CC) 1.80 1.29 1.60
AREAL WEIGHT (GR/SQ M) 755.9 346.5 1102.3 32.48
THICKNESS (MILS) 16.5 10.6 27.1 ^OZ/SQ YD ^
VOLUME FRACTION (%) 45.5 29.1 74.6
WEIGHT FRACTION (%) 51.1 23.4 74.6

TOTAL AREAL WEIGHT (GR/SQ M) 1013.5 464.6 1478.1 43.55

FABRIC THICKNESS (MILS) 22.2 14.2 38.3 *OZ/SQ YD ^
VOLUME FRACTION (%) 61.0 39.0 100
WEIGHT FRACTION (%) 68.6 31.4 100
DENSITY (GR/CC) 1.60

328
 CONSOLIDATION/FORMING MANDRELS

It was decided to consolidate/form the commingled AS4/PEEK 150g Y-spar preforms using matched
monolithic graphite mandrels. Monolithic graphite tooling was chosen because of the good performance
of the mandrels used to consolidate the previously discussed commingled AS4/PPS 0°/900 I-Beams.

It was recognized that these mandrels are bulky and would act as heat sinks during autoclave
consolidation of the Y-spar preforms. An optimum tooling approach would be to use integrally heated and
cooled monolithic graphite mandrels. The cost of this type of tooling for the consolidation of three
Y-spars was inconsistent with the scope of the subject program. It was considered more cost effective to
use the relatively less expensive solid graphite tools with relatively longer autoclave consolidation cycles.

The figure below shows the matched monolithic graphite mandrels fabricated by Coast Composites,
Inc. The tools are configured to accommodate Y-spar preforms with a web height of 10.70 in.

329
 UPPER AND LOWER BASE PLATES

Monolithic graphite upper and lower base plates were provided to ensure optimum consolidation/
forming of the Y and 90-degree flanges.
 WOVEN 0°/900 Y-SPAR PREFORMS

Three sections of Oo/900Y-spar preforms (each in excess of 40 in. long) were woven by 'IT1 using the
web and flange architectures previously described. A shorter section of Oo/900Y-spar preform was also
provided for physical and mechanical properties characterization testing.

Visual inspection of the Oo/900Y:spar carcasses revealed the presence of floaters (maximum length 2.5
in.) in the radii of the Y and 90-degree flanges. The floaters were due primarily to the use of 12K tows
l
and the requirement that only 5 percent of the fiber reinforcement be used in the Z direction of the preform.
The length and number of floaters could be reduced significantly by the use of 3K tows which would
provide more fiber ends per in. (but at increased cost) in the Z direction to tie down the Y direction fibers.
The percentage of fiber reinforcement in the Z direction could also be increased but with a corresponding
reduction of in-plane mechanical properties.

Potential reduction in mechanical properties of the consolidated Y-spar due to the presence of floaters
in the radii of those elements was minimized by the subsequent stitching operation for installation of
45'/135" fabric reinforcement.

The radii of the preforms (with 45'/135' fabric reinforcement located on the Oo/900Y-spar carcasses)
were stitched with three rows of stitches using an 1/8 in. row spacing. The remainder of the preforms
were stitched using a cross hatch pattern with 1/4 in. row spacings.

I The floaters are visible in the radius of the section of 0°/900woven preform shown in the figure below.
 CONSOLIDATION OF SECTION OF 0°/900Y-SPAR PREFORM

A section of woven commingled ASWEEK 150g Oo/900Y-spar preform was consolidated for
120 minutes at 720 f l O O F and 200 psi in an autoclave. The section of preform was consolidated using the
monolithic graphite tooling and the autoclave/vacuum bag process previously described for the
commingled AS4PPS 0"/90" I-beams.

The consolidated section of woven AS4PEEK 150g Oo/900 Y-spar preform was visually excellent. A
16 in. (0" direction) by 6 in. section of the Y-spar web was cut and inspected using ultrasonic NDI. The
section of Y-spar web was free of sonic discrepancies. The average thickness of the consolidated panel
was 0.140 in. Based on an as-received web thickness of 0.340 in., the percentage reduction in preform
thickness was 58.5 percent.

Resin content and fiber volume determinations for the consolidated panel were
Percent Fiber Volume: 55.8%
Percent Resin Volume: 43.6%
Percent Void Volume: 0.6%

. .. ..... .
.I
..
..
..
..
..
..
.
.._____._______I.
I .
.. I

332
 WOVEN/STITCHED Y-SPAR PREFORM

Required 4 5 O / 1 3 5 O fabric reinforcement was stitched to the woven Oo/900 commingled AS4PEEK 150g
Y-spar carcasses by SMX. As stated earlier, the preform was stitched using a cross hatch pattern with a
row spacing of 1/4 in. In the radius areas, however, three rows of stitches were installed with a row
spacing of 1/8 in.

It was intended that the preform be stitched using only Toray T-900-100 50A carbon fiber; SMX,
1 however, required the use of fiberglass loops in combination with the carbon fiber thread in the radii and
flanges of the preform. The carbon stitching equipment was too large to be conveniently used for the
Y-spar flanges. In addition, this equipment lacked the sensitive feeding characteristics required for the
flange stitching operation. Ultimately, the Y-spar preform flanges were stitched manually.
 CONSOLIDATION OF
Y-SPAR PREFORM S/N-l
Near Net Trimming

D19B8220-13 Y-spar Preforms S/N -2 and -3 will be fabricated using the procedures to be described
for Preform S/N-1.

Preform S/N-1 was installed on one of the tool mandrels and manually trimmed to near net dimensions
using scissors.

J
 INSTALLATION OF Y-SPAR PREFORM
BETWEEN TOOLING MANDRELS

The Y-spar preform is installed between the matched monolithic graphite mandrels. The mandrels are
undercut to accommodate the flanges of the Y-spar preform. The width of each of these undercut surfaces
is oversize with respect to the final mm of the corresponding Y-spar flange.

The 90" flange of Y-spar Preform S/N-3, as shown in the figure below, has been trimmed to a near net
dimension that accommodates that of the tool but is larger than the final dimension of the consolidated
Y- spar.

335
 INSTALLATION OF UPPER AND LOWER TOOLING PLATES

The near net trimmed Y-spar preform is installed between the matched mandrels. As shown in the
figure below, the mandrels are positioned on the autoclave table so that the web of the part is parallel to the
surface of the autoclave table. The upper and lower closure plates are then placed against the outboard
surfaces of the preform Y and 90" flanges.

336
 CONSOLIDATED (TRIMMED) Y-SPAR

The consolidated commingled AS4PEEK 150g wovedstitched Y-spar was consolidated for 4 hours at
720-L10°F, 160psi fluid pressure plus full vacuum bag pressure. The prolonged hold at elevated
temperature was required to accommodate the relatively large mass of the monolithic graphite mandrels
which acted as heat sinks. In production, integrally heated and cooled tools would be used in combination
with cold-wall autoclave procedures to provide a low-cost consolidation methodology. The high-
temperature autoclave run was performed without any processing difficulties. The consolidated Y-spar
was visually acceptable. Tap testing indicated a satisfactory consolidation.

337
 TARGET, PREFORM & FINAL PART DIMENSIONS
OF WOVEN COMMINGLED 3-D AS4/APC-2 Y-SPAR 1

The results of a preliminary dimensional inspection (the Y-spar will be sectioned to provide in-depth
dimensional analysis) are presented below. Target values for the consolidated oversize Y-spars were
provided by TTI. These values are estimates based on corrective inputs to their Framework computer
program to accommodate the greater mass of AS4 fiber used to fabricate the Y-spar preforms. (The
Framework is an interactive computer program that predicts the thickness of a consolidated preform based
on the physical characteristics of the fibers used in the fill and warp directions of the preform.) The table
also provides initial preform thickness values and corresponding consolidated thickness measurements.
Comparison of the last two measurement groups permitted calculation of percent reduction in preform
thickness values.

Target values for the web (A) and outer sections of the Y-spar flanges (D1 and D2) were essentially
realized by the consolidated structure. Target values for the inner sections of the Y-spar flanges (HI and
H2) and the 90 ° flanges (B 1 and B2) were not achieved. Indeed, the consolidated thicknesses for the two
inner Y-spar flanges were considerably different (0.170 vs 0.158 in.) indicating that the corresponding
side mandrel may have hung up during consolidation. Percent consolidation values for these flanges
support this hypothesis (H1 vs H2, 53.6 vs 60.2% and D1 vs D2, 58.2 vs 64.5%).

Woven Commingled 3-D AS4/APC-2 Y-Spar S/N-1

Target, Preform & Final Part Dimensions
CONS.
TARGET PREFORM CONS.
Y-SPAR
(IN.) (IN.) (IN.) (%)
_'_[ E1 _G'__/ E2 ]_ -
A 0.215 0.463 0.242 47.7

B1 0.151 0.381 0.134 64.8

B2 0.151 0.350 0.127 63.7

C 2.5O 2.74 2.50 N/A

F D1 0.151 0.366 0.156 58.2

--_ _"-A
D2 0.151 0.397 0.141 64.5

E1 1.25 0.9O

E2 1.25 1.00
91 92
F 11.65 11.65
G 3.40 3.34

i--c-i,
M R90.4125-034
H1

H2
0.151
0.151
0.366
0.397
0.170

0.158
53.6

60.2

338
 COST STUDIES FOR COMMERCIAL FUSELAGE CROWN DESIGNS 1

T. H. Walker, P. J. Smith, G. Truslove, K. S. Wiliden,

S. L. Metschan, C. L. Pfahl

Boeing Commercial Airplane Group

ABSTRACT

Studies were conducted to evaluate the cost and weight potential of advanced composite design
concepts in the crown region of a commercial transport. Two designs from each of three design
families were developed using an integrated design-build team. A range of design concepts and
manufacturing processes were included to allow isolation and comparison of cost centers. Detailed
manufacturing/assembly plans were developed as the basis for cost estimates.

Each of the six designs was found to have advantages over the 1995 aluminum benchmark in cost and
weight trade studies. Large quadrant panels and cobonded frames were found to save significant
assembly labor costs. Comparisons of high- and intermediate-performance fiber systems were made
for skin and stringer applications. Advanced tow placement was found to be an efficient process for
skin layup. Further analysis revealed attractive processes for stringers and frames. Optimized designs
were informally developed for each design family, combining the most attractive concepts and
processes within that family. A single optimized design was selected as the most promising, and the
potential for further optimization was estimated. Technical issues and barriers were identified.

INTRODUCTION

During the 1970s, high fuel prices dictated a focus on reduced weight in aircraft design. The lower
fuel prices in recent years, in conjunction with a highly competitive aircraft marketplace, have forced
airframe manufacturers to consider the affordability of weight savings. Past applications of composite
materials have demonstrated their ability to reduce weight, but typically at significantly higher costs.

The relative lack of experience with composite materials results in increased risks in both cost and
performance. Boeing, therefore, requires potential composite applications to not only have significant
weight savings, but also to have costs less than or equal to aluminum alternatives. NASA has also
recognized the need for affordability and funded the Advanced Composite Technology (ACT)
program, with the objective of developing the technology required for cost-effective application of
composite materials to primary aircraft structures. The specific goals of the program are to obtain 25-
30% cost savings and 40-50% weight reductions from current airframes for a resized all-composite
airframe.

The emphasis of Boeing's Advanced Technology Composites Aircraft Structures (ATCAS) contract,
funded under the ACT program, is on pressurized commercial transport fuselages. The approach is to

1This work was funded by Contract NAS1-18889, under the direction of J. G. Davis and W. T.
Freeman of NASA Langley Research Center.

339
develop and demonstrate innovative composite fuselage structural concepts that are cost and weight
effective. Boeing selected the fuselage section immediately behind the wing box and main-landing-
gear wheel well area for this technology development and verification effort. The 767-X development
airplane, targeted for production in 1995, was selected as the metal benchmark to provide a
comparison with state-of-the-art aluminum technology.

Recurring labor is a major cost center in metal aircraft structure, due primarily to the low raw material
costs. The large amount of assembly required in aluminum structure results in assembly activities
accounting for nearly half the recurring labor costs, as shown in Figure 1. This, combined with
indications of strong interactions between design details and manufacturing costs, led to a decision to
consider assembled structure during concept selection. Manufacturing and cost personnel were
included in the design team to ensure these areas were addressed early in the design cycle, where
changes have the largest impact. The design-build-team (DBT) process, which is discussed more fully
in Reference 1, involves detailed cost- and weight-sensitivity studies (referred to as "global
optimization") to determine the best overall design concept. These studies are followed by "local
optimization," which includes subcomponent and element tailoring within the selected design concept.

Element Fabrication

Figure 1: Recurring Labor for a Typical Boeing Metallic Aft-Fuselage

The study section of the fuselage was divided into four quadrant panels: crown, keel, left side, and
right side. Each of these panels was treated separately. This paper addresses the detailed cost and
weight studies of the global optimization process for the crown quadrant panel, which was
accomplished during the period of January 1990 through July 1990.

340
 COSTING PROCESS

An industrial engineering approach was used to develop cost estimates of each design concept. The
completed design layouts contained information needed to generate detailed manufacturing plans.
These plans defined each individual process required to fabricate and assemble the finished quadrant
panel, and the tooling, labor, and recurring material requirements to support each. In-house historical
data and vendor-supplied estimates were used to define machine capabilities, process limits, process-
based material utilization rates, materials costs and labor for individual operations. Process variables,
such as learning curves, and shop variances, were also developed from historical and vendor-supplied
data. Detailed costs for each operation were then generated and summed to provide various levels of
cost visibility.

In developing manufacturing plans and cost estimates, an automated factory was assumed for
definition of equipment and tooling. Reduction of part handling and the combination of operations
were significant considerations. Estimates were based on a production run of 300 airplanes at a rate
of 5 per month. Costs for materials were based on only this 31 ft. crown panel at these production
rates. Labor rates of $100 and $75/hr. were used for recurring and non-recurring labor, respectively;
1995 dollars were used in all cases. Capital equipment costs were not included.

DESIGN DEVELOPMENT

Design Conditions

Designs developed for this study were sized to loading conditions representative of the baseline
metallic fuselage. Hoop and axial damage tolerance, Euler stability, and bolted joint strength were all
considered. A minimum gage of 10 plies (0.074 in.) was used to minimize repair requirements after
severe hail storms.

Predictions of damage tolerance strength were made using a method based on References 2 and 3, and
Boeing-internal studies. Analysis of crown configurations using this method indicated that "failsafe"
tension damage tolerance conditions are more critical than "ultimate" or "safe-flight" conditions. An
axially oriented 8 in. narrow slot in conjunction with a hoop tension loading of 1260 lbs./in, was used
for the hoop damage tolerance. A circumferentially oriented 8 in. narrow slot in conjunction with an
axial tension load distribution, ranging from 2900 to 3600 lbs./in, on the forward end and from 1400 to
1800 lbs./in, on the aft end, were considered for axial damage tolerance.

"Ultimate" tension loads were used for determining joint strength. The ultimate hoop loading is 2200
lbs./in. The ultimate axial loads range from 5400 to 6800 lbs./in, on the forward end and from 2700 to
3500 lbs./in, on the aft end.

Column stability was determined with ultimate compressive loading. These axial loads range from
2300 to 2700 lbs./in, on the forward end and from 1100 to 1400 lbs./in, on the aft end.

Design Families

Early developments by the ATCAS DBT prompted a need for an efficient method of studying
candidate fuselage design concepts and manufacturing processes. Initially, 30 candidate fuselage
panel concepts were produced by design personnel. The number of concepts was increased to 159
during subsequent brainstorming sessions. Schedule constraints would not allow cost and weight
evaluations of all concepts. Instead, concepts were classified into eight design families, each having
common manufacturing characteristics.

341
Three of these families are permutation of the skin/stringer/frame concept, with differing amounts of
cocuring and cobonding of individual elements. Two families are variations on sandwich construction
with circumferential frames. Other families include geodesic stiffening, integrally stiffened skins, and
continuous 360 ° fuselage concepts.

Crown Designs

Three families were selected for consideration in the crown panel studies, based on their perceived
potential for cost- and weight-effectiveness. Family B is a traditional skin/stringer/frame geometry,
with the stringers cobonded or cocured to the laminate skin. The frames are mechanically attached to
the stiffened panel. Family C is also a skin/stringer/frame geometry, with both the stringers and
frames cobonded or cocured to the laminate skin. Family D is a sandwich geometry, with cobonded
frames to provide hoop stiffening.

Two designs were developed from each of these three families. In developing the concepts, cost-
minimization was a major consideration. In addition, a range of concepts for each element (i.e., skins,
stringers, frames) was included within and across the design families to isolate costs.

Table 1 itemizes the important features of the two Family B designs. The major differences are (1) the
skin and stringer materials, (2) the stringer geometry and fabrication process, and (3) the frame
fabrication process. Design B 1 uses IM62/3501-63 for the skins and drape-formed hat stringers. The
frames for this design are compression-molded fabric. Design B2 uses AS44/3501-6 for the skins and
pultruded non-tapered blade stringers. The frames are fabricated by resin-transfer-molding (RTMing)
braided preforms.

Design B1 Design B2

Skin Mat'l, Form IM6/3501-6, Tow AS4/3501-6, Tow

Mfg. Process Batch Tow-Placement Batch Tow-Placement

Shape Hat Blade

Stringers Mat'l, Form IM6/3501-6, Tape AS4/3501-6, Prekitted Tape
Mfg. Process CTLM/Drape Form Pultrusion

Shape Z Z
Frames Mat'l, Form AS4/3501-6, Fabric AS4/DPL8625, Braid
Mfg. Process Compr. Molded Batch RTM (4)

Table 1: Major Features of Family B Designs

2IM6 is a graphite fiber system produced by Hercules, Inc.

33501-6 is a resin system produced by Hercules, Inc.

4AS4 is a graphite fiber system produced by Hercules, Inc.

5DPL862 is a resin system produced by Shell Oil Corp.

342
Table 2 compares the two designs of Family C. The primary variables between the two concepts are
(1) the stringer geometry, (2) the frame geometry, and (3) the frame manufacturing process. In Design
C1, the hat stringers are constant-height and thickness-tapered. The frames for this design are
manufactured using braid/RTM technology, and the frame flanges are bonded to the skin and entire
stringer cross section. In Design C2, the hat stringers are constant-thickness and height-tapered. The
frames are fabricated from a long discontinuous fiber (LDF6)/PEKK using a stretch forming operation.
The frame flanges bond only to the skin and attached stringer flanges. "Mouse-hole" cutouts in the
frame at the stringer-frame intersections allow the stringers to pass through.

Design CI Design C2

Skin Mat'l, Form IM6/3501-6, Tow IM6/3501-6, Tow

Mfg. Process Batch Tow-Placement Batch Tow-Placement

Shape Thickness-Tapered Hat Height-Tapered Hat

Stringers Mat'l, Form IM6/3501-6, Tape IM6/3501-6, Tape
Mfg. Process CTLM/Drape Form CTLM/Drape Form

Shape Contoured-Flange J Mouse-Holed J

Frames Mat'l, Form AS4/DPL862, Braid AS4/PEKK, LDF
Mfg. Process Batch RTM (16) Stretch Form

Table 2: Major Features of Family C Designs

The major features of the two Family D designs are shown in Table 3. Primary differences include (1)
the skin material, (2) the panel edge concept, and (3) the frame manufacturing process. Design D1
features AS4/3501-6 skins, with a ramped-down panel edge. The frames for this design are fabricated
using braid/RTM techniques. IM6/3501-6 is used for Design D2 skins. A square-edge panel concept
is employed, and the frames are manufactured using a dry-fiber pultrusion process.

Design D1 Design D2

Skin Mat'l, Form AS4/3501-6, Tow IM6/3501-6, Tow

Mfg. Process Batch Tow-Placement Batch Tow-Placement

Shape Ramped-Edge Square-Edge

Core Form Aramid Honeycomb Aramid Honeycomb
Thick., Density 0.5 in., 4 lb./ft 3 0.5 in., 4 lb./ft 3

Shape J J
Frames Mat'l, Form AS4/DPL862, Braid AS4/3501-6, Raw Mat'ls
Mfg. Process Batch RTM (2) Dry-Fiber Pultrusion

Table 3: Major Features of Family D Designs

6LDF is a trademark of E. I. du Pont de Nemours & Co.

343
 RESULTS

From the design studies conducted, several generalizations can be made concerning design drivers.
Tension failsafe damage tolerance controls the majority of the panel in all designs. Stringer
thicknesses are determined by Euler stability considerations. Skin and stringer thicknesses at the edges
of the panels axe controlled by the fastener bearing requirements. Minimum gage requirements affect
the skin thicknesses near the aft (lightly-loaded) end of the panels.

Figure 2 illustrates the approximate weight breakdown observed in the six design concepts. The
stiffened panel (skin and stringers, or sandwich panel) accounts for 70 to 80% of the total crown
quadrant weight. In the skin/stringer designs, the skin is approximately 50% of the total, and the
stringers 20%. In both concepts, frames account for 10 to 15%, and the splices 5 to 10% of the total
weight.

100

Skin

8o _=!_Stringers

U Frames

O i!!!i!!iSplices
I-
6O
O

¢.. 4O

2o

Skin/Stringer Sandwich

Figure 2: Approximate Weight Breakdown of Crown Designs

Figures 3 through 5 illustrate cost results specific to Design C1, but they reflect trends inherent to all
the designs. The relationship between recurring and nonrecurring costs are shown in Figure 3.
Recurring costs comprise approximately 75% of the total costs and are divided nearly equally between
material and labor.

In Figure 4, the recurring and nonrecurring costs are each separated into fabrication, panel bonding,
and assembly/installation costs. About half of the nonrecurring costs are related to element fabrication
(e.g., skins, stringers, frames, etc.), with costs relating to bonding and assembly operations comprising
the other half. In contrast, approximately 70% of the recurring costs are related to element fabrication,
with the remainder related to bonding and assembly operations.

344
 N/R LABOR 24%
REC MATL 35%

N/R MATL 3%

REC ASSY 6%
REC FAB 32%

Figure 3: Recurring and Nonrecurring Costs of Design C1

ASSY & INSTL 13%

ASSY & INSTL

PANEL BOND

3 AB49%

PANEL BOND12% FAB 63%

TOTAL NONRECURRING
LABOR & MATERIAL I I LABOR
TOTAL RECURRING
& MATERIAL

Figure 4: Breakdown of Recurring and Nonrecurring Costs of

Design C1

345
 25%

[_
[] ECURRING
RECURRING MATERIAL
LABOR /
20%
17.3% 16.9%
Note: The percentages are of total costs
I- (recurring plus nonrecurring)
Z 15% 13.2%
iii \\\

O
tr
iii
O. 10% 6.2%

4.1% 3.8%
5%
2.2%

1.1% _
0%
SKIN I PANEL BOND I'RAMESPUCES
IS'RORSPL'CES
I BODYJO,N
STRGRS FRAMES CIRC. SPLICES PANEL INSTL

MAJOR OPERATIONS

Figure 5: Design C1 Recurring Costs by Major Operation

C2 _
140 Family B ........i................ Oi ..... _ ...... i....................... _.........
[] ! : ,. !

120
F-n,.y° ........
i.......................
i................
_ ................
_.
.........
O i B10 ° _ A _ !
_ 0 B2 _ D2i _ i 1
Family D :: C1 :: D1 i _i. /
X, 100
f_
¢0
80
I-I Aluminu_ (767-X) 1 :: ! :: ::
o_
u) 60
J
O

40

20
-....................
i...................................................................
i.........
-
0 i i
0 20 40 60 80 100
Weight, % 767-X

Figure 6: Crown Design Cost/Weight Results

346
The recurring material and labor costs for each major manufacturing step are shown in Figure 5. The
most significant recurring cost centers are (1) the fabrication of the skin, stringers and frames; (2) the
panel bonding operation; and (3) fasteners required for installation.

A comparison of the cost/weight results, normalized to that of the 767-X aluminum baseline, is
contained in Figure 6. The costs, in general, are within 100 to 120% of the baseline value, with Design
C2 being the notable exception. The skin/stringer concept weights are approximately 50 to 60% of the
baseline, with the two sandwich concepts (i.e., Designs D 1 and D2) being somewhat higher.

The sloped line through the 767-X baseline point in Figure 6 represents a typical performance value of
weight. This value is the amount that customers are willing to pay for reduced weight, and therefore is
a measure of the life cycle costs of this weight. All designs falling on a single line parallel to this are
of equal value. Those designs falling below this line are more desirable, and those falling above, less
desirable. As shown, all composite concepts, although not optimized, are more attractive than the
aluminum baseline. Note that two of the designs (i.e., Designs C2 and D2) are less attractive than the
other four composite designs.

ANALYSIS OF RESULTS

Cost Comparisons

As shown in Figure 2, structural weight is dominated by that of the skin. All designs studied used
advanced tow placement and a quadrant panel batch process to layup the skin plies (see Ref. 1). This
process was found to be cost efficient in several ways. First, large quadrant panels minimize labor
costs for final assembly (see costs for panel installation and body join in Figure 4). As shown in
Figure 1, assembly labor is significant for metal structure. Another advantage to advanced tow
placement in a batch quadrant process is the relatively low percentage of recurring labor costs.
Finally, prepreg tow is projected to be a low-cost material form in 1995 (e.g., $25/lb. for AS4/3501-6
tow prepreg).

Since the stiffened panel is a major cost and weight center, the impact of the fiber system used in these
locations is of extreme interest. In all designs, either AS4/3501-6 or IM6/3501-6 material was used for
the skins and stringers. An analysis of the cost trends involved in replacing IM6 fibers with AS4 is
shown in Figure 7. The bars in this graph show the ratio of AS4 to IM6, with the adjacent bars
addressing the behavior observed in the skins and the stringers. The lower performance of the AS4
fiber system results in a 10 to 15% weight penalty due to added material. However, the cost per pound
of the AS4 is significantly lower, ranging from 55 to 65%, depending on the material form (i.e., tow or
tape). The final material-cost ratio is the product of these two values, and is in the 70% range.
Additional recurring labor costs, which are incurred due to the increased number of plies, are not
included in this analysis. However, even when they are included, the cost advantages of AS4 clearly
outweigh its performance penalty, for these applications. A performance value of weight in excess of
$130/lb. would be required to select the IM6 system, significantly higher than normal values.

Similar results are seen in an analysis of the sandwich designs. In Design D1 and D2, a hard
(0°-dominated) AS4 concept and a soft IM6 (+_45°-dominated) concept are used, respectively.
Additional studies of soft AS4 and hard IM6 concepts indicate that, while the hard concept is slightly
more cost effective than the soft concept, the large cost reductions are gained by changing to the less
expensive AS4 system.

347
 1.2
• Skins

iiiiiii!iStringers

ID
0.8

0.6
<
._o
t_ 0.4

0.2

o
Weight (Performance) Cost/Lb. Structural Cost

Figure 7: Comparison of AS4 and IM6 Fiber Systems

A comparison of stringer fabrication processes is contained in Figure 8, which illustrates the costs per
pound of each stringer design. The pultrusion process used on the Design B2 blade stringers results in
cost of $175/lb. primarily due to the cost of the prekitted prepreg material form. The hot-drape-formed
hat stringers of Designs B 1, C 1, and C2, are less costly on a per pound basis (= $110/lb.), although
recurring labor was higher than the pultrusion process. In a modified Design B 1, where AS4 material
is substituted for IM6 (including additional plies to meet the performance requirements), the cost per
pound is reduced to approximately $75/lb.

Several interesting conclusions result from this comparison. First, the pultrusion process is inherently
more efficient than tape laminating with hot-drape-forming, but the material costs must be reduced to
approximately $30 to $40/lb. for the end product to be a viable option. This suggests that dry fibers
with an in-line resin wetting process is likely required. Secondly, material is a major cost center in
stringer fabrication. Further reductions could possibly be obtained for the drape-formed stringers by
switching from a tape form to prepreg tow in fabricating flat charges for draping.

The frame manufacturing process used for each design is the primary variable for which costs can be
isolated. Costs per pound are shown in Figure 9. The stretch-formed LDF (Design C2) and the
compression-molded fabric (Design B 1) designs both result in relatively high costs ($212 and $170/lb.,
respectively). The LDF frames are purchased as finished parts, so their costs are considered as
recurring material costs. Because of this, the major cost centers of this process cannot be identified.
High recurring labor costs for the compression-molded fabric frame are due to hand layup of the fabric
charges. The costs per pound of batch processed braided/RTM frames exhibit a considerable range.
This is attributed to the number of frames in a batch process, with the costs reducing from $160 to
$90/lb. as the quantity of frames fabricated in a single operation increase from 2 to 16. The $126/lb.
costs of the dry-fiber pultrusion concept (Design D2) is lower than all other concepts, except the 16-at-
a-time batch RTM process.

348
 200

Recurrlng Matl
i_ Nonrecurrlng Total
Recurring Labor
150

JD

100
O
O

50

B2 i
B1 C1 C2 i
,Modified B1, i

AS4/3501-6 Blade IM6/3501-6 Hat AS4/3501-6 Hat

Prekitted Material Tape Material Tape Material
Pultrusion CTLM/Drape Form CTLM/Drape Form

Figure 8: Cost Comparison of Stringer Fabrication Processes

25O

• Nonrecurring Total

iiiiliiiil Recurring Matl

200
• Recurring Labor

J2 150
m

(n
O
0 lOO

50

C2 B1 D1 B2 D2 C1

LDF Fabric Braid/RTM Braid/RTM Dry Fiber Breld/RTM

Stretch Form Compr.- 2-at-a-time 4-at-e-time Pultrusion 16-at-a-time
Molded

Figure 9: Cost Comparison of Frame Fabrication Processes

349
Global Optimization

After understanding the cost estimates derived for individual design concepts, an "optimum" design
within each family was developed. Efficient materials, fabrication process, and element design
concepts that are included in the six designs described above, were combined to provide the most cost-
and weight-efficient crown panels for each design family. This process was significantly less
formalized than the original cost estimates but was necessary to provide a basis for determining the
best design.

The globally optimized Family B design is primarily the hat-stiffened concept from Design B 1. The
skins, stringers, skin splices and stringer splices are all converted to AS4/3501-6. The thickness of the
skins and stringers are increased by 2 plies and 1 ply, respectively, to maintain adequate damage
tolerance. The braided/RTM Z-section frames from Design B2 are used, although it is assumed that
the frames are RTMed 16-at-a-time. The results of the study show that this design concept is 98% of
the cost of 767-X baseline concept and 54% of the weight.

The globally optimized Family C design is a slight modification of the continuously bonded frame
concept of Design C1. The skin, stringer, and associated splice materials are converted to AS4/3501-6,
with 2 plies and 1 ply being added to the skin and stringers, respectively, for damage tolerance
requirements. The J-section frame with the contoured outer flange is maintained without modification,
since it already assumes an RTM batch process of 16-at-a-time. The results of the study show that this
design concept is 99% of the cost of 767-X baseline concept and 55% of the weight, nearly identical to
those for the globally optimized Family B design.

The globally optimized Family D design is close to that of Design D1. The only modification is the
cost-efficient frame batch process that RTMs 16 frames in one process. The skins are a hard
AS4/3501-6 concept. The ramped edge of the quadrant panel is maintained, although it has not been
established as clearly superior to the square-edge design. The results of the study show that this design
concept is 94% of the cost of 767-X baseline concept and 64% of the weight.

Selection Rationale

The cost and weight results of the globally optimized designs for each family are shown in Figure 10
with the original results. The Family B and C costs and weights are nearly identical, both being
approximately equal in cost and 50% of the weight of the 767-X. The sandwich design (Family D) is
slightly less costly, yet significantly heavier than either of Families B or C. The sloped line through
the Family B and C designs reflects a typical performance value of weight. When this is considered,
Family D is clearly not an optimum design concept.

For all concepts, damage tolerance requirements control much of the design. It was therefore a major
consideration in choosing the baseline crown concept for further development in local optimization. In
all designs, it is assumed, based on limited existing data, that additional skin padding (i.e., tear straps)
are not required for tension damage tolerance. Since longitudinally oriented cracks appear to be the
more critical condition, the Family B concept seems to be at most risk from this assumption. Families
C and D have integrally bonded frame flanges to provide some crack stopping capability, where
Family B has no such features. Family D design appears to be at least risk, since the sandwich
construction increases the local bending stiffness at the crack tip, which in turn reduces the localized
bending stresses. Significantly higher strengths are realized in sandwich construction for pressure
loads with the 8 in. crack size considered for the failsafe condition.

350
 140 Family B
[]

Family C
120 O

X 100
CO
r_ 80
o_
u_ 60
O
o
Glob.

Glob.
Family

Opt.

Opt.
A

S
D

Family

Family
B

C
..............
!iiiiiiiiiiiiii
Glob. Opt. Family D

40
Aluminum (767-X)

20

o I i i I f
o 20 40 60 80 100
Weight, % 767-X

Figure 10: Globally Optimized Designs

Manufacturing risk was also a significant consideration in selecting the baseline concept. Family B
has the lowest perceived risk, since manufacturability of a mechanically fastened concept has been
previously demonstrated, although on wing/empennage-type structure. The Family C manufacturing
risk was judged to be the highest. Both Families C and D carry substantial risk associated with joining
very large, stiff sections to other quadrants and to other body segments. High local stresses can be
induced by forcing compatibility between warped panels. The overall stiffness of these built-up
designs magnifies the localization of these high stresses. The Family C design, however, has
additional complexity of splicing the stringers at the body-join operation. Maintaining the very small
locational tolerances required for these splices at both ends of a very long panel adds additional risk.

Family C was selected as the baseline crown concept. It demonstrates excellent cost/weight
performance, clearly superior to Family D. Its manufacturing risk was judged to be higher than that of
Family B, but it also carries significantly less performance (i.e. damage tolerance) risk.

Due to damage tolerance uncertainties in both Families B and C, Family D was selected as a backup to
the baseline. This provides a fall-back position if the apparent cost/weight performance erodes as
additional data on damage tolerance becomes available, or if the manufacturing concerns of Family C
cannot be overcome.

Local Optimization Potential

The local optimization process provides the opportunity to further refine the selected concept within
the cost constraints defined by global optimization. Material, geometric and laminate variables
affecting cost and weight are considered in the local optimization, as well as improvements in the
manufacturing processes. An effort was made to evaluate the magnitude of the potential cost/weight
changes that might occur during that process.

351
The use of less costly materials will be assessed, especially in the skins and stringers, where the
majority of material resides. Fiberglass, which exhibits a very large strain-to-failure, appears attractive
as a low-cost material for use in the skins and stringer flanges, where tension damage tolerance is an
important consideration. Intraply hybrid concepts using S-glass and AS4 material could provide
similar residual strength as compared to an all AS4 concept, resulting in a cost savings and a weight
increase. This intraply hybrid concept is also ideally suited for the efficient tow-placement process.
Other reductions in material cost may be realized by using dry-fiber pultrusion processes in the
stringers, and by tow placing the stringer charges for hot-drape-forming.

Major geometric variables to be considered include stringer spacing, frame spacing, stringer height and
width, and frame height and width. Major laminate variables include ply orientations and stacking
sequences. A software design tool that incorporates cost and structural mechanics constraints with an
optimization algorithm is being developed to support studies on the effects of material, geometric, and
laminate variables.

Several manufacturing improvements provide opportunities for cost savings. Tow placement
efficiency rates can be increased by simply enlarging the current band width or using multiple tow
placement heads. These technologies are considered to be low risk and could conceivably increase
rates up to 100%. A major cost center is the process for locating and bagging the quadrant assembly
for subsequent curing. The technology of form-fit reusable bagging offers significant cost savings by
reducing locating-tooling, recurring material costs, rejection rate, and assembly/bagging labor.

Panel splicing provides additional opportunities for improving both cost and weight. Composite
fasteners and rivets may be able to reduce the recurring fastener costs. Stringer splicing at the fore and
aft ends of the quadrant is also a major concern for Family C. Splice concepts that do not require
precise stringer alignment are attractive as a method of reducing the cost and risk in this area.

The potential cost and weight improvements in the local optimization process are shown in Figure 11,
along with similar potential for the aluminum baseline design. These aluminum improvements relate
primarily to breakthroughs in assembly technology, including high-speed robotic fastening. The
broader width of the composite potential is an indication of the wider range of materials and other
variables available in the local optimization process. The cost-reduction potential appears to be
significantly greater for the composite design than for the aluminum baseline.

TECHNICAL ISSUES AND BARRIERS

The most critical manufacturing issue associated with both the Family C (skin/stringer, bonded frame)
and Family D (sandwich) concepts is the effect of locational tolerances and panel warpage on final
assembly. Quadrant panel designs have high local stiffness due to bonded stringers/frames or
sandwich construction. This stiffness magnifies panel fitup and stringer splicing difficulties.

Another critical manufacturing issue is the control of fabrication processes to yield quadrant panels of
acceptable quality. Quadrant panel cost benefits assume that large panels will not be rejected due to
manufacturing defects. Further refinement of the designs will include robust designs and processes
which minimize potential problems due to fabrication anomalies.

352
 140
Glob. Opt. Family C ......................................................................................

120 Aluminum (767-X) ............................ .- .......................................................

X 100

r,o
I_ 80

8o
8
40

20

0 I I
0 20 40 60 80 1O0

Weight, % 767-X

Figure 11: Local Optimization Potential

Several critical performance issues require further understanding to ensure adequate performance and
properly refine the design. These include:

.
Hoop tension damage tolerance of panels with large penetrations. The most critical damage
geometries for hoop loading are expected to be slender notches oriented along the longitudinal
axis (Reference 3). The effectivity of bonded frames as "tear straps" needs to be determined.
The scenario of a penetration that severs a frame and skin must also be studied.

,
Axial tension damage tolerance of panels with large penetrations. The critical damage
geometries for axial loads are expected to be slender notches oriented along the circumferential
axis and severing a stringer. Both hoop and axial tension damage tolerance in the crown are
expected to yield lower strengths for unidirectional loading cases since the Poisson effect
reduces ply stresses for the biaxial tension case (Reference 4). However, the complex stress
distribution near a hard point could be most severe.

.
Axial compression stability of panels with and without damage. Euler stability and post-buckled
performance of the panel must be demonstrated with and without damage.
.
Minimum skin gage required to satisfy hail impact criteria. Structural tailoring in the crown is
limited by the minimum skin thickness requirement used to suppress visible damage due to
severe hail storm conditions.

,
Fiberresin distribution of frames after RTM processing. The performance, warpage, and
dimensional tolerance control of complex geometries such as curved frames is expected to relate
to fiber/resin distribution. Cost efficient methods of controlling the quality of RTM parts must
be established.

.
Performance of the stringerframe intersection. The intricate bond concept, with the frame
bonded to the entire stringer cross section, requires precise alignment of all parts. The mouse-

353
 hole concept alleviates some of the alignment concerns, but adds an additional stress
concentration in the frame. Sufficient damage tolerance for skin penetrations located near the
intersection must also be established.

. Durability of design details (cobonded frames and mechanically fastened splices). Cyclic
pressure load conditions are expected to drive the design of frame-to-skin adhesive joints in the
crown. Creep/fatigue interactions must be addressed. Potential bond-surface contamination,
resulting from poor handling, can also affect the durability of the frame-to-skin bond. Combined
cyclic load conditions also pose a significant problem for longitudinal and circumferential
mechanically fastened joints. The effects of environment and real time on damage accumulating
in material surrounding the bolt hole will need to be considered.

CONCLUSIONS

Several conclusions can be drawn from the above studies. First, and most important, composite
designs can be developed for fuselage crown applications that are significantly more attractive, from a
cost and weight perspective, than current aluminum concepts. Composite designs that are
approximately equal in cost and half of the weight of aluminum designs are possible.

Recurring costs account for approximately three-fourths of the total crown panel costs, with material
splitting this amount nearly equally with fabrication and assembly labor. Element fabrication (i.e.
skins, stringers, frames), panel bonding, and fastener costs are the major recurring cost centers.

Batch advanced tow placement layup of quadrant skin panels was found to be cost-efficient for all
designs studied. Large quadrant panels minimized assembly labor costs by reducing the number of
splices; however, manufacturing concerns about panel warpage and element locational tolerances need
to be addressed in future work.

In the crown applications considered, high performance fibers, such as IM6, do not appear to justify
their increased cost when compared with intermediate performance fibers, such as AS4. This
conclusion cannot be generalized to other applications.

Drape forming of flat uncured charges appears to be an efficient method of stringer fabrication.
Automated methods to create the flat charges from either tape or tow are important. Pultrusion also
appears attractive for stringer fabrication, but only using a dry-fiber method.

Resin-transfer-molding of braided preforms is an effective method for fabricating body flames. Batch-
RTMing provides significant cost advantages as the number of frames per operation increases. Dry-
fiber pultrusion also has potential for this type of structure.

A stiffened panel design concept with cobonded frames was selected for future studies that address
additional cost and weight savings. The local optimization potential of this design was judged to be
greater than that of aluminum designs. This relates to the larger number of composite design variables
and the relative maturity of aluminum technology.

Damage tolerance is the primary technical issue to be resolved for crown applications. Several issues
required further understanding to ensure adequate structural performance and allow proper
optimization. These issues include bonded and bolted joint durability, panel stability with large
damage, and the effects of fiber-resin distribution in RTM parts.

354
 REFERENCES

.
Ilcewicz, L. B., et al, Advanced Technology Commercial Fuselage Structures, First NASA
Advanced Composites Technology Conference, NASA CP- 3104, Part 1, 1991, pp. 127-156.

,
C. C. Poe, "A Unifying Strain Criterion for Fracture of Fibrous Composite Laminates,
Engineering Fracture Mechanics, Vol. 17, No. 2, pp. 153-171, 1983.

.
Chang, S. G. and J. W. Mar, "The Catastrophic Failure of Pressurized Graphite/Epoxy Cylinders
Initiated by Slits at Various Angles", AIAA Paper 84-0887, Submitted at 25th Structures,
Structural Dynamics, and Materials Conference.

.
Smith, P. J., L. W. Thomson, and R. D. Wilson, "Development of Pressure Containment and
Damage Tolerance Technology for Composite Fuselage Structures in Large Transport Aircraft,"
NASA CR 3996, Contract NAS1-17740, August 1986.

355
 A Unified Approach for Composite Cost Reporting
and Prediction in the ACT Program

W. Tom Freeman
NASA Langley Research Center

Louis F. Vosteen and Shahid Siddiqi

Analytical Services & Materials, Hampton, VA

Abstract

The Structures Technology Program Office (STPO) at NASA Langley Research Center has
held two workshops with representatives from the commercial airframe companies to establish a
plan for development of a standard cost reporting format and a cost prediction tool for conceptual
and preliminary designers. This paper will review the findings of the workshop representatives with
a plan for implementation of their recommendations.

The recommendations of the cost tracking and reporting committee will be implemented by
reinstituting the collection of composite part fabrication data in a format similar to the DoD/NASA
Structural Composites Fabrication Guide. The process of data collection will be automated by
taking advantage of current technology with user friendly computer interfaces and electronic data
transmission.

Development of a conceptual and preliminary designers' cost prediction model will be initi-
ated. The model will provide a technically sound method for evaluating the relative cost of different
composite structural designs, fabrication processes, and assembly methods that can be compared to
equivalent metallic parts or assemblies. The feasibility of developing cost prediction software in a
modular form for interfacing with state of the art preliminary design tools and computer aided design
(CAD) programs will be assessed.

Introduction

Boeing Commercial Airplane (BCA) Group and Douglas Aircraft Corporation (DAC) use
approximately 400,000 pounds of composites per year in spoilers, rudders, elevators, doors, and
other secondary structure. The rate of application of composites to empennage, wing, and fuselage
commercial airframe primary structure has been disappointingly slow. Composite materials are an
obvious choice for performance optimization, corrosion resistance, and fatigue suppression, but
before a bold leap toward more extensive use of composites can be expected in commercial applica-
tions, accurate cost prediction methods and confidence that production costs can be predicted accu-
rately must be demonstrated. The Advanced Composite Technology Program's goal is to establish
design concepts, develop manufacturing approaches, and demonstrate the structural integrity and
cost effectiveness of innovative low cost composite assemblies, providing confidence for production
commitment to primary structure by the turn of the century.

357
 The need to unify cost reporting and prediction methods for the Advanced Composites
Technology (ACT) program has been identified by industry participants during program reviews.
Accurate cost prediction is considered a high priority issue to assure a valid comparison of cost
effective structural concepts, material forms, and assembly methods being developed by the partici-
pants. The Structures Technology Program Office (STPO) has hosted two workshops with represen-
tatives from the commercial airframe companies to define
(1) a standard cost tracking and reporting format, and
(2) a development plan for a conceptual and preliminary design cost prediction model.

The preliminary design process has been identified as the most critical period of opportunity
for substantial cost reduction during an airframers hardware production cycle. Boeing has experi-
enced that 70% of airplane fabrication costs are fixed by the time the design is frozen and that the
influence of engineering on fabrication cost reductions is significantly reduced once the design is
completed. Concurrent engineering interdisciplinary teams are emphasizing cost evaluation at the
early stages of the development cycle in the preliminary design process. The advent of CAD/CAM
on powerful work stations provides the designer with the possibility of including cost as a compli-
mentary variable in the design process. A comparative cost algorithm, which can function purely as
an engineering design tool to evaluate different design concepts, would be exceptionally valuable to
concurrent engineering teams. As part of the overall NASA effort to improve the economic viability
of composite structures, the STPO plans to implement two activities related to composite costs:

1. Reinstitute and automate the collection of composite part fabrication costs in a format
similar to the DoD/NASA Structural Composites Fabrication Guide (Fab. Guide) (Ref. 1).

2. Determine the feasibility for development of a universally accepted academically rigorous

theoretical method for predicting the relative cost of different composite structural designs
in the preliminary design process.

The NASA-Industry Workshops

The first workshop on cost reporting and prediction (Ref. 2) was held in Norfolk, VA in
December 1989. The purpose of the workshop was to

.
Determine the procedures currently used by the industry to predict the production cost of
composite components and to determine if there was a need to develop or modify existing
methodology to account for new composite manufacturing processes such as tow place-
ment, resin transfer molding, and filament winding.

2. Establish a uniform procedure for reporting the costs of parts developed in the ACT
program.

Participants at this workshop were divided into cost reporting and cost prediction committees
and concluded that

1. Development of standard cost reporting and prediction methodologies were desirable.

2. Each company would identify representatives to serve on a steering committee to draft a

plan.
358
 3. The STPO should strive for implementation of unified reporting and prediction methods
by the last quarter of 1990.

The second workshop (Ref. 3) was held at Douglas Aircraft Corporation in Long Beach, CA
in February 1990. The purpose of the second workshop was to

1. Establish standard forms for cost collection and reporting.

2. Establish written requirements for a conceptual and preliminary designers' cost prediction
model.

Participants at this workshop were divided into cost reporting and cost prediction committees
and requested to report their recommendations to STPO by July 15, 1990. L.E. Meade (Lockheed
Aeronautical Systems Co.), chairman of the cost reporting group, indicated that representatives of
the three commercial companies agreed that the Data Abstraction Form developed for the Fab.
Guide adapted to a Lotus 123 spreadsheet format would be an acceptable form for the ACT program.

G. Swanson (BCA), chairman of the cost prediction group, prepared a committee report
(Ref. 4)* with the following recommendations:

.
NASA should take an active role in updating the composites data base with current state-
of-the-art cost data and manufacturing processes. A "subscriber" approach, wherein
contributors to the data base would have access to it, was suggested as one approach for
obtaining data in addition to the ACT program participants' hardware cost results.

2. NASA should ensure that the data base be kept current with long term support.

.
NASA should develop a producibility guide to assist design-build teams in making deci-
sions on a design concept. This document would supply information on selected manufac-
turing processes and provide information to the designer on types of design details to
avoid that would adversely affect cost. At the same time, large cost drivers would be
delineated. An implementation plan to address CAD interfaces would be required to
accompany the development of the producibility guide.

4. NASA should establish standard material costs (including future costs) to be used for
comparative costing studies and include them in the data base.

Cost Tracking and Reporting

As part of the ACT program, various airframe manufacturers will be designing and fabricat-
ing composite components that are more cost effective than previous composites or equivalent
aluminum structure. The components, of various sizes, will be made using low cost and automated
fabrication processes. In order to assess the cost effectiveness of the designs and their fabrication
processes, cost information must be acquired on the fabrication process. As noted above, the work-
shop committee suggested that the form originally developed for the Fab. Guide included all the
essential information and was familiar to the industry. Most Government programs on composite

*(Letter Report. See Ref. 4)

359
 structure development during the 1970's and the early 1980's included requirements for the comple-
tion of the "Fabrication Guide Data Abstraction Form". The effort to collect fabrication information
ended about 1983. More recent contracts have not had that requirement.

In revitalizing the data collection activity, STPO will attempt to automate and simplify the
process. A standard, unified cost collection Data Abstraction Form will be implemented via a
software module that easily allows the relevant manufacturing data to be collected and formatted for
subsequent inclusion in a fabrication cost data base.

The proposed cost tracking program will proceed in two stages:

1. The procedure for entering data will be standardized to a user friendly software interface
which is MS-DOS ®, Macintosh TM , and UNIX TM compatible.*

.
A data base with an appropriate data base management system (DBMS) will be estab-
lished to store the existing fabrication data as well as data acquired in the current pro-
grams. The DBMS will be selected so that the data base can be easily updated and sorted
to provide a variety of forms, charts and graphs. The data base will be accessible to
companies that contribute fabrication data.

The first task is essentially an evolution in the technology of the Fab. Guide Data Abstraction
Form (DAF). Interactive software will be developed to run under MS-DOS, Macintosh, or UNIX
systems. The software will be "intelligent" enough to prompt the user for only required input, and
present the user with a flow diagram of a composite structure manufacturing process. Figure 1
shows the hierarchical structure of the DAF. The diagram will be displayed to the user and boxes

D • _Fabrication_.___.
Gener-I -_ (_Structural_ (Material
= _Applicationjl T es,gn __
I companyI--i Military I- i Fiber h I Skin IL
Part Info I IC°mmerdal Matrix Sti.ener
I1
I NC Type I Tape lq I Sandwich [] I Bagging J-I
I Textiles
[ etc etc L.[ CUe..e_c

Cost )= _Experience_-7--'--_ _Control

Quality _-_
_:T_ T°°ling
Program
Repair Q Tooling 17 I Function
No.of Units
I Rework ] I Raw Marlh Shape [7
Material
I SamplingI-] Type/Use
I CureI- I Material
h
IMhr/Task
Mhr/ft2 ] I etc : 1 etc ]
mR&D
Prototype
-- Production
Figure 1. - Hierarchial structure of data abstraction form.

360
will serve as menus and "buttons" that allow the user to move directly to any section of the form.
Figure 2 shows screen images of a demonstration version of a portion of the data input form which
runs under HyperCard TM* on the Macintosh. The program is configured so that the cursor moves to
the next appropriate field after input has been entered.

Fairs M_rJl
C_mpany
0 Carbon 0 [poxg
Dh.Ulon
O Keuiar-49e Q) Polglmlde

0 E-Gloss 0 Thermoplasllc
i._ _, nm Iv,_d _ _ ItmdNr
0 S-Glen 0 Palgesler

Material Form

Fiber Ollm [---7 mils @ Tape 0 Totu

• Con'_nercial 0 Military
FIb4HI per tow E_.I( 0 Sheet 0 Broadgeods

Feb¢ic S_k* I

Appllcatlol__ _AppllcaUon_
(_)--(Experlence_

P_onn Cross Section

ADolkutlon & NC TyIEt 0 Rectangular O Rectangular

0 Square ® Square
O Transpol_
0 Trapezoidal 0 Trapezoidal
0 HeUcopler
0 Triangular 0 Triangular
O ExecuHve/Jel
0 Irregular 0 Rlrfoll
® Executive/Prop 0 Round 0 Round

0 GA

Idax veklh _ In Max Th_ness __ In

Wax len_h____ in W_ area __ sq in

,,_opllcatlon_
App_
_Experlence_

Figure 2. - Screen images of automated Data Abstraction Form.

After the software form has been filled out, the user will have the option of electronically
transmitting these data to a NASA host mini-computer via an electronic mail system, by calling an
800 number to log in directly, or by mailing a disc.

Another software module that will be resident on the host computer will be a data input
parser (checker). This will verify that the user inputs are within "reasonable" and "acceptable"
ranges. Once developed, this software could be made available directly to the user. It will also be
available for interactive use when the data is directly transmitted to the host computer. Companies
will not interface with the data base directly, but only with a host data collection file. Data will be
entered into the data base by NASA only after the source and the acceptability of the data are veri-
fied.

The last software module that will be developed will be one which allows a user to interface
with the fabrication cost data base in a "read only" mode. A user friendly interface is envisioned
that will allow the user to extract information based on specified queries such as "provide the labor
hours required for manufacturing hat stiffeners of any composite material by all manufacturing
processes." A schedule for the development and distribution of the data abstraction form and the
establishment of the data base is given in figure 3.

361
 Calender Year
Activity
1989 1990 1991 1992 1993 1994

Cost Tracking/
W°rksh_ F C°mmittee Re[:)°rts
Prediction Workshops
l ,-- Final Release
Prelim. Release -71 / I
Develop Automated Cost
_-, _---- Install & Update DB
Tracking/Reporting Form I-kV_/'-kV / F Prototype On-Line
/ / Query Capability
Fabrication Cost Select DB Struct. _ / / /- Data base maintenance
Data Base

Figure 3. - Fabrication cost data base development.

The selection of a DBMS requires careful attention in order to ensure that it is both user
friendly and versatile. The user interface must be structured so that the user will not have to learn
and understand details of the data base structure in order to access it and obtain information from it.
A survey will be made of the DBMS packages available on the market with the following attributes:

1. Wide acceptance/use in the field.

2. UNIX based or demonstrated on a number of platforms.
3. A demonstrated MS-DOS and Macintosh interface capability.

For the future, one can envision including additional information in the data base such as
tables of available material forms and their current costs, manufacturers' property data, digitized
drawings and images of parts, contractor reports, video images, audio reports, etc.

Composite Cost Prediction

"All costs are based on facts that may or may not be true" (Ref. 5). The word "cost" has a
variety of meanings to different disciplines. Designers, accountants, estimators, managers, manufac-
turing engineers etc. are interested in different levels of detail and economic conditions that imply a
numerical value to the term "cost". Often price is confused with cost. This lack of uniform, concise
description of the elements and time-valued rate constants that make up recurring cost, nonrecurring
cost, etc. leads to confusion and debate. Unifying the way the composites community represents
hardware cost for composites and metallics is perhaps as much a communication problem as it is a
demanding engineering challenge. This program will determine the feasibility of establishing
theoretical cost functions that relate geometric design features to summed material cost and com-
puted labor content in terms of process mechanics and physics.

Figure 4 provides a flow chart form of the detailed cost bookkeeping elements that should be
considcred when comparing composite aircraft cost to a metallic equivalent. The ability to fabricate
very large one piece composite structure to eliminate thousands of fasteners in equivalent aluminum
hardware requires assembly level cost estimating to establish a fair comparison during preliminary
design. The exceptional fatigue life and resistance to environmental degradation of composites
should be considered since they provide favorable maintenance and supportability comparisons.

362
r- ................................
' Part Fabrication
!
!
!

A©quleltlOncoat
t

Operntlone
Fuel Cost

$upportoblllty
Cost

Figure 4. - Detailed cost bookkeeping elements.

Large weight savings associated with extensive use of composites in wing and fuselage structure
would result in significant fuel savings over the operational life of each aircraft. Ideally the designer
should be aware of the cumulative effects of operational and supportability cost savings, but his
influence on lowering the acquisition cost generally dictates the success of a replacement part or new
design being committed to a production application.

STPO's objective in attempting to develop a designer's cost model is not to replace company
accountants or estimators, or to develop more efficient bookkeeping tools, but rather to develop a
cost model to provide the designer with a user friendly tool that relates cost to terms the designer
normally uses. The cost related issues a designer can influence usually are related to selections of
materials, tolerances, simple versus complex shape or geometry, and process dependent features that
contribute to automation potential and tooling complexity. The designers model should provide
definitive assistance in identifying the cost implications of these choices, but it should not be ex-
pected to replace the professional cost analyst that has to interpret company policy and historical
pricing practices. Managerial decisions affecting actual program costs related to availability of land,
unused company facilities, future labor rates, return on investment, etc. are not issues the designer
can be expected to consider. The primary goal for a design-with-cost model is to provide the de-
signer with a produciblilty data base and theoretical cost model that relates a new composite design
to an equivalent aluminum structure using elements of the design process that the designer can
realistically influence.

Figure 5 illustrates the standard methods used for cost/price estimating. Variations of these
methods are used routinely by estimators and price analysts to forecast or compare the relative value
of materials, automated processes, and projects. Figure 6 shows four state of the art cost models
used for estimating composite hardware program requirements. The current state of the art models

363
 Estimating Resource
Methods Technique Category
Parametric • Correlatesdesignand historicalcost data Conceptual
• Estimatesapplicableat the componentand subsystem Preliminary
design
Usedwhen details of item are not available
Factors Conceptual
Comparesnew item to a priorsimilar item Preliminary
design
andthen factorshistoricalcost

Newitem very similar to a currentitem in production. Conceptual

Analogous Specificanalogiesare drawn as the basisfor defining
the costof the new item Preliminary
design

TrendAnalysis • Usehistoricalproductiontrend data to estimate Conceptual

future productioncosts Preliminary
design
• Actualcost of the item is known. Unitsare in Design
IdenticalUnits Development
productionfor other applications.
Production

• Estimatesaccomplishedby definingtasks, characteristics, Design

Detailed manhours,materialcostsfor specificwork packages Development
• These types of estimatesare based on previousdetailed Production
labor/materialhistory
Figure 5. - Summary of estimating methods•

• A detailed breakdown of the production process into its

• An interactive spread sheet to include cost
component parts based on a time and motion study
conducted in 1976
• Starts with basic part of a simple form then adds:
• Palate of manufacturing methods Advantages: • A data base of limited production
• Builds up descrete parts to give airframe structure processes for AI and composites
Each production process is based on
Advantages: • Allows both metal and composite the average labor time, average
technologies productivity factor, QC, etc.
Disadvantages: • Requires a mainframe and does not
have user friendly data base or CAD Disadvantages: • Requires a mainframe and does not have
interface a user-friendly data base or CAD interface

(a). - Battelle Manufacturing Cost/Design Guide (b). - Northrop ACCEM & FACET

• A commercial model with nation-wide subscribers

• Lotus 123 spreadsheet models for many composite • Resident on a mainframe
processes
Advantages: • Includes extensive mathematical
Advantages: • Production methods compared & methods for manufacturing cost
evaluated on the basis of cost of estimation
materials, scrap, QC, tolerances, & • Includes standard risk models
overhead
• Econometrics module available
Disadvantages: • Requires a highly expert user with a large
Disadvantages:, Can only handle parts, not assemblies
training time investment
• Code and equations practically
unavailable

(c). - MIT/IBIS model. (d). - GE PRICE model

Figure 6. - State of the art cost estimating models used for composite structures•

364
used to estimate the cost of composite fabrication for hand layup and automated tape laying are the
ACCEM (Ref. 6) and FACET (Ref. 7) programs. Northrop developed the ACCEM program in 1976
based on a time and motion study of different composite material manufacturing processes. Equa-
tions were developed to estimate recurring composite part manufacturing costs. FACET has been
developed as a Fortran language mainframe computer program that evolved from ACCEM with
updated Air Force project data bases. New material forms and manufacturing processes that can be
evaluated for production of the most cost effective structure are considered in the MIT/IBIS model
(Ref. 8). These spreadsheet models estimate individual cost elements and enforce consistent ac-
counting assumptions. The G.E. PRICE H (Ref. 9) model is very complex and requires extensive
training with terms and concepts that best suit the needs of a cost analyst or accountant.

The Boeing, Douglas, and Lockheed workshop participants all currently use a preferred
composite fabrication cost estimating methodology. Boeing estimators rely heavily on the G.E.
PRICE Model, the only model that cost analysts from all three companies use routinely. Lockheed
uses both ACCEM and FACET and has developed parametric equations based on in-house fabrica-
tion experience. Douglas is developing an expert system model based on the GURU (Ref. 10)
artificial intelligence program and in-house experience. None of the current cost models used by the
airframers contain information on newer fabrication processes (e.g. RTM, pultrusion, filament
winding, braiding and stitching). All of the available composite cost models appear best suited for
fabrication experts, and none are independently used by designers. The tools that are available are
not suitable for preliminary and conceptual designers who will not spend their time filling out forms
and collecting material or process specific data that must be input to existing cost models. Modify-
ing existing models has been considered. Establishing accounting consistency with Lotus 123 (or
equivalent) spreadsheet forms and contractually requiring all ACT program participants to uniformly
report with these forms has been discussed. Holding an additional workshop to develop unified
equations, factors, and standard constants to be used in the G.E. Price model also has been consid-
ered. The major concern for these approaches is that designers will not use a tool that is unfamiliar
and unrelated to the design process. If a cost estimating system is to be useful to the designer, and
helpful in the selection of design concepts with their associated fabrication processes, the system
must be relatively transparent to the designer. A model for designers must be structured to have
input that can be coupled directly to a preliminary design module. Such input relates cost to panel
thickness, stringer spacing, stiffener height, laminate ply orientation stacking sequence, etc.

The primary thrust of the designers' cost model development would be to use a first principles
approach to establish building block unit cell elements (e.g. prepreg tow, 12" prepreg tape, cloth,
etc.) that represent different material forms, and to use basic principles (mechanics, dynamics, physics,
etc.) to describe labor content in terms of machine feed rates, accelerations, and material deposition
efficiencies that characterize processes and the effectiveness of automation. Modeling concepts of
cost per inch, materials cost per cubic inch, and layup man hours per square inch for a unit cell
representative of each material form are concepts that would suit the designers'needs. Engineers
customarily express cost comparisons as S/pound or man-hours/pound. Ratioing comparisons with
respect to geometric properties and dimensions of length, area, or volume would provide a means of
incorporating geometric complexity in the comparison. Complexity factors determined as theoreti-
cal relations for radii of curvature, degree of double curvature, tight dimensional tolerances, number
of stiffening elements, etc. would provide equations to uniformly express theoretical cost of materi-
als and labor for simple or difficult to fabricate designs. The designer employs laminated plate
theory to sum lamina properties that are experimentally determined through the thickness of a

365
 laminate to develop smeared stiffnesses that account for ply orientation and stacking sequence. A
similar approach for treating cost as a lamina material property (S/square inch) that's summed
through the thickness, accounting for material length associated with part topography and process
dependent scrap of off angle plies, would "allow development of a totally theoretical cost representa-
tion. Panels could be designed to calibrate the labor content of automated processes by measuring
man hours per square inch to apply lamina to a simple and complex shape mandrel providing pro-
cess dependent coefficients similar to the lamina modulus measurements now used to evaluate
composite materials.

Assemblies could be considered by describing the material and man hours associated with
fastener installation and added structural joint complexity. A metallic structural part or assembly
cost could be used for comparison to provide ratios (index of value) that nondimensionalize the cost
representation and remove issues of proprietary labor rates, time value of money, etc. Initial efforts
should concentrate on recurring costs which are most amenable to a detailed breakdown and are
directly related to design features. Recurring costs should be a function of the physical description
of the part and the fabrication process related to the part. As the term "recurring costs" implies,
these costs are incurred for every part made and should be consistent from part to part. Since the
recurring cost elements of a fabrication process are amenable to a time and motion study, equations
can be developed that predict cost from material volume, part geometry and the physics that de-
scribes the time and resources required to perform each step of the of construction, machining, and
assembly with fasteners or adhesives. This approach would provide designers with an technically
sound, academically rigorous, and universally accepted model to describe a theoretical material cost
and labor content for comparison of their designs. These models could be calibrated with actual
corporate experience to provide bounds of theoretical versus actual process efficiencies. Ideally the
model can be a module to existing design software that will compute cost from the geometric fea-
tures derived from design software, geometry generator programs and, eventually, CAD programs.
Such a model would allow the designer to use cost equations as minimization functions in optimiza-
tion models now used in designing to minimum weight. Figure 7 provides a flow chart for a proto-
type designers' theoretical cost optimizer concept with emphasis on the data base elements that must
be developed.

Baseline I /
Design& I _ liii_i_C_f:iP_t_i_i_i]

Analysis L_'_ptimiz__--_

I Dataease
MaterialsI
_ S_il_--_l [Cost-Effective
Structural /

Structural _
! oes, n
[B_iSeiine Dalai BaSes _1
Design I _ l_!SOxAi:Ai_mi_um::::::i::!::::iiiii::
I
Constral__p_i_[_i!_i!_!:_!ii_::i_ii:_:l
........................................ :.!::iA_ocl:aved::::icomp_::_::l

Figure 7. - Optimization of design process with cost as a design variable.

366
 Unit We,ght
" Cost

.01 .05 0.1 0.5 1.0

Weight Efficiency
W-Wopt
( Wopt
)
Figure 8. - Influence of design/manufacturing integration (D/MI).

The designers' cost model must have sufficient fidelity to distinguish between concepts if the
concepts have significantly different costs. This fidelity implies the need for adequate detail in both
the description of the part and the associated cost methodology. Figure 8 provides a schematic of the
results of the design-with-cost process for a simple stiffened skin compression panel. The influence
of design concept on cost and weight is the product of this process. The D/MI zone permits the
designer the opportunity to increase weight efficiency with cost as a primary variable. A cost meth-
odology that sums the cost of each element of the fabrication process and allows for parallel as well
as serial operations may be required to achieve the needed fidelity. Figure 9 shows a flow chart for
manufacturing an elevator including panel, rib, and spar details. Developing equations representing
economic relationships in terms of energy, power, thermodynamics, mechanics, process physics, etc.
for each step would sum to a theoretical cost for performing each operation. Statistical bounds
appfied to each operation could establish theoretical maximum and minimum cost values. One
model concept would be to treat cost as a control theory or chemical engineering process problem
where time dependent cost functions were inputs to be integrated through process steps to comple-
tion as a part or assembly. Participants in the ACT program will be generating cost data related to
new processes and the state of the art for manufacturing large composite structures, providing the
required data base to formulate a theoretical cost model approach and the coefficients and constants
necessary to calibrate or verify the model.

367
 Panel
I' I I r l.
Honeycomb Core Machining =_

Spar _ Assembly
Cure [Tr

Honeycomb Core Machining =:_

RibI're're'
utll a, 'a'l- Cure IT r _ff,.,.,... I
Honeycomb Core Mac_=:_

[_CutlL_ B

Figure. 9 - Elevator manufacturing flow.

Nonrecurring costs are important since the need to build new tooling often adds large start up
expenses to a project. Tooling costs should be considered because they vary according to the se-
lected design concept or fabrication process and are therefore an element of cost that is directly
related to design. Tooling cost should be predictable in relation to the physical, dimensional and
geometric complexity of the part to be made. Tooling costs are a function of the production rate and
the total number of parts to be made. A program that includes tooling costs should have the flexibil-
ity to consider changes in production rate and the total number of units over which the costs of
tooling will be amortized. The feasibility of a theoretical tooling cost model related to tool material
type and geometric complexity of the part to be made will be evaluated.

Conclusions

The remarkable advances in computer hardware and commercial software technology have
led to low cost data storage and sophisticated data base management systems. These developments
make it economically feasible to track the cost history of numerous projects and provide the historic
opportunity for bringing cost into the preliminary design process as an engineering variable.

Recommendations of the cost reporting and cost prediction workshop committees will be
implemented by

1. Continuing an established task with AS&M to develop an electronic data base that will unify
formatting and automate the collection of composite part fabrication costs provided by ACT pro-
gram participants. A "subscriber" approach, wherein contributors to the data base would have

368
 accessto it, will beimplemented.The databasewill includestandardmaterialcosts(includingfuture
costs)for consistentcomparativecostingstudies.Thedatabasewill bekept currentthroughthe
durationof the ACT programandmethodsfor long-termmaintenancewill be considered.

2. Developmentof an academicallyrigorousmodelfor predictingthecostof different composite

designsduring thepreliminary designprocesswill beinitiated. This effort will includedevelopment
of a producibility guide(a manufacturingdatabasesoftwaremodule)to provide thedesignerwith
informationon selectedmanufacturingprocessesandtypesof designdetailsthatadverselyaffect
cost. Large costdriverswill beidentifiedandsoftwareapproaches to couplethe databaseto design
optimizationandCAD interfaceswill bepursued.

References

1. Meade, L.E., "DoD/NASA Structural Composites Fabrication Guide-3rd Edition", pages A7-
A15, Volumes 1-2, Air Force Wright Aeronautical Laboratories, 1982. Presented.

2. "First Workshop on Cost Tracking and Prediction", Structures Technology Program Office-
NASA-LaRC, Norfolk, Va., November 30-December 1,1989. Presented.

3. "Second Workshop on Cost Tracking and Prediction", Structures Technology Program Office-
NASA-LaRC, Long Beach, CA., February 21-22,1990. Presented.

*4. Swanson, G., Cost Model Recommendations, Letter Report-dated July 13,1990, "Second Work-
shop on Cost Tracking and Prediction", Structures Technology Program Office-NASA-LaRC, Long
Beach, CA., February 21-22,1990.

5. Davis, H.V., LASC Present Approach To Cost Tracking, "First Workshop on Cost Tracking and
Prediction", Structures Technology Program Office-NASA-LaRC, Norfolk, Va., November 30-
December 1,1989. Presented.

6. LaBlanc, D.J., "Advanced Composites Cost Estimating Manual", AFFDL-TR-76-87, WPAFB,

August 1976.

7. Meade, L.E., "DoD/NASA Structural Composites Fabrication Guide"-Final Administrative

Report, AFML-TR-82-4137, Air Force Wright Aeronautical Laboratories, 1982.

8. Krolewski, S.and Gutowski, T., Economic Comparison of Advanced Composite Fabrication

Technologies, 34th International SAMPE Symposium, Reno, Nevada, April 1989. Presented.

9. PRICE H TM Reference Manual, Price Systems, General Electric Co., Dayton, Ohio, 1988.

10. Cost/Benefits Analysis System User's Guide Ver. 2.0, Advanced Program Development,
Douglas Aircraft Company, March 21,1989.

* MS-DOS® is a registered trademark of Microsoft Corporation, Macintosh TM and HyperCard TM

are trademarks of Apple Computer, and UNIX TM is a trademark of AT&T Bell Laboratories.

369
 F-15 COMPOSITE ENGINE ACCESS DOOR +

Ramaswamy L. Ramkumar
Northrop Corporation, Aircraft Division

James C. Watson
McDonnell Aircraft Company

SUMMARY

This paper presents a summary of the successfully concluded

Phase I of the two-phase Design and Manufacture of Advanced
Thermoplastic Structures (DMATS) program. It addresses the
design, manufacture and validation testing of a thermoplastic
F-15E forward engine access door and includes lessons learned
during the concurrent product and process design development
phases of the program.

INTRODUCTION

The F-15 forward engine access door is moderately sized (42 in.
x 38 in.) with a contour curvature that varies from gentle to
relatively severe. The door is a built-up, channel stiffened,
titanium structure on the F-15C/D and a superplastically formed/
diffusion-bonded (SPF/DB), hat-stiffened titanium structure on
the F-15E (Figure i).

The F-15 door is located approximately eight feet aft of the

main landing gear, directly underneath the engine compressor
section. The component requires structural integrity and
durability in a high temperature (300°F) and severe acoustic
(156 dB max.) environment. Due to its location, the door is
impacted with runway debris and is removed frequently by
maintenance personnel during routine maintenance. Engine
equipment located above the door includes the main oil tank and
fuel filter; thus the door is exposed to aircraft fluids. The
door is a secondary structure and is loaded primarily in shear.

+ Work reported in this paper was performed in Phase I of an

ongoing Northrop/WRDC contract F33615-87-C-5242, titled "Design
and Manufacture of Advanced Thermoplastic Structures". Ms. D.
Carlin and Ms. T. B. Tolle are the WRDC technical monitors.

371
 Major longerons located inboard and outboard of the door carry
the majority of the fuselage bending loads. Air loads are
relatively low (3.75 psi). The door structure contains two non-
structural service doors for oil and fuel filter service. When
open, the door is attached to the surrounding structure by two
metal gooseneck hinges.

MATERIAL SELECTION

Thermoplastic matrix composite (TP) material selection for the

F-15E forward engine access door was dependent primarily on the
maximum continuous service temperature the structure is exposed
to during aircraft operations. Recorded flight temperatures for
the critical speed and altitude show a maximum temperature of
280°F. Therefore, a material system with a 300°F service
temperature was deemed adequate to meet temperature
requirements.

Originally, Imperical Chemical Industries' (ICI) AS4/HTX was

selected as the material system for the door. The AS4/HTX
material's 350°F service temperature met the temperature
requirements, had the necessary mechanical/physical properties,
and was available. However, micro-cracking and processing
problems that occurred during trial manufacturing runs led to
the withdrawal of the HTX material system from the program.

ICI's AS4/ITX was subsequently selected to replace AS4/HTX. The

AS4/ITX material's 300°F service temperature met the
temperature requirements for the door, and it had mechanical
properties comparable to AS4/HTX. The processing
characteristics of ITX are significantly better than those of
HTX, and no micro-cracking was identified in the parts.

TP DOOR DESIGN

The initially selected coconsolidation concept for the TP F-15E

door was finally replaced by an amorphous-bonded structure
containing autoclave-consolidated outer mold line (OML) and
diaphragm-formed inner mold line (IML) skins (see Figure 2). To
facilitate diaphragm forming of unidirectional TP, the hat
stiffeners on the TP door are shallow and the bend radii of the
hat stiffeners (at the corners of the bays) are generous. The
two non structural service doors were redesigned to be

372
compression molded using chopped AS4/HTX tape. Due to the
difference in the stiffener cross-sectional geometry between the
titanium and the TP access door, all attachment and
reinforcement hardware were redesigned to fit the TP hat
stiffeners on the door. This included changes to the aluminum
hinges for the service doors and the titanium gooseneck hinges
that attach the access door to the aircraft. The gooseneck
hinges were also designed to permit adjustment during
installation. Two brake-formed sheet metal parts attach the
hinge to the hat stiffener (Figure 3). This allows for
manufacturing tolerances in the hat-stiffener geometry.

The OML and IML skin layups result in a laminate design that is
a series of five-ply symmetrical sublaminates. The individual
components (OML and IML skins) are symmetrical to reduce
fabrication-induced warpage. Also, the ply build-ups in the OML
skin are symmetrical to reduce warpage, and ply drop-off
spacings are no smaller than 0. I inch to enhance performance and
producibility. Titanium fasteners were used, and all
non-titanium metallic hardware were isolated from the composite
door through the use of non-metallic shims.

The door was designed to fit to an existing metal substructure.

Therefore, the edge distance (E) on the peripheral Milson
fasteners was only two fastener diameters (D). To obtain an
allowable bearing stress for an E/D of 2, bearing tests were
performed. An allowable ultimate bearing stress of 40 ksi was
determined from the test. In order to meet the bearing stress
requirements, the door thickness around the edges was increased
to 35 plies.

The door was designed to buckle below limit load and operate in
the postbuckling range as an additional weight savings measure.
The SS8 computer code was used to predict the initial buckling
of individual door bays. Assuming that the hat stiffeners
provide clamped supports, the center bay was found to be
critical at an initial buckling load of 319 ibs/in shear flow.
Therefore, at a maximum shear load of 850 Ibs/in, the
postbuckling ratio was 2.66.

The nonstructural service doors were designed using chopped

AS4/HTX material (Figures 4 and 5). The basic shape of the
service doors remained similar to the current aluminum service
doors. However, because of the difference in material
properties between aluminum and chopped AS4/HTX, the thickness
of the door was resized to withstand air pressure and handling
loads. Material properties for chopped AS4/HTX were determined
from test results.

Amorphous bonding of AS4/ITX skins with polyetherimide (PEI) was

shown to have strengths equivalent to coconsolidation. This is

373
 because PEI is a low melt temperature thermoplastic material
with matrix properties compatible with thermoplastic materials
including PEEK and ITX. A typical amorphous bond process has
one 0.005 inch PEI ply consolidated to the joining surface of
each detail during the consolidation of the details. When the
two details are joined together, two additional 0.005-inch PEI
plies are inserted between the details to effect joining/
subassembly. The resultant consolidated bondline is typically
0.010 inch thick.

Based on the results from two subcomponent tests, a design

change was made to install fasteners around the hats in the
amorphously bonded door full-scale assembly (Figure 6). This
was required because of the inconsistency in the bondline
quality, resulting in a premature failure during one of the
subcomponent tests. However, a second subcomponent went to 150%
of the design ultimate load, with no failure in the bond
region. Therefore, the fasteners were installed as a safety
precaution since a reliable database on amorphous bonding was
not available. The fastener pattern design for the door (see
Figure 6) uses 4D and 6D spacing, depending upon the proximity
to highly stressed areas. The fasteners chosen were 3/16 inch
tension head hi-loks with self-aligning washers and collars.

MANUFACTURING PROCESS DEVELOPMENT

Significant technical challenges were encountered in

transitioning from the preliminary manufacturing processes at
the initiation of the program to the processes that were
eventually applied on the flightworthy TP F-15 engine access
doors. The challenges were posed by material-, design-,
tooling- and equipment-related issues, and are summarized in the
"Lessons Learned" section of this paper. The manufacturing
processes used for the "production" TP doors are described
below:

Fabrication of AS4/ITX IML Skin

Diaphragm forming, using an SPF aluminum diaphragm, was chosen

as the manufacturing process for the IML skin because of the
proven ability of this process to fabricate high quality,
complex, efficient composite parts (Figure 7). The skin was
fabricated in a press on a female carbon/ceramic tool. Dual
diaphragms were utilized to maintain location of the ply pack
during forming.

374
 Fabrication of AS4/ITX OML Skin

Autoclave consolidation was the selected manufacturing process

for the OML skin (Figure 8). Diaphragm forming in a press with
a contoured vacuum frame was recommended for a future production
scenario and was used in the cost analysis for the fabrication
of the OML skin. The fabrication process consisted of envelope
bagging on the OML carbon/ceramic tool utilizing an aluminum
caul sheet. Processing parameters were set as close as possible
to those used in the diaphragm forming process. The caul sheet
was roll- formed to the approximate contour of the OML tool.

Amorphous Bonding of AS4/ITX OML and IML Skins

The IML skin and the OML skin initially had one layer of PEI
film coconsolidated on the bonding surface. These surfaces were
cleaned prior to bonding using cheesecloth and isopropyl
alcohol. The OML skin was then placed on the OML tool. Twelve
layers of dried .005 inch PEI film were placed over the skin to
ensure the filling of any gap due to IML/OML mismatch. The IML
skin was then placed on top of the PEI film layers and envelope
bagged to the OML tool. The periphery of the part was
periodically taped with high temperature tape to allow air to
escape and to control PEI squeeze-out. To aid in air/gas
removal, fifteen 0.040-inch diameter holes were drilled at
strategic locations through the IML skin into the bondline.
Amorphous bonding was effected during a 45 minute hold at 20 psi
and 575°F. A 3-5°F/minute cool down was used to minimize
any resulting thermal stresses.

TESTING AND TP DESIGN VALIDATION

An extensive test program was conducted to validate the TP F-15

door design. Results of the test program were used to evaluate
the flightworthiness of the door. A building block approach was
used, as evidenced by the testing of elements, subcomponents,
and finally the full-scale door.

One full-scale production door (without hardware, access doors,

or finishes) was subjected to structural verification testing
consisting of static, spectrum fatigue, and residual strength

375
 tests. All tests were performed under room temperature
ambient (RTA) conditions. The door, to be structurally
qualified for flight test, had to meet the following test
criteria:

(i) No delaminations or disbonds occur that result in permanent

set of the composite material at or below 100% design limit
load (DLL). No yielding or fracture of the the composite
material at or below 100% DLL. Disbonds between the hat and
the skin would be allowed if the disbond were arrested by the
fasteners placed around the hats.

(2) The door must survive two (2) lifetimes of spectrum fatigue
loading (I lifetime = 8000 spectrum flight hours).

(3) After completing two lifetimes of spectrum fatigue testing,

the door must reach 150% DLL with no catastrophic failure.

The full-scale door was subjected to shear loading using a

picture frame arrangement. The shear loads applied were based
on loads taken from the F-15E aft fuselage finite element
model. This shear load is slightly lower than the original 850
lb./in, shear that the door was designed to.

Before the door was tested, a baseline ultrasonic A-scan and

C-scan were performed. The door was A-scanned after being
subjected to the limit load and during the fatigue tests to
determine if any nonvisible damage had occurred. A C-scan was
taken of the door after the ultimate test was performed to
determine the extent of the damage.

The full-scale AS4/ITX F-15 door test was successful. Test

results validated the TP door design and established its
capability to withstand 100% DUL after two lifetimes of F-15
fatigue loading.

LESSONS LEARNED

The Phase I effort described in this paper culminated in the

successful delivery of flightworthy TP F-15 forward engine
access doors to the U.S. Air Force. However, the path to
success was punctuated by challenges, some of which could not be
overcome, in technology areas that included materials and
processes, manufacturing technology, tooling technology,
processing equipment and structural design. A summary of the
lessons learned is presented below:

376
 Materials

(i) Materials-related challenges faced on the DMATS program

included high melt viscosity, narrow processing window and
limited reprocessibility. The last issue (limited
reprocessibility) was primarily introduced by long process
cycles, due to a lack of rapid heating/cooling capability in
the processing equipment. Consequently, the preferred
assembly process for the OML/IML skins was changed from
coconsolidation to amorphous bonding.

(2) Selected Phase I TP materials (AS4/HTX and AS4/ITX) were

only available in the unidirectional tape (UDT) product
forms. This posed a forming challenge for the complex-
contoured, stiffened IML skin. The use of a Kapton/Upilex
polyimide film in Northrop's diaphragm forming press setup
increased diaphragm rupture occurrences at forming pressures
in excess of i00 psi. But, the IML skin did not completely
form at i00 psi, exhibiting small regions of "bridged"
plies. This necessitated the use of the SPF aluminum
diaphragm in MCAIR's diaphragm press setup, and the
application of a 200 psi forming pressure. If drapeable
AS4/HTX and AS4/ITX product forms had been available,
forming of the IML skin and the concurrent OML/IML
coconsolidation could have been demonstrated on the program.

Manufacturing Processes

(i) Diaphragm forming of small flightworthy F-5 and T-38 details

had been established by Northrop as a producible and cost-
effective process prior to the DMATS Phase I effort.
However, the scaleup of this process to the F-15 door IML
skin was a major challenge. The replacement of the
polyimide diaphragm by the SPF aluminum diaphragm, and the
use of a 200 psi forming pressure, overcame the IML skin
forming challenge and resulted in a producible process.

(2) Autoclave consolidation of the gently contoured F-15 door

OML skin, using an envelope bagging technique, required the
use of a caul sheet to eliminate wrinkles.

(3) The use of a washaway mandrel to coconsolidate the OML and

IML skins was successfully demonstrated on the program,
though its application was considered risky due to the
complexity in the final OML/IML cross-sectional cavity shape

377
 compared to the simple hat cross section in the preliminary
design.

(4) The implementation of the amorphous bonding process was

challenged by three factors: (a) the unknown interlaminar
(bondline) properties for the ITX/PEI combination; (b) the
lack of dimensional control of the IML and OML bonding
surfaces, due to the design of the forming tools for the
preliminary coconsolidation process; and (c) the use of a
low amorphous bonding pressure (20 psi) to eliminate the
need for washaway IML cavity mandrels. Significant process
development efforts were expended to accommodate these
factors, and to establish the reprocessability of the
amorphous bonding operation to eliminate any
PEI/moisture-induced porosity in the bondline.

(5) Compression molding of the oil and fuel filter service

doors, using AS4/HTX offal, was successfully demonstrated on
the program.

Tooling

(i) The complex pan-stiffened F-15 door IML skin design and the
high processing temperature (700-800°F) requirement for
TPs, established the need for a match in the coefficient of
thermal expansion (CTE) between the tool and the part.
Though only partially successful in the reported effort, a
matched CTE tooling concept that lends itself to rapid
forming is essential for successful application of TPs to
composite structures.

(2) Polyimide diaphragms (Kapton and Upilex) were barely

adequate for processing AS4/ITX under 720-750°F, i00
psi conditions. SPF aluminum alloy diaphragms performed
very well and withstood a 200 psi forming pressure at
720-750°F.

(3) For the F-15 door details, the forming tools were designed
to control the OML side of the OML skin and the IML side of
the IML skin, based on the original coconsolidated assembly
concept. This created a mismatch between the bonding
surfaces when the coconsolidation process was replaced by
amorphous bonding, requiring a variable thickness of PEI
along the bondline and raising the issue of bond strength
variation with PEI thickness. Had the OML and IML tools
controlled the bonding surfaces, this issue may not have
arisen.

378
 Equipment

(i) The diaphragm forming process is unique to TP applications,

and has only been demonstrated in a press setup for small
parts. Transitioning this process to the F-15 door
application required considerable equipment development
efforts at Northrop and MCAIR. In extending the process to
larger primary structural parts A the capability to adapt
existing high temperature (>800_F) autoclaves to this
process is essential.

(2) The feasibility of economically rapid forming TP parts has

been established in many programs. However, the development
of large rapid thermoforming equipment has been slow.

Design

(i) A concurrent design development effort was successfully

demonstrated on the program and contributed to the
selection of the preliminary coconsolidated F-15 door
concept, and its change to the final amorphous bonding
process. However, the use of developmental materials,
manufacturing processes and tooling concepts, and the
ambitious Phase I schedule constraints, forced the concur-
rent design development effort to falter at times and accept
too many changes without the necessary supporting data.

(2) In transitioning from the preliminary design (uniform

thickness OML and IML skins) to the final design (multiple
drop-offs, local doublers, etc.), producibility was
partially sacrificed in favor of structural performance, and
a large weight reduction (39%) over the baseline SPF/DB
titanium door was aimed for. This played a major role in
introducing drop-offs and local build-ups, and in changing
the assembly process from coconsolidation to amorphous
bonding. The program demonstrated that considerable cost
and weight savings are realizable with TP designs, but these
have to be achieved without compromising producibility.

379
 CONCLUSIONS

Phase 1 of the DMATS program was concluded very successfully

with the structural qualification/design validation of the TP
F-15E engine access door via full-scale static and fatigue
tests and the delivery of flightworthy TP doors.

For the selected F-15E engine access door application, the TP

design developed and validated on this program was projected to
yield significant weight (39%) and cost (25%) savings over the
SPF/DB titanium production doors. The viability of using TP
materials in secondary structures, and their potential for
primary structural applications, was established beyond doubt.
Lessons learned from this Phase I effort are being incorporated
into ongoing Phase II tasks to identify cost-effective primary
structural applications for thermoplastic matrix composites.

380
 Figure 1 F-15E S P F / D B Titanium Door

phragm-Formed
IML Stiffened Skin

OML Buildup Ply Pack

+
Autoclave-
Consolidated
Toget her
I +
[A
Thermoplastic F-15E Forward Engine Access Door
(Amorphous-Bonded)

OML Skin

Figure 2 T P F-15 Engine Access Door Design

381
 Figure 3 Redesigned Gooseneck Hinges

Figure 4 O M L S i d e of t h e Compression Molded A S 4 / H T X Oil

S e r v i c e Door

382
 Figure 5 E l e c t r o m a g n e t i c I n t e r f a c e (EMI) F i n g e r s o n t h e
I M L S i d e of O i l Service Door

Figure 6 F a s t e n e r P a t t e r n for Amorphous Bonded F-15 Door

383
 Figure 7 Diaphragm Formed F-15 Door I M L S k i n

Figure 8 A u t o c l a v e C o n s o l i d a t e d F-15 Door OML Sk in

384
 FABRICATION OF THE V-22 COMPOSITE AFT FUSELAGE
USING AUTOMATED FIBER PLACEMENT

Robert L. Pinckney

Boeing Helicopters, Philadelphia, PA

SUMMARY

Boeing Helicopters and its subcontractors are working together under an Air Force Wright
Research and Development Center (WRDC)-Manufacturing-Technology Large-Composite
Primary Structure Fuselage program to develop and demonstrate new manufacturing
techniques for producing composite fuselage skin and frame structures. Three sets of aft
fuselage skins and frames have been fabricated and assembled, and substantial reductions
in fabrication and assembly costs demonstrated.

INTRODUCTION

Advanced composite structures are lightweight, strong, stiff, and resistant to fatigue and
corrosion. These features make composites highly attractive for major applications in
military and commercial aircraft. Currently, industry is producing a number of secondary
structure applications, and a significant amount of R&D effort is being expended to further
develop engineering and manufacturing technology for wing primary structure. To
maximize the benefits derived from using composites on aircraft, there is a need for a
corresponding level of development activity on fuselage primary structure.

The Boeing Company, in a team effort with other major companies, conducted a WRDC
sponsored program entitled "Manufacturing Technology for Large Aircraft Composite
Primary Structure (Fuselage)." This program addressed the need to establish
manufacturing capability to produce primary composite fuselage structure for large aircraft
at predictable and reasonable cost. The program objective was to establish and validate low
cost manufacturing methods for efficient production of such structure which could be
applied to a variety of future aerospace systems.

The Boeing team effort combined the technical skills, automation technology, resources,
and production experience of six companies in a concerted effort to achieve maximum cost-
effectiveness in composites fuselage manufacturing. While Boeing had primary
responsibility for carrying out the program, five other companies participated in Phase I
and H. They were

Northrop Corporation
Hercules, Inc.
Teledyne-Ryan Aeronautical
Rohr Industries, Inc.
Xerkon, Inc. (formerly Proform, Inc.)

In order to evaluate various manufacturing methods and assure incorporation of the latest
state-of-the-art technology in advanced composite manufacturing, the six-member team
produced manufacturing test hardware during the first two phases of the program.

385
Innovative tooling and manufacturing methods were used to fabricate the test hardware.
These test components were compared for manufacturing cost, quality, strength, and
weight. The methods found to be superior were selected to fabricate full-size verification
and demonstration hardware in Phases HI and IV.

The planned down selection at the program decision point led to the start of Phase III in
September 1985 with the team narrowed to Hercules, Teledyne-Ryan, and Xerkon.

A pictorial overview of the program test and demonstration components is shown in figure
1.

MANUFACTURING CONCEPTS

The manufacturing concepts selected for the production demonstration of the V-22 aft
fuselage were those developed by the Hercules, Teledyne-Ryan and Xerkon Companies.
Hercules fabricated the one-piece aft fuselage skin with its 157 cocured stiffeners on a male
mandrel using their multiaxis fiber-tow placement machines.

The Xerkon Company

The Xerkon Company fabricated the J-section stiffener preforms and two of the five
fuselage frames using their dry-fiber stitching and resin infusion processes. Teledyne-
Ryan fabricated the remaining three frames using their press molding (Quadrapress ®)
system. Design changes were made to accommodate each of the new manufacturing and
material forms as required. Changes were validated by means of detail stress analysis and
predicted fuselage dynamic responses.

The Xerkon Company replaced square weave fabric and unidirectional tape preimpregnated
materials with dry fiber tows stitched into preformed shapes for the frames and stiffeners.
The preforms were then impregnated with epoxy resin and cured in integrally heated and
cooled, closed molds using their Autocomp ® process. A high degree of automation in the
stitching process was employed. Robotic tool loading and final part trimming was
developed, figure 2.

Teledyne-Ryan

The Teledyne-Ryan Company fabricated the three forward frames required for final aft
fuselage assembly using their Press Cure Quadrapress® technique. No major design
changes were required to utilize the press cure system rather than the more conventional
autoclave cure technique; however, the fabric material forms, splice areas and other design
details were changed to reduce layup hours. The use of 60-inch wide, 5 harness weave, 13-
mil thick carbon fiber fabric significantly reduced preform layup time.

The use of graphite fiber-epoxy tooling inserts to transport, load, and unload the frame
components to and from the curing press allowed the press platens to be maintained at a
constant temperature of 182°C (360°F) thus appreciably reducing the cure cycle time. In
addition, the frame configurations were such that the three individual frames could be
nested and simultaneously cured during one press cycle. The tooling configuration is shown
in figure 3 and a press load of cured frame sections in figure 4.

386
 PHASE I & II LARGE FUSELAGE SECTION

SKIN PANELS

231-CM (91-INCH) DIAMETER

FULL CIRCUMFERENCE
76 x 121-CM (30 x 48°1N.) VERIFICATION

ZEE FRAMES
213-CM (84-1N.)

ONG SECTION
OGIVE PANEL
101 X 152-CM (40 x 60-IN.)
152 x 152-CM (5 x 5 FT) FORWARD
SIDE PANEL WITH
REPRESENTATIVE
OPENING

152 x 152-CM (5 x 5-FT) STIFFENED

OGIVE KEEL PANEL
WITH FRAME SEGEMENTS
FABRICATED AND
FASTENED

PHASE III & IV V-22 AFT SECTION

i,

60 x 121 CM (2 x 4 FT) (3)

PROCESS COUPON TESTS (72)

Figure 1. Program Test Components

387
 Figure 2. Robotic Handling of Center Tool Section

Figure 3.

388
 Figure 4.

The Hercules Company

The Hercules Company; using CADAM data, tooling master models, and technical
assistance supplied by Boeing; designed and fabricated the necessary tooling to produce the
V-22 aft fuselage shells using their multiaxis fiber tow placement machines. Fiber tow
placement offers reduced cost through automated, multitow placement (up to 32 tows of
prepreg simultaneously onto a variety of tool geometries). Thickness and fiber angle control
and in-process compaction, using the lowest projected cost material form, are key elements
of this technology. Sectional, internal tooling provides precise detail part location, internal
mold line (IML) control, and allows internal doubler pads and stiffeners to be cocured to the
skin. Following autoclave curing of the part, the tooling is disassembled from within
providing a single unit fuselage section. This results in the elimination of assembly joints
and fasteners, and reduces the part count from eight sections to one.

The fiber placement machines utilize a computerized mathematical model of the part to
generate a fiber path of a specified width, thickness, and orientation and to control cut and
add functions. The software and hardware provide synchronization control and movement.
The material delivery system processes, delivers, and compacts the prepreg tow material on
the mandrel a s demanded by part geometry.

389
Tooling for fiber tow placement must meet the following three requirements: It must form
and apply pressure to all of the internal details of the finished part; it must have a center
shaft to enable it to be rotated in the fiber placement machine; it must be able to withstand
the pressure and temperature of repeated cure cycles; and it must be capable of being
removed from the finished part interior.

Fiber tow placement offers many improvements over hand layup which contribute directly
or indirectly to cost savings. Tow width control allows for nonstandard ply thicknesses
which optimizes part design while maintaining constant band width. Gaps and overlaps are
kept within a tolerance of 0.75 mm (0.030 inch). Constant ply thickness can be maintained
by adding or dropping tows as the part changes cross section. Tow and band cut/add
features reduce material scrap to as low as 5% by placing the material only where required.
Fiber placement also utilizes prepreg tow which is projected to be the lowest cost material
form available.

During the fiber tow placement process, a conformable roller rides directly on the part or
tool to deliver the tow material while providing in-process compaction. This minimizes the
need for intermediate compaction steps. The placement head flexibility allows fiber
placement on convex and concave surfaces. The delivery head delivers individual tows as a
flexible band to minimize material distortion. This flexibility provides fiber angle control
which allows for fiber placement of non-geodesic shapes. There are no limits on the winding
angle. Purely axial (0 degree) plies can be readily placed with this process.

Hercules currently has two fiber placement machines (FPM). FPM1 is a 6-axis,
development machine which has been in use since 1982. This machine was used to
manufacture the first three V-22 aft fuselage sections. FPM2 is a production rated, 7-axis,
fiber placement machine to be used for production programs and to fabricate the final
deliverable V-22 aft fuselage section. Figure 5 shows a typical FPM2 setup with an
explanation of the various axes of motion.

LONGITUDINAL

Figure 5.

390
The panelized design for the V-22 FSD aft fuselage consists of a basic skin of four layers of
AS4 fabric for a total thickness of 0.75 mm (0.030 inch). The fiber placed design consists of a
skin with five plies of IM6 prepreg tow (equivalent to grade 145 tape) for a total thickness of
0.71 mm (0.028 inch). In subscale testing, fiber placed test panels were the equivalent of the
baseline test panels.

The panelized FSD design utilized outside mold line (OML)tooling, which required building
the fuselage component from the outside in. Once laid up, they are bagged on the IML
surface and cured. Each component is limited in size by the workers’ ability to reach the
middle of the tool during processing.

In contrast, the fiber placed design utilizes a knockdown, male IML mandrel. The internal
stiffeners are assembled into the tooling which locates them exactly. The stiffener doublers
are laid up in sections on recesses built into the surface of the full-size composite mandrel so
that the resultant inner skin surface is smooth. See figure 6. The mandrel can be rotated so
that workers can lay up sections of doubler material from a convenient position. The skin is
then fiber placed on the mandrel surface using continuous, full-length, 9 cm (3.50 inch) wide
bands of prepreg tows.

Figure 6.

391
The hand layup design of the V-22 aft fuselage consisted of 10 skin panels; 2 in the ramp and
8 in the upper fuselage. The fiber tow placed aft fuselage was cured in one piece, with the
mandrel components being disassembled from within. For ease in manufacturing, and since
the ramp is a separate assembly from the upper fuselage section, the ramp is cut before cure.
Both components remain on the mandrel through the cure cycle. Manufacturing the aft
fuselage in fewer pieces results in fewer subassemblies and fewer fasteners. This reduces
handling, part control, and traceability costs as well as component, subassembly and
inspection costs. The completed aft fuselage one-piece shell with the ramp skin and mandrel
removed is shown in figure 7.
BOEING HELICOPTERS AFT FUSELAGE ASSEMBLY

The three aft fuselage shells and frame details supplied by Hercules, Teledyne-Ryan and
Xerkon were trimmed to net shape, ultrasonically and dimensionally inspected, and shipped
to Boeing Helicopters for final assembly. The detail components were reinspected by Boeing
and the additional components required, (frame splice plates, angles, and fasteners) were
procured. A new assembly fixture was fabricated to accommodate the large one-piece shell
and to locate the frames and other details.

Figure 7.

392
Frame-fastener hole locations were pilot drilled on the bench, and the frameswerelocated inthe
assembly fixture by means of tooling holes defined in the CAD data supplied to each frame
contractor. The one-piece shell was positioned over the frames and pinned to the assembly
fixture by means of tooling tabs located at each end of the shell. Fastener holes were located
in the one-piece shell skin by back drilling through the frame pilot holes. All fastener holes
were drill reamed to final size and titanium fasteners installed with wet polysulfide sealant.
The one-piece shell eliminated eight individual hand layup panels and reduced from thirty-
two to eight the number of trim lines required to be hand fitted, pressure sealed, and
electrically connected for EMI protection.

Since the frame/skin mating surfaces were all tool controlled, no shims were needed for fit
up and thus no need for fasteners of various lengths to accommodate shimming. Substantial
assembly cost savings were realized. A completed aft fuselage skidframe assembly is
shown i; fig :8.

Figure 8.

393
 COST PROJECTIONS, COMPARISONS AND SAVINGS

At the completion of design modifications in 1987, we projected a cost savings in production

basic factory labor for the V-22 Aft Fuselage as a result of automating key operations• In
1988 we performed time studies for fiber tow placement of the stiffened skin to validate the
earlier projections. The time study costs have been evaluated and are first broken down as
cost by operation, then as production basic factory labor savings. The latter includes a direct
comparison with the 1987 projections.

1987 Cost Savings Projection. We projected a cost savings for the V-22 production aft
fuselage section of 54% in basic factory labor. This cost savings is due to automation of the
frame and stiffener forming and skin fabrication (see figure 9). Thisprojection also included
material cost (converted to a labor equivalent) and assembly labor. During the program, we
performed studies to validate the projection.

1988 Time study by Operation. Figure 10 illustrates the results of the time study
performed on automated stiffened skin fabrication. This study included all fabrication
operations for the stiffened skin on the first two deliverable units (SS-001 and SS-002).

In referring to figure 10 note that the mandrel assembly and fiber placement are the largest
cost contributors. However, expectations for labor and material savings for this operation
were not fully realized on SS-001 and SS-002. Further improvements to the fiber placement

•,,100% Aft section detail fabrication

Assy labor
12%

Material
equivalent 10%
• Overall
•""", "',.54% reduction

$50/hr material equivalent

46% of baseline
Fabrication Assy labor
labor 78% material Automated methods
equivalent • Fiber placed skin
Fabrication labor only • Autocomp (Xerkon Inc.)
57% reduction stiffeners and two frames
Fabrication
labor • Quadrapress ® (Teledyne-
73% Ryan) frames - three
• Hand laid up stiffeners
doublers and misc padups,
Hand Layup Automated Geber cut details
Baseline Methods

Figure 9. Automated Technology Application to V-22 Production

Hand Lay Up Versus Automated Methods

394
 m Doubler fab-5%

Stiffener prep-3%

Stiffener fab-10.8%

Mandrel assembly
& fiber placement-
Cure, trim & 55.3%
disassembly-
25.9%

Figure 10.Automated TechnologyApplication to V-22

Time Study Costs by Operation

machine as well as process impovements were integrated into the production rated fiber
placement machine (FPM2) and upgraded tow placement heads. These were used to
fabricate the final deliverable aft fuselage SS-003. Projected lower prepreg tow costs also
contribute to a lower projected production cost.

PROGRAM RESULTS

Significant direct labor cost savings were demonstrated throughout the program. Direct
labor savings, when compared to the V-22 FSD hand layup costs, were more than 50% for
the detail composite component fabrication and their assembly. The demonstrated direct
labor savings for the Xerkon Autocomp ® and Teledyne-Ryan Quadrapress ® frame processes
are shown in figures 11 and 12. Figure 12 shows the Hercules Fiber tow placement process
results.

A segmented learning curve based on prior composite component fabrication experience was
used to calculate the average cost for the 912 production units: Units 1 through 165 follow
an 83% curve, and units 166 through 912 a 90% curve for an average of approximately 84%.
In an effort to more equitably compare the benefit of the fiber tow placement process in
making large cocured fuselage skins, the actual data obtained in the hand layup fabrication
of the first seven shipsets of V-22 components was collected and used to recompute the
baseline cost, figure 13. The reduction found was due to improvement in the manufacturing
process, reduction in the number of stiffener configurations, and improved adhesives and
copper shielding materials. Several of these changes were also incorporated into the fiber-
tow automated shell.

The final assembly operations, mechanical fastening of the five frames, frame splices, and
the fiber-placed one-piece cocured skin also resulted in substantial savings in direct labor

395
costs, as shown in figure 14. These were due to reasons previously presented. An 80% curve
was used through unit 240 and a 90% through unit 912 to calculate the savings.

Preliminary analysis of the data currently available shows that the 84 kg (185 pound) aft
fuselage structure fabricated by the advanced manufacturing techniques demonstrated in
this program can be produced in quantity for the equivalent direct labor cost of
approximately 400 hours. This represents a 64% savings when compared to the hours
required for the conventional fabrication and assembly of the multicomponent baseline
structure.

5.3 MH/Kg
(11.6 MH/LB)

1.9 MH/Kg 2.1 MH/Kg

(4.8 MH/LB)
(4.3 MH/LB)

Hand Layup Phase II1-1 Phase II1-1

Baseline Timestudy Actual

Figure 11. Manufacturing Costs--

Frame Fabrication: Xerkon Autocomp Process
Average for 912 Production Units

4.9 MH/Kg
(11MH/LB)

2.4 MH/Kg
(5.7 MH/LB) 2.19 MH/Kg
(4.85 MH/LB)

Hand Layup Phase II1-1 Phase 111-2

Baseline Actual Actual

Figure 12. Manufacturing Costs-

Frame Fabrication: Teledyne Quadrapress Process
Average for 912 Production Units

396
737 MH
2.04 MH/Kg
(4.5 MH/LB/

Stiffened Skin

529 MH
1.84 MH/Kg
(3.4 MH/LB)

398 MH
1.08 MH/Kg
(2.4 MH/LB)
298 MH
0.81 MH/Kg 243 MH
(1.8 MH/LB) 0.72 MH/Kg

I (1.6 MH/LB)

1987 1990 1987 1989 1990

V-22 Hand Layup Estimate Phase 111-2Actual
Baseline Automated Methods

Figure 13. Manufacturing Costs --

Skin Fabrication: Hercules Fiber Tow Placement
Average for 912 Production Units

!59 MH

Station 559.0 to station 724.5

23 MH 18 MH

V-22 AFML No.1 AFML No.2

Baseline

Figure 14. Costs of Aft Fuselage Assembly by Boeing

Average for 912 Production Units

397
 LESSONS LEARNED FOR COMPOSITE STRUCTURES

R. S. Whitehead
Northrop Corporation

SUMMARY

Lessons learned for composite structures are presented in three technology areas:
materials, manufacturing and design. In addition, future challenges for composite
structures are presented.

Composite materials have long gestation periods from the developmental stage to
fully matured production status. Many examples exist of unsuccessful attempts to
accelerate this gestation period. Experience has shown that technology transition of a
new material system to fully matured production status is time consuming, involves risk,
is expensive and should not be undertaken lightly. The future challenges for composite
materials require an intensification of the science based approach to material
development, extension of the vendor/customer interaction process to include all
engineering disciplines of the end user, reduced material costs because they are a
significant factor in overall part cost and improved batch-to-batch pre-preg physical
property control.

Historical manufacturing lessons learned are presented using current in-service

production structure as examples. Most producibility problems for these structures can be
traced to their sequential engineering design. This caused an excessive emphasis on
design-to-weight and schedule at the expense of design-to-cost. This resulted in
expensive performance originated designs, which required costly tooling and led to
non-producible parts. Historically these problems have been allowed to persist throughout
the production run. The current/future approach for the production of affordable
composite structures mandates concurrent engineering design where equal emphasis is
placed on product and process design. Design for simplified assembly is also
emphasized, since assembly costs account for a major portion of total airframe costs.
The fututre challenge for composite manufacturing is, therefore, to utilize concurrent
engineering in conjunction with automated manufacturing techniques to build affordable
composite structures.

Composite design experience has shown that significant weight savings have been
achieved, outstanding fatigue and corrosion resistance have been demonstrated, and
in-service performance has been very successful. Currently no structural design show
stoppers exist for composite structures. A major lesson learned is that the full scale
static test is the key test for composites, since it is the primary structural "hot spot"
indicator. The major durability issue is supportability of thin skinned structure. Impact
damage has been identified as the most significant issue for the damage tolerance control
of composite structures. However, delaminations induced during assembly operations have
demonstrated a significant nuisance value.

The future challenges for composite structures are threefold. Firstly, composite
airframe weight fraction should increase to 60%. At the same time, the cost of
composite structures must be reduced by 50% to attain the goal of affordability. To
support these challenges it is essential to develop lower cost materials and processes.

399
 Agenda

............................................................
_ii_i!!iiii!iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii!il

• Manufacturing

• Design

• Future Challenges

Material Development Process

POLYMERS

PERFORMANCE CANDIDATE RESIN QUANTITY

• TOUGHNESS
REQUIREMENTS POLYMERS
CUSTOMER _._ REVIEW H • SELECT
HOTANET PROMISING
CAPABILITY (APPROX. 5SMALL
Ibs.)
H FABRICATE _._
• AVAILABILITY

CUSTOMER II PREPREG AND RESEARCH

PREPREG BATCH
• TENSILE STRENGTH

ACCEPTABLE? SCREENING TESTS • PERCENT ELONGATION

• FABRICATE
STANDARD SMALL
FIBERS RESIN TESTS
REVIEW _ LAMINATE • TOUGHNESS
• STANDARD SIZING

ADJUST RESIN CHEMISTRY AND PROCESSING

INTENSE VENDOR/CUSTOMER INTERACTION

L NO
CHARACTERIZATION FABRICATE
INTERMEDIATE TESTS SCALED-UP
RESIN BATCH RESIN BATCH
• PREPREG
YES t FABRICATE _"_ SELECTED ___
(APPROX. 10 Ibs.) (50 TO 80 Ibs.)
• LAMINATE L1 ADJUST

ADJUST PROCESSING I

| PROCESSING

TESTS [ SELECTED _./ FLEXIBILITY

STRUCTURAL VALIDATION IqM. _
• FULLY CONSOLIDATED | CHARACTERIZATION

PANELS _ TESTS

" STIFFENED PANELS / 1__

400
Material Performance Requirements

MATERIAL PROCESSES STRUCTURAL PROPERTIES

LOW COST
PROCESSABILITY CAI

LE-UP M DULUS

PRODUCTION ES_STANCE I THERMAL

PROCESS DEGRADATION

SUITABILITY

MICROCRACKING BEARING

Technology Transition Milestones

• MATERIAL AVAILABLE IN APPROPRIATE QUALITY,

QUANTITY AND FORMS, SCALE UP BY VENDOR COMPLETE

• PRELIMINARY M & P SPECIFICATIONS, MANUFACTURING INSTRUCTIONS

• GOOD AVERAGE PROPERTIES FOR ACCURATE WEIGHT ESTIMATION

• ELIMINATION OF MATERIAL SHOW STOPPERS

• FINAL DESIGN ALLOWABLES

• MATURE TECHNOLOGY

• COMPLETELY MATURED TECHNOLOGY

-4 -3 -2 -1 0 2 a 4 5 YEARS
I I I I P I r

401
 Technology Transition Problems
Examples

• TOUGH BISMALEIMIDE

• HIGH TEMPERATURE THERMOPLASTIC

GALVANIC CORROSION INDUCED DEGRADATION

OF BISMALEIMIDES AND POLYIMIDES

Lessons Learned

• ALL MATERIALS HAVE LONG GESTATION PERIODS FROM

THE DEVELOPMENTAL STAGE TO FULL PRODUCTION
STATUS

ATTEMPTS TO ACCELERATE THE GESTATION PERIOD

INVOLVE SIGNIFICANT TECHNOLOGY TRANSITION RISKS

THERE ARE MANY EXAMPLES OF UNSUCCESSFUL

TECHNOLOGY TRANSITION FOR MATERIALS

• TECHNOLOGY TRANSITION OF A NEW MATERIAL TO FULL

PRODUCTION STATUS IS TIME CONSUMING, INVOLVES
RISK, IS EXPENSIVE, AND SHOULD NOT BE UNDERTAKEN
LIGHTLY

402
 Future Challenges

• INTENSIFY SCIENCE BASED APPROACH TO

MATERIAL DEVELOPMENT

• EXTEND VENDOR/CONTRACTOR INTERACTION

PROCESS TO INCLUDE ALL DISCIPLINES OF END
USERS

• REDUCE MATERIAL COSTS BECAUSE THEY ARE A

SIGNIFICANT FACTOR IN OVERALL COMPOSITE
PART COST

• DRIVE TOWARDS TIGHTER PRE-PREG PHYSICAL

PROPERTIES CONTROL ON A BATCH-TO-BATCH
BASIS

Agenda

• Materials

..........
_ _ iii_i_ii
i_z_iiii_!_i_i_i
i_i_i_!_!i
i iii_i_ii
fill iii_
_i_Z
_i_i!i_i_i_i_i_iiii_ii_i_i!ii!_i_iiiii
ili
::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
ANUFACT URI NGi_:::_i_:_i_ii_i_i_i:_iiiii_.:_i

• Design

• Future Challenges

403
 Analysis of Current Approach

PART PRODUCIBILITY NOT DESIGNED IN

Manufacturing Problems With Initial Production

EXTENSIVE MRB ACTION REQUIRED FOR PART DISPOSITION

Substantial Increase in Support Costs

PART "BUY-OFF" WITH AN EXPENSIVE TEST PROGRAM

Inadequate QA and Effects of Defects Database

RESISTANCE TO REDESIGN FOR PRODUCIBILITY

Avoid Design Cost Increase

Schedule Driven Production Program
Avoid Process Development Costs

• PROBLEM PERSISTS THROUGHOUT PRODUCTION

Example F-18 Landing Gear Door

LANDING
GEAR DOOR

404
 F-18 Landing Gear Door

t--_ 060

_._oo

o L,
A

F-18 Landing Gear Door Stiffener

,jr_,-T' ! "i _-=--

/ / L,-c

SECT D-D SECT E-E SECT G-G SECT H-H

405
 Door Stiffener Analysis

NON-PRODUCIBLE SHAPE, I.E., COULD NOT BE PRODUCED

REPEATABLY AND COST EFFECTIVELY

RADIUS DESIGNED FOR FLIGHT LOADS WITHOUT

ACCOUNTING FOR TOOL/PART INTERACTION

PART DIFFICULT TO REMOVE FROM FEMALE TOOL

(DICTATED BY ASSEMBLY REQUIREMENTS)

HIGH REJECT RATE AND MRB ACTIONS DUE TO

DELAMINATIONS AND POROSITY IN NARROW REGION

Historical Lessons Learned

• PAY-AS-YOU-GO RATHER THAN SUBSCRIPTION PRICE APPROACH

• DESIGN-TO-WEIGHT AND SCHEDULE WERE PRIMARY DRIVERS

• DESIGN-TO-COST NOT EMPHASIZED DUE TO A LACK OF COST

MODELS/METHODOLOGY

• PERFORMANCE ORIENTED DESIGNS REQUIRED EXPENSIVE

TOOLING AND SEVERELY AFFECTED PART PRODUCIBILITY

• M&P SPECIFICATIONS DEVELOPED WITHOUT ACCOUNTING FOR

3-D CONFIGURATION

* INSUFFICIENT LEAD TIME FOR TOOL DESIGN AND PROCESS

DEVELOPMENT

• QA PLAN NOT INCORPORATED IN PART DESIGN

• NO MECHANISM AND IMPETUS FOR INTERDISCIPLINARY

COMMUNICATION AT DESIGN STAGE

406
 Current/Future Approach

IMPLEMENTATION OF CONCURRENT ENGINEERING

Co-Located Multi-Disciplinary Dedicated Teams

Personnel Skills Are a Significant Contributor
Systematic Checks and Balances

PROCESS DEVELOPMENT UNDERTAKEN INDEPENDENTLY

Separate Design and Manufacturing Development Articles

DESIGN FOR SIMPLIFIED ASSEMBLY

Reduced Part Count

Shimless Designs in Certain Areas

REDUCTION OF POST DRAWING RELEASE CHANGES

DESIGN-TO-WEIGHT AND DESIGN-TO-COST USUALLY IN CONFLICT

Design-To-Weight Dominates
D/MI a Key to Balanced Design

Agenda

• Materials

• Manufacturing

• Future Challenges

4O7
 Historical Factors

• SIGNIFICANT WEIGHT SAVINGS

• EXCELLENT FATIGUE AND CORROSION RESISTANCE

• MIXED CERTIFICATION EXPERIENCE

SUCCESSFUL IN-SERVICE EXPERIENCE

(Except Early Vintage Sandwich Structure With Metallic Core)

• NO STRUCTURAL DESIGN SHOW STOPPERS

Lessons Learned- Static Strength

• INHERENT PROPERTY DIFFERENCES EXIST BETWEEN

COMPOSITES AND METALS

• COMPOSITE STRUCTURES ARE SENSITIVE TO OUT-

OF-PLANE LOADS

• MULTIPLICITY OF POTENTIAL FAILURE MODES

• FAILURE MODES OF FULL-SCALE STRUCTURES ARE

DIFFICULT TO PREDICT

• STATIC-STRENGTH TEST IDENTIFIES STRUCTURAL

"HOT SPOTS"

4O8
Wing Component Failure Loads

150 - _ Intermediate Spar/

Lower-Skin Failure [_ Upper-Skin Failure

122 126 124

d 118 119
E3
100
//YA
/
0
d TS1 TS 6 /TS 9///
50

U_

Static RSS Static RSS RSS

#
o

RTA 250 F/1.3% Moisture

409
 ...... . I . . . . -
b
*-
41 0
41 1
 Building-Block Approach
Wing Structure
!

Root Rib q-
Root Root R_b Aft
R_b/Rear Spar Trunnion I
Trunnion Joint Subcomponent i
1
I
I
I

Front Spar/
Skm Joint , r

Torsion Box
Lower
Skkn
"_ j I
Coupons Intermediate Inlermedtate Outboard Fuel I
Spar/Lower Spar/R_b Bay Area I
Skin Jotnt Joint Subcomponent I Wing Component

I
I
I
Upper _ I
Skin Spar Flange/Skin
Coupons
Bolted
Jo,nt Design Verification Testing I Full-Scale Test

Lessons Learned - Durability

• DURABILITY IS A MEASURE OF ECONOMIC LIFE

THIN COMPOSITE STRUCTURES ARE SENSITIVE TO

LOW-LEVEL (<10 ft-lb) IMPACTS
-- Visible Skin Damage
-- Nonvisible Skin/Core Damage
-- Accelerated Moisture Ingression
High Repair Frequency
Part Replacement

HIGH MAINTAINABILITY COSTS

412
Defect/Damage Severity Comparison
Compression
1.0

0.9
_D
(D _ RT/Ambient
0.8
IE "-"
O') 0.7
_3 c-
0.6
(-- .,_,
O9 _'_-_,.\ "_•_'o
0.5
O'_ ctl 0.4 • _ Delamination
co OO
c- 03 O ...... Flawed Hole
O--- Porosity Ba, ely "i"_- _- i
0.2 Visible
• .... Impact Damage
0.1 Easily
Visible
0 i i 1. i
0.5 1.0 1.5 2.0 2.5
Damage Diameter (in) or Percent Porosity

Lessons Learned- Damage Tolerance

• IMPACT DAMAGE IS THE MOST SEVERE DEFECT/DAMAGE TYPE

• IMPACT-DAMAGE AREAS AND STATIC STRENGTH ARE STRONG-

LY DEPENDENT ON STRUCTURAL CONFIGURATION

• FAILURE MODES OF IMPACT-DAMAGED BUILT-UP STRUCTURE

ARE SIGNIFICANTLY INFLUENCED BY STRUCTURAL CONFIGURA-
TION

• SIGNIFICANT IMPACT-DAMAGE TOLERANCE SCALE-UP

EFFECTS EXIST FOR BUILT-UP STRUCTURE

• IMPACT-DAMAGED STRUCTURES ARE RELATIVELY INSENSITIVE

TO FATIGUE LOADING

• DELAMINATIONS INDUCED DURING ASSEMBLY OPERATIONS

HAVE DEMONSTRATED A SIGNIFICANT NUISANCE VALUE

413
 Lessons Learned- Design Criteria

• STATIC STRENGTH, FATIGUE/DURABILITY AND DAMAGE

TOLERANCE

• THERE ARE SIGNIFICANT DIFFERENCES BETWEEN

USAF, NAVY, ARMY AND FAA REQUIREMENTS

• RATIONALIZATION OF DAMAGE TOLERANCE REQUIRE-

MENTS INTO A SINGLE DOCUMENT IS IN PROGRESS

BY J. JAEB (BOEING) THROUGH AIA

• BEWARE OF THESE DIFFERENCES WHEN DESIGNING

AN AIRFRAME FOR MORE THAN ONE AGENCY

Agenda

• Materials

Manufacturing

• Design

FUTURE CHALLENGES _

414
Future Challenges

• REDUCE THE COST OF COMPOSITE AIRFRAME STRUC-

TURES THROUGH MULTIDISCIPLINARY (MATERIALS, MAN-

UFACTURING, DESIGN) CONCURRENT ENGINEERING

• INCREASE COMPOSITE AIRFRAME WEIGHT FRACTION

TO 60%

• DEVELOP LOW COST MATERIALS AND PROCESSES

ESSENTIAL IN ORDER TO MEET THESE GOALS

415
 National Aer onaullcs and Report Documentation Page
Space Adm_nlslra1_on

1. Report No. 3. Recipient's Catalog No.

NASA CP-3104, Part 1 2. Government Accession No.

4. Title and Subtitle 5. Report Date

First NASA Advanced Composites Technology Conference January 1991

6. Performing Organization Code

7. Author(s)
8. Performing Organization Report No.
John G. Davis, Jr., and Herman L. Bohon, Compilers
L-16889
10. Work Unit No.
9. Performing Organization Name and Address
510-02-11
NASA Langley Research Center
Hampton, Virginia 23665-5225 11. Contract or Grant No.

13. Type of Report and Period Covered

12. Sponsoring Agency Name and Address
Conference Publication
National Aeronautics and Space Administration
Washington, DC 20546-0001 14. Sponsoring Agency Code

15. Supplementary Notes

John G. Davis, Jr.: NASA Langley Research Center, Hampton, Virginia.

Herman L. Bohon: Lockheed Engineering & Sciences Company, Hampton, Virginia.
16. Abstract

This document is a compilation of papers presented at the firstNASA Advanced Composites

Technology (ACT) Conference held in Seattle,Washington, from October 29-November 1, 1990.
The ACT program is a major new multiyear research initiativeto achieve a national goal of
technology readinessbefore the end of the decade. Conference papers recorded resultsof research
in the ACT Program on new materials development and processing, innovative design concepts,
analysisdevelopment and validation,cost effectivemanufacturing methodology, and cost tracking
and prediction procedures. Papers presented on major applications programs approved by the
Department of Defense are also included in this document.

17. Key Words (Suggested by Authors(s))

Thermoplastics Aircraft
Thermosets Composite design
Review for General Release
Stitching Manufacturing
Graphite fibers Analysis January 1993
Processing Subject Category 24

19. Security Classif. (of this report)

Unclassified Unclassified
20. Security ClassiL (of this page) 21. 426No.
of Pages 22. Price

NASA FORM 1626 OCT S6 NASA-Lan_, IHI