NAVAL POSTGRADUATE SCHOOL

Monterey, California (
AD-A283 El 618
I111111 INE~UIH
I

DTIC
ELECTE
AUG 2 4 1994

THESIS
Course Module for AA2021: Wing Structural Design Project
(V- By
CD Stephen A. Bunis, P.E.
N._ March 1994

O} - Thesis Advisor: Gerald H. Lindsey, Ph.D.

Approved for public release: distribution is unlimited

_ _ _94
8 23 025
Unclassified
SECURITY CLASSIFICATION OF TMS PAGE

REPORT DOCUMENTATION PAGE O °.07°"°15

Is. REIVK IOURIfT CEMASSI1AIC IIt). MY..3ICIUIVII M"
Unclassified
Ze. sm.URilrul A I,.ILAI]A14 AUIITIUI 1 3. L31I1UD•IJUNIAVADIL[AU.I Y Ut K1MV •U
_________________I________________________ Approved for public release: distribution is unlimited

4. PERFORMING URUANIZA111014 KEPuIT NUMBER(S) 3. MUNFIURINU ORUAPIIZA-HON E•EUOR N13MERS))

6L NAME UWV12KML1U UOUAANFLAI jUN 6b. UrT(lC 3YMBOL A I& NAM OFt"
MU MUrlUxU
UKUANIZAIIUIN
(If applicable)

Naval Postduate School Naval Postaduate School

6C. AUUKE33 (Uty, !,Mte. Alld Zr Code0 In. AUDVMSS (LUity. aitate, aod "rp f-nc)
Monterey, CA 93943-5000 Monterey, CA 93943-5000
BT.NAMO FUNDI
04ONUISIUNURINO 9b. OFFICE SYMBOL 9. FRitt2 KILNST/ Ru MEN UJENTUILA-InN NUMBER
.I
ORGANIZATION (If applicable)

3c. AD . (Uty.
i late, and ZIP It. SOURICE l'-I INFINUNUMBIRS
rU'URAM IPROJECT IWOK UNrT
ELEME rNO. NO. N ACCESSION NO.

11. ILl1± (Include Security Ulaaildication)

Course Module for AA202 1: Wing Structural Design Project
17- PIRSUNAL AUTORK(S)
Stephen Allen Burris, P.E.
1-a. TYMP0F REIVKI 13b. 1114 UVFREUD 14. ^ATE O1 REFUK"IYear.M4ondi.Dy) 15. PAGE COUNT
Master's Thesis IFROM TO March 1994 132
it. SNUTMIAR NIIAIIUN
The views expressed in this thesis are those of the author and do not reflect the official policy or position of
the Department of Defense or the U.S. Government.
17. CUSAII COVES 1. SUIBJECTI IAMS (Continue on reverse it necessary and identify by block number)

Course Module for AA2021: Wing Structural Design Project

19. A CTUontinue
(B• on reve -it necesaary and identify by block number)
This thesis defined a fundamental approach for aircraft wingbox design appropriate for an introductory
course in aircraft structures based upon material strength and stiffness requirements. The process developed
sought to encompass major conceptual engineering design considerations that ranged from load estimation at
various points in the subsonic flight envelope, to initial structural sizing and layout. The goal was to present a
process that could be readily conducted via hand calculations and applied by any student entering basic aircraft
structures design.
The sequence of analysis began with application of a comprehensive panel code method developed by
NASA Ames Research Center known as PMARC. Loads obtained from the code were then used to formulate
a strength of materials study of the structure subjected to combined bending, shear and torsion. The static load
approach allowed initial estimation of component sizing based upon material or buckling allowable stress
selection. Finally, the study demonstrated a strength to weight ratio comparison.
Several calculation examples and computer-based spreadsheets were prepared for rapid analysis of multiple
option design scenarios. Since the study employed analysis methods that could be performed without the aid
of a finite element routine or extensive computer programming knowledge, it serves as a good introduction for
the entry and intermediate level structural engineer.

7U. IJKDUINIRUHU14AVAII-AMUIJTY Ur AMI R-AL.-1 21. AJSNI-KAU'I 3BC'UKrI Y

I L:,AS3•II*:A'IIUN
[R] LNLA&wmw)wNII [] SAME AS RPr. -DnC USERS Unclassified
27. NAME U• IDLE I "IVUJUAL, ZA. T- UI.MJUN
(anclude Area Code) alc. UFFIC. SYMB(A.
Prof. Gerald H. Lindsey, Ph.D. (408) 656-2311 AA/Li
DO Form 1473, JUN UWevidus editions are obsolete. '-Iz W ry U AM MII.I(-IICUIIM IUM
ll
S/N 0102-LF-014-6603 Unclassified

i
 Approved for public release: distribution is unlimited

Course Module for AA202 1:

Wing Structural Design Project

by Stephen A. Burris, P.E.

Lieutenant Commander, United States Navy
B.S., United States Naval Academy, 1982

Submitted in partial fulfillment

of the requirements for the degree of

MASTER OF SCIENCE IN AERONAUTICAL ENGINEERING

from the

NAVAL POSTGRADUATE SCHOOL

March 1994

Author: .____
hen A. Burris

Approved by: G rl T0 n " ~ ei

Accesion For
NTIS CRA&I
DTIC TAB w Wu, Second Reader
Unannounced
Justification

By-Distribution
.........
I
Daniel J. CollinsVChairman
Department of Aeronautics and Astronautics

vnil
V'It and/or
Dist Special

fiC
 ABSTRACT

This thesis defined a fundamental approach for aircraft wingbox design appropriate

for an introductory course in aiicraft structures based upon material strength and stiffness

requirements. The process developed sought to encompass major conceptual engineering

design considerations that ranged from load estimation at various points in the subsonic

flight envelo, - o 1c,'al structural sizing and layout. The goal was to present a process

that could be readily conducted via hand calculations and applied by any student entering

basic aircraft structures design.

The sequence of analyis began with applicat;3on of a comprehensive panel code

method developed by NASA Ames Research Center krown as PMARC. Loads obtained

from the code were then used to formulate a strength of materials study of the structure

subjected to combined bending, shear and torsion. The static load approach allowed initial

estimation of component sizing based upon material or buckling allowable stress selection.

Finally, the study demonstrated a strength to weight ratio comparison.

Several calculation examples and computer-based spreadsheets were prepared for

rapid analysis of multiple option design scenarios. Since the study employed analysis

methods that could be performed without the aid of a finite element routine or extensive

computer programming knowledge, it serves as a good introduction for the entry and

intermediate level structural engineer.

 TABLE OF CONTENTS

I. INTRO D U CTION .................................................... I

11. LOAD AN ALY SIS ................................................... 3
A. REQUIREMENTS ............................................... 3
B. REVIEW OF AVAILABLE METHODS ......................... 4
C . PM A RC .......................................................... 5
1. Validation Study Summary ................................... 6
2. U ser Interface Issues ......................................... 7
D. SAMPLE PROBLEM RESULTS ................................ 10
1II. STRENGTH ANALYSIS ............................................ 15
A. BOUNDARY CONDITION DEFINITION ...................... 15
1. Shear Diagram Construction ................................. 18
2. Bending Moment Diagram .................................. 19
3. Torsional M oment ......................................... 23
B. BENDING STRESS ANALYSIS .............................. 27
1. Area Determination ......................................... 28
2. Determination of Allowable Stress ........................... 30
3. Spar Placement Considerations .............................. 31
4. Straight Wing Approach ..................................... 33
5. Swept Wing Considerations ................................. 34
C. SHEAR STRESS ANALYSIS ................................. 37
1. Shear Flow due to Torsion .................................. 37
2. Shear Flow due to Applied Loads ........................... 38
D. VARIATION IN OPERATIONAL CONDITIONS .............. 41
E. SKIN SIZIN G .................................................. 42

iv
IV. BUCKLING CONSIDERATIONS ................................... 46
A. STIFFENER SPACING ........................................ 46
B. STIFFENER SIZING ............................................ 49
1. Initial Size Estimate ......................................... 50
2. Uniform Thickness Approach ................................ 54
3. Stringer Cross-Section Stability ............................. 55
V. STRENGTH TO WEIGHT .......................................... 58
A. BENDING STRENGTH TRADE STUDY ....................... 58
I. M aterial Selection ........................................... 58
2. Forward Spar Location ...................................... 60
B. INSTABILITY IMPLICATIONS .............................. 60
C. ADVANCED MATERIAL CONSIDERATIONS ................ 61
VI. SU M M AR Y ......................................................... 63
A. FUTURE AREAS OF STUDY ................................ 63
1. Aerodynamic Investigations ................................. 63
2. Advanced Structural Analysis ................................ 64
B. CONCLUSION ................................................ 66
APPEN D IX A ......................................................... 68
A. WAKE INITIALIZATION ...................................... 68
B. ANALYTICAL VALIDATION ................................. 71
C. COMPUTER-BASED VERIFICATION ......................... 76
D. EXPERIMENTAL DATA COMPARISON ...................... 77
APPEND IX B ........................................................ 79
APPEND IX C ....................................................... 114
A. LOCATING THE FORWARD SPAR ......................... 114
B. RIB SPACING CONSIDERATIONS .......................... 116
APPEND IX D ....................................................... 121
LIST OF REFERENCES ............................................... 122

V
INITIAL DISTRIBUTION LIST ....................................... 124

vi
 I. INTRODUCTION

This thesis provides a systematic approach to the preliminary design of an aircraft

wingbox structure that employs a two spar support base. The method chosen was based

upon an applied elasticity approach as described by Allen and Haisler [Ref. 1:pp. 145-233]

using strength and stiffness specifications as the design target. The intent of this effort

was to provide insight into the flow of engineering design decisions that begin in load

determination and end with preliminary structural layout. As a result of this work, several

study sections were developed for incorporation into the current aeronautical engineering

curriculum at the Naval Postgraduate School (NPS) to be presented in the undergraduate

and graduate core structural engineering courses.

The organization of this thesis begins with determination of the static load boundary

condition applied to an airfoil of arbitrary geometry. To solve the low speed problem for

designs encountering flight speeds up to and including Mach number 0.6, a potential flow

panel code method was chosen that allowed user definition of wing geometry, and total

aircraft, if desired. The code was used in this research to calculate spanwise force and

moment information; however, it is also capable of providing pressure distribution

information along the surface of the structure as well.

Once the loads were obtained along the span from the code, they were used as a

series of discrete point loads. From this information traditional bending and shear

diagrams were constructed for use in the strength analysis. As part of the instructional

1
module for the Introductory Aircraft Structures, AA 202 1, a teaching guide and

automated spreadsheet routine were developed to aid in presenting this material in a four

hour lecture series.

The thesis continues with a review of the buckling stability of a wing cover subjected

to the stresses projected from a shear flow solution. Using the methods described in

Sechler and Dunn, an analysis of a stiffened wing section was conducted to determine

potential stringer size, geometry and total number required to support the structure at a

given cross-section [Ref. 2:pp. 154-258]. As these chapters are presented, several design

questions arise from the structural engineer's perspective that are presented and answered.

At the conclusion of the research, several follow-on topics were presented for future

thesis work by students at NPS. Since this work does support an ongoing design effort

that is a fundamental segment of the curriculum for all students within the department, this

work can serve as a valuable reference for use in future aircraft design reviews and

competitions.

2
 II. LOAD ANALYSIS

A. REQUIREMENTS
The very nature of design requires tools that support creative and interactive

development of aircraft structures. Interactive flexibility demands that the designer must

be able to quickly analyze several options feeding back to arrive at a configuration that

best suits the requirements at hand. The boundary condition imposed upon an aircraft

structure is the result of an external pressure distribution, internal structural weight and

internal pressure distribution associated with a given flight envelope. These components

of the overall boundary condition can then be idealized in a number of ways. Any method

chosen to idealize the boundary condition to allow solution of its impact on the strength of

a structuie, necessitates the capability to adapt to arbitrary geometries and provide

meaningful, accurate data in a reasonable amount of time.

Several Computational Fluid Dynamics (CFD) methods exist today that provide very

reliable data for nearly all aspects of the flight envelope. Unfortunately these methods also

require extensive training and experience in grid generation, input data preparation,

interpretation of results, etc.; in addition to tremendous computer processing time.

Another drawback to CFD is the time associated with development of an individually

tailored analysis grid which can require several weeks of effort.

Earlier estimation methods employing conversion of 2-d airfoil data into 3-d wing

results, as described by Nicholai [Ref. 3:pp. 2-1:2-9] and Raymer [Ref. 4:pp. 342-344]

3
tend to leave many students searching for specific solutions to problems that are not

covered by generic tables or charts.

B. REVIEW OF AVAILABLE METHODS

For the reasons listed above, the CFD approach was rejected for use in an

introductory to intermediate level approach to load determination. Since most tactical

naval aircraft operate in both the supersonic and subsonic flight regimes, methods of load

approximation are required in both arenas. The problem of estimating supersonic loads

was believed to be reasonably approximated by the application of 2-d gas dynamics to

obtain the external resultant pressure. Since all supersonic aircraft also operate in the

subsonic regime, the load calculation method chosen for this work was for those loads

associated with flight operations in the 0.0-0.6 Mach number regime. One area this thesis

did not address concerns the load approximation for flight in the transonic regime. For

first order approximations at those speeds, the reader is encouraged to review the methods

depicted in Raymer [Ref 4:pp 293-297].

Several analytical and empirical approximations in the subsonic range have been

developed throughout the years to aid the designer in a quick approximation of the loads

experienced along the span of a wing. Two of the more recognized approaches include

DATCOM and Schrenk's approximation. Unfortunately during previous work, several

students found these methods lacking in sufficient detail when confronted with changing

geometries, (consider the case of high-lift devices), or actual changes in airfoil

cross-section, (i.e., wash-out). However these methods did prove to be useful in

4
validating computer solution methods. A key design lesson learnedin the applicationof

any computer solution is a thorough validationof expected resultsfrom a second

approachat one or more known benchmarks.

Students are exposed to potential flow theory early in the aeronautical engineering

curriculum at NPS and generally understand the underlying principles of most pane! code

solution methods. Prior to replacement of the department's VAX computer network

system, a panel code produced by NASA Langely known as SUB had been used

-extensively as a design tool by many students. During the interim between the loss of the

VAX and complete system integration of a UNIX based Silicon Graphics. workstation

network, the department sought to incorporate a new panel code routine for use by *he

aerospace engineering student body.

C. PMARC
During 1993 LCDR Dave Porter, USN, obtained a panel code method from NASA

Ames known as Panel Method Ames Research Center (PMARC), for use in his thesis

research. The software was accompanied by an additional routine, also developed by

NASA Ames, known as General Visualization System (GVS). This second routine allows

a 3-d graphical interpretation of the output from PMARC, which includes color display of

pressure distribution, streamline generation, wireframe modeling and solid surface

rendering.

PMARC enabled the designer to rapidly construct a complex geometry and analyze

flows around that geometry in the subsonic regime. Since the code allowed for assembly

5
of several component geometries, it allowed for load analysis of difficult layouts that were

previously considered unworkable by most students in the preliminary design phase (a

large fowler flap assembly is one example).

Installation of this code for use by the entire department did not occur without some

additional effort beyond simple file transfer to the system file server. The first obstacle

encountered with PMARC was a lack of adequate information for the first time user

regarding the generation of input and interpretation of output. To reduce the learning

requirements placed upon the user, an installation and introductory instruction manual

[Ref, 5] was developed for use by all department members as a part of this thesis work.

The second hurdle to overcome was development of a means to benchmark the

performance of the code.

1. Validation Study Summary

This section highlights a few of the tests that were conducted to validate the

output generated by PMARC. Although one set of validation results was provided by

NASA Ames with the original documentation, they involve a rather complex wing-body

combination that is not well-suited for generation and duplication by the average student

user. One example of PMARC's capability to produce an acceptable 3-d lift curve slope

using a NACA 4 Digit airfoil cross-section is presented in Figure 1.

The comparison depicted is made from a 2-d transformation method described by

Nicholai [Ref 3:p. 23] that uses the airfoil lift curve slope adjusted for overall geometry.

Another comparison performed to verify the results was on a wing based using a

6
NACA 6 Digit airfoil. The comparison was made against wind tunnel data presented by

Bertin and Smith using an unswept trapezoidal wing at low mach number [Ref. 6:pp.

253-257]. Those results are summarized in Figure 2.

Several validation studies were performed to check the accuracy of output, in

addition to learning the correct data input procedures. These studies demonstrate a

Comparismn of PMARC and Transformation of

2-D Section Data Into 3-D Plot
1.2

0.8 _

S0.6 _
.0.4 -
"0.2
0 -
(5) 0 5 10 15
AOA (degrees)

3-D Approximation PMARC Solution

Figure I

method for constructing wing shapes of various cross-sections using standard airfoil

shapes, in addition to modeling individually tailored designs. A complete summary of

these validation studies can be found in Appendix A.

2. User Interface Issues

Since PMARC allows for a wide range of analysis including boundary layer

investigation, ground effect corrections, multiple geometries and oscillatory analysis, it

7
requires a sizable portion of input data in a FORTRAN format. The code does allow for

quick generation of any wing shape that employs a NACA 4 Digit airfoil by employing an

automatic panel generating scheme. However all other cross-sections require a

specification of the geometry in 3-d space for all dividing sections; i.e., the root and tip

cross-section. To provide easier body generation, the code allows several different axis

orientation systems to build component geometries. During the development of a user's

guide, the benefit of templates proved invaluable in reducing data input file construction

time and minimization of data entry errors. A model template directory was constructed

on the IRIST computer system that was accessible to all users. This directory contained

VWind
Tunnel Test Compared
to PMARC EstimaUon
1.4
1.2 00
! 1

0.6
0.4
0.2
0
(0.2).'°" , . , I
(4) 0 .4 8 12
AOA (degrees)
EXPMImO PMARC POW-I Fow
Rssal SoluAin Theoy
0 6 -

Figure 2

examples of the construction of several NACA 4 Digit airfoil sections and one NACA 6

Digit custom cross-section design. With less than a few hours of instruction and practice,

the average user can construct a NACA 4 Digit wing in approximately 10 minutes. Given

8
the coordinates of a custom cross-section, most users can construct an input file using one

of the templates in approximately 20 minutes. This short construction time proves to be

very useful in a time-constrained design syllabus.

For introductory design work, the force and moment coefficient output of

PMARC was determined to be easiest to use for most students investigating strength

requirements of a given design proposal. However, the availability of pressure data offers

some interesting possibilities for advanced study that will be addressed later in this report.

The lift and drag coefficient data provides information for each individual panel, a

summary of all columns of panels along the span, component summary (the right wing for

example) and complete assembly summary in the case of an entire aircraft. For this

project the column summaries were used to develop an approximate discrete load acting at

the centroid of each column section. The column in this output is based upon the resultant

force and moment that is obtained by grouping all of the panels around a section of the

airfoil between two neighboring spanwise station locations. For instance, if the wing

panels divided the wing into ten equally spaced segments starting from centerline, the first

column corresponds to the first segment, and all panels acting on the upper and lower

surface of that segment. In this analysis, the resultant loads due to lift and drag for the

column have been assigned to discrete locations within each column, i.e., the center of the

column in the spanwise direction and the center of pressure in the chordwise direction.

9
D. SAMPLE PROBLEM RESULTS
The geometry chosen for analysis is depicted in Figure 3, including operating

condition data. To demonstrate the procedure used in developing the instructional

modules, one aircraft configuration was chosen to be used as a repetitive example for the

design sections of coursework encountered throughout the structures discipline at NPS.

This allowed students to compare results using different methods of analysis as their skill

level and exposure to advanced methods increased.

This wing was chosen to support an aircraft operating at a gross weight of

150,000 lbs subjected to a maximum allowable load factor of 3.5 g's. To demonstrate the

methodology involved, one flight operating condition was used for anldysis and sizing

throughout this paper, with a discussion to follow on how to apply these techniques to

multiple flight operating conditions.

Given the allowable load factor, the condition chosen for analysis was slightly above

the comer velocity. The wing was assumed capable of delivering a maximum lift

coefficient of 1.35 at 572 ft/sec. Since the panel code is based upon potential flow, the

analysis point was moved out away from cu in an effort to avoid substantial bias due to

a lack of non-linear lift analysis capability in the routine. Using sea-level conditions of

0.53 mach and standard day density, the required cL was determined via equation (1).
CL= 2nW
CL- 2
PV Aref (1)

The result was a minimum cL of 1.269 with the corresponding AOA to be determined

from the 3-d lift curve versus Angle of Attack (AOA) diagram. To obtain a lift curve

10
 Planform 1

Cr = 15ft Airfoil: NACA 4418

ct = 5ft AR = 10.0
= 10.83ft Re = 38,378, 000
IMN=0.5
A 10.33 Temp =60°F
A ref = 1000ft2 Altitude = Sea Level
b = 100ft AOA=2 0

Figure 3

11
diagram, a PMARC input file for this wing was created and evaluated at several different

angles of attack. Table I is a summary of those results over various alphas. Using the

data of

Table I and simple linear regression analysis, the corresponding AOA was found to be

9.025 degrees at the desired operating condition.

TABLE 1

Column CL
-2.0000 0.2022
0.0000 0.3972
2.0000 0.5905
4.0000 0.7841
6.0000 0.9755

This angle of attack was then used as the input for PMARC to obtain the spanwise

load. The code's panel generation routine was employed for model construction. The

routine provided a spacing technique that increased panel density near the leading edge,

trailing edge and wingtips for improved solution fidelity. Figure 4 depicts the resultant

column lift coefficient distribution along the semi-span of the right wing. In addition to

the lift coefficient output, PMARC also provided extensive coefficient information in both

the wind axis and body axis system. From a structural standpoint, the load analysis was

based upon the normal and axial forces. Since these forces were computed to act through

the center of pressure on the aircraft's chord line, some modification of the results was

12
required to determine structural boundary conditions acting at the structures centroid.

Table 2 is a summary of the coefficients that were required along the span.

Spanwise Lift Coefficient Distribution

1.4

1.2

A1
0.8

0.6 0 , , , , , , I ,
20 40 60 80 100
Percent of Semi-Span

Figure 4

To simplify the following load analysis, the load was assumed to be a discrete load

acting at the center of each column. To obtain the actual load, the force coefficients were

multiplied by the planform area of each column and the freestream dynamic pressure. The

moment acting at each column was obtained by multiplying the coefficient by planform

area, dynamic pressure and the wing Mean Aerodynamic Chord (MAC, also referred to as

cbar). The moment is the resultant torque due to the normal force applied at the column

center of pressure. The location of moment computation was based upon a user defined

reference line specified within the input data file.

13
 TABLE 2
Column FtI@,
S CL CN CA C3

1.0000 0.0780 1.2723 1.2621 -0.1723 0.1256

2.0000 0.2330 1.3515. 1.3398 -0.1877 0.1934
3.0000 0.3820 1.4066 1.3932 -0.2017 0.2641
4.0000 0.5210 1.4430 1.4283 -0.2121 0.3315
5.0000 0.6470 1.4604 1.4451 -0.2171 0.3903
6.0000 0.7580 1.4492 1.4343 -0.2137 0.4344
7.0000 0.8500 1.3906 1.3779 -0.1970 0.4537
8.0000 0.9210 1.2482 1.2396 -0.1619 0.4330
9.0000 0.9690 0.9951 0.9933 -0.1020 0.3569
10.0000 0.9940 0.7335 0.7597 0.0722 0.2451

14
 III. STRENGTH ANALYSIS

A. BOUNDARY CONDITION DEFINITION

From the discussion in Chapter 1I of the panel code outputs along the semi-span of

the wing, it is possible to obtain a reasonable approximation of the load pattern. This

discussion will treat the wing as a composite of four bar elements as depicted in Figure 5.

Note that these elements are not constrained to be symmetrical about the X or Z axis.

Right Wing Semi-Span

Figure 5

Figure 5 also depicts the orientation of the axis system that was used throughout most

of the analysis. This sense of direction is consistent with the output from the panel code

where coefficient data was obtained.

Since force and moment computation requires knowledge of the reference area, the

planform area for each spanwise column must be determined. The panel code does

provide the location in x,y,z coordinates for the corner points of each panel if instructed to

do so within the input file. From these locations and simple trigonometric relationships,

15
the planform area of each column can be determined for use in the calculation of applied

forces and moments. Figure 6 serves to describe the orientation of a positive load

condition. Since this project was directed towards preliminary, vice detail engineering

Normal Force, +Nl

•.-..... Chord Une

Axial Force, +A

Figure 6

design, the problem was analyzed at the expected maximum load points near the wing

root. However, this method can be applied throughout the semi-span to determine initial

minimum sizing to meet strength requirements.

At this point in the design, a common question among new design students pertains to

the choice of a two spar system. Since this four element structure is being initially

designed to carry the bulk of normal stress due to bending, it follows that a two spar

wingbox employing stiffened wing skins will generally provide a low cost, easily

manufactured, lightweight structure. Another advantage as pointed out by

Niu, [Ref. 7:pp. 247-263] was large interior volume for increased fuel capacity. These

benefits seem well suited for large commercial and military carriers.

16
 This project began by assuming the simplest possible configuration based upon a two

spar arrangement. As loads increase and size constraints decrease in a design project, the

feasibility of the two spar structure begins to diminish. The designer needs to look at his

results during analysis and determine if they appear reasonable. For example, if the load

factor requirement is increased, the spar cap cross-section will also increase. Additionally,

the wall thickness of the front and rear spar webs must also be rising to meet the increased

shear requirements. As these components grow in size, the designer needs to consider the

effect of adding one or more additional spars to begin reducing the structure's internal

cross-section dimensions (and ideally total wing weight).

Operating requirements may steer the designer in other directions, as evidenced in

current tactical military aircraft structural design. Generally these aircraft demand high

strength, while also minimizing vertical wing depth to allow for high speed operation.

Most tactical aircraft also require multiple load paths for enhanced survivability if

subjected to hostile fire. Consider the ramifications of a fore and aft spar structure that

has been severed at the leading face of the wingbox (forward spar) under a maximum

loading condition. The unexpected load path may prove too weak to carry the new load

distribution, resulting in catastrophic failure. Such a design problem may seem

overwhelming at this point to the beginning structural engineer. Hence, this project has

been constrained to a problem that readily allows solution via hand calculations within the

classroom without reliance on finite element or energy methods.

17
 1. Shear Diagram Construction

Returning to the normal and axial force coefficients, the resultant loads acting at

each spanwise station of the wing are depicted in Figure 7. These loads provided the basis

for calculation of the shear in the z and x direction across the face of each section. Due to

the assumption that the load acted at discrete points, the shear was assumed constant

across each column element. The means of estimating shear along the semi-span was to

start at the wing tip with zero shear and sum the loads moving inboard to the aircraft

centerline. Figure 8 represents the shear loads for the two axes along the semi-span of the

wing. For modeling considerations, the wing was assumed to be a continuous structure

Load Distribution
60,000
50,000
" 40,000
30,000
20,000

_j 0
(10,000) 0 15.45 29.39 40.45 47.55 50
Semi-Span Locailon (ft)

* Axial Force 0 Normal Force

Loading Loading

Figure 7

extending through the aircraft centerline; however, strength analysis modeling was based

upon a cantilever beam fixed at the aircraft centerline and extending out the right wing.

18
All shear values were assumed to change at the left face of each column section moving

along the y axis.

2. Bending Moment Diagram

The bending moment diagrams for the load condition previously selected are

presented in Figure 9. These diagrams prove a bit more challenging to construct due to

the number of repetitive calculations involved. Since the design engineer is likely to

consider hundreds, or even thousands of load and geometry combinations during the

design sequence, construction of these diagrams is a prime candidate for automated

calculation.

These diagrams proved quicker to build with the aid of a spreadsheet by starting

at the fixed end of the wing, on centerline. Figure 10 serves as an example of the

calcuiations for the moment on the left face of column 8 near the right wingtip. The

resultant moment for each face was calculated by summing up the product of all forces to

the right of a cut section and their respective distances to the cut. These sums provided

the data for the approximate moment at the left face of each column section.

19
 Vertical Shear, Vz
400,000

300,000

200,000 -

100,000

O0I 10 20 30 40 60
Semi-Span Location (fi)

Axial Shear, Vx
2000
oo
00

-6000
'-8O00O

-10000

Figure 8

20
 Bending Moment, Mxx

10,000,000

8,000,0001

•, 6,000,000

S,000,000

2,000,000

0 10 20 30 40 50 60
Semi-Span Location (ft)

Bending Moment, Mzz

250,000

200,000

9150,000-

j100,000 4

50,000 ->

(50,000) 0 ' '

0 10 20 30 40 50 60
Semi-Span Location (ft)

Figure 9

21
 12,196 lb
+z;• 59352 lb
1,333 lb

44.55 ft +x

46.05 ft
48.45 ft
49.70 ft
Measured from Centerline

Mxx = 12196(46.05-44.55) + 5352(48.45-44.55)

+ 1333(49.70-44.55)
= 46,032 ft-lbf

Figure 10

22
 3. Torsional Moment

Production of the torsional moment diagram depicting Myy along the semi-span

requires the transfer of several couples prior to the final diagram construction. PMARC

provides a value for the pitching moment coefficient for each column of panels based upon

the planform area of each column and referenced to the Mean Aerodynamic Chord (MAC)

of the overall wing structure. Equation (2) serves to illustrate the relation between cM,

and the actual moment produced, M.

M= VA fC-cm (2)

It is important to understand that cM and M can be used to specify moments

anywhere on the wing cross-section. However the structural engineer is generally

interested in the resultant moments found at the centroid of the cross-section under

investigation. Anderson [Ref 7:pp. 15-30] provides an extensive discussion of the

integral approach to resolving the pressure force acting on the surface of the wing into

normal and axial components. These forces can then be resolved into a single pair of

forces acting at the center of pressure on the structure. If these forces, N and A are

applied at the center of pressure (xcp), there is no induced pitching moment along the

chord line. Figure 11 depicts the order of transformation required to obtain a usable

relationship.

Since the center of pressure is unlikely to be located at the cross-section's

centroid, a relationship must be determined that can relate the moment arm distance

obtained between xcp and the selected reference. Returning to Anderson's [Ref 7:p. 30]

23
discussion of normal forces and resultant moments, the center of pressure is defined via

equation (3).

XCP= N (3)

Note that this relationship only applies at the Leading Edge (LE), which is the basis for the

transformation depicted in Figure 11. Another important aspect to equation (3) is the

definition of positive orientation for the moment produced. Conventional aerodynamic

notation directs that any force acting on the wing cross-section which causes a nose-up

4Ma ML=
4*ft (-RMPX)lt)+MI,

Figure 11

rotation of the wing leading edge is deemed positive, (note that positive lift results in a

negative ML). It is also important to realize that aerodynamic sign covention for

moments and forces may not coincide with the convention required of structural analysis.

It is the design engineer's responsibility to properly translate the appropriate sign when

moving among the two disciplines.

Since any moment can be expressed as the sum of two or more vector

components (the components being moments themselves), any moment obtained along the

24
chord, where its location and corresponding normal force is known, can be used to obtain

the center of pressure. Since PMARC allows specification of the line about which it

calculates cM (specified as RMPX), and provides cN as an output, it is possible to locate

the center of pressure. Figure 12 highlights the relationship graphically among all relevant

measurements for a swept wing (referenced to the X=0.0 datum line). Equation (4) is

used to solve for the location of XCP when provided with the necessary input and output

data from PMARC.

XCP = RMPX- Mppx (4)

N
In equation (4), N is the normal force obtained from the normal force coefficient cN, (note

that the output format of the code labels cN under the format of cL in the body axis system

Alowing for Leading Edge Sweep:

X0.o0

XR.PX
Caenr of

Contmldal Axis
',, ,.o . . .....
/ .

Figure 12

and reserves cL as an output of the wind axis system). Likewise, MRx is the moment

obtained from equation (2) using the output of the code labeled cm (for the pitching

25
moment output in the code there is no difference between cM in the wind axis and the body

axis system).

The remaining translation required involves the effect of the normal force acting

at the center of pressure in relation to the centroid of the cross-section. The resultant

moment on the centroid is readily determined as listed in equation (5).

m YN = N(Xn..t, - Xcp) (5)
Where X..,, is based upon the location of the centroid, also referred to as xbar in later

sections. Note that equation (5) specifies my vice Myy, this is due to the fact that the

chord line in many airfoils does not pass through the centroid. To account for this line of

action, equation (6) solves for myA, which can then be summed with my,, to yield the

desired resultant torque for an individual column cross-section, myy. Once the resultant

moment, myy, has been obtained at each discrete column section, the resultant torsion

experienced at a cross-section can be determined. As previously depicted in the bending

moment discussion, beginning at the right wingtip and summing the column torsion values

yields the resultant torsion, Myy, experienced along the span.

m YA = (Z,1 .. mi - Zc)A (6)

Due to the format of the output of PMARC, the torsion solution is obtained for each

column center, vice the left face; recall that shear analysis was performed at each column's

inboard face. Figure 13 is the final result for total torsion applied to the center of each

column along the right semi-span of the wing. Since Figure 13 was derived from a

solution of equations (5) and (6), an interesting dilemma arises for the designer. Each of

26
these equations assumes that the centroid is already known. Therefore, Figure 13 cannot

actually be developed prior to completion of the next section involving solution of the

Total Torsion at a Cross Section, Myy

0
(20.000)
((40.000)
(60.000)

* (2ooo
(120,000)
(140.000) ' .
0 10 20 30 40 50 60
Swnm-,pan Locaion (IN)

Figure 13

bending stresses due to N and A. However, it is presented at this time to provide the

reader with a sense of relative magnitude in relation to the shear and bending diagrams

previously developed.

B. BENDING STRESS ANALYSIS

The four spar caps represented by the four areas in Figure 5 will be assumed to

initially carry the primary bending loads that give rise to Mxx and Mzz. Since bending

occurs in both directions, and none of the areas were assumed to be symmetrical, the

bending stress can be found from Allen and Haisler [Ref, l:pp, 166-170]. Assuming the

centroid is known and the size of each area is known, equation (7) serves to calculate the

normal stress, 7,,, carried at each spar cap.

27
 Application of equation (7), in conjunction with an allowable working stress can be

used to determine the location of the centroid and size of the four spar caps.

E Maz
= Aim~r -[ Mh M4 ]z (7)
Although optimization routines do exist in industry today that can quickly identify the

preerred location and distribution of the four areas [Ref. 8], observing the interaction of

the four areas in relation to an allowable -stressconstraint proves very beneficial in the

learning environment.

The operational envelope requirements set for this exercise included a +3.5 to -1.0 g

load factor boundary. Analysis of the upper surface of the wing reveals that at the

maximum load factor of +3.5 g's, the structure will experience a significant compressive

load. Based upon information that will be presented in the buckling analysis section, the

maximum allowable stress level for the upper surface is the material yield stress limit,

which represents the maximum buckling stress for the stringers. However the lower spar

cap allowable stress level for the material chosen was set at the ultimate tensile stress.

Buckling instability of the lower surface did not present the same challenge as the upper

surface because of the differences in design load requirements.

1. Area Determination

As mentioned above, an iterative solution performed by the student in design

proved very rewarding in understanding the demands placed upon various points within

the cross-section. Accomplishment of this task was performed with the aid of a

spreadsheet. By initializing all four areas to some arbitrary amount, and then beginning a

28
sequence of variations, the average student could arrive at an acceptable solution within

approximately ten minutes for a given flight condition. Appendix B provides a sample of

the spreadsheet used in this manner. Figure 14 is a review of key dimensions employed by

the spar cap area calculation spreadsheet.

Al Measured at the Root

-I X2 •

xc ,

Figure 14

The spreadsheet requires the input of Al through A4, the values of Mxx and Mzz,

and the geometry constraints of the wing cross-section at the spanwise location under

investigation. As the areas are input, xbar, zbar, Ixx, Izz and Izx are calculated. Having

obtained these values, the routine goes on to evaluate the stress at points 1 through 4,

using equation (7). While observing the total area and the stress levels for the four points,

the designer brings points I and 3 to their respective allowable stress levels, and minimizes

the total area by varying areas at points I through 4. The most productive method

discovered for minimizing the time involved, employed a rapid change of areas in sections

I and 3 until these sections are close to meeting the appropriate allowable stress levels. At

that point the rear sections are varied until their change begins to significantly alter the

stress carried in each respective forward member. At this point it becomes important to

29
start monitoring the total area to determine which changes in individual section areas leads

to a reduction in total area. The process is terminated when the solution with the smallest

total area is found.

This problem was worked on the basis of a prior material selection and the

assumption that construction of stiffeners, skin and spar caps would be done with the same

material. The method will also work for different material build-ups as long as a modulus

weighted moment of inertia approach is applied. The problem can also be worked from

the position of picking an allowable stress first and then determining what materials can

sustain the expected stress levels.

2. Determination of Allowable Stress

For those regions of the structure that are subjected to maximum loads that are

tensile in nature, it is of prime importance to ascertain what load condition and stress level

should be chosen to achieve the desired factor of safety. The following discussion was

applied to this project in determining the allowable stress level for maximum tensile loads

for any material where yield and ultimate strength test data were available.

Begin with a comparison of two-thirds of the ultimate tensile stress to the yield

stress value. The smalle of these two values will determine the eventual allowable stress

and limit load analysis conditions, assuming a factor of safety of 1.5. When two-thirds of

the ultimate stress is the lower value, analyze all loading problems at the ultimate load

(I.5*Limit Load). For this case set o = , However, if initially the yield stress

value proved to be the smaller value, the problem should still be analyzed at the ultimate

30
load. Although, in this situation the allowable stress level is reduced such that
3
Oallowable = Oyied. Recall that the limit load is the maximum load used for design

purposes that the structure is expected to encounter in its normal operational routine. The

ultimate load is derived simply from the concept of ensuring the existence of a margin of

safety in design calculations.

Since the upper surface of the wing was subjected to large compression loads and

stringer buckling was the design constraint, the material yield stress was chosen as the

allowable stress. If, after a buckling analysis was performed, the column length between

ribs was determined to be unacceptable (resultant weight increase), one option would be

to incrementally reduce the allowable stress below the yield value. This reduction in stress

carried will tend to increase the weight of the spar caps, forcing a strength to weight trade

study to compare spar cap increase in weight, to weight reduction due to increased rib

spacing (which results in fewer ribs).

3. Spar Placement Considerations

During the calculation of the stress levels at each corner point in the

cross-section of the root chord wingbox, the location of the front and rear spars was

assumed to be known. These parameters are not fixed at the outset of the design process,

but rather represent one additional consideration and potential optimization that will

require the designer's attention. Reviewing Figure 15 can provide a little insight on the

considerations at hand.

31
 Figure 15 depicts the location used for the analysis in this design. However, it is

important to understand the range of flexibility involved, and associated implications. In

his presentation on layout techniques, Niu [Ref 9:p. 254] presents his range of forward

spar location to be 12-17% of the chord. He also suggests the rear spar requires

positioning at 55-60% chord to accommodate installation of a 30% aileron and flap

section.

Thesis Project
Spar Layout
15% Chord

65% Chord

Figure 15

Given this range of values, where does the designer place the spar locations?

Often the aft spar is positioned based upon aerodynamic considerations like attachment

requirements to accomodate a flap or aileron system that is necessary for control of the

vehicle. For this project the rear wall of the wingbox was fixed at 65% chord. Again

reviewing Figure 15, as the forward spar moves aft, the overall height of the wingbox is

increased while the distance between spars is reduced. Locating the front spar then

becomes more of a structural optimization problem that strives to maximize 1,, I and I.

32
to give minimum total spar cap area. Also, moving the forward spar aft, generally tends to

reduce overall wingbox cross-sectional area. This serves to increase the shear flow

around the surface for a given torsional loading. For the beginning designer, the

recommendation is to make an initial comparison at the potential fore and aft locations of

each spar to evaluate the magnitude of the differences found. For a swept wing design,

torsion may prove to be the design driver as one proceeds out the semi-span. Appendix C

offers further insight into the options available in a study on various locations for both the

front and rear spar.

4. Straight Wing Approach

Once the initial effective area has been determined, allowing computation of the

three moments of inertia, it is possible to enter all required elements into equation (7) for

determination of the effective stress at a particular location. The only caution offered in

the use of equation (7) involves maintaining a consistent use of orientation. For analysis

of the wing root, the evaluation method chosen in this attempt involved determination of

all forces and moments applied to a positive face at the desired analysis location.

Figure 16 serves to illustrate the orientation chosen. The cut depicted is intended to occur

at an infinitesimal distance from the centerline of the wing.

Since all axes pass through the centroid for this portion of the analysis, it is

important to monitor the sign convention for both the distance measured from the neutral

axis and the allowable stress levels used during area determination. The sign orientatiorn

for positive x and z was based upon Figure 16. The sign convention employed for the

33
allowable stress assumed all compressive stresses to be negative and tensile stresses as

positive. Appendix B also lists the results of the effective stress determination for several

operating conditions.

Load Orientation

S
+Vx

Figure 16

5. Swept Wing Considerations

One of the key advantages in using PMARC for initial load determination was

that it allowed analysis of virtually any geometry in wing design. So the initial procedures

for obtaining the loads applied to the exterior of any structure operating subsonically, will

always remain the same. However, as Figure 17 illustrates, application of those loads will

dramatically depend upon wing geometry chosen. In fact, it is a function of both external

geometry (wingsweep and dihedral) as well as internal geometry (rib and spar orientation).

Figure 17 is taken from a broader discussion presented by Niu, where he depicts

the internal structural layout of a wide range of current commercial transport aircraft that

34
incorporate swept wing designs [Ref. 9:pp. 278-281]. Since the panel code output is

consistent with alignment shown in Figure 17-(a), one might think that particular layout

would prove quicker to analyze. However, structural analysis involves working with an

axis of bending and cross-sections perpendicular to it. The axis of bending will coincide

with the line of centroids along the span, which can be taken as a first approximation as

Potential Orientations
(a) Rib Aligned (b) Rib placement
with A/C Centerline perpendicular to
the Leading Edge

Figure 17

40% of the distance between spars. The axial forces from PMARC must be resolved into

a component transverse to the bending axis and along it as shown in Figure 18. The

transverse direction will be along the rib, and the spanwise component will be along the

stringer. The latter will add a tensile component to the bending stresses.

35
To accomplish this task the following methodology is proposed:

"* Resolve the loads obtained from PMARC into normal and axial components
oriented with the rib direction.

"• Modify equation (7) to account for the axial load component.

"* Update the spanwise position of areas one through four in stress calculations.

Figure 18 hopefully serves to illustrate transforming the applied load into meaningful

component requirements. This information provides the basis for modification of the

analysis routines presented in Appendix B. At this level of complexity in the analysis, the

amount of effort and time consumed rapidly begins to multiply. This should provide the

Axial Force, A
V1

Al18,A3

Rb locted a ~
center of Pmessue
A2 Afor Spanwise Station Y1

Figure 18

reader with an understanding of the requirement for more sophisticated techniques for

analysis as the structure begins to depart from conventional beam appearance.

36
C. SHEAR STRESS ANALYSIS
To allow for quick computation of the expected shear on the spar webs and wing

covers, a lumped mass approach, assuming constant shear flow between centers, was

employed. Figure 19 illustrates the combination of shear flow due to torsion, and shear

flow resulting from transverse loadings.

1. Shear Flow due to Torsion

Since the torque across the cross-section was solved in an earlier section as the

resultant moment Myy, the remaining solution requires minimal effort. Equation (8) forms

qm Al +ZAl q A
A A2 A

A4 AL4 ]A A4 -a,

Figure 19

the basis of a constant value of shear flow that is imposed on all faces between the lumped

mass centers. Note that equation (8) requires A to be the cross-section area located

within the region bordered by centers Al through A4.

q=M - (8)
2A
Again, consistent sign convention is essential to a correct solution. Unlike bending,

multiple sign errors throughout this analysis can result in potentially unrecognizable final

solution errors, that in turn can generate a structurally deficient design. The next chapter

37
will reveal the importance of accurate shear flow estimation, since q^ serves as the basis

for the buckling analysis.

2. Shear Flow due to Applied Loads

The middle segment of Figure 19 also has a straight forward, although longer

method of solution. For both single-cell, as the current illustration depicts, and multicell

analysis, a set of linear simultaneous equations must be solved. To obtain the appropriate

number of equations in a multicell problem, the additional constraint of all cells adhering

to the same angular rotation or deformation must also be applied. The starting point in

determining q0 is the relation between qj and q,+, as defined in equation (9).

qia= q,7+ vI + Q
Jx
q,+t~JzzL7cc-fz]Q VXJ= - VZIZX.2 ]QZ (9)
1=1=~~
1=, I 1=1= - (1=)2

Allen and Haisler provide a discussion on the derivation of equation (9) and

related multicell problems [Ref. I :pp. 205-228]. Applying equation (9) to the middle

drawing in Figure 19, qj, q2 and q3 can be related to the shear flow qo. All that remains for

the single-cell problem is the formulation of one last equation allowing solution of q0 . At

this point one has the option of analyzing a single face at a given cross-section, or a

differential element containing a combination of opposing faces. Figure 20 has been

provided as a simplified schematic of the options available. The solution involves either

equating the moments produced by the shear flows in the webs and skins to the equivalent

moments produced by the resultant shear forces at that cross-section; or summing the

moments to zero for equilibrium of the shear flows and shear forces acting on the

differential element. The results presented in Appendix B are based upon analyzing the

38
shear flow due to concentrated shear forces on the same (positive) face. In this situation

the moments are required to be equal.

The moments were considered about A3 since it provides no torque effect from shear

flows q2 and q3. Having solved for q0, equation (9) can again be applied, solving in

sequence shear flows q, through q3. These results are algebraically summed to obtain the

final shear flow in each spar web and wing cover (again, note the importance of consistent

Front Face or Rear Face

/+Vz +Vz

a 2 x3 A4
A4 I q2

Figure 20

orientation). Figure 21 is the final solution with proper orientation for the operating

condition picked thus far.

39
Sýhear Flow due to Shear Flow due to
Torsion Transverse Shear
311 lb/in 654

+ 36241 t10266

45536
Combined Shear Flow

39351i t9954

847

Figure 21

40
D. VARIATION IN OPERATIONAL CONDITIONS
The previous figures illustrated the shear flow developed due to the operating

conditions that exist at point I in Figure 22. Before attempting to size the skin thickness,

several other operating conditions warrant investigation to determine which condition

develops the largest shear flow within the entire flight envelope. The four points

illustrated in Figure 22 encompass flight at high and low angles of attack, in addition to

maximum positive and negative load factors. In developing the curves associated with the

maximum lift, a simple rule of thumb was employed to relate c,. of the 2-d airfoil section,

to c,. of the 3-d wing structure. Obtaining both the positive and negative values of c,..

Sample V-n Diagram

(Assumes Cimax df +1.35 and .0.765)
4

j2
3Cmm a I IV

6
(1)
(2)10 10 2;0 300 400 500
Velocity at Sea Level (lt.)

Figure 22

for a NACA 4418 airfoil from Abbot and Doenhoff [Ref 10:p. 492], 90% of those values

were then substituted as cL. into equation (1) to determine the corresponding load

factors. The data compendium (DATCOM) offers more extensive methods for estimating

41
the maximum lift of a wing operating in the subsonic through hypersonic flight regimes

(Ref. 11: Sect 4.1.3.4].

Appendix B provides all stress levels and shear flows for conditions I through IV.

For this particular design problem, condition I was the primary design constraint since all

other operating points at sea level conditions developed stress levels below those of

condition 1. If, in a different design problem, one or more flight conditions develop stress

levels that exceed the allowable limits after having sized the spar caps for a given

condition, a logical sizing sequence must be employed to ensure the structure is properly

sized to support flight at all locations. The following analysis routine is provided to aid in

that form of design problem:

" Begin at condition 1, note that if PMARC is used condition I is offset a small
amount from the maximum lift line to avoid estimation errors in the nonlinear
regime of the lift versus alpha curve.

" Having sized the structure, move to condition I. Maintain all spar cap sizes from
condition I as the minimum values for condition I1. If the allowable stress is
exceeded at any point in condition II, increase spar cap areas as required. Set these
new areas as the minimum spar cap sizes.

" Move to condition II; use the same procedure as that at condition II. Establish the
new minimum areas and move to condition IV. Having completed condition IV,
the structure is now assured to carry the applied loads throughout the expected
flight envelope.

E. SKIN SIZING
Based upon a review of conditions I through IV, condition IV was found to develop

the largest shear flow in the upper skin, where qA was 1238 lb/in between spar caps. This

value of shear flow was used to determine the minimum skin thickness required to carry

42
the applied load along the upper wing skin. If this value is substituted into equation (10)

and . is known, it is possible to solve for the minimum required skin thickness on

the basis of strength alone.

Tafiowable = q
1 (10)

Determining the value of Tao,,, is accomplished in the same manner as that employed to

obtain the allowable normal stress which was calculated earlier. To review this method,
first consider the values of'Twd and Ifthe proves to be the lower

value, then set the allowable stress to the ultimate shear stress of the material. However,

if the yield shear stress proved to be the lower number, then the allowable stress should
be set to j2 . In both of these cases, the load used for analysis should always be the

ultimate load using this criteria. For the calculations presented in Appendix B, material

properties for 2024-T4 extruded aluminum were taken from Popov [Ref 12:p. 554],

however, a better approach is to develop specific material properties from the

MILHANDBOOK-5F series. Using the information within Popov, the allowable shear

stress was taken to be 32,000 psi, which requires a minimum skin thickness of 0.0387 in.

Having solved for the skin thickness t, it is also a responsibility of the designer to

investigate any other constraints placed upon the vehicle by outside agencies or the final

customer. Consider some of the following requirements as set forth by the Navy. Table 3

lists the minimum skin thickness requirements for a variety of materials that may be

considered during the design process [Ref. 13 :p. 17]. From Table 3, the minimum gauge

requirement for alminum construction is 0.026 in, since this is less than that required due

43
to the loads carried, the minimum thickness remains at 0.0387 in. Had the material

considered been high strength, the minimum thickness may have been required to increase

to comply with minimum gauge requirements, this can prove to be a major factor in

considering a strength to weight study.

TABLE 3

Material Minimum Gauge (in)

Corrosion-resistant Steel 0.008
Alluminum Alloys (Exterior) 0.026
Superalloys
Titanium Alloys 0.016
0.015
Graphite / Epoxy Skins 0.020
Boron / Epoxy Skins 0.020

One last consideration in the manufacturing effort concerns the availability of

commercial stock material. For instance, if a roll of aluminum sheet is manufactured in a

standard gauge thickness of 0.028 inches, yet the minimum thickness required calls for

only 0.026 inches, what is required of the designer? To answer that question the design

team would probably meet with the material supplier to determine the cost effectiveness of

initiating a special production run, to provide a limited amount of material in the lighter

gauge. The alternative to a special production run is acceptance of the associated excess

material (and weight) to achieve a lower manufacturing cost. Table 4 is an excerpt from

Appendix D which lists some of the dimensions used in production of commericial stock.

44
Note that several standards of measurement still exist today, and different manufacturers

adopt one or more of these standards for use in their production runs.

In the analysis performed as part of this thesis, it was assumed to be cheaper to

acquire commercially available aluminum sheet. After a review of the load requirements

and the Navy's minimum thickness requirements for aluminum sheet, a minimum thickness

of 0.0387 in is used to enter Table 4. With the information available, it appears that the

smallest available gauge for use in the upper wing skin is on the order of 0.040 in, which

corresponds to a number 18 gauge sheet in the American Standard. Using this value as

the skin thickness t, and inserting it into equation (10) with the orginal shear flow, reveals

an effective working shear stress of 30,948 psi in the upper wing cover.

TABLE 4
Gauge No. American Std Washburn Std Birmingham Std U.S. Standard
18 0.040 0.047 0.049 0.050
19 0.035 0.041 0.042 0.043
20 0.032 0.034 0.035 0.037
21 0.028 0.031 0.032 0.034

45
 IV. BUCKLING CONSIDERATIONS

Having developed the initial estimate for spar cap sizes and approximated overall

shear flow in the upper wing cover, it is now possible to begin to estimate the stability

requirements of the upper surface. This process takes on two distinct apects. The first

phase involves determining the number and spacing of stiffeners to be employed. Once

that has been accomplished, the designer must then choose the shape and size of the

individual stringer elements. It is in this portion of the design that many interactions occur

during calculations that open the door to a wide range of trade-offs that the designer must

review. This diversity of possibilities in the design is perhaps one of the most challenging,

yet rewarding aspects of the process for the apprentice designer.

A. STIFFENER SPACING
Recall that during the bending analysis, the allowable normal stress level for th,-upper

surface was set at the material yield stress. One should also observe that the upper cover

deals with two distinct forms of primary buckling analysis, column buckling of the stiffener

and plate buckling for the skin itself. The plate buckling problem becomes the starting

point, where the operating stress in the skin is kept below the buckling stress of the skin

when treated as a flat plate. The methodology presented here is based upon an attempt to

46
preclude buckling throughout the wing structure in all flight conditions operated at or

below ultimate load conditions.

The working shear stress now becomes the defining buckling shear stress for the

wing skin. The dimensions of the plate shown in Figure 23 will be selected to prevent

shear buckling. The equation for shear buckling of flat plates is given in equation (11)

[Ref, 2:V. 172]. The desired output of equation (11) is the ratio of plate thickness to

width.

rc, 12(1-V2)](b (11)

'ince the critical buckling stress is known, plus the skin thickness, t, it is possible to solve

for plate width, b. The plate width is the distance between stringer attachment lines.

Referring to Figure 23, if the stringers are attached with a single row of closely spaced

rivets, then b would represent the distance between these rivet lines, which are depicted by

dashed lines along the span.

Figure 23

47
 Before one solves for b explicitly in equation (11), several key constants have to be

obtained. The first is the modulus of Elasticity, E. The results presented in Appendices B

and C were based upon 2024 aluminum possessing a uniform modulus of elasticity of
10.6x106 lb/sq-in for aluminum plate. The constant v is Poisson's ratio, and was taken to
be 0.33. The final constant, K., is known as the shear buckling constant for use with the

Euler buckling equation. The value of K, reflects the type of support the plate has at its

boundaries. It ranges from 5.35 for simply supported edges to 8.98 for four rigidly held

edges in plates that are rectangular in shape [Ref 2:p. 172]. Since ideal boundary

conditions are seldom the rule in actual structures, review of several design handbooks can

provide a range of values obtained from experience to use for K,. Niu provides several

illustrations of design charts for K, that depend upon how each side of the plate is

supported, the ratio of the plate length to width and the curvature of the surface of the

plate [Ref. 9:pp. 138-140].

For simplicity of calculations, the value chosen in this analysis was based upon the

discussion of Sechler and Dunn in their development of shear buckling for a flat plate.

They provide a figure that theoretically estimates this constant for two different end

supports. One curve is based upon simply supported sides, while the second employs

clamped edge support. Since the actual aircraft skin panel support structure falls

somewhere between these two locations, an average was taken for K,, assuming the length

of the plate (which will eventually determine the rib spacing) would be at least 5 times

longer than the width of the plate. For these parameters the value of K, was found to be

7.17. [Ref, 2:pp. 164-173]

48
 Determination of the total number of stringers required along the upper cover

necessitates the knowledge of the plate width between stringer attachment fines. Since the

locations of the leading edge spar and trailing edge spar have already been fixed, the

distance between them is also known. If the stringers are placed at equal distances

between these two spars, equation (12) can be used to solve for total number of required

stringers for the upper wing cover between the two spars. Note that the number of
n Total Width 1 (12)
b
individual plate elements exceeds the number of stiffeners by one if the fore and aft edges

are assumed to be bounded by the spar caps themselves. Solving for flight condition IV,

b=-1.42 in and n=62 total stringers at the wing root.

B. STIFFENER SIZING

This portion of the buckling analysis offers many possible variations to arrive at an

acceptable solution. Rather than provide the student with an optimal solution, it was again

determined to be more beneficial for the student to explore the effects of variations in a

wide range of parameters. The first obvious question to arise in sizing of a stiffener stems

from the geometry of the stringer itself. Niu is quick to point out the importance of the

design engineer understanding more than optimum strength or minimum strength. In his

description of the Y stringer, he indicates that it is a very strong support that can be easily

attached during manufacture. However this same device is expensive and provides an

inaccessible region on the inside skin surface that may become susceptible to corrosion

49
damage during the life of the aircraft. This region also precludes a visual inspection

without complete removal of the stringer. Another concern of the designer is the interface

of manufacturing processes involved in physically attaching a stringer. Questions to be

asked include, "Does this design hinder the shop's ability to rivet or attach jig and

alignment hardware?" Awareness of these factors at the earliest point in the process can

produce enormous savings over the life of a program. [Ref 9:pp. 141-142]

Keeping these ideas in mind, the Z section was chosen to demonstrate one method of

stringer sizing. The Z geometry does not place enormous computational requirements on

the designer, is readily attached and allows for visual inspection.

1. Initial Size Estimate

In an attempt to incorporate previous design experience and reduce the time for

solution, two important ratios were taken from Niu as minimum weight rules of thumb and

applied to the stringer cross-section build-up. The first, listed in equation (13), ratios the

total cross-sectional area of the stringer to the cross-section of the plate located between

two stringers.
A, =1.5(bl) (13)
The second relation in equation (14) is taken specifically for use with the Z cross-section

geometry and relates the skin thickness, t, to the thickness of the vertical web of the

stringer, t,.
1,= 1.051 (14)
Returning to one last manufacturing consideration, actual construction of the

stringer. One more cost study that may prove beneficial early in the process, is the choice

50
of stringer build-up. Possible options include commercial stock, limited production

extrusions or bending operations to produce the desired section. If it is determined to be

cost effective to bend a flat sheet into the appropriate Z section, one should observe that

the value taken for t, now becomes constant for all three segments of the Z. Turning to

Figure 24, it depicts one possible labeling of the required dimensions used in this analysis

to arrive at the final geometry.

r IW

tt

12

H_ _1-

Figure 24

Also observe the segment defined by 2We in Figure 24. The value of 2We is

taken from the concept of an effective width of skin acting in support of the attached

stringer to prevent its compression buckling. It should not be confused with the width of

the plate between stringers, which is greater than the effective width. In Figure 25, the

general pattern of stress variation can be seen as one moves in a chordwise manner along

the wing cover. As expected, the stress level rises in the vicinity of the stringers, which

51
must resist compression buckling. The stress carried then falls off to its low point midway

between stringers where only the skin acts to carry the resultant load.

Compressive Stress Distribution

Along the Cross-Section
2We b

mc;

I I
t qA

Figure 25
The effective width relates the stress carried at the supported edge, labeled ca,,
to the normal buckling stress of the plate section, referred to as the critical stress, ae,.
Since the upper wing cover structure was initially sized to ensure the maximum normal
stress level remained below the material yield stress, the value of a,. is taken to be ayw.
In the case of 2024 aluminum, the yield stress was assumed to be 44,000 psi. Using

another form of the Euler buckling equation, the critical buckling stress for a flat plate can
be used to determine s,,, as described in equation (15). Again, the values of K
-K E 2(b) (15)
1 2 ( 1 vW
-ci=c
will vary depending upon the type of edge supports the flat plate encounters. The values

of K, range from 4.0 for the simply supported case, to 7.5 for all sides rigidly clamped.

Again, taking an average of these values, K,=5.75 was used for the solution of equation

(15) in determing the critical stress of the plate sections between stringer attachments.

52
Substituting this value into equation (I5) results in a buckling stress of 24,181 psi for

flight condition IV in the upper skin plates along the wing root. It should be noted that

the largest value the critical stress can become is the material's normal yield stress.

[Ref. 2:pp. 164-169]

Returning to the relation of the critical stress of a plate section and the stress

carried at its edge supports, equation (16) defines this relationship in terms of the effective

width of the skin attached to one stringer. Sechler and Dunn provide an extensive

discussion on the development of equation (16) and determination of the period of

oscillation associated with the loading pattern depicted in Figure 25 [Ref. 2:pp. 203-234].

We = A(16)

Using these inputs, the effective width is readily obtained as 0.74 in or 1.48 in for

2We as depicted in Figure 24. This information is now used in another Euler buckling

problem that can determine the ratio of column required length to radius of gyration for

the section shown.

53
 2. Uniform Thickness Approach

If the stringer thickness is assumed to be uniform throughout (as in the case of a

section produced by bending up a flat sheet), a starting point in estimating the required

size is quickly obtained. Since the overall area of the stringer was determined via equation

(13) and the thickness in equation (14), it becomes possible to relate the total length of the

Z stringer sub-sections in the form of equation (17).

A, =(11 +t2+13)t, (17)
Turning to prior design experience, the relative dimensions of the legs that produce a

structure for optimum strength are specified in equations (18) and (19).
3 = 0.3 (12) (18)

3 = 0.7 (11) (19)

Once the length of each segment is known, it becomes possible to locate the

center of gravity for the entire stringer cross-section that is depicted in Figure 26. This

Effective Buckling Colunmn

Cross-Section

2We It.' " I'.

- to

Figure 26

composite stringer includes the skin section of 2We and all three stringer segments.

Following determination of the centroid, the next requirement is to calculate the moment

54
of inertia, 1. for the composite cross-section. It is the combination of effective skin width

and the attached stringer that will act in unison as an individual column subjected to

compressive buckling along the span (between the wing ribs). Recalling that the moment

of inertia is related to the area of the cross-section, as illustrated in equation (20), it

becomes possible to solve for the radius of gyration, r.

A w.1t (20) S=

All that remains for this first approximation is determination of the required

column length or rib spacing. For buckling beyond the proportional limit, The tangent

modulus is inserted into the Euler column curve, and it is possible to develop equation

(21), which can be used to solve for the column length, L.

(Fcr+ = MCycrM
X2(21)
(JF

The constant m in equation (21) represents the inverse of the strain hardening exponent

for the given material. In this analysis Of 2024-T4 aluminum, m was 5. Similarly, K also

represents a material property, it is known as the strength coefficient. It should be noted

that K is generally provided in units of ksi; this requires consistent matching of units

among the values substituted into equation (21). Throughout this analysis a value of

K= 117 ksi was used.

3. Stringer Cross-Section Stability

Although the results obtained above were based upon approximate size

relationships obtained from previous design experience, it is still important to verify the

55
integrity of the stringer itsel Fsubjected to ultimate loading. The easiest means of analysis

is accomplished by treating each individual segment of the Z stringer as a plate element

subjected to compressive loading. Again returning to the Euler buckling relation, this time

in the form of compressive plate loading, results in equation (22).

KCX 2 E ( , ')2
12(,=
12- -v2) (22)
The value of K,, varies depending upon the location of the indidual stringer element.

Referring to Figure 24, note that the upper flange and lower flange pieces (described as

parts I and 3), are assumed to act as a simply supported plate on three sides, with the

fourth side free. Using the plate normal stress buckling results provided by Sechler and

Dunn [Ref, 2:p. 168], reveals K,=K 3=0.50. Similarly, the web section (part 2), is

considered to be a simply supported on all four sides; thus K2=4.00.

Now substituting these values into equation (22) allows calculation of the

expected buckling stress for each individual plate element composing the stringer. For
each element, the result is then compared to the yield stress. The lower of owu and ac, is

then referred to as O>ro.bi., which is used to find the crushing stress, cr, for the entire

stringer. To determine this crushing strength begin by finding the allowable effective load

in each element given by equation (23).

Pi , = Oia, (23)
Once the allowable load in each element is determined, the crushing stress is determined

via equation (24).

(C = (PI +P2 +P3) (24)

(AI +A 2 +A 3 )
The importance of this check is to ensure that the crushing strength of the stringer is in

56
fact greater than the material yield stress. If the value comes up short, then the designer is

forced to change the values of A, through A, until that goal is met. Recall that if the

manufacturing intent is to construct the stringer from bending a single sheet of material,

this places a restriction on area changes due to length changes only. However, if custom

extrusions are being considered, then both length and thickness of the individual

components are available for customizing to meet design strength requirements.

Considering just these few variables that are open to design selection at only one operating

condition, begins to portray the expansive nature of optimization involved with an

aircraft's structural design life as thousands of variables surface for use in trade studies.

57
 V. STRENGTH TO WEIGHT

This chapter offers the beginning designer some ideas and options of trade studies

that can be used to optimize the previous design. These ideas are but a few of the many

available. The distinction between this academic project, and actual manufacture of an

aircraft, arises from the need to develop and understand principles on a small scale prior to

their application in the field. The topics will be presented in the order of bending stress

considerations, followed by instability constraints. However, the order of these studies

also occur in a chronological sequence that supports the current sequence of teaching

within the department.

A. BENDING STRENGTH TRADE STUDY

1. Material Selection

Since each material has a unique set of ultimate and tensile load properties, the
value of cn.u.bl. varies by a large amount over different types of conventional metals.

Referring to equation (7) and inserting different values for the allowable stress will

generate significant variations in the resultant areas that exist in the four spar caps. For

each case considered, every material has a unique density associated with that material.

For ease of comparison, consider a wing section at any given location along the span and

assume it to be of unit length in the y direction. For instance, in this problem consider a

chordwise segment that includes the wing from y=0.00 out to y=1.00.

58
 Next assume the cross-section areas of the four spar caps, (AI - A4), remain

uniform throughout the entire section (y=0.00 : y=1.00). Consider the case of

2024 aluminum alloy, assumed to weigh 0.100 lb / cu-in. For the problem solved in

Chapter III, the total weight of the spar caps for a unit depth at the wing root is indicated

by equation (25).
W,,p = (AlI +A2 +A3 +A4)p *ag~W,,,, (25)
Note that in equation (25), the density is assumed to be in units of mass / cu-ft. When the

density is specified in that fashion, an adjustment for dimensions must be included to arrive

at the correct units for total weight per unit span. For the density value listed above, the

acceleration due to gravity has already been included. Table 5 is a quick summary of

representative values for a few common engineering materials.

TABLE 5

Material Weight / Volume Ultimate Tensile Yield 36

Strength
(lb/in3) Strength
42 (ksi) (ksi)
2024-T4 Aluminum 0.098
0.100 60 44
Aluminum
6061-T6
5Cr-Mo-V Steel 0.281 240 200
17-7PH Stainless 0.276 170 140
Titanium 0.162 115 110
Inconnel 718 0.297 180 150

One should also notice the lack of two additional columns in Table 5, these

include the cost per pound to acquire the material and a similar reference of cost per

finished foot. Consider the case of titanium, for certain applications, the tooling required

to shape a particular component may prove unacceptable in the overall cost constraint of

59
the project for a limited production run. The average student is unlikely to possess such

detailed cost data, but should understand its impact when progressing to an industrial

design team or contract review team.

2. Forward Spar Location

Earlier in chapter III, it was pointed out that the front spar location would

generally occur between 12 - 117% of the chord. Since there is an available range to

anchor the forward spar, this offers another opportunity to evaluate the effect of changing

the spar's location. A series of spar locations might be investigated for a given operating

condition, aircraft configuration and material selection to determine the optimum location

of the forward spar. Again the optimum point will be identified by the lowest total

cross-sectional area required of the spar caps. An example of this type of investigation has

been provided in Appendix C.

B. INSTABILITY IMPLICATIONS
The shear flow in the skin, and thus minimum required skin thickness, will change

similar to the change in spar cap size given a variation in material choices. However, the

shear flow problem offers an additional consideration not associated with bending. The

earlier analysis finished with solving for the skin thickness that sustained the ultimate

expected load and complied with Navy design specifications. Now, the designer needs to

consider the effects of increasing the skin thickness beyond the required minimum. As

skin thickness rises, so does the load carried by the stiffeners. This increase tends to

reduce the column length in response to increased buckling potential. However, as the

60
buckling column lengths are reduced (implying reduced rib-spacing, thus more ribs

required along the span), the total number of stiffeners required is also reduced.

Aside from changing the skin thickness directly, rib spacing can also be affected by

the value specified for the upper surface allowable stress design level. Additionally, for

any given thickness chosen, if the stringer crushing strength proves to be unsatisfactory,

the areas of the individual stringer elements can be varied. All of these changes impact the

ratio of skin thickness to rib spacing. As both of these quantities change, so does the

overall weight of the wing. For educational purposes it is recommended that at least one

trade be conducted by the students and investigated over a reasonable range of parameters

to introduce the concepts and gain an understanding of the extent of possibilities involved

over the entire flight envelope of an aircraft.

C. ADVANCED MATERIAL CONSIDERATIONS

As a student's understanding and exposure to structural considerations grows, one

more consideration that weighs heavily in all aviation designs remains for review. The

design engineer must consider the effects of cyclic loading. This arena may force the

complete exclusion of various materials for failing to meet minimum required life cycles.

As a student completes his or her study in the area of fracture and fatigue, he or she can

then analyze effectively the likelihood of a given design satisfying the Navy's expected life

cycle requirements. This stage of the design could prove to be the most challenging in

attempting to improve the strength to weight ratio of the structure while subjected to a

61
wide range of loads over a long lifetime. Structural fatigue has been a central issue in all

Naval Aviation programs for the past 30 years.

62
 VI. SUMMARY

A. FUTURE AREAS OF STUDY

1. Aerodynamic Investigations

During the validation and testing of PMARC, all of the work was completed

without the use of the boundary layer routines that are built into the analysis portion. This

was done to prevent overwhelming a student in his or her first exposure to such a complex

code. The boundary layer routine may prove however to offer an interesting and effective

means of determining the approximate cL. based upon a reversal in the boundary layer

shear stresses. A separate research project in itself would encompass a validation of the

boundary layer routine and eventual incorporation into the user's guide that was developed

as part of this thesis project.

Since most high lift devices are employed well below Mach 0.5, PMARC remains

a good solution for those unusual design problems. However, the supersonic, transonic

and high subsonic regimes cannot be addressed currently through the employment of

PMARC. These regions comprise a major portion of the operational envelopes of today's

tactical jet aircraft. To support the structures design of wing sections operating at high

speeds, with all high lift devices stowed, a simplified CFD routine could be developed for

student use. Such an undertaking would require earlier introduction of CFD methods into

the current syllabus, a grid generation system that is readily adaptable to geometric

iterations, and analysis routines that can respond to large variations in freestream

63
conditions. Having met all of those conditions, the output needs to be transformed into a

usable format for structural analysis, in a fashion similar to that presented in this thesis, or

like the pressure distribution analysis described in the next section. Given the magnitude

of that challenge, the distinction between educational and industrial goals needs to be

addressed at length prior to undertaking such a project.

2. Advanced structural Analysis

Several areas of improvement were identified throughout this project and

presented for consideration. The first point introduces the incorporation of Finite Element

Analysis (FEA) into the advanced aircraft structures course. This is the first introduction

to the subject for most students in the aerospace discipline at NPS. A logical building

block on the original design project would be verification and comparison of results using

the layout, sizing and material choices from the AA2021 class project.

If the initial class project were to be modified to include design and analysis of a

swept wing, as depicted in Figure 27, the wing could be analyzed over the period of two

quarters. The first quarter would be done using the methods presented here with an

adjustment for loading based upon the leading edge sweep. The root section would

require the use of FEA. The root section analysis would also necessitate the incorporation

of the results of the straight wing approach applied at the boundary of the two regions.

This project probably represents both the most extensive of follow-on suggestions, and the

most beneficial to future design students.

64
 If fewer fundamental concepts are presented in future instruction to

accommodate increased FEA skills, there are two investigation areas to aid in expanded

use of FEA. The first is oriented towards the reduction of time consumed to generate a

finite element model. Since PMARC is already used to determine a wireframe model,

incorporate the means to port that information into a FEA routine as a part of an input

file. PMARC is written in FORTRAN 77; thus, only a minor modification should be

required within the source code to generate an additional output data file. This project

would only be available to students of U.S. citizenship however, due to restrictions on

accessing the source code.

Finite Element
Analysis

•- Straight Wing
Approach

Minimal Load Carriage

(Considered in Detail Design)

Figure 27

The second project entails the boundary conditions that are solved under FEA.

As mentioned earlier, PMARC can generate output on the basis of pressure distribution

(cr) or force and moment coefficients. Since most FEA routines can solve an external

65
pressure problem, a modification to the source code to provide yet another output data

file, could be used to input the boundary conditions to a particular FEA application. This

project would parallel the previous one in development, with only slight differences

resulting from the requirement of converting cp to actual pressure.

B. CONCLUSION
The computer code developed to produce the outputs displayed in Appendix B

represents a significant step forward for the aircraft design course offered to NPS

students. It offers a means to reduce required computational time without sacrificing the

importance of careful consideration of the impact of engineering decisions in establishing a

baseline design. Hopefully the methods presented in the body of this thesis in conjunction

with the sample calculations listed in Appendix C offer insight to the beginning design

student that can build one's confidence level in that first design undertaking.

The intent for the software development on a spreadsheet format was to reinforce the

student's understanding in a guided environment. As each portion of the series of

spreadsheets becomes relevant in a student's classroom discussion, exact portions are

worked out in the classroom. Ideally the project will support about four hours of in-class

instruction (for all core structures courses), to allow the student the opportunity to derive

the underlying equation used in every calculation found throughout the worksheets.

Additionally, the classroom exercise should introduce the concept of working in

engineering teams to achieve a final result in a reasonable amount of time. By constraining

the project to an in-class basis, the instructor can ensure participation by all students and

66
provide valuable insight on stumbling blocks as they arise among individuals. This

intermediate approach to design, should serve students extremely well as they prepare for

eventual placement in the most demanding acquisition jobs the Navy can offer, both today

and on the horizon ahead.

67
 APPENDIX A

A. WAKE INITIALIZATION
The panel code used for this analysis, PMARC, allows for a wide range of

computations. One of its most complex features includes the analysis of flow past an

oscillating airfoil. To accomplish this task, the code requires the assembly and analysis of

a wake. Unfortunately, that type of computation generally demands extensive CPU time

on most computers. Recognizing that handicap, the creator of PMARC wanted to allow

experienced users of the code several means by which they could reduce the overall run

time during program execution, depending upon their individual requirements. As a result

the user has the option to:

" Specify the number of timesteps over which the wake, is calculated. This assumes
analysis begins at time t=O, where the body under investigation is about to enter
the freestream flow.

" Specify the length of the time increment used for completion of one time step.

" Allow PMARC to initialize the trailing wake and determine its development in
direction and magnitude.

" Override the code and allow the user to specify the wake's initial size and direction
upon startup. This option is designed to reduce overall run time if the user is
simply trying to analyze a complex flow geometry at a steady-state condition.

Because of the availability of these two features, and the intent of developing several

templates that would satisfy a wide range of student applications, a determination of the

optimum settings for the student design requirement had to be accomplished. To review

68
the effects of varying the number of timesteps, a test matrix was conducted on a consistent

wing geometry to observe when the outputs converged within a reasonable tolerance to a

steady-state value. In reviewing Figure 28, the steady-state solution was found to occur

to within four decimal place agreement following 10 timesteps. This test was performed

using the geometry described as Planform 3 in Reference 5. The operating conditions for

the test were the defaults listed for Planform 3 and AOA=0.00 degrees.

Variation in Number of Timesteps.

Convergoe to a Steady-State Solution
0.3
0.28
0.26
0.24
0.22
0.2
0.180
0 0.2 0.4 0.6 0.8 1
Location on the Right Wing (%Semi-Span)
I fTmlep 5m 15 ~M
slaps 10 TkmuplSo m I
The solid le repre•be• the *ietconvergence kiowng 10 Utimelp

Figure 28

Turning to the question of which initialization method to employ, the user

specification of wake attachment, or allowing PMARC to initialize the wake, prompted

the development of a second test scenario. The first concern in this test questioned the

existance of reasonable agreement between a user defined wake, and PMARC initiation of

the wake. Since the wake's departing orientation is generally quite complex, it will not

69
immediately adopt the direction of the freestream, which can lead to errors in the analysis.

Following that result came the question of the existance of any time savings afforded to

the typical NPS student using the various initiation options. The final review would then

focus on the accuracy of the results. Figure 29 is useful to demonstrate the first point.

Using the same airfoil from the earlier test, the two methods of wake generation were

compared at AOA=O.00 and AOA=8.00 degrees.

Comparison of Wake Generation Methods

1
0.8 2 •

0.6

0.4

S 0.2 0.4 0.6 0.8 1

Locsan an Uh RI, Win (%Swnl-.spin)

Figure 29

Reasonable agreement was obtained between the two methods, therefore a second

test was conducted to determine the time savings involved by specif~ying the wake's initial

direction and magnitude. Figure 30Ois a summary of those results. At first inspection of

Figures 29 and 30, one might assume a distinct advantage exists in user specification of

the wake initial positioning. However, it should also be noted that the code rendered

70
good convergence for the problem after only 10 timesteps. At that iteration level, the time

savings offered by the user specification mode is not considerable. An additional point

involves how much time is required to actually develop the required information that is

specified in the input data file to accomplish user initialization. From that aspect, coupled

with the intent to develop templates to be available for general use, the determination was

made to allow PMARC to specify the initial wake orientation and magnitude. The penalty

of additional run time does not outweigh the advantage of significant potential errors that

may be induced due to user specification of the wake.

Program Execution Time Comparison

12

10
8
6
4

2
0 10 : 20es l
NsMXr d Wke Tumftp

Figure 30

B. ANALYTICAL VALIDATION
Having established the desired input format, the next step in the process involved

checking the output against other methods of calculation. The verification test first

attempted to match the value of total lift for a wing against a prediction based upon a

71
conversion technique described in Nicholai. Knowing the characteristics of a

two-dimensional airfoil (infinite span), Nicholai offers a means to convert those

characteristics into a finite, three-dimensional wing. Equations (26) and (27) were

employed to accomplish this conversion of data. In these equations the constant C I was
assumed to be 0.6, while the value for P3 was taken as 0.866. Note that j3 is a correction

for compressibility which is not considered within PMARC. Equation (27) is actually the

first to be solved since it determines the lift curve slope. [Ref. 3:pp. 2-8:2-9]
CL = [CL 6.]o(a-aohf) + CI(x -0f•) 2 (26)
CL. = 21/AR (27)

2+ 14+AR22( +tan2A )

The results of this comparison are depicted in Figure 31. Since the approximation

accounts for compressibility, it shows a slightly larger value of lift for any given AOA at

the operating condition of 0.5 Mach.

Having attained a good correlation between total lift and AOA, the next task was to

look at the lift distribution the code predicted along the span at a given AOA. Planform 3

was again chosen as the baseline configuration for the test. The analytic model chosen for

the reference was based upon a modification of Schrenk's Lift Approximation as described

by Raymer. The process begins by first solving for the average between a predicted chord

72
length determined via equation (28) and the actual (or trapezoidal) chord length at various

stations along the span. The station points chosen were increments of 10% of the

semi-span.

C(y) =LS) (28)

The value of S is obtained from equation (29), note all other parameters are taken from

the characteristics of Planform 3.

S -C,(l +X) (29)

2
The trapezoidal chord length, C(y), at a particular station location, y, can be found from

equation (30).
C(Y)=C,[- 3 -L iX)] (30)

Carmam n of PMARC o04 wth

3-0 Appraodank frm MNem
1.2

S0.4-

1 -0.20
.
0%.O10
-0050 1 015 0. 0.25
AOA (rajr)

Figure 31

Once the elliptical chord length and trapezoidal chord length have been determined,

Schrenk's Approximation states that the distribution of total lift along the span is

proportional to the average of these two chord lengths. If this average chord length at a

73
given station is plotted against the station value, the total area under the resultant curve

must sum to be equal to the total lift. [Ref 4 :pp. 342-343]

To accommodate a comparison between PMARC and Shrenk's Approximation, a

somewhat different curve was developed using the same principles. Once the average

chord values were determined for the approximation, they were normalized by dividing

each individual value by the root chord length. Since this approximation was based upon

lift, vice lift coefficient, the output of PMARC also required modification. Taking the

local coefficient of lift for each spanwise station, that value would have to be multiplied by

the dynamic pressure (q) and the area of that spanwise station. Having obtained the lift at

each station, all values were then normalized by dividing the lift at the first output location

(7.8% of the semi-span) of PMARC into all station lift values.

Since the lift distributions for both methods were based upon a location described by

percentage of semi-span, it was then possible to overlay the two results for accurate

comparison. These results are depicted in Figure 32. For reference, Table 6 provides a

brief synopsis of the characteristics asiociated with Planform 3.

TABLE 6

-b 36.7 ft
c, 20.0 ft
-b 0.1445
S,,f 420 ft2
ALE 25 deg

74
Spanwise Load Distribution Comparison
1.2
0 -
00.2

t.0.02 04 062. .
0.4- --
Location on the Right Wing (% Semi-Span)

Schrenk's Approx PMARC

Based upon Planform 3

Figure 32

75
C. COMPUTER-BASED VERIFICATION
Having attained reasonable results from generalized approximations, the next step

entailed comparing the output of PMARC with an independent panel code. The choice

for this comparison was a panel code named SUB, which was developed by NASA

Langley. The test involved nearly identical geometry and freestreamn conditions.

However, two small differences did exist between the two codes. First, SUB did not

permit easy identification of the wing's cross-section, as a result the airfoil considered in

SUB was a flat plate. The cross-section used for PMARC was a NACA 0012 airfoil. The

second difference was due to the selection of independent variable used by each code in

their analysis routines. In SUB, the user specifies the desired total output cL, whereas

PMARC evaluates the input on the basis of operating angle of attack. Figure 33 depicts

the actual wing chosen for this portion of validation testing.

Test Case Wing Design

14 ft y

2 ftE
+x

Figure 33

The results of those tests are illustrated in Figure 34. The runs were compared along the

right wing for two different choices of AOA. Even with the subtle differences in the

codes, the results demonstrated good correlation.

76
 Comparison of SUB and PMARC
0.6
10.5
0.4
S0.3

0.1
0
0 20 40 60 80 100
RIght Wing Location (% Semi-Span)

PMARC SUB PMARC SUB1

2 dg AOA 2deg AOA 6deg AOA 6dfgADA
•--.- -- • B

Figure 34

D. EXPERIMENTAL DATA COMPARISON

The last validation study included a check against actual wind tunnel results for a

wing section that was built with a NACA 6 Digit airfoil cross-section. Data for the study

were obtained from Bertin and Smith in their discussion of panel code methods

[Ref 6:pp. 253-256]. The test cross-section was a NACA 65-210 forming a 15 ft span

wing similar in shape to Planform I used throughout the body of this thesis. Comparison

of the wind tunnel results and PMARC output is displayed in Figure 35. The effects of

non-linearity due to viscous effects can be observed at the higher values of AOA. With

the conclusion of this test, PMARC was determined to be a good introductory tool for use

by design students at NPS, that could also be readily validated by available references.

77
 Wind Tunnel Comparison
1.4
1.2 -
0Q
10

A 0.8

S0.6

S0.4

0.2

0
(0.2) 0
(4) 0 4 8 12
AOA (degrees)
Experimental PMARC Prandtl's Lifting-
Results Solution Line Solution
0

Figure 35

78
 APPENDIX B

This appendix provides the output of the spreadsheet routines developed to support

design students faced with the task of repetitive calculations based upon the principles

presented in this thesis. The results provided are for flight conditions I, II, III and IV,

which were previously described within chapter III of this thesis. These values are based

upon the analysis of an aircraft wing geometry represented by Planform I throughout the

thesis. The routines used to determine normal and axial loadings allow the inclusion of

planform drag through the specification of c,,.. The values obtained at all four flight

conditions used the airfoil section data at the appropriate velocity and Reynold's number

from Abbot and von Doenhoff[Ref. 10:p. 493]. Anderson provides an example of using

the section drag characteristics to solve for the expected lift and drag associated with a

finite wing [Ref 14:pp. 221-222].

Each page represents a complete output from individual worksheets that have been

linked together in a 3-D spreadsheet format. The code was developed within

LOTUS® 1-2-3 release 4.0 for Windows,. The code runs without delay on a 386-based

machine or better that employs a math co-processor and at least 8 Mb of RAM. For those

users with a color VGA or better monitor, the input cells have been shaded to further

reduce the time required to conduct a preliminary design study. Those inputs dealing with

a wing's physical geometry are indicated in light blue. Yellow cells pertain to material

properties, while purple is used for those items that should always be updated with any

79
change to the operating conditions. Due to the reproduction limitations of this thesis,

within Appendix B, all input cells are indicated by a grey background.

80
 O~-'8 88
aoo 4)ClV Ii
"a 0 VL)t -a D m0 wW
*~~~ - ~ -6- -q -i
0

o LU

jo -
.. .... .... .

o 1 6

I_
CLC

0* i
S .al-
 * __acaae
q* q

mme- e4 Uu

I
q w8000

s; ;a;a m

.3C -I d
A
q *

,apuioou. -

g
31 I1ICO0C

d -

eOCeO

I -

I 88888
ii q

11 0
 0 coo

NPo

~~a oo~a
0

000 ao co

C4 a)4 0

, ca

Cs OW 6 C; 46 4

I I I

- - 3
 aIO
0 oo

.~ o0iooo

ac a

a0 aw ace ac

1 800 a

"~ ~~~~.
• . . ",. ' v-; C4¢

•H ~~r,, ~ ~ ivlr coo. CM,,9

I • .

11 84
 CD~
49 x

V)

C44

x Ni

on I
XN
SOQ
11

VC

cmv

if,

a Nm

ca

oc
C CL

s5
 -,- C-

Ik t

I
0999999

a i C4ftN o
cd00Iw
Fý
t

0 V aq
Vý I
C4)

9 0Iic ?C a QW C1

C? CýCý 9

Igi.v v
q19
N C"cc

I,-m ,a
a O-

k ~3 flNN -Q .. r 86
 U)
U:l~- II%

a-a
CV

-a.(a

+ +

0 +

C4 C
LA. 'V

Oi

+ 87
 (1) Go

LU

0 E

+I i

4L
II ý

E 0 D

I li

E g*

COC
w co 0

08
 C a ~) 0

C UN
tI 9

CNN
O>
C

0l c InI
'd6- c

08
 cq 00 I

jn ... L...

CiC

-r V r P. M

cl~ ~ ~~f .) C.........

...
.- ...
...

......
..-
.-..
aa I is 8 8 1
9~ o~~

evP l l

(J 0qn 0 W.. In

900
 >j
+

0 0 @2 me

lo -D - I

In N~

Sam=e

Jai Iticco
R8 a8 I

vg~o V V

91- -
 +

0 o

0 0 4=e

o 0a O

90 080 a
Cal

116 co

Li I
ai l W:'r': --
II
BO -0
Oq'naaO n

S~xsg ga

wii
IIgrit
92
 U
x Uv

C -i

0.4

C.'

*14
t,~

C'4-

ii U
CL
"

93
 Is~------

U ~~
S IiI
~

94£
 Go r- c0-

ICI
CMC

go
-

LL= 4

22! *4

I- C C 4 ý C4
+ C4 1- N N -

C.
)~ a a a,~

95
 5.~

w
> E

UL

II I

uJl

0U

=1 IIE0

r.m co

ii: 96
 Icu 4-4

oa 0 acc

0 0

N NJ
U) 0)

CL ~

~ N '97
 ga

....
......
o Aw- -

Aoi
II too Mn
CRr----L!

ApoEL
Cý~~ 0 IN' 0

ou~ -- _9-
 +J
£ I ~

+ m

-•o ... oom

** *1111+°-
g "+ ,

ii!
I 'a!
a m

, ,. -+
-- --- , a s::•.+m
r+€,,t
s++_-

.+
.,
...
. U+._.

99
 a e31

a aco

a a a al

0 0 a aaI

accooamaa

*fla 0 00 amq

Rfea aa a o

0o
A

It d

--- m
P. s8s6c - -

ac o

100
 0

xgx
C-4

Ct <

r- -

V.V

In cn~
NiN
NI Z

cv,

~N

C,,

0 0.

101
 ,lG• cl li,, i•l - C.",.•

'II C.0€!m•••i~
Ui

1 .. .......

...... V
C4 00 N N N

I !i i y- ..-

Il_ '<"

.- , ,., , .. ....;.-..
........
.,
...
>

102
 II II

r.CV CU 0)

CC

c= CCU c

a. C) 80 9
Cc) C% 0)O
D

I Ii

I 103
 Ile

00

EE

IIi

•n co ._
9 CI-

ii '10"

110.)4
 .70 CDC

C CLd

coo a N

U)

CD CD N~

00105
 co Ia 0

IM

r- Cq CD M

*", * * 0

CO)~
IIi

& F1
JU0 cm0 W P 4V et

10
 cocoae

I eeem.

co o S eceemel

- co o I 11 1 o oo f

cocoa

coc Aba

~ e e c @@mil
i A l@og
:~E1ee
i

It! II107
 Ncoo

00=

000 a0

9t 0000=a

OMMMU spo- u

in V~I

SIC;!

sI~. -- coc

108
 A40-

X6 N
R.

S CM

*10
 1FIE

!lj ,i ~ -j!
Z.
~
S", •', 9 Cd

I E-i

II %II g3gqgR2R8

a ll •llllam
tem.••

110
 II II
OF
da

lie
r o

II _r
1
IL11L 27U9

c
o

C, M
 04 ;
I t

CLC

III

US m

112
 ji -

NN

a-C 06U

NO 0

C4 to 0-
a aD 0 N C4

~~113
 APPENDIX C

This appendix was developed to demonstrate the process of design variation for

optimization, more commonly known as a trade-study. This appendix will consider two

specific cases that support study within the current core structures curriculum at NPS.

The first study is intended for use within the introductory structures course, while the

second study is aimed at the advanced structures student.

A. LOCATING THE FORWARD SPAR

Earlier in Chapter III, it was presented that the forward spar location will generally

fall between the 12% and 17% chord position, assuming a conventional two spar

configuration. Since this range of decisions is available to the designer, he or she must

determine the final location. Since the average introductory level engineer lacks

considerable design experience, a reasonable and quantitative process must be devised to

lead one to that final decision.

One recommendation is to begin first by reviewing what parameters have been used

as inputs in determing the strength of the wing box. Considering the spar cap area sizing

spreadsheet, one observes that several inputs are affected by changing the location of the

forward spar. These inputs include X1, X2, Z 1 and Zc. Through the use of a scale

drawing of the wing's cross-section, it is possible to locate the inputs required for a wide

range of spar locations.

114
 However, before entering a long trial and error process of arbitrarily entering a series

of spar locations into the spreadsheet, a short review of anticipated outputs can prove very

beneficial in reducing computation time and increasing one's understanding of the

principles at work. One or more specific outputs should be identified as being uniquely

related to specific inputs. For this problem, the first obvious output to consider is total

area of the F3ar caps. Since maximum strength in comparison to structural weight is

sought, the smaller the total area at a fixed allowable stress, the greater the overall

strength-to-weight ratio. However, moving the structure's centroid may also affect the

shear flow due to torsion around the body. As shear flow increases, so does skin

thickness, which serves to increase total weight. Table 7 is a summary of data for moving

the forward spar within the range of 12%-17% of chord.

TABLE 7
Spar Location Total Area (in2) Shear Flow, qA (lb/in)
12% 152 998
13% 150 1,007
14% 148 1,015
15% 147 965
16% 142 1,018
17% 142 960

As observed in Table 7, moving he spar aft to 17% reduces the total area without a

major fluctuation in shear flow in the upper wing cover. Returning to the discussion on

anticipation of results. Consider the NACA 4418 airfoil used in the wing cross section.

For the 12%-17% chord range, as one moves aft a sizeable increase in the wingbox height

115
occurs. This implies that zbar is also increasing in the calculation of the bending stresses.

Yet, xbar is decreasing at the same time, so why the inconsistancy in results? The answer

lies in the sensitivity to change of the two parameters. The chord change from 12% to

13% produces a relative increase in zbar compared to the height of the forward spar that is

much greater then the reduction in xbar compared to the overall length of the wingbox.

This revelation should lead the designer to two additional considerations. The first

issue that should come to mind is the quesion of consistency. Does this trend in improved

structural performance hold for moving the forward spar aft for all airfoils? Hopefully, the

beginning student would respond no, since the rate of vertical change to horizontal change

is a function unique to every airfoil design. Second, what other assumptions were made in

the development of Table 7? Recall, the rear spar was fixed at 65% chord. That fact

should warrant at least on investigation of the points in Table 7 at a different rear spar

location. A good place to start might be the forward limit of travel for the rear spar,

which is generally taken to be 50 % of the chord.

B. RIB SPACING CONSIDERATIONS

In Chapter IV of this thesis, a method was developed based upon previous design

experience to optimize a stringer size and location for material strength. When the issue

of total weight enters that process, often very different design decisions will result that will

change one's recommendations rather drastically. Recall that as the rib spacing is

increased, the stringer cross-sectional area must also increase to preclude column buckling

of the stiffener. Also an increase in the wing skin thickness reduces the effective working

116
shear flow carried in the skin. As the shear flow drops, the distance between stringers

increases. This distance increase will imply a reduction in the total number of stringers,

and hence lower total stringer weight.

These points should lead the designer to consider the effect of increasing skin

thickness on total weight of the wing. Does an optimum condition exist that will carry the

applied loads for lowest total wing weight? Additionally, what input parameters must be

considered in such a study? These types of questions are fundamental for efficient and

effective work in the field of engineering design.

To determine the input parameters, consider those variables in the current structure

that add weight to the wing. The components considered thus far include the spar caps,

the wing skin and the ribs. Knowing the density of the material and the dimensions of

each structure, it is possible to estimate the weight of each of these components. In the

case of the spar caps, the cross-section area times the length of the semi-span yields the

volume. Yet this volume does not change when the skin thickness is varied, therefore it

does not enter into this particular trade study.

For a quick example of trade study calculations that are relevant, the following

assumptions are made for analysis:

" Only the wing skin on the top of the wing between the forward and rear spar will be
considered for total wing weight estimation.
" The initial study will consider a wing of constant chord, equal to the root chord
dimension (a rectangular wing).
" The only wing skin associated variable affecting total weight will be skin thickness.

117
 Note that to properly determine the location of the rib spacing, the actual wing chord

at various spanwise stations must be accounted for. Turning to the only remaining

variable, some estimation process must be derived for the weight of the ribs. For the

results depicted in Figure 36, the following steps were used to calculate individual rib

weight:

"Determinethe cross-sectional area bounded by the four spar caps to use as the
external boundary of the rib.

" Assume only 70% of that area contains sheet metal, that allows for cutouts and
weight reduction.

" The rib weight is the product of the reduced cross-sectional area, material density
and specified rib skin thickness. For illustration purposes, three different skin
thickness were used, beginning with the minimum gauge value. Actual
determination of the required rib skin thickness is generally a detail design
procedure.

Figure 36 is the result of the spreadsheet that follows at the end of this appendix. The

calculations are based upon the same equations developed in chapter IV of this thesis.

However, one distinction is made for this method that differs from the previous

development. Recall earlier, the rib spacing (L), was determined by first specifying the

stringer cross-section geometry, and then solving for I., and the radius of gyration. When

L is specified first, the designer must then solve for the radius of gyration and required

moment of inertia. The geometry must then be developed to satisfy the resultant moment

of inertia and radius of gyration. This method appears to require a bit more iteration on

the part of the designer, but does lead to an overall reduced weight. Hopefully, these

118
studies offer some insight into the nature of optimization that serves to build a skill that

can be applied to a wide range of engineering decisions.

Total Upper Wing Weight vs Rib Spacing

1,200
Rib Thickness
8, 0.028 In
1,100
lb Rib Thickness
1,000 0.040 In

900 Rib Thickness

0.060 in

800

700 , ,
5 10 15 20 25 30
Rib Spacing On)
Assumes uniform chord length
Only accounts for skin weight on upper wing surface

Figure 36

119
 cot
c
6
CO) 3

* ", a

If 04P f cm

Li

Ct ~a)
100
mull 5q)i!EM m-

12-

44 ni
CM~o~ ~o w Cm -M C C

IOD
i •| . ... . .-. .. .

10
 APPENDIX D

Common Sheet Metal Gauges

m/

7-_ 04 .
6-0 .580 .461 .481-05 .4 02 .4
5-0 .516 .430 .500 .437 20 .032 .034 .035 .037
4-0 .460 .393 .454 .406 21 .028 .031 .032 .034
3-0 .09 .62 .425 .375 - 22 .25 .28 08 .3
2-0 .364 .31 .8 343 23 .022 .025 .025 .028
0 .324 .306 .340 .312 - 24 .020 .023 .022 .025
1 .289 .283 .300 .281 - 25 017 .020 .020 .021
2 .257 .262 284 .265 26 .015 .08 .018 .018
3 1 229 243 .259 .250 27 .04 .07 .16 .1
4 .04 .225 238 234 28 .02 .01 .04 .015
5 .181 .207 .20 .218 29 .01 1 05 .01 1 014
6 A62 A92 .203 .203 30 01 .014 .02 .1
7 " .177 180 . 3 008 .
8 A2 .162 .65 .1 - 32 .008 .012 .009 .010
9 .114 .148 .148 .156 33 .007 .011 .006 .009
10 A0 .35 .134 .40 34 .006 .01 .007 .008
11 .090 .120 .120 .12 35 .005 .009 .005 .007
12 .080 .105 .109 .109 1 36 .005 .009 .004 .007
13 .072 .9 .095 .093 37 .004 .006 .006
14 .064 .080 .083 .078 38 .004 .009 .006
15 .057 .072 .072 07 39 .003 .007 - -

16 .050 .0621.065 .062w 40 .003 . 7 ~

17 .045 j .054 1.058 .0566E00

121
 REFERENCES

1. Allen, D. H., and Haisler, W.E., Introductionto Aerospace StructuralAnalysis,

John Wiley & Sons, Inc., 1985.

2. Sechler, E. E., and Dunn, L.G., Airplane StructuralAnalysis andDesign,

John Wily & Sons, Inc., 1942.

3. Nicholai, L. M., Fundamentalsof AircraftDesign, Domicone Printing Services,

1975.

4. Raymer, D. P., Aircraft Design: A ConceptualApproach, 2d ed., American Institute

of Aeronautics and Astronautics, Inc., 1992.

5. Burris, S. A., Subsonic LoadAnalysis Manual, Technical Manual prepared for the
Naval Postgraduate School, Department of Aeronautics and Astronautics, 1994.

6. Bertin, J. J., and Smith, M. L., Aerodynamicsfor Engineers,2d ed., Prentice Hall,
1989.

7. Anderson, J. D. Jr., Fundamentalsof Aerodynamics, 2d ed., McGraw-Hill, Inc.,

1991.
8. Interview between Mr. Dave McNally, and others, Northrop Aircraft Corporation,
and the author, 30 July 1993.

9. Niu, M. C., Airfrmne StructuralDesign, CONMILIT PRESS LTD., 1988.

10. Abbott, I. H., and von Doenhoff A. E., Theory of Wing Sections, Dover Publications,
Inc., 1959.

11. McDonnell Douglas Corporation, DA TCOM, USAF Stabilityand ControlData

Compendium, by Fink, R. D., and Hoak, D. E., April 1978.

12. Popov, E. P., Introductionto Mechanics of Solids, Prentice-Hall, Inc., 1968.

13. Naval Air Systems Command, General Specification for Design and Construction of
Aircraft Weapon Systems, v. 1,Fixed Wing Aircraft, 2 June 1982.

122
14. Anderson, J. D. Jr., Introductionto Flight, 3d ed., McGraw-Hill, Inc., 1989.

123
 INITIAL DISTRIBUTION LIST

No. of Copies
1. Defense Technical Information Center 2
Cameron Station
Alexandria VA 22304-6145

2. Library, Code 052 2

Naval Postgraduate School
Monterey CA 93943-5002

3. Department Chairman, AA 1
Department of Aeronautics and Astronautics
Naval Potraduate School
Monterey CA 93943

4. Prof. Gerald H. Lindsey 3

Code AA/Li
Department of Aeronautics and Astronautics
Naval Postgraduate School
Monterey CA 93943

5. Prof Edward M. Wu 2
Code AA/Wu
Department of Aeronautics and Astronautics
Naval Postgraduate School
Monterey CA 93943

6. Prof M. Platzer 2
Code AA/Pl
Department of Aeronautics and Astronautics
Naval Postgraduate School
Monterey CA 93943

7. Stephen A. Burris 2
c/o George R. Burris, Jr.
2223 Vdxumum Lane
Worthington OH 43235

124
8. Prof. R. Ball 2
Code AA/Ba
Department of Aeronautics and Astronautics
Naval Postgraduate School
Monterey CA 93943

9. Prof. C. Newberry 4
Code AA/Ne
Department of Aeronautics and Astronautics
Naval Postgraduate School
Monterey CA 93943

10. Dr. 0. Lynn Webb, Ph.D.

30 NW Quail Creek Drive
Lawton OK 73501

125