PDF Bicycle

A 001 012.
AD AnMJM
TECHNICAL REPORT
75-31
ANALYSIS OF STRESSES AND

DEFLECTIONS IN FRAME SUPPORTED TENTS
1
by
■
Paul J. Remington
■
John C. O'Callahan
■
and
■
Richard Madden
Bolt, Beranek and Newman Inc.

50 Moulton Street
Cambridge, Massachusetts 02138
I
Contract No. DAAG17-73-C-0107
Approved for public release;
distribution unlimited.
I
April 1974
UNITED STATES ARMY

NATICK LABORATORIES
Natick, Massachusetts 01760
Aero-Mechanical Engineering
■
Laboratory
I
mn»tiMinwii:„iliini'nmi?»rtl
mimt !F STSESSES HNS

siriE&Tsnis lü FUME mmiB IEKTS
WH'BilllWMLliiHIIHHIIBil»« llillUWMWllWlMlllllll IHM
by
a
Payl J. Remington
•
John C. Q'CöilehcR
i
und
t
;: '
III ütchord t&Qtld&n
I
*HWIti ieronek and Newmca Inc.
50 MeutJort $!r«®t
Cssmbridge, Moisochu$*tts 02131
Contract No. DAAG?7-73~C-0*O7

^«r@v«d f®'ij public r®fscs$©j
i if §1 »*r b eô » ~v% M mitoct.
I
April 1974
ÄiFi-ltoiükil Eifiififiij
s
jlfif Stiff
Afppcrod for pAlie releaaei difltrüration unlimited.
Citation of trsde names in this report 4c,* ro«
ooastlttrte an official indcrsejöat or pjrgi m 1 of th«
««e of «%K& itesta.
Destroy t&Ls report vaen no longer naeded. Do not
return it to tbe originator.
*Tf^!*jr,*~E*i ,
ANALYSIS OF STRESSES AND DEFLECTIONS

IN FRAME SUPPORTED TENTS
by
Paul J. Remington
John C. O'Callahan
Richard Madden
Bolt Beranek and Newman Inc.

50 Moulton Street
Cambridge, Massachusetts 02138
Contract No. DAAG17-73-C-0107
April 1974
Submitted to:
U. S. Army Natick Laboratories
Natick, Massachusetts 01760
/DDC
DEC
II • in i*■■
UlJEEEinna
B
■
FOREWORD
This report was prepared by Bolt Beranek and Newman

Inc. under U.S. Army Contract No. DAAG17-73-C-0107. The
work was carried out under the direction of Drs. Constantin
J. Monego and Earl Steeves of the U.S. Army Natick Labora-
tories acting as project engineers.
iii
^Mtmmmtmttl^mm
TABLE OF CONTENTS
page
FOREWORD 11 i
LIST OF FIGURES vi
ABSTRACT x
1. INTRODUCTION 1
2. COMPUTER PROGRAM 3
2.1 Program Capabilities 3
2.2 Program Organization 4
2.3 Program Input and Output 4
2.3.1 Input requirements 4

2.3.2 Output descriptions 6
3. SCALING LAWS FOR FRAME-SUPPORTED TENTS 8
3.1 Fabric Scaling 8
3.2 Frame Scaling 10

4. MODEL FABRIC 12
4.1 Screening 12
4.2 Biaxial Fabric Testing 24
4.2.1 Biaxial testing apparatus 24

4.2.2 Biaxial properties of the model fabric. 27
4.2.3 Mathematical model of the biaxial
stress/strain data 29
4.3 Fabric Force Sensor 38

4.3.1 General description 38
4.3.2 Calibration 38
iv
—
• ' ""^rr: rr^tyro^m*- i,„w J.^^JS.»l»:#*!»Pf»î!W(BWp^ •«
TABLE OF CONTENTS (Cont'd.)
page
5. MODEL TENT FRAMES 43
5.1 Slant Roof Frame 43

5.2 Arch Roof Frame 46
6. PRELIMINARY VERIFICATION OF THE COMPUTER CODE 52
6.1 Fabric Strip Membrane 52

6.1.1 Test set-up 52
6.1.2 Computer program model 55
6.1.3 Test results 55
6.2 Tent Frame Deflections 62
6.2.1 Slant-roof frame computer model 62
6.2.2 Arch-roof frame computer model 64
6.2.3 Model tests 64
6.3 Two-Dimensional Fabric Memb it 69
6.3.1 Testing procedure 69

6.3.2 Computer model 71
6.3.3 Results 73
7. FINAL VERIFICATION OF THE COMPUTER CODE ., 80
7.1 Computer Models 80
7.2 Model Tent Tests 80

7.3 Results 87
8. CONCLUSIONS 105
APPENDIX A: SIMPLIFIED MEMBRANE ANALYSIS 106
LIST OF SYMBOLS 116
DD FORM
M^B MMMMMI
...:.!,...,.|....v. ..... .-..^...„^ -.-.--;;■■:, ,.■:-, ~..,;,-: ■■ -. ■
LIST OF FIGURES
icjure page
2 1 Flow Chart of Program 5

4. 1 Uniaxial Test Apparatus 13
4 2 Stress/Strain Properties of 9.85-oz Cotton
Duck (Fill Direction) 15
4 3 Stress/Strain Properties of 9.85-oz Cotton
Duck (Warp Direction) 16
4 4 Stress/Strain Properties of 3.1-oz (78/78)
Bleached Cotton Muslin (Fill Direction 18
4 5 Stress/Strain Behavior of 3.1-oz Cotton Muslin
(Warp Direction) 19
4 6 Stress/Strain Behavior of 2.6-oz Typewriter
Ribbon Cloth (Fill Direction) 20
4 7 Stress/Strain Behavior of 2.6-oz Typewriter
Ribbon Cloth (Warp Direction) 21
4 8 Stress/Strain Behavior of 2-oz Balloon Cloth
(Fill Direction) 22
4 9 Stress/Strain Behavior of 2-oz Cotton Bal'oon
Cloth (Warp Direction) 23
4 10 Comparison of Scaled 2.6-oz Cotton Typewriter
Ribbon Cloth and 9.85-oz Cotton Duck 25
4 11 Biaxial Fabric Testing Machine 26
4 .12 Fabric Cross Test Section 28
4 .13 Biaxial Fabric Test Data for 2.6-oz Cotton
Typewriter Ribbon Cloth 30
4 .14 Biaxial Fabric Test Data on 2.6-oz Cotton
Typewriter Ribbon Cloth 31
4 .15 Comparison of Uniaxial Data With the Biaxial

Stress/Strain Mathematical Model 36
vi
L, MM
f - n
LIST OF FIGURES (Cont'd.)
Figure page
4 16 Load Link in the Fabric Force Sensor 39
4 17 Fabric Force Sensor Drawn Actual Size 40
4 18 Fabric Force Gauge Calibration Curve 41
5 1 Tent Maintenance Shelter Beam Cross Section ... 14
5 2 One-Eighth Scale Approximate Model of TMS 45

5 3 Tent Frane Model Foot Detail 47
5 4 Fritche Shelter Frame Members 48
5 5 One-Eighth Scale Approximate Mode of the FrS .. 50

5 6 Model Tent Frames 51
6 1 Test Configuration 53
6 2 Stress/Strain Data for 1.8-oz Dacron (Coated)
in the Fill Direction 54
6 3 Fabric Strip Computer Model 56
6 4 Deflection of Initially Flat Membrane 57
6 5 Strip Deflection as a Function of Position .... 59
6 .6 Strip Deflection With an Initial Deflection

of 0.93 in. at the Center 60
6.7 Comparison of Theoretical and Measured Fabric

Membrane Strip Tensions 61
6.8 Slant-Roof Frame Computer Model 63
6.9 Arch Roof Frame Computer Model 65
6.10 Slant Roof Frame Deflections 67

6.11 Arch Roof Frame Deflection 68
6.12 Fabric Frame Test Arrangement 70
vii
--■——■
i^jipjgp ■- ^^^m^%m^mssiiff!fi^l^K 'ÎPBiWWWHJ1!".-''
Figure page
6.13 Computer Model of the 2-D Fabric Membrane 72
6.14 Comparison of Predicted and Measured Deflec-
tions 76
6.15 Comparison of Measured and Predicted Deflec-
tions Along the Center Line of the Fabric
Noise 1, 2, 3, 4, 5 77
6.16 Comparison of Measured and Predicted Stresses . 79
7.1 Slant Roof Tent Computer Model 81

7.2 Arch Roof Frame Computer Model 82

tions in the Slant-Roof Tent at Node 8 88
tions in the Slant-Roof Tent at Node T 90

7.7 Z Deflection in the Slant-Roof Tent Along
Nodes 5-8-11-14-17 With 30-1 b Load 92
7.8 X Deflection With Slant-Roof Tent Along

Nodes 5-8-11-14-17 With 30-lb Load 93
7.9 Z Deflection in the Slant-Roof Frame Along

Nodes 9-10-11 With 30-lb Load 94
7.10 X Deflection in the Slant-Roof Frame Along

Nodes 9-10-11 With 30-lb Load 95
tions in the Arch-Roof Tent at Node 15 97

viii
■MM. -"———»" MM
. . ■ - ■ : .
Figure page
.13 Comparison of Measured and Predicted Deflec-
.14 Z Deflection in the Arch-Roof Tent Along Nodes
11-15-19-23-27 With 21-lb Load 100
.15 X Deflection in the Arch-Roof Tent Along Nodes

11-15-19-23-27 With 21-lb Load 101
.16 Z Deflection in the Arch-Roof Tent Along Nodes
16-17-18-19 With 21-lb Load 102
.17 X Deflection in the Arch-Roof T^nt Along Nodes

16-17-18-19 With 21-lb Load 103
A.l Coordinate System for Initially Flat Membrane

Strip 107
A.2 Displacements of Elemental Portion of
Membrane 107
A.3 Coordinate System for Initially Displaced

Membrane Strip 112
A.4 Displacements of Membrane With Initial
Deflection 112
ix
■ ill i mmam
ABSTRACT
A finite element computer code has been developed for

predicting the stress and deflection in frame-supported
tents under static load. The code can accept geometric non-
linearities due to large deflections and nonlinear biaxial
stress/strain fabric properties. The predictions of the
computer code are validated by tests performed on two 1/8-
scale model cents that approximately model two existing Army
tents, the tent maintenance shelter and the Fritohe shelter.
The 2.6-oz cotton typewriter ribbon cloth used in the models
was tested on a biaxial testing machine to obtain its bi-
axial stress/strain behavior for input to the computer code.
Preliminary comparison of computer predictions and measure-
ments from tests on the deflection and stress in a fabric
strip, on the deflections of the tent frames under a point
load, and on the deflections and stress in a rectangular
fabric membrane in a rigid frame demonstrated the validity
of the code predictions. These comparisons also pointed out
the need to include joint efficiencies in modeling the frame,
Comparison of computer predictions and measurements of the
deflections in the tent frame models showed the computer
predictions to agree adequately with measurements.
«■■t ■MHM „__-, ■ttHMM

* ■ -mm>~m
SECURITY CLASSIFICATION OF THIS PACE (Whtn Dim Entered; /fPMnnzi 07.<b

READ INSTRUCTIONS
REPORT DOCUMENTATION PAGE BEFORE COMPLETING FORM
t. REPORT NUMBER 2. 3QVT ACCESSION NO, 3. RECIPIENT'S CATALOG NUMBER
TR 75-31
4. TITLE (and Submit) S. TYPE OF REPORT ft PERIOD COVERED
Analysis of Stresses and Deflections in Final

Frame Supported Tents
6. PERFORMING ORG. REPORT NUMBER
BBN Report No. 2802
7. AUTNORfiJ B. CONTRACT OR GRANT NUMBERS«;
Paul J. Remington
John C. O'Callahan DAAG17-73-C-01701
Richard Madden
9. PERFORMING ORGANIZATION NAME AND ADDRESS 10. PROGRAM ELEMENT, PROJECT, TASK
AREA ft WORK UNIT NUMBERS
Bolt Beranek and Newman Inc. 6.2 Exploratory Dev
50 Moulton St. Project AE 98
Cambridge, Mass. 02138 Task 16AE
tl. CONTROLLING OFFICE NAME AND ADDRESS 12. REPORT DATE ,
Aero-Mechanical Engineering Laboratory (AMEL) 22 April 197^
Engineering Science Division (ESD) STSNL-UE 13. NUMBER OF PAGES
US Army Natick Laboratories, Natick, MA 01760 W ill
14. MONITORING AGENCY NAME ft ADORESSf/f different from Controlling Office; IS. SECURITY CLASS, (ol '.hit report;
NA Unclassified
15a. DECLASSIFICATION/DOWNGRADING
SCHEDULE
16. DISTRIBUTION STATEMENT (of «lie Report;
Approved for public release; distribution unlimited,
17. DISTRIBUTION STATEMENT (of tho ebafract entered In Block 20, It different from Report;
Same
is. SUPPLEMENTARY NOTES ^ ^ ^3^^ by the Technical Information Service

of the Department of Commerce
19. KEY WO, llnuo on reveree aide If neceeeary and Identity by block number;
Frame Supported Tents

Fabric
Membranes
Biaxial Fabric Testing
20. ABSTRACT (Continue on reveree eide II ntcoeeaiy and identify by block number)
A finite element computer code has been developed for predicting the
stresses and deflections in frame-supported tents under static load.
The code can accept geometric nonlinearities due to large deflections
and nonlinear biaxial stress/strain fabric properties. The predictions
of the computer code are validated by tests performed on two l/8 scale
model tents that approximately model two existing Army tents, the tent
maintenance shelter and the Fritche shelter. The 2.6-oz cotton type-
writer ribbon cloth used in the models was tested on a biaxial test-
DD,:FORM
AN 73 1473
EDITION OF 1 NOV 65 IS OBSOLETE •
Unclassified W
SECURITY CLASSIFICATION OF THIS PAGE (Wien Data Entered;
*"*"* ■ - 1 mitt^ätmmmlttmmmmiimm SSE

PWJJW^.#J«,iJI).Ji--!-.«!(WUiWll"«.»"J.tl»W.MH!llJ«l.y -tMIU.UIM.IWKU W.U!1U"MW■W >ll^lWII!P^^»IJl»lll|!iPlip.«.l..l:l«|il, l. M IUI..HU .*W
SECURITY CUWIf ICATIOM OF TH!S EAOEfWIH» Data tot—«Q
20. cont'd
ing machine to obtain its biaxial stress/strain behavior for input to
the computer code. Preliminary comparison of computer predictions and
measurements from tests on the deflection and stress in a fabric strip,
on the deflections of the tent frames under a point load, and on the
deflections and stress in a rectangular fabric membrane in a rigid
frame demonstrated the validity of the code predictions, ltoese com-
parisons also pointed out the need to include Joint efficiencies in
modeling the frame. Comparison of computer predictions and measure-
ments of the deflections in the tent frame models showed the computer
predictions to agree adequately with measurements.
fr
SECURITY CLASSIFICATION OF THIS PAGEflWlMi Data Bnltnd)
mf"tiJfi«VLl lilj-li i • •-|n-rilT:ha"r''"''A Hi iJm i'röiMJt Wn-*1f1iWii,il-1ltf--|-i|- ■ i

f CO
V» 4J
I
T- f-
4-* C
f- 3 >» * in
s. ca i | * t*
cai-i o O o 1 m
CO >, H rH H o O
+J •— M rH
i- O CL ■^-
• • ro •
01 4-»T- rn LO in en CD • •
> +J in in CM o H in ON ■=r C\J t—
co C (/» •— ■=r -=r O m ON -=r o VO o t— vo H cr\
OPS t— • • CO
U<PZ o .=3- O o O \o o H o OJ CO H -^~ vo ^r
c
0=3
K
co
O Ü
I- •H 0 0
XI M H M
co CD CD CO CO
o U P 3
CO 0 a1 cr
V) to CO M 3 E to to
■P M u CO CO a» cr
•p» a> 0 H u a CO M M M
c -p p (Ü CD CD 0 0
=> a> co a * a a
CO
I—1
i
to
a co
E
CD
E
P
CD
E
P
0
E
E
CO M
M
CD
a M CO
^^.
co to u CO U
cc
CO übO c 0 cu M CD CD c M C 0 C 0
o M M M M M o CJ bO-P CO -P o 0 O -P o .p
ca o p CD flj 0 a) CtJ •H •H O CD E CD -p P p 0 4-> 0
H S -P p P 3 3 £1 Xl H E c0 E £ 0 5 E S E
s: •H CD CD 0 0 o* o* 3 3 •H U cu E 0 0
o « 53 E E E CO CO Ü o bO c v^ c
ac
0 0
o M M
CO •H X! CO CO
CO +J XI Ü 31 3
cc •p* 3 c o D"
C 0) CO o •H co co
Z> CO O Ä p CD
CO M Ü o x; p M H H M
o x: CO O C o O 0 CD 0 J- 0 0
o (A E CM ■H CM C CD a a ^^\ a a
•r— •H CM co
a: +> CO CO XI 3 -H C 3 c 3 -H 3 CM
o o C o CO o< a* 3 3 o 3 o ■H O o
a a •H CM >» CO co O o a o a K_^ a a
ca
» CM TH
4-> CM CO o o
•i— O iH •H
■P >> c ■P CO P 60
C X! 0 p o P H 3 CO c
a CD -p E •H •H C 0 H CO ■H
3 CO O bO CO 3 CO co 0 C 3rH TJ
co- CO M c 0 H c c EH nW CO
CO O 0 M O OJ 0 O o o
S fe ►J < > Q fr s s 1-5
xi
1 INTRODUCTION
One type of shelter extensively used in Army field

operations is the frame-s upported tent, essentially a metal
frame to which is attache d one or more layers of fabric,
This type of shelter has many attractive features. It is
light, easily transportab le, and provides a reasonably secure
shelter from the weather, Although these tent structures
are fairly simple to cons truct and erect, a good understand-
ing of their static respo nse to snow loads is lacking,
Clearly, if lighter, more efficient, structurally sound,
frame-supported tents are to be designed, such an understand-
ing must be developed.
The program described in this report has concentrated
on the development of a computer code for predicting the
stresses and deflections in frame-supported tents under dead
weight loads and the verification of that code through com-
parison of the predictions with measurements on model frame-
supported tents. The resulting computer code is a finite
element code capable of predicting stresses and deflections
in both tent fabric and frame, including the effect of any
geometric nonlinearities due to large deflections. The code
will accept any configuration for the frame which can be
modeled as a number of simple beams and will also accept
fairly general nonlinear stress/strain properties for the
fabric. A detailed description of the code is given in
Sec. 2.
Once the computer code was developed, it was necessary
to verify its predictions. To this end, two model 1/8-scale
tent frames were constructed and a model fabric was selected
so as to simulate approximately the tent maintenance shelter,
a slant roof tent, and the Fritahe shelter, an arch roof
tent, whose geometries are typical of Army frame-supported
tents. It should be emphasized that these model tents were
not exact scale models of the above-mentioned Army tents.
All that was required was to show that the computer code
could adequately predict the stresses and deflections in a
frame-supported tent. For this reason, any frame-fabric
model would have been sufficient. It was felt, though, that
the models should roughly scale to existing full-scale tents,
so that once it has been demonstrated that the code can deal
with a small-model tent, the user will have confidence that
the code can also deal with the full-scale tent. Therefore,
selection of the model fabric, scaling of the applied loads,
and design of the frames were based on the crude scaling laws
developed in Sec. 3. These laws result in (1) comparable
strains in the model tent fabric and the full-scale tent
Ill LAAWp11 ■■M!H!I!IV simwtwiJ.WSJ«•'»* mipj»pi»KlM..W|W«p-•-- u i. I.- MWi-Jmill,..MJM"MJUIMIJUM>jM ---'' <.»»iJP.Hmw»;«m»M *"J»TO,1pi,M|jU,,i.i '■
fabric and (2) deflections in the fabric of the model in the

same proportion to the model frame deflections as in the
full-scale.
Section 4 describes the screening process for selecting
a model fabric, the ensuing biaxial stress/strain test on
the chosen fabric, and the adaptation of a fabric force
sensor for measuring the low level loads in the modes. The
tent frame design is described in Sec. 5.
Before testing on the model was begun, it was felt that
a number of simpler tests would be useful to point up any
difficulties in predicting stress or deflections in the
fabric or the frame of the model tents. Section 6 compares
computer predictions with results from three of these tests:
the stresses and deflections in a thin fabric strip., the
deflections in the model tent frames under a point ioad, and
the stresses and deflections in a rectangular piece of fabric
in a rigid frame.
The final comparison of computer predictions and mea-
surement of deflection in the two model tents is made in
Sec. 7. Section 8 presents the conclusions.
ill!
2. COMPUTER PROGRAM
In this section a very brief description is given of

the computer code developed during this program. No attempt
is made in this report to provide detailed information on
the code. For such information the reader is referred to
the User's Manual.*
2.1 Program Capabilities
The computer code developed to solve problems involving
nonlinear geometry and material properties has been named
NONPESA (Nonlinear Finite Element Structural Analysis). The
code was designed to determine deflections and stresses in
frame-supported tents by discretizing the continuous frame
and fabric into "finite elements" such that straight beam
elements represent the frame and flat triangular membrane
elements may be assembled for the fabric.
The beam element is the standard, straight, finite ele-
ment beam with the ability to include shear deformations.
It is considered linear in the code; i.e., it is assumed
that structural deformations are small and can be modeled
by linear beam theory. Therefore, the stiffness and strain-
displacement matrices for the beam elements need be calcu-
lated only once.
Because the membrane elements involve large displace-
ments, the program must calculate the nonlinear stiffness
matrix by breaking up the full load into small increments or
steps and by using an iterative scheme to update the dis-
placements and stresses. The material properties may be
linear (stress proportional to strain) or nonlinear (treated
as piecewise linear assuming incremental stress proportional
to incremental strain).
Loads on the membrane elements may be in the form of
local surface tractions, pressure, gravity, or point-
concentrated loads. As described above, for dealing with
nonlinear problems, the load is divided into small incre-
ments, or steps.
*0*Callahan, J.C., "NONFESA - Nonlinear Finite Element Struc-

tural Analysis Program for the Analysis of Stresses and De-
flections in Frame-Supported Tents," BBN Report No. 2803,
June 1974.
wm**m. ••■ m ""wmmmQtmm ..„ -J*^.-„™-T nampîiiuiv-nHi1 ' :
».IHvl ' 'wnJiAWJiUM
2.2 Program Organization
The program, a derivative of SAP II,* is very versatile.

It is designed such that all subroutines effectively use the
common storage arrays without even knowing a fixed dimension
for the arrays. This feature allows the user easy assembly
of a very large or a very small problem.
The flow of the present program can best be described
by the chart given in Fig. 2.1. The program has the potential
of mixing many different elements, although, at present, only
the beam and membrane element are available.
The solution scheme used i n NONPESA is accomplished by
a direct reduction technique — i.e., a Gaussian elimination,
The algorithm allows for an in- core as well as an out-of-core
solution of the linear aigebrai c equations. The decision
about which method to use is ma de internally and is particu-
iarly affected by the user allo cation of common block
storage. The larger the alloca ted space, the bigger the
block of equations that will be in-core for solution.
Input/output processing will be minimized If the blocks are
kept as large as possible.
2.3 Program Input and Output
This section contains a brief description of the

input data required to run the program and the type of out-
put the program generates. Further details of the input
and output may be found in the User's Manual.
2.3.1 Input requirements
The program requires that the user input information

which will set maximum parameter conditions, such as number
of nodal points, element types, load steps, and convergence
criteria. In addition, one must input the fraction of the
total load used In the incremental solution technique.
The geometry of the structure is modeled with nodal
point cards. These cards also contain pertinent information
about the equilibrium equations to be solved. Any zero
displacement boundary conditions are set by these cards.
Following the nodal cards, sets of element cards are input
»Wilson, E.L., "SOLID SAP - A Static Analysis Program for

Three-Dimensional Solid Structures," SESM Report 71-19,
Dept. of Civil Engineering, Univ. of California, Berkeley,
1971.
1
MMH* "- -~- "**"* MM
UV I mi!« liBwii.j .if.^n^np
riUj.iiy-IMIIiptpMtWWJf.>W.«iii!Jp.i) ..iiyij»^.ji-uj .i».-..! ipw'Timi— ■'BHJl
MAIN PROGRAM FLOW
READ NODAL POINT DATA AND DISC

ESTABLISH EQUATION NUMBER STORAGE
BEGIN ITERATION OR
LOAp STEP CYCLE
(IF NONLINEAR)
STRESS-DISPLACEMENT
TRANSFORMATION
ENTER ELEMENT PROPERTIES MATRICES - USE
AND ELEMENT GENERATION LATE? IN STRESS
CALCULATION
INPUT FIXED ELEMENT STIFFNESS

NODAL LOADS MATRICES
ASSEMBLE ALL ELEMENTS LOAD VECTORS

INTO MASTER STIFF-
MESS MATRIC
STRUCTURAL STIFFNESS
MATRIX AND ASSEMBLED
SOLVE LINEAR EQUATIONS LOAD VECTOR
DISPLACEMENTS
UPDATE SYSTEM PARAMETERS,
DISPLACEMENTS, STRESSES
CHECK CONVERGENCE
(~ET
EN SOLUTION
FIG. 2.1. FLOW CHART OF PROGRAM.
■xaAMMMMn ■Mill
r-«i.iw>m)ft.. ■u..yjyw;-nwy-«.:. •■- "-■ mm ■ --»m^mmm - IWIBF'
to define the element connectivity to the nodal structure

and material properties. Presently, the program contains
two element types: the linear beam and the nonlinear mem-
brane. From these cards, the element and such properties
as stiffness and stress-displacement matrices are calculated
and written to disc storage.
The last type of input required is the loading function.
If point loads are used in the solution, then the list of
cards containing the nodal forces are input to the program
and a load vector is generated and placed in disc storage.
If distributed loads are used on the elements, the informa-
tion is calculated and stored with the element Information.
When assembly occurs, all load vectors are added together.
2.3.2 Output descriptions

The output data calculated in the program is printed
for each cycle in the load step.* Information such as
incremental nodal displacements and relations, beam and
fabric stresses, beam forces and moments, accumulated nodal
displacements and rotations, p.nd new nodal global coordi-
nates may be selected in the cycle printout.
The first page contains the start of cycle information:

the cycle number, load step number, and load fraction. The
second section is a table of Incremental nodal displacements
and rotations calculated during the given cycle. This table
contains all nodes and possible degrees of freedom. If a
degree cf freedom is fixed or deleted from the solution, a
zero is printed.
Next, beam stresses are printed in tabular form. They
are listed by element with reference to both ends of the
beam. Principal bending and average shear stresses relative
to the local axes of the beam are printed. The next section
is a table containing the resulting beam forces and moments.
This table, which is similar in form to the beam stress
table, contains the internal forces and moments at each end
of the beam.
The membrane stress resultants are printed next by

element number. The resultants are described relative to
*As described above, the total applied load is divided into

small increments called steps. The iterations within each
step are called cycles.
■ - ■
■ "»■"•'■ ~r- J HWU HOT ■^■P »üffWIiP«
the local axes of the membrane. A table containing the

accumulated (all previous steps plus the given cycle) incre-
mental nodal displacements and rotations is then printed.
The new global nodal coordinates of the structure
(including deflections) are printed next. This table is
helpful when a deformed configuration of the structure is
desired.
«... .»m^^Can nm..».^«,«jjr ^npi
■PSWPHSP" "■"■"' «m.mmwmmim: .<■•' mmmvu. L
3. SCALING LAWS FOR FRAME-SUPPORTED TENTS
The primary objective of this study is the development

of a computer program to predict deflections and stresses in
frame-supported tents under static loads. The purpose of
the model tent test program is simply to verify the validity
of the computer program. Strictly speaking, then, accurate
scaling of the model tents with full-scale tf;nts is not
necessary, because agreement of the computer program and
test results is all that is required to verify the computer
model. In fact, though, we have approximately scaled the
tent models with real tents so as to obtain realistic fabric
strain and frame deflections in proper proportion with fabric
deflections.
3.1 Fabric Scaling
If the relevant partial differential equations govern-

ing the deflection and stress in tne membrane and frame are
known, development of the scaling laws is almost a trivial
problem. Referring to Appendix A, we find approximate
equations for the deflection and stress of a strip membrane.
We can use these equations for developing the scaling laws
for two-dimensional membranes, if we recognize that certain
conditions must be satisfied. First, the models must be
geometrically similar to full-scale tents; i.e., the length
and width of the membrane must be in the same ratio for the
model and the full-scale tent. Second, shear stresses in
the fabric must be negligible. Last, the stiffness matrix
relating tension and strain in the real tent fabric must be
proportional to the same matrix for the model fabric. Tak-
ing these conditions to be valid, we find from the equations
of equilibrium that
3T
"BE" = o (3-la)
3
3xx \l 3x/ (3-lb)
where T is the tension per unit length, f is the load per

unit area, w is the r..embrane deflection perpendicular tc a
plane connecting the end supports, and x is a spatial co-
ordinate. From the strain displacement relationship, we find
-'■ - - -
UUMMMW ,..-. ...... ■■■-a.J^..»-.
- ""^w^ ■-»«
e + i /aw\2
-n 2 \Sx/ »
(3-2)
where e is the strain and u is the displacement in the mem-

brane parallel to the plane connecting the end supports.
Assuming a linear stress/strain relationship, which is
approximately true for cotton fabrics*, we find that
T - Ke (3.3)
where K, a constant, is the slope of the fabric stress/

strain curve.
From the equilibrium equations, we can show that at
geometrically similar points in the model and full-scale
tents
fe) "fei
>fL /m \fL /FS
(3.4)
where L is the distance between membrane supports and the

subscripts m and FS refer to model and full scale, respec-
tively. From the strain displacement relationship and the
equilibrium equation, it is easy to see that at geometrically
similar points in the model and the full-scale tent
m-{% (3.5)
Since we want the strain in the model tent and full-

scale tent to be the same, we can show from Eq. 3.5 that
a-a (3.6)
*As will be seen later, at very low stress levels cotton

fabrics exhibit a stiffening character. Under increasing
levels of stress, the stress/strain relationship eventually
becomes linear.
■MÜ HW
■ ■. ■ . ■ ■ ■■ ■
■ ■ ■■■■■■■ .■.■■■:;-::■-..■■.■. .-• -.-■rr:- ■ ■ ■ ■
and from this equation, and Eqs. 3«3 and 3*4 that
(3.7)
Equation 3.7 tells us that if we make the model 1/10

full scale and choose a fabric with 1/10 the stiffness, then
we must load the model with the same load per unit area as
the real tent to obtain the same strains. Once a model
fabric is se' „cted, then Eq. 3.7 can be used to choose an
appropriately scaled load.
3.2 Frame Scaling
We follow a similar procedure to the one above to scale

the frame. By modeling the tent frame as a beam in bending*,
we can show that
(3.8)
x2 V 9x2 /
where E is the modulus of the beam material, I is the moment

of inertia, wß is the bending deflection of the beam, and T
is the tension per unit length in the fabric. Prom Eq. 3.8,
we can show that a geometrically similar points in the model
and full-scale tent frame
EIW
/ B\ _/EI"B\
u* i w ;FS' (3.9)
To ensure that the deflections in the model and full-

scale tent are geometrically similar, we must have
*If the fabric is mounted to the beam n such a way that the
force on the beam due to the fabric acos through the axis
of the beam, there will be no torsional deformation of the
beam.
10
mm ■Mm
< mammmmsM ■ >mA*mvmMmmmMmmmmmmmmm>imm>mvv*#»:-ta*e,
Comoination of this equation and Eqs. 3.3 and 3.6 yields
(K_\ =(^\ . (3.io)
By using Eq. 3.10, we can select the frame material and

frame beam cross section such that the ratio of fabric de-
flection to frame deflection is the same in both model and
full-scale tents. We will use this scaling lav/ in a later
section to design the model tent frames.
11
-'■■ i ma - - vmumä
"■■ "?*^B^^HH?f!9'77T.' ' ■ '"; WW "
4. MODEL FABRIC
In this section we describe the process for selection
of a model test fabric to be used in the model tents, the
testing of the biaxial properties of that fabric, and the
development of a sensor to measure fabric stress.
4.1 Screening
There is a wide variety of light fabrics available for
use as a model fabric in scale model tents. We decided to
limit our search to cotton fabrics, since most existing
Army frame-supported tents presently use 9.85-oz cotton duck
or 8.5-oz cotton sateen. Also (as the scaling equations in
Sec. 3.1 show), the stiffer the fabric, the higher will be
the required load applied to model to scale with the full-
scale load. Therefore, we decided to search for as compliant
a fabric as possible so as to keep the model loads small and
manageable.
Ideally, one would like to base fabric selection on the
biaxial stress/strain properties of the candidate fabrics,
but selection by that means iz prohibitively expensive. In-
stead, we limited ourselves to uniaxial fabric stress/strain
tests on the candidate fabrics in both the warp and fill
directions and compared these results with similar tests on
9.85-oz cotton duck.* To simplify the comparison further,
we considered only the fabric stiffness < in the region
where stress and strain are linearly related, K is defined
in that region by
K = AT
Ae
where AT is the change in tension per unit length of the

fabric and Ae is the change in the strain.
The uniaxial test setup is shown schematically in Pig.
4.1. The fabric is held by clamps that consist of two
aluminum plates with grooves on their inside surfaces. The
grooves accept two strips of 1/8-in. diameter drill rod.
*We could also have used 8.5-oz cotton sateen for comparison,
but cotton duck is used in the tent maintenance shelter and
the Fritahe shelter, which are the tents we model In later
sections of this report.
12
•mm -——I ■Mia ■MBllllriii i wm

"
411
-~r
I- I 4-
-IB-
FIG. 4.1. UNIAXIAL TEST APPARATUS.
13
m - HlfMlB MjaMMM-i...
Three bolts (shown as center lines) clamp the fabric between
the rods and the aluminum plates. This holding method is
excellent for light loads but does lead to some reduction in
breaking strength (10 to 20£) at high loads.
We applied loads to the fabric simply by hanging weights

for light loads (up to about 20 lb) or by placing the holding
brackets in a tensile testing machine. In general, creep
with the cotton fabric tested here was not a serious problem.
Deflections were measured using vernier calipers having
accuracy to within a few thousandths of an inch. The 4-in.
by 1-in. fabric specimen size was found to be convenient
for both loading and deflection measurement.
Cotton duck (9»85-oz) and a number of candidate model

fabrics were tested in this manner in both warp and fill
directions. Among the most promising were 3.1-oz cotton
muslin (78/78), 2.6-oz cotton typewriter ribbon cloth, and
2-oz cotton balloon cloth.
All of the fabrics tested required repeated loadings

before reproducible stress/strain data could be obtained.
Since it was anticipated that repeated loadings would be
required during the tent model tests, we based our compari-
son of fabric stiffness on the repeatable values; i.e., we
loaded each candidate fabric a sufficient number of times
such that additional loadings produced no change in the
stress/strain behavior.
The stress/strain properties of 9.85-oz cotton duck in

the fill direction are shown in Fig. 4.2. After loading the
fabric up to 70 lb/in. five times, we found that the fabric
stress/strain properties were essentially repeatable. Note
that the second loading of the fabric is not shown as the
data was not properly taken. Fitting a straight line through
the points from the last three loadings, we find a fabric
stiffness of 1800 lb/in. in the fill direction, a value which
we will use in future calculations.
Figure 4.3 shows that, after about five loadings, the

stiffness in the linear range for 9°85-oz cotton duck in the
warp direction is about 1900 lb/in., giving a warp-to-fill
stiffness ratio of 1.05. We wanted not only to have a very
compliant model fabric but also to have the same warp-to-
fill ratio as the cotton duck. With these two criteria in
mind, we examined the stress/strain behavior of the candidate
model fabrics.
14
mmtm «Mm .■■...

!^«i;:ll«HW.WB &*<-*<■**•??■*-*■""- - - - ,— -,,-, ^—,„,.**,*.*v)^. .... .u,.,w.i,wgyw,,ff?,t>r,..,.» -
1800 LB/IN.
5 6 7 8 9 10 11 13 14 15
STRAIN (PERCENT)
FIG. 4.2. STRESS/STRAIN PROPERTIES OF 9.85-OZ COTTON DUCK

(FILL DIRECTION).
15
- LWBBtlW! !l|l,IM!«l«piU!lllll«l.Jjl»lU,"ll'W,^W»l|liuy.'» J 'I«, U '"'
tß tu
cc
1—(
* Q
Q.
QC
K> <
^* 3
N
O
o
o
M
O
I
bJ in
oo
• (t
bJ
Ü.
o
<
(0 CC
o
a:
CL
on
oo
oo
oo
a:
I—
oo
CO
(NI/81) N0ISN31
CJJ
16
riHH .nMrîMM *1aaiîü-ii»iii ■ r f

-",,"WJ>iah-iinil!iftilir
.lîii^Ji.Jh|iU.,W-)< •" "
;'_iT#*fl?»*rv
The first candidate fabric shown is 3.1-oz (78/78)

bleached cotton muslin. Four loadings showed that its
stiffness in the fill direction is 370 lb/in. (see Fig. 4.4).
Three subsequent loadings (see Fig. 4.5) showed that the
stiffness in the warp direction is 620 lb/in. Thus, the
muslin's warp-to-fill ratio of -1.7 is considerably differ-
ent from that of cotton duck. Figures 4.6 and 4.7 show the
stress/strain behavior of 2.6-oz cotton typewriter ribbon
cloth in the fill and warp directions, respectively. Three
loadings on this fabric showed the fill stiffness to be
about 320 lb/in. and the warp stiffness to be about 400 lb/
in., giving a warp-to-fill ratio of 1.25 — very nearly the
same as cotton duck.
The last fabric to be tested was 2-oz cotton balloon

cloth. Its stress/strain behavior in the fill and warp di-
rections is shown in Figs. 4.8 and 4.9, respectively. The
fill stiffness of 435 lb/in. and warp stiffness of 830 lb/in.
gives a warp-to-fill ratio of 1.9. The results of the fabric
testing are summarized in Table 4.1.
TABLE 4.1. FABRIC SCREENING SUMMARY.
Warp Fill Warp/Fill

Fabric Stiffness Stiffness Ratio
9.85-oz Cotton Duck 1900 lb.'in. 1800 lb/in. 1.05

3.1 -oz Muslin 620 lb/in. 370 lb/in. 1.7
2.6 -oz Typewriter
Ribbon Cloth 400 lb/in. 320 lb/in. 1.25
2 -oz Balloon
Cloth 830 lb/in. 435 lb/in. 1.9
Based on the ratio of stiffness in the warp and fill

directions in the linear region of stress/strain behavior,
the 2.6-oz typewriter ribbon cloth is most similar to cotton
duck. Also, the typewriter ribbon cloth is the most com-
pliant of the fabrics tested.
LooKing for further similarities, we decided to compare

the stress/strain behavior of the 2.6-oz fabric with that of
cotton duck. For example, if the 2.6-oz cloth were five
times stiffer (i.e., if five times the stress were required
17
__„
--*■-■ -—-
MMMI
370 LB/IN.
10
1 1 1 1 -p~ 1 1 1 | 1 I 1
A
1
_
9—
8 — a A __
A
7
>—
6— ffn A —
3
A __
— CB
<n A _
Z 4 — QJ
~~
3|— ^P ▲ ▲ FIRST LOADING _
• SECOND LOADING
2 A 0 THIRD LOADING _
Q FOURTH LOADING
1 - % ▲ j
1 1 /1 1 1 1 1 1 I 1 i 1 1 1 1
5 6 7 8 9 10 11 12 13 14 15
STRAIN (PERCENT)
FIG. 4.4. STRESS/STRAIN PROPERTIES OF 3.1-OZ (78/78)

BLEACHED COTTON MUSLIN (FILL DIRECTION).
18
MM. Ml MM
i .4WW-.»l ""..* .Hi»* - JWIMi' " ' u iWilim-ÂI. HLHI.IIHIH1,. ijwyîpji
620 LB/ IN.
T"l—I I I I I I I
A FIRST LOADING _
• SECOND LOADING
O THIRD LOADING .
J I I I I I L
5 6 7 8 9 10 11 12 13 14 15
STRAIN (PERCENT)
FIG. 4.5. STRESS/STRAIN BEHAVIOR OF 3.1-0Z COTTON MUSLIN

(WARP DIRECTION).
19
mm rn-m IMM "

'j/mm mm . - - -MIIJW!WWH( '
JL i—i
I I I I I I f LL.
*
o o -*
*" a:
O? (-
o
z5?g 10
*"
_i
o
llfS N
z
o
CO
CO
hggtt
wg£
(—1
0£
3 r—
9=ü£p OS
UOQ _o
-<*H UJ
Q_
M
O
I
VD
CM
<
X
LU
CQ
< •
\- Z
OO O
"■s.l-1
ooi—
oo o
LU LU
CCCC
O
I
0»
'
OD (0 m * K»
ICM l-l-H
00 Q
(NI/81) N0ISN31
«a-
20
«MHWlaMIMM«ttM ■ - "-•■•■- -■■

raw*'
o
_l
K> o
^»
2:
0
* CQ
co
1—1
at
10
•" at
— CM
a.
—. O
o> 1- 0
z 0
Ul
_CO u
Q»
M
O
1
U VC
0.
K —* C\J
Z
Li_
< O
(0 K
H on
(/) 0
— m
co
— K> <: o
CC LU
t— Qi
in t-H
— CVJ ^Q
CO Q-
LlJ Oi
Oi fit
I— 3
00 —*
o
N <0 IO * W o
(NI/81) N0ISN31
CD
21
L. tmmm ■—-■ ■ ■ ■ ■ -,.--.-... _ .

■^K..,m*.'Wm\.,mmvj!.wp<*4K**-v;*vrrm-:u.! .im. v. .. W«tIW.'»■■.'I'i)|. 'in - --, "™-w . n..jj..„,..,. u .*,.,,.„...,„,,... .»..,.-„ •-.uJIgWMMWj
435 L8/IN.
10
' 1 I I r i ! 1 1 i 1 i i jr i
9 —
r# X ~"
8 — p# yf —
yf —
Z
7
P#
CD r# VSB» ^~
A —
Z f>
O T •
V)
Z 4 ^k -~
u T#
ns0 ▲ FIRST LOADING —
3
• SECONO LOADING
2 0 THIRD LOADING _
D FOURTH LOADING
1
-/rii i 1 1 1 1 1 1 1 1 1
'0 1 3 4 5 6 7 8 9 10 11 12 13 14 15
STRAIN (PERCENT)
FIG. 4.8. STRESS/STRAIN BEHAVIOR OF 2-OZ BALLOON CLOTH

(FILL DIRECTION).
22
mm i-lrtlMi Mtl ....,.;,. ~.^.,,.^ ■.. ■ :~-.. -,

wawujfr WiJiWHJMiuawB"1^^ ■
83C LB/IN.
T^TT I I I I I I I
▲ FIRST LOAOING
• SECOND LOADING
O THIRD LOADING
I I I I J I l_l L J L
5 6 7 8 9 10 11 12 13 14 15
STRAIN (PERCENT)
FIG. 4.9. STRESS/STRAIN BEHAVIOR OF 2-OZ COTTON BALLOON

CLOTH (WARP DIRECTION).
23
irwM-ti. -VI i
Ml ■ mm mm— mtmittimmm
-*>mr<m "■""" '■■ ' '■-•'''*''y''!J,"l*.M»ll'*T^*'/".^>U1.':«^"^^^
to produce the same strain), the stress/strain behavior of

the two fabrics would be as shown in Fig. 4.10. Clearly,
even the nonlinear behavior of the two fabrics in uniaxial
strain is quite similar, although the cotton duck is some-
what more compliant in the fill direction.
We should emphasize that the selection criterion used
here is a very approximate one. For example, we have made
no real effort to match the nonlinear behavior of the model
fabric to the equivalent behavior of cotton duck, although
the two are actually quite similar. In addition, selecting
a fabric based on uniaxial properties (an economic neces-
sity) when it is biaxial properties that we want to match
has some drawbacks. However, the problem is not to develop
an accurate tent model but rather to provide a means for
validating the computer model predictions.
4.2 Biaxial Fabric Testing
In this section, we discuss the design and construction

of a device for measuring the stress/strain behavior of
fabrics under biaxial load as well as the results of tests
on specimens of 2.6-oz cotton typewriter ribbon cloth, the
model test fabric.
4.2.1 Biaxial testing apparatus

The apparatus for obtaining the stress/strain proper-
ties of fabric under biaxial loa" is shown in Fig. 4.11.
The device consists of two aluminum channels (3 in. x i% in.)
welded together to form a cross. Two adjacent arms of the
cross have ball-bearing pulleys mounted at their ends. The
other two adjacent arms have angle blocks through which eye-
bolts are passed. By adjusting the nuts which hold the eye-
bolts to the angle blocks, we could vary the length of eye-
bolt protruding through the block and align the fabric with
the cross .
A cruciform of fabric approximately 3 ft square in which
each arm is 4-in. wide (one arm parallel to the fill direc-
tion and the otner parallel to the warp direction) was tested
in the apparatus. Clamps similar to those used in the uni-
axial fabric tests (see Sec. 4.1) were attached to each arm
of the cross. Nylon lines wera attached to the clamps on
two adjacent arms and passed over the pulleys. Steel wire
was attached to the clamps on the other two adjacent arms
and connected to the eyebolts.
24
NMMaa «MMMnaM
jWVW ".»VKn<*f*x 11 iinnf ii,,y. ,j „...,,..., i.^w^r^. *rr?Vn*^»?>T'?<'^''>''*'<WPI<^ ll
r'l.«»»«W«M"«» ■'^,
o
T <
—r"i—i i i " x
- . u> o
* * _l
o o
^.3
- go . s: •z.
o
CO
lf»2 00
cop -
♦2
I—1
- a:
2 ^ o>h
1— I-
o
o - </>
-
et
-
a.
>-
- cvi<" ^V
,_. 1. -.-.. ,. Jill
o o o o
00 CM o
o
CNI/81)SS3cJlSddVM rsi
z——1 I 1 11 1
^
©
a
■"
N3 -
- «0
v OO
a •
^V IOZ LU ^
" X. coo —1 o
< :D
oa
v. \ o z «/)
z
QlD Ü z:
U. o
\ \N o:
ts N H
CO
Oh
h-
ZO
8R- *^\
oo
CO
SCALE
6 OZ
- - N,- a:o
< :
a. in
S CO
' CM o •
\\ o en
O o
1
o o
1 o
CO u> 04
■
(•NI/8-|)SS3HlSTliJ
IS
25
MM MaMgHüMgüM^
FIG. 4.11. BIAXIAL FABRIC TESTING MACHINE.
26
Small buckets were connected to the nylon lines passed
over the pulleys. Weights (small bags containing 1/2 or
1 lb of sand) were placed in the buckets. In general, the
load was increased in 2- or 3-lb increments on that arm of the
cross carrying the major load. The load on the other arm
was increased simultaneously in the proper ratio. Each 1-lb
load generates a fabric stress of 1/4 lb/in.* Tests were
conducted in which the ratio of warp-to-fill load was 1/6,
1/2, 1, 2, and 6> After applying each load increment, we
adjusted the eyebolts so that the arms of the cross would
remain straight and perpendicular.
The fabric deflection was measured in a 3-in. * 3-in.

section of fabric in the center of the cross. Small hard-
ened steel tabs with a cross scribed on each were glued to
the fabric in the test section as shown in Fig. 4.12. By
using a vernier caliper with pins glued to the jaws, we were
able to measure the distance between the scribed lines on
opposite tabs to about 1 mil."'" Knowing the gauge length be-
tween the tabs (3 in.) and the change in spacing between the
tabs as the load was applied, we were able to calculate the
strain in both the warp and fill directions.
4.2.2 Biaxial properties of the model fabric

To characterize the mechanical properties of 2.6-oz
cotton typewriter ribbon cloth, we tested it in the apparatus
described in the previous section at stress levels up to
4 lb/in. As discussed in Sees. 6 and 7, this load is some-
what above the maximum levels to be encountered in later
testing. In general, the first time the fabric was loaded,
its stress/strain behavir-1 was different from subsequent
loadings. As a result, we loaded the fabric a number of
times at each stress ratio until the deflections obtained
were reproducible (within 1 or 2 mils). The first time a
piece of fabric was mounted in the apparatus, ten repeated
loadings were required for reproducibility. Generally, after
•Tests with a sprii g scale showed that passing the line over
the pulley re suited in no change in applied load; i.e., the
load in the b uckets was the load applied to the fabric
There was ess entially no difference in reading if the load
was directly supported by the scale or if the load was at-
tached to a 1 ine and passed over the pulley and then the
line was atta ched to the scale (with the scale parallel to
the ground pi ane).
+
Measurements of this distance by two different technicians
were never mo re than 1 mil apart.
27
MfcMto. KMMMMW ^ ' -^^.,J:^.J,^V...,.^..-It.hl)t,:^...^'.:..,1:':...-... ■„.^■Û^ VJ.g .~~**

LIGHT, HARDENED „
STEEL TABS ~ 1/4x1/2
WITH SCRIBED CROSS
WARP
FILL
FIG. 4.12. FABRIC CROSS TEST SECTION
28
mummmmutmmauâmiimmjM ^t^^ggütttmmmmaaimm,,, „,.._....i,..^,,.^

MüHMH
this first cycling of the loading, it would only require a

few repeated loads (at most three) to get reproducible de-
flections at other stress ratios.
Data were obtained on three separate pieces of fabric,

all from the same bolt. Unfortunately, the fabric was sub-
ject to breaking at the higher loads. The breaking usually
occurred after repeated loads and usually at the center of
the cross. As a result, we do not have three complete sets
of data.
Figures 4.13 and 4.14 present the results of the bi-

axial fabric tests. Comparison between two specimens of
fabric are included when available. The solid lines are the
mathematical models of the stress/strain data (described in
the next section). The circles, triangles, and squares
(either open or closed)* refer to the first, second, and
third specimen of fabric, respectively. There is some in-
evitable scatter in the data, but, in general, the con-
sistency is satisfactory.
4.2.3 Mathematical model of the biaxial stress/strain data
In this section, we discuss a mathematical model of the

biaxial stress/strain properties of the fabric for use in
the computer code. The model is based simply on a curve
fitted to the data rather than on any theoretical model of
yarn interaction. A very simple model matches the data
quite well. We found that at a stress ratio of 1 (Tw/T„ = a
= 1), functions of the following form fit the data in Pigs.
4.13 and 4.14 quite well:
e C T
w (4.1)
w " W W
£
F ~ C T
F F (4,2)
*To prevent confusion the open symbols refer to the stress

ratio Tw/Tp = 2 in Fig. 4.13 and to the stress ratio Tw/Tp
1/6 in Fig. 4.14.
■
29
0BJ4 aumi^ânii. ..-a-.-.,..J,.^.,..,.„J..-'».«.i,„a..i.iT<ifliiiw<it

mm i ■mfiai. ■ —
I
o
—I
o
o
ca
a:
ui
o
M
O
I
10
CM
CM a:
ii
1^ CVJ o
0<D
> II
\r u. ^^*
»- <
>*
in*™ »■ «««« O0
UI
^S^ h-
• <m o
I—<
IJP< a:
co
<(/) oo
I _L
lO * to (SI
UJ
CNI/S81) SS3H1S ddVM
30
-■— ■""-"
I
o
-I
o
CO
ca
>—I
a:
ü. LÜ
Q.
>-
O
o
M
O
u>
V
I
F^
vo
ii
(M
l> 0<D
> II o
h^
^ <
$ *s
««ÜJ'T H* Q
ÖE^
a I-
s^ 00
UJ
«8z • «B H
o
-Jh-< i—i
<:..^ -*:,,~... v: :-*.,-.^,-.;—.

-- .......
where e is the strain, T the stress, C and P are constants,

and the subscripts W and F refer to the warp and fill direc-
tions, respectively. At other stress ratios, we found, quite
surprisingly, that Eqs. 4.1 and 4.2 would still be valid if
one simply multiplied them by a constant that depended only
on the stress ratio a. The resulting biaxial stress/strain
model that emerges is of the form
W
6u
•w = C„T„
WAW Gu(a)
"Wv (4.3)
e = C T Fa (a) (4.4)
F F F F '
where one must now select the four constants C Ws W» and

Pp and the two functions Gw(a) and Gp(a) by fitting Eqs. 4.3
and 4.4 to the data points in Figs. 4.13 and 4.14. Before
doing this, however, we need to examine a number of other
requirements that will constrain the allowable values of the
constants and the allowable forms of the functions.
The finite element computer code described in Sec. 2

requires at least piecewise linear material properties. In
particular, we require a stress/strain law of the form
= B
*W WWTW + B
WFTF
B
FWTW
+ B
FFTF (4.5)
where the B's are material constants that may be a function

of the fabric tensions T.*
*Note that we have neglected shear. In general, this ap-

proach is reasonable because uncoated fabrics are very weak
In shear. The computer code is capable of accepting some
shear stiffness in the fabric and, at present, a value of
5 lb/in. shear stiffness is used in the program. This
value has very little effect on the fabric deflections out
does help the code to converge. Using a shear stiffness of
zero causes considerable problems with convergence.
32
mmmm •—mttM 11
IM II MIM IM« 1 r-**0V*m * <M4 V»^
We can put our fabric model into this form on a piece-

wise linear basis by expanding Eqs. 4.3 and 4.4 in a Taylor
series about a given stress operating point (TWQ, Tpo). The
resulting equation takes on the form
9e 9e
e e T T + W T T + W (T T J (4,6)
w " tr wo» F<r W^ ^ w" wo^ WZ P~ PO *
3eT 9e
F
= e v(T T ) + - (Tl -T ) + £. (<v _T ) (4.7)
-p *WO- PO 3T,W W "WO' 3Tp ViF XF0;
Since ew(Tw0,TF0) and ep(TW0,TF0) are the warp and fill

strain, respectively, at the operating point (TW0,TF0), we
can rewrite the above equations as
9e 9e
Ae = W W AT
w w:W ATw + ifrF F • (4.8)
9e 9e
p T?
A£ = AT + AT (4.9)
F w:W w STT
F
F
Equations 4.8 and 4.9 agree with the form of Eq. 4.5, but
now the Ac's and AT's mean increments in strain and stress,
respectively, relative to the chosen operating point. Cal-
culating the derivatives called for in Eqs. 4.8 and 4.9, we
can express the fabric material properties called for in
Eq. 4.5 as
3G
wl (V1J
B.WW CW|PWGW(OO + «TSTK
9G
3G„1 (P
B
rP G (a) F1 T (P..-D
»*
FF - C
p[ F F " ° "Sij F
W W 2
B -CWTW a
WF 3a
33
«■HHHMI -J^..-^,:.. ,,,„., „,,.,.■, ■» ■— -- ----- .l^m.

1 3G
B = C T
V F (4.10)
FW F F 3a
The major difficulty with implementing this model in

the computer code is that most existing finite element
"solver-i" require symmetry in the material property matrix;
i.e., BWF = Bpw. As a result, we need to examine what con-
straints the requirement that Bwp ■ Epw places on Eqs. 4.3
and 4.4 and then try to fit those equations to the data in
Figs. 4.13 and 4.14. Symmetry in the material property ma-
trix requires that
( 9G (PF-D 3Gp
V^
c T w 2 C T (4.11)
'Vw o = p F ~"5a
As described above, the G's can be approximated as being

functions only of a. One means of satisfying this condition
in Eq. 4.11 is to take Pw = Pp = P, which gives
9G W C
F -(1+P) 3G
F
a (4.12)
3ct cw "So"
In earlier attempts at fitting Eqs. 4.3 and 4.4 to the

data in Figs. 4.13 and 4.14, it was found that Gp(a) was
approximately a linear function of a; i.e.,
Gp(a) = a(a-l) + 1 . (4.13)
where a is a constant obtained from linear regression. Sub-

stituting Eq. 4.13 into Eq. 4.12 and integrating the result,
one obtains
v.)-%»(if*(i-^)- (4.14)
Using a trial-and-error process, we have fitted the

data in Eqs. 4.3 and 4.4 to the data in Figs. 4.13 and 4.14
subject to the restrictions that PM
W = P^ and that G'W
w(a) and
34
MWHttl .AM* IM
Gp(a) have the mathematical form shown in Eqs. 4.13 and 4.14,
The resulting values of the constants (for Ty and Tp in lb/
in.) have been chosen to be
Cw = 3.5 • 10-3
C., = 4.6 • 10-3

F
P = 0.5
a = -1.54 ,
which gives
ew = 3.5 • IQ"3 Tj 0w(o)
£p = 4.6 • IQ"3 T* Qp(o) , (4.15)
where
Gw = -4.06 -r + 5.06
a
Gv = -1.54a + 2.55 . (4.16)
Equation 4.15 is plotted as the solid curves in Pigs.

4.13 and 4.14. The curves match the biaxial stress/strain
data quite well at most stress ratios. However, as is ap-
parent in Fig. 4.13, Eq. 4.15 overestimates the warp strain
for TTW,/T„F = 6 and underestimates these strains for TTTw/T„F =
1/2. In addition, from Fig. 4.14 one can also see that Eq.
4.15 tands to overestimate the fill strain (at least at low
stress) for T../T-,
W r
= 1/6. Using" Eqs. 4.15 and 4.16 to pre-
diet the stress/strain relationship for uniaxial stress
yields quite good agreement with measurement for uniaxial
stress in the fill direction; i.e., G„ = 2.55 for a = 0 and
e„r = 1.75 • 10~3T„r (see Fig. 4.15). For uniaxial stress in
35
MHMH yafefraaifl *.»*■-■ —

£ z
z i—<
<
< tc
CO
(0 co
co
-i LLI
_i cc
CO
I-H
00
X •
zo
O _l
zo
CO I-
des:
a. nz
srt-
o<
o 2:
in
(NI/91) N0ISN31 ddVM
36
the warp direction (a ■+ °°), Eqs. 4.15 and 4.16 predict exces-
sively high strains. However, if we limit C-w(ot) to the
largest value measured (i.e., 3-2 at a = 6), then quite good
agreement with uniaxial data is obtained, as shown in Fig.
4.15.
In general, though, with the above exceptions, Eq. 4.15
agrees well with the measured biaxial stress/strain data and
will constitute our mathematical model for the biaxial
stress/strain behavior of 2.6-oz cotton typewriter ribbon
cloth.«
Equation 4.16 requires some special treatment, because

it is clearly in error not only for predicting uniaxial warp
strain under stress in the warp direction but also for pre-
dicting the Poisson effect under uniaxial stress (i.e., warp
strain due to fill stress or fill strain due to warp stress).
Equation 4.16 as presently constituted would predict infinite
warp strain under uniaxial fill stress (Gw ->■-«> as a •*■ 0)
and infinite fill strain under uniaxial warp stress (G„ ■*■ -«
as a +•)• To avoid this problem, we have limited the value
of Gw and G„ in the computer code to those values obtained
for the range of a over which we have biaxial stress/strain
data (i.e., a = 1/6 ■*■ 6).** This approach results in
-6.2 < Gw < 3.2
-7 < Gw < 2.25 ,
which yields good agreement with uniaxial stress data.
«The mathematical model in Eqs. 4.15 and 4.16 was used in

the computer code to model the biaxial stress/strain tests
of typewriter ribbon cloth as a check on the code. The
resulting predictions of the biaxial stress/strain behavior
were in good agreement with the solid curves of Pigs. 4.14
and 4.15.
**In effect the computer code does not allow a to go out of
the range 1/6 to 6 when Gw or G- is being calculated.
37
-——■
- •■ - -i 111 l II IBM ttHihihlii itti i *.~*a~*-e~w . . . .
_ .. ■ ■ ■ . ■..- ■ •■ . ■■■• .
4.3 Fabric Force Sensor
4.3.1 General description
The sensor for measuring the tension forces in fabrics

was developed at BBN and was designed to measure those
forces independently of the fabric's properties. During
this program, the sensor was refined so that it could mea-
sure small loads. Such measurements are distinct from
strain gauge measurements in that one must know the usually
nonlinear biaxial stress/strain properties of the fabric to
translate strain measurements into tension forces. The
sensor is much stiffer than the fabric and, as a result,
carries all the load in the yarns to which it is attached.
This is analogous to a soft spring and a hard spring at-
tached in parallel; the hard spring carries all the load.
The disadvantage of the enlargement is that the sensor is a
rigid inclusion in the fabric which distorts the strain
field. This distortion can be minimized by making the sen-
sor as small as possible.
The sensor used in this program has two stainless steel

load links like those shown in Fig. 4.16. The load links
are fastened (one above and one below the fabric) to ~0.25-
in. diameter stainless steel buttons, which are glued to the
fabric (shown actual size in Pig. 4.17). The load links
consist of a measuring beam (see Fig. 4.16), to which a
strain gauge is attached, and flexures, which tend to de-
couple the measuring beam from all but axial deformations of
the load link. A second, dummy, strain gauge is attached to
a nondeforming surface of each load link for temperature
compensation. The four sensor strain gauges are then wired
together into a full bridge. The resulting instrument is a
rugged, reasonably stable device whose one disadvantage is
low sensitivity.
4.3.2 Calibration
A number of fabric force sensors have been constructed
from a variety of materials. In general, we have found the
invar sensors to be the most sensitive and the least subject
to drift. Figure 4.18 is a typical calibration curve show-
ing a sensitivity of 10 yV/2V/(lb/in.); the scatter is caused
primarily by drift at these extremely low voltages. The
most reliable means that we have found for calibrating these
sensors is to assemble the load links with no fabric between
them, apply a known load, and measure the output voltage of
the strain gauge bridge. Since the gauges are attached to
1/4-in.-diameter "buttons" on the fabric, a 1-lb load in the
38
mmm - ■»-•-- MMMUHH

MEASURING
BEAM
LOAD MEASURING DUMMY
STRAIN GAUGE TEMPERATURE
COMPENSATING
STRAIN GAUGE
1 O
0.15"
0.85"
FLEXURE
FIG. 4.16. LOAD LINK IN THE FABRIC FORCE SENSOR
39
■^ I,, ■ -- • ■*"—»
LOAO LINKS
STAINLESS STEEL FABR.C

BUTTON GLUED
TO FABRIC
FIG. 4.17. FABRIC FORCE SENSOR DRAWN ACTUAL SIZE,
HO
,..,,..■■,..,.... Hüjföto ^...., :,.,.•-;■,• ...,:..v, ■;:.■:■.

40
>
9 20
on
O
W
10^V/2V/LB/IN. ▲ FIRST LOADING
z
• SECOND LOADING
10 O THIRD LOADING _J
LB/IN.
FIG. 4.18. FABRIC FOP.CE GAUGE CALIBRATION CURVE.
41
■lÜiiifltflliiiilll'iliTil kimibUm^Miäi^.-.,^ "^^BBiAi'iifcrrtlii. m'^r'ii'iiiiiairitMtrii ii:-rt ;-i; mi^-*-^™-'^™-***^* ■ ^&^

calibration procedure corresponds to a fabric stress of
approximately 4 lb/in.* when the gauge is attached to the
fabric.
^Because of the low shear stiffness of most fabric, nearby

threads may be thought to act independently. Thus, the only
threads applying load to the sensor are those attached to
the buttons.
42
MMHÜ - -
5. MODEL TENT FRAMES
In this section we discuss the design of two model tent

frames based on the scaling laws of Sec. 3 and the model
fabric properties of Sec. 4. The two full-scale shelters to
be modeled are a slant-roof frame tent and an arch-roof
frame tent.
5.1 Slant Roof Frame
The slant-ro of frame model is based on, but by no means

is an exact scale model of, the tent maintenance shelter
(TMS). The frame ha? been chosen to deflect in approximately
the same ratio to th? model fabric deflections as the full-
scale frame does to the full-scale fabric. We have selected
a scale factor (b ased on a convenient model size) of 1/8.
All of the arches and frames in the full scale TMS are con-
structed from ste el box beams whose cross section is shown
in Fig. 5.1. The bending stiffness of our model beam should
be related to the bending stiffness of this box beam by (see
Sec. 3.2)
(El)
m m / m \
TETT FS K
(5.1)
FS \ LFS /
where E is Young's modulus, I is the bending moment of iner-

tia, K is the fabric uniaxial stiffness, L is a character-
istic length, the subscript m refers to the model tent, and
the subscript FS refers to the full-scale tent. The model
cloth is 2.6-oz cotton typewriter ribbon. From uniaxial
load tests on the model fabric and on the full-scale fabric
(9.85-oz cotton duck), we have shown the ratio < m /K-,
r o to be
1/5 (see Sec. 4, Fig. 4.10). This value, along with the
scale factor of 1/8, gives a ratio of bending stiffnesses
(El)
m 10 -it (5.2)
TETT FS = 3.9
Modeling the full-scale beam as a box beam, we find that

L
FS = 0.4 in.1* A 1/4-in. square aluminum beam, which very
closely satisfies the above ratio, is used in our model
frame. A drawing of the model is shown in Fig. 5.2. The
model is scaled to one section of the TMS (in reality there
43
"-"■■--
— 2 »
ll
0.074 — 2
_JGf_
FIG. 5.1. TENT MAINTENANCE SHELTER BEAM CROSS SECTION.
*-"-- •"-"■-- ;
--- -— t^mämatu^maamimkä, ■-■—■ -^■■■- -- — ma^^-Mt^.—- ■■■■■;—^
ts>
x
o
cc
a.
o.
<
o
t/1
ca
CVJ
45
■fei n^^M yM ■HHMHUMi -— -

are three arches) and models primarily the bending stiffness
of the arch and purlins and their relative lengths. Little
effort has been made to scale the interconnection between
beams* or the bending stiffness of the ridge pole. This
additional complexity is not justified as these details vary
greatly for different designs of slant roof tents. Note,
however, that the connection between the model frames and
the ground (a hinge joint, see Pig. 5.3), has been modeled,
because this is a fairly common geometry in existing Army
frame-supported tents.
5.2 Arch Roof Frame
The arch roof frame model is based on one section of

the Fritche Shelter (PrS) and models two arches with their
respective purlins. (The actual shelter contains up to
eight arches.) We again selected a scale factor of 1/8 to
give a convenient model size. Again, the model fabric is
2.6-oz cotton typewriter ribbon cloth. Since the full-scale
cloth is 9.85-oz cotton duck, the ratio Km/<ps remains 1/5,
and Eq. 5.2 applies to this frame also.
Unlike the TMS, the FrS has different cross sections
for the arcn beams and the purlins. These cross sections
are shown in Fig. 5.4. Since the arch cross section is not
square, we need to match the stiffness for bending about the
two possible axes of the beams. The purlins are square and,
hence, one bending stiffness is sufficient to characterize
them. The full-scale moments of inertia are given below.
Arches
J
xx 2.42 in."
I 0.63 in."
yy
Purl ins
r
xx=Iyy 0.30 in.*
*A11 connections between arches and purlins, except at the

ridge pole, are drilled and tapped (for ease of assembly
and disassembly) to model a rigid interconnection similar
to the full-scale interconnections.
46
II-
oo
Oii-ij
(Eo0
X§ffl
1
P"
LU
Q
O
O
Q
O
s:
<
CO
LT)
47
0.118 1.548
PURLINS
0.29
ARCHES
FIG. 5.4. FRITCHE SHELTER FRAME MEMBERS
48
^^
MttHMM ■^.a:-^...A.ji«.!—■,.■■■,—■.■■,■■; „^,;u..-iJM.„ ..,;::..'.^... .a .._
W- ""
Both purlins and arches are magnesium (E = 6 • 106 lb/

2
in. ). We find that a 3/l6-ln. square aluminum beam satis-
fies the criterion in Eq. 5.2 for the purlins and that 5/16
in. x 0.34 in. aluminum beam satisfies Eq. 5.2 for the
arches. The latter beam can be readily machined from a
standard 5/16 in. * 1/2 in. aluminum bar.
A drawing of the resulting model frame is shown in
Pig. 5.5. Again, the primary objective was to model lengths
and bending stiffnesses of the arches and th e purlins. For
this particular model frame, the interconnec tions between
the purlins and arches are quite representat ive of the full-
scale tent. In both cases, the purlins are connected to the
arcr s by "bolts" that run through the arche s. Again, the
hinge-like connection in the full-scale tent between the
arches and the ground is also used in the mo del. Photographs
of the model frames are shown in Pig. 5.6.
49
MR
o
UJ
I
sOOZin
,W
^.
n
CJ
CO
Q
O
s:
X
o
a.
<:
<:
o
00
i
LU
en
CD
50
■•-^•"-'""'îKliiiÜiiii-tiii
F
IG. 5 6. MODEL TENT FRAMES
51
6. PRELIMINARY VERIFICATION OF THE COMPUTER CODE
Before beginning detailed measurements on the model

tents, we decided to check the performance of the computer
code by comparing the code predictions with the results of
some simple tests. Three tests were performed: the stress
and deflection of a thin strip of fabric uniformly loaded
perpendicular to its plane, the deflections of the two tent
frames under a point load, and the stress and deflection of
a two-dimensional fabric membrane rigidly supported at its
boundaries. We next describe the results of those tests.
6.1 Fabric Strip Membrane
Measuring the deflection of a strip of fabric uniformly

loaded perpendicular to its plane is a particularly useful
problem, because it can be solved analytically (see the
Appendix) as well as experimentally; both calculated and
measured results can then be compared to the computer code.
6.1.1 Test set-up

A 24-in. long by 5-in. wide strip of 1,8-oz coated
dacron fabric* was mounted in a rigid wood frame attached to
the bed of a milling machine. By moving the bed up and down,
we were able to measure the deflection of a point on the
fabric strip from a fixed reference. Also, we could select
any point on the fabric for testing by moving the milling
machine bed from side to side and from front to back. We
applied the load at five discrete points along the 24-in,
length of the strip, as shown in Fig. 6.1. Each of the
loads was attached to a 5-in.-long rigid wooden strip, which
in turn rested on the fabric so as to distribute the load
evenly across the 5-in. fabric width. Deflections were mea-
sured near each of the loading points. Two configurations
;.ere considered: (1) an initially flat membrane and (2) a
membrane with some initial deflection (i.e., free hanging
deflection).
Uniaxial stress/strain tests of the 1.8-oz coated dacron

fabric were performed to provide the fabric data required
for the computer code. The results of these measurements
are shown in Fig. 6.2. At low load, the fabric behaves
approximately linearly with a stiffness of 50 lb/in.
*These tests were performed early in the program before the

2.6-oz cotton typewriter ribbon cloth was available.
MMMMI fa ffcrn'tt-lilirnriniit
MILLING
MACHINE
HEAD
FABRIC STRIP
RIGID
FRAME
rTTTTTTTTTTTTTTTl
MILLING MACHINE BED
FIG. 6.1. TEST CONFIGURATION,
53
- 11 mOiiteü*^^Wm>#i^«MM*^xi-^-^.-±,,îi_,\^.-.:^ ,-,J- ■ .„.:. fi. t

CNI/ai) N0ISN31
a?
54
ia*-i...-itMA-,t,J.,Jj,.-.
6.1.2 Computer program model
The computer program was exercised using the models of

the fabric membrane strip shown in Fig. 6.3. Two sets of
input were used: one for an initially flat membrane and the
second for a membrane with an Initial displacement. If the
x and y membrane curvatures are zero, the computer program
will predict, at the first iteration, very large displace-
ments. The program will then try to reach an equilibrium
state from this first iteration but will take a very long
time to converge. Therefore, in both models, a slight y
curvature was introduced, and in the initially flat membrane
model, a slight x curvature was used as shown in Pig. 6.3.
The x,y grid of the strip model is shown in Fig. 6.3,

where the numbers at the triangular vertices represent the
global nodal numbers and the circled numbers represent the
element numbers. Since the model and loading are symmetric,
only one-half the model had to be used. The loading is uni-
form and is applied in the z-direction. The boundary condi-
tions are that
• nodes 1, 2 and 3 are fully restrained,
• all nodes are restrained in the y-direction, and
• nodes 10, 11, and 12 are restrained in the x-direction

(symmetry condition).
The membrane material was taken to be Isotropie with a

fabric stiffness of 50 lb/in. Although the fabric is, in
fa^t, orthotropic, the fabric strip is in uniaxial strain;
c.s a result, the isotropic assumption and the initial curva-
i"'rcd are simply means of initializing the computer code so
that it will run efficiently. The results of the computer
calculations are described below.
6.1.3 Test results

The initially flat membrane deflections are shown in
Fig. 6.4. The theoretical results, calculated as described
in the Appendix with no initial tension, agree poorly with
the measured results. Note, however, that the computer re-
sults and theoretical results agree well, indicating that
the computer program is performing satisfactorily and that
the discrepancy between theory and measurement is some un-
controlled factor in the measurement. A possible explana-
tion for the discrepancy might be the tension required in
the fabric to make the strip initially flat. We did not
55
—^f*—» ^.^teMäHMü&MHxMh^ lit

H—LINE OF
! SYMMETRY
12
MEMBRANE GRID
4,5,6 7,8,9 10JJ.12 I

"*~ d=0.05"
/
i—*X
/
/
\ NODES 1,2,3
INITIALLY FLAT MEMBRANE
-NODES 1,2,3
INITIALLY DISPLACED MEMBRANE
FIG. 6.3. FABRIC STRIP COMPUTER MODEL
III Ill« ■ — ■■ -—
(S81) lNIOd 9NIQV01 HDV3 IV QVCTI
1—4
Ll-
57
measure the tension in the fabric for the initially flat
case, but if we make a guess of 0.14 lb/in. and use that
value in the theory of Sec. 2, we obtain the solid curve of
Fig. 6.4, which agrees quite well with the data points. The
deflection then appears to be quite sensitive to initial
tension and the large discrepancy between measurement and
theory appears to be due to neglect of that tension.
Also of interest is the deflection as a function of

position in the membrane. Plotting the measured deflection
divided by the measured center deflection in Fig. 6.5, we
find that theory, computer program, and data all agree quite
well.
The results of tests on an initially deflected membrane

are shown in Fig. 6.6. The membrane's initial shape was
nearly parabolic (in theory, it should be catenary) with a
center deflection of 0.93 in. Note that the loads applied
here are somewhat higher than in the undeflected membrane
tests. The agreement between theory (see the Appendix) and
measurement is not bad, although the measured points appear
to be consistently higher. This may be due partly to creep.
We noticed that after unloading the strip, the center deflec-
tion was -100 mils greater than before loading, but that this
deflection rapidly decreased to its original value with time.
Again, the computer calculations agree closely with the
simplified theory.
For the initially deflected membrane, we also measured

the stresses in the fabric strip at two locations — at 1% in.
from the support (position No. 1) and at 8 in. from the sup-
port (position No. 2). Both sensors were mounted in the
center of the width of the strip.
The comparison of predicted (see the Appendix) and

measured fabric stresses is shown in Fig. 6.7. Theoreti-
cally, the fabric tension should be the same at positions 1
and 2. In fact, the fabric was supported only over 4% in.
of its 5-in. width at the end supports. Since the fabric is
coated and can be expected to have some shear strength,
proper distribution of the load requires the tension per unit
width at the center of the span of the strip (position 2)
to be lower than that near the end supports. The dotted
line in Fig. 6.7 corrects for this "apparent" width change
by increasing the tension per unit width by a factor of
5/4.25 = 1.2.* The open circle (position 1) should coincide
*Note that the measured tension at position 1 (rear supports)
is a factor of -1.3 larger than the tension at position 2
(near the center of the strip span), which agrees fairly
well with this number.
58
MülMMlIMli
,8
HS
J 2
|N Q: j_
O -
Q. 00
o
z
O
UJ o
00
z
<
<c
Q oo
<
o
1—1
o
LU
«■
_J
LU
Q
a.
i—i
cc
1-
co
in
o VD
M31N33M/{X)Aft CD
U
59
ro
en
O
U-
o
—. ■z.
z
■■■
o
1—4
I—
z o
o UJ
1- u_
u UJ
-1 Q
b.
tu _J
O
cr
UJ z
H 1—1
Z
UJ z
Ü <c
u_
UJ
Q
Q-
i—i
00
o • • *
- o o o <5 ■
CD
(S81) iNIOd 9NI0V01 HDV3 IV 0VO1
60
to
z:
o
i—i
01
LU
I-
I Q-
i—•
a:
H-
</->
Ul
<
Q:
o
a:
CO
01
<
•z.
O
a:
o
o
oi
i—i
<t
CNI/81) QV01 "IVNU31X3 CO
■■-~""^-"~;~ -- magjägAgjlJljmäBUm^
with the dotted line and the solid circles should coincide
with the solid line. We find, in fact, that prediction and
measurement agree to within 10 to 20 percent.
6.2 Tent Frame Deflections
To check the ability of the computer code to deal with

the deflections of the model tent frames, we performed a
very simple test in which point loads were applied to both
the slant-roof and the arch frames and the deflections were
measured at the point of loading. The computer code was
then exercised, and its predictions were compared with the
measurements. In the following sections, we describe the
computer code and the deflection measurements.
6.2.1 Slant-roof frame computer model
The detailed dimensions of the slant-roof model frame
are described in Sec. 5. Figure 6.8 illustrates the com-
puter model used. The dots represent the nodes of the
structure and the adjoining number represents the node or
joint number. Node 7 defines the point of load application,
The boxed numbers are the beam element numbers.
Four computer models of the frame were set up to study

the effect of beam-end conditions. In Model 1, it is
assumed that the structure is completely built-in and that
all joints can support torsional and bending moments. Model
2 differs from Model 1 in that Model 2 is assumed to be pin-
jointed in the global z-direction at the base of the four
column supports. In the actual scaled tent model, the joints
that are screwed together are probably not 100$ efficient.*
Therefore, Model 3 was designed to simulate the frame be-
havior when screwed joints cannot support torsional or bend-
ing moments. Hence, beam elements 5 and 6 are truss elements
and beam elements 11 and 12 have been moment-end released at
nodes 5 and 6. Model 4 was developed to simulate a very
flexible structure. It was assumed that the base of the
columns were built-in and that all other beam joints were
ball joints. This model can only support a global x and y
load. The material properties for all models were E = 10
psi and v = 0.333. All beams in the frame have the same
relevant geometric properties:
*This implies that the joints are not rigid; i.e., the angle
between adjoining beams is not constant under load.
62
jitUi^^^ .,,i-ii..-..J.1i^n.-.J.fc,-J.>,^.;j|
—■■ ■■ ■ ■
cross-sectional area 0.0625 in."
cross-sectional dimension 0.25 in. x 0.25 in.
bending moment of inertia 3.26 x io-" in.1*
6.2.2 Arch-roof frame computer model
The detailed dimensions of the arch-roof model frame

are given in Sec. 5. Figure 6.9 illustrates the computer
model used to predict the frame deflections. Note that we
can make use of symmetry and thus model only one half of the
frame. The numbers at each node are the node designations,
and the numbers in boxes at each beam represent the beam
designation.
We consider three different models to study the ef-
fects of joint efficiencies. Model 1 has all joints built
in. Model 2 is similar to Model 1 except that joints 1 and
16 are pin-released about the global y-axis; i.e., these
joints cannot support a bending moment about the y-axis.
Model 3 is similar to Model 2 except that node 8 is "ball-
jointed"; i.e., it cannot support a moment in the x, y, or
z direction.
The material properties are the same as for the slant

roof model. Different geometric properties are required for
the arches and purlins:
arch cross-sectional areas .045 in.2

purlin cross-sectional areas .035 in.2
arch cross-sectional dimensions 0.34 in. x 0.187 in.
purlin cross-sectional dimensions 0.187 in. x 0.187 in.
arch bending moment of inertia 6.12 x 10"" in." and
1.87 x 10-" in."
purlin bending moment of inertia 1.03 10 -"♦ in.
The thinnest dimension of the arch cross section is in the
y-direction. For this case only two loads were applied in
the computer model (in the x- and z-direction), as shown in
Fig. 6.9.
6.2.3 Model tests
The tests performed on the scale model frames were
quite simple. A weight was hung by a wire from the point on
the model frames corresponding to the loading point in the
64
MuaauuwUMIMUU ■MHM _.j

FIG. 6.9. ARCH ROOF FRAME COMPUTER MODEL
65
- IIMI ■■■---"■■■ f-.^.^±.„^n .-.,„.^t ,,..■■■.„: V-,,.,-,L :,--■ .,■■ .. : ■ r: -■■-. - ..■:v^Ja
computer calculations. By changing the orientation of the
frame, we could vary the direction of application of the
load to correspond with that in the computer simulation. A
scale was used to measure the defle ctions relative to a fixed
reference in the direction of load application. Loads up to
6 lb were applied to the frames and since the deflections
were linearly related to the loads, all deflections were
normalized to a 1-lb load. The res ults normalized to a 1-lb
load for the slant roof frame in th e x, y, and z directions
of Pig. 6.8 were
6 = 22 mils
x
6 = -25 mils
y
<5 z = 100 mils
The same results for the arch roof frame were
6 A = 22 mils
6 = 20 mils
y
Comparison with computer predictions is shown in Figs. 6.10

and 6.11. Although we tried to make the feet of the frames
appear pinned, the x-direction loads in Pigs. 6. 10 and 6.11
indicate that they are somewhere between pinned and built-in,
The y-direction deflections in Fig. 6.10 and the z-direction
deflections in Pig. 6.11 indicate that the arch- to-purlin
connections for both frames are somewhere betwee n rigid and
pinned. At the present time, the computer code has no pro-
vision for dealing with this problem. Clearly, then an
extension of the present code is needed to allow for "joint
efficiencies",* since real tent frames will prob ably present
similar modeling difficulties.
*A joint with an efficiency of 1 is rigid; i.e., a rotation

of one beam at the joint produces an equal rotation in the
adjoining beam at that joint. A joint with an efficiency
of zero is pinned and can transmit no moment.
66
Mm M— mm .„■^.-im.to^D ami
X LOAD YLOAD Z LOAD
100
!
V)
=!80
2
UJ
0 60
O
2 40
g
i-
o
kJ 20h
A MEASUREMENT
• PREDICTION
FIG. 6.10. SLANT ROOF FRAME DEFLECTIONS.
67
■ ■
X LOAD ZLOAD
«3
loo r
</> (»2
!80
0>
£60
o
z
z 40-
g
H
O
u- 20
▲ MEASUREMENT
0 PREDICTION
FIG. 6.11.. ARCH ROOF FRAME DEFLECTION.
68
6.3 Two-Dimensional Fabric Membrane
To assess the effectiveness o f the computer code in

dealing with the mechanics of the fabric (including the non-
linear biaxial stress/strain formu las of Sec. 4.2), we de-
vised a test in which a rectangle of the 2.6-oz cotton type-
writer ribbon cloth was suspended from a rigid frame. The
fabric was then loaded with a dist ributed load perpendicular
to its plane, and the resulting st resses and deflections
were compared with the computer co de predictions. In this
section, we discuss the results of that comparison.
6.3.1 Testing procedure
The test used a membrane of 2.6-oz typewriter ribbon

cloth suspended from a rigid rectangular (17 in. * 20 in.)
frame constructed from 2 x ^ lumber as shown in Fig. 6.12.
To clamp the fabric to the frame, we screwed a wooden strip
to the top of the frame at all four sides and compressed the
fabric between the strip and two lengths of 1/8-in. drill
rod laid along the full length of the frame on all four
sides.
We applied load by laying bags containing lead shot on

the fabric. The bags were approximately 4 in. x 20 in. long
and were sewn together so that they contained many small
compartments for maintaining an even distribution of the shot
along the length of the bag. Each bag weighed -2.5 lb; four
bags were required to cover the entire surface of the fabric.
One, two, am" three layers of bags were used, resulting in
total loads of 10, 20, and 30 lb (a maximum load of 13 lb/
ft2).
The fabric w as suspended from the frame in such a way
that there was an unloaded center deflection of ~1.3 in.
The computer code requires the ini tial (unloaded) shape of
the membrane in o rder to predict i ts deflection under load.
Before measuring this quantity, we fully loaded the fabric a
number of times t o obtain a repeat able initial shape. This
initial shape was measured with a depth gauge from an alumi-
num rod (1 in. sq ) laid across th? frame. We defined the
initial deflectio n by measuring th e depth from the rod to
the fabric at poi nts interior to t he fabric and appropriately
subtracting the d epth from the rod to the fabric where the
fabric joins the frame (see Sec. 6 3.2).
The same procedure was used for measuring the deflec-

tion under load. Deflections were recorded at each of the
three load levels (10, 20, and 30 lb). In addition, the
69
-WOOD
T FRAME
*2
8^_L
—T W
STRESS GAUGES
17 __Ji wk
"1E 2
10
•20-
2x4
FRAME
FABRIC
DRILL ROD FABRIC
FIG. 6.12. FABRi: FRAME TEST ARRANGEMENT,
70
"^m»ep™™<
stress gauges described in Sec. 4.3 were attached to the
fabric at the 4 locations shown in Fig. 6.12. The gauges
were oriented in the warp direction (measuring warp stress
only) and the lead shot bags were laid between them. The
computer model and the results of these measurements are
discussed below.
6.3.2 Computer model
The grid for the finite element computer code is shown

in Fig. 6.13. The numbers in circles at the center of each
triangle designate the element number. The numbers at the
junction of the triangles designate the node numbers. Note
that only one-quarter of the membrane has been modeled. The
symmetry of the initial membrane shape during the testing
described above makes this simplification possible. The
warp direction of the fabric corresponds to the x a::is in
Fig. 6.13. The x and y axes define the rigid boundary of
the membrane where all deflections are defined as zero. At
the other two boundaries of the membrane, symmetry deter-
mines the boundary conditions; i.e., for
nodes 2, 3, 4 — the y deflection is zero,
nodes 10, 15, 20 — the x deflection is zero,
node 5 — both x and y deflections are zero.

In order to apply the distributed load to the membrane,
the computer applies point loads at each node interior to
the membrane, No loads are applied to the nodes at the
rigid boundary. If the entire membrane had been modeled,
there would have been 49 interior nodes. As a result, at
the full load of 30 lb, 1/49 * 30 is applied to each of
nodes 2, 3, 4, 5, 7, ••• 18, 19, 20.
One difficulty was encountered in interpreting the com-
puter predictions. The computer code outputs deflections at
°ach node. During the testing, deflections were measured at
positions corresponding to nodes 3, 5, and 15. This arrange-
ment makes comparison of measurements and predictions quite
convenient. Stresses are referred to an element, and the
membrane element is such that the stress throughout the ele-
ment is constant. Thus, there can be step changes in stress
at each element boundary. The stress gauges described in
the previous section were attached to the fabric at positions
that correspond to nodes 3, 5, 13, and 15. Relating these
measurements to computer predictions requires averaging the
stress in each computer element contiguous to those nodes.
71
-LINES
" OF
, SYMMETRY
*X
RIGID BOUNDARY
. 6.13. COMPUTER MODEL OF THE 2-D FABRIC MEMBRANE
72
UJII1 .IIILIIIl JI^WI^WW

Discrepancies can result, if there are large changes in
stress across an element boundary.
A further difficulty wes encountered in inputting the
initial deflections of the membrane. With the measurements
used to define the initial shape, the computer code had
difficulty in converging. By smoothing this data, we were
able to obtain good convergence, although the error in the
initial deflections of the membrane introduced by this pro-
cedure resulted in deflection errors under load. The initial
shape of the membrane is given in Table 6.1, where the nodes
refer to Pig. 6.13. The measured data from which these
values were derived are given in Table 6.2. All the z posi-
tions are referenced to node 1.
6.3.3 Results
In general, the deflections measured during the tests

described above agree well with computer code predictions,
as is shown in Figs. 6.14 and 6.15. For the three positions
shown in Fig. 6.14, the computer code consistently under-
estimates the measured deflections — primarily because of
the smoothing procedure applied to the initial deflections.
Figure 6.15 indicates that the code adequately predicts the
shape of the fabric membrane at the centerline of the fabric.
The predicition of stresses is considerably less reli-

able. The measured and predicted stresses at node 5 in the
center of the membrane agree quite well, but, unfortunately,
this is the only place where the agreement is good. It is
apparent that after smoothing the Initial unloaded shape of
the membrane, we should have expected this result., For ex-
ample, suppose that in the unloaded state there is a dimple
in the fabric and, unknowingly, we place a stress sensor in
this dimple. Since we do not place the lead shot-filled
bags directly on top of the sensor, it is conceivable that
the dimple will not have much load applied to it: Yarns
surrounding the dimple must be stretched cor.3iderably before
the yarns running through the dimple can develop any stress.
It is therefore to be expected that under such conditions
our sensor will (initially at least) measure very little
stress.
In setting up the computer program, we used a fairly
coarse grid and smoothed measurements of the initial un-
loaded shape of the membrane at the nodes of that grid. The
coarseness of the grid and the smoothing process would prob-
ably prevent any such dimple being entered into the computer
calculations, thereby making it impossible for the code to
73
——
o o o o o o o o o o o o o o o o o oo o o o o o o
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
0)
-t-> o o o o o o o o o o o o o o o o o o o o o o o o o
•0 o mo >-ON O O O 0\H o in CVJ co o o o o mco o o o o o
c Nl OCO CM CM CM OCO H H CM O VO OtONO o mvovovo o o o o o
*r-
-a O H r-f rH O H H rH O H o o o o o o
i- 1 1 1 1 1 1 1 1 1 1 1 1 lilt
o
o
o
in in in in in VO vo vo vo vo CO CO CO CO CO 0\ <J\ ON CT\ CT\ o o o o o
c c—1~~ e^-1^- e~- in in in in in m co oo oo co rH H H H H o o o o o
•r- >- 00 CO CO CO CO vo vo vo vo vo CM CM CM CM CM o o o o o
O
Ou CO CO CO CO CO vo vo vo vo vo JT^r^r ^--3- CM CM CM CM CM o o o o o
(0
! O o o o o o o o o o o o o o o o O o O o o o o o o o
z o o o o o o o o o o o o o o o O O O O O o o o o o
X o ino mo o ino mo o mo mo o mo mo o mo mo
o CM int--o oojinso owinso O CM 1AKO oaiü"\NO
H rH H
M
M rH rH rH rH 'H rH rH rH rH rH rH rH rH rH rH H H H H H rH rl rH rH rH
V)
0)
o >-
o >- rH rH rH rH rH rH H H H H rH rH rH rH rH H H H H H :~\ rH rH rH rH
r-
< c
o
•M X
•r- X rH rH rH rH rH H H H H H rH rH rH rH rH rH H H H H rH rH rH rH rH
c
o
o
M H O O O O H O O O O rH O O O O H O O O O rH rH rH rH rH
>>
i-
O (0
a. ■a
c >- rH rH rH rH rH H O O O O H O O O O rH O O O O r~i rH r*H rH PH
O
Q co
O
z X H O O O H H O O O H H O O O H H O O O H H H r-> H H
Node
No.
vo H CM ro.^- in VO ^co o\o H CMmzr m VO !^-CO 0\O H CM m^r m

H H H H H H H rH rH rH (V CM CM CM CM CM
LU
—I
co
<c
r-
74
TABLE 6.2. MEASURED UNLOADED INITIAL FABRIC SHAPE.
Node No. z Location of Node (in.)
1 0
2 -0.80
3 -1.24
4 -1.27
5 -1.29
6 +0.01
7 -0.84
8 -1.19
9 -1.20
10 -1.21
11 +0.06
12 -0.80
13 -1.03
14 -0.98
15 -1.00
16 -0.01
17 -O.65
18 -0.67
19 -0.61
20 -0.58
21 -
22 +0.01
23 -0.01
24 0
25 -0.01
75
NODE 15
1 -I
10 20 30
u 2
I
u
z NODE 5
z
o
h-
U
UJ
_J
UJ
o
-J
NJ 10 20 30
2r
NODE 3
±
10 20 30
TOTAL LOAD (LBS)
▲ MEASUREMEN "
• PREDICTED
FIG. 6.14. COMPARISON OF PREDICTED AND MEASURED DEFLECTIONS
76
^^•jttgjtniiamm^ t titüiinfcM m ■ ■-■■i. 'nn .vn.fc-t——^— r , „ inînfrt jjliiitfii1lJlfll^^^-^--'"teMaw....:»c..^.W; .--■,:■&

o
■z.
o
—I
<=c
i/o
■z.
o
LU
Q
I—I ^
Q CO
LU
I en »
CL CM
Q «
Z r—
<
LU
O Q
LU O
Q: Z
:D
00 <_>
< 1-1
Lu Q:
<
u. u.
o
LU
z: :r
o h-
oo
>—l LL.
ce o
<
Q- LU
S z:
o <—
O _l
( NIS N0llD3~ld3Q Z
C3
77
?^îl
predict local regions of low stress in the dimple or possible
high stress on the periphery of the dimple. These arguments
have considerable implications for using computer codes for
the design of frame-supported tents. Clearly, a detailed
knowledge of the unloaded shape of the fabric is necessary
if stress concentrations in the fabric are to be predicted.
Since this detailed shape is extremely difficult, if not im-
possible, to predict a priori, some sort of a safety factor
must be introduced for sizing a fabric based on the computer
predicted stresses.
Finally, a brief word about the stress measurements at

node 13 is in order, Note that Fig. 6.16 says that very low
(essentially zero) st resses were measured there. If we
carefully examine the computer predictions, we find (see
Fig. 6.13) that the s tresses in elements 21, 22, 23, and 24
(4 membrane elements in a row beside node 13) are also very
low. This result pro vides considerable confidence in the
computer code. By th e same token, however, nowhere does the
computer code predict the high stresses measured at node 15.
78
IM mmtm MM MSMMmmmVfH
■■■■■-■'
NODE 13
u
CD
VERY LOW
MEASURED
STRESS
10 20
TOTAL LOAD (LBS)
▲ MEASUREMENTS
• PREDICTIONS
FIG. 6.16. COMPARISON OF MEASURED AND PREDICTED STRESSES
79
"-*■—-a--^—— ■■-—■- - mmjmmmmttm eagjdjsaaBMimM . ■.-, am I'I niliAiMl

7. FINAL VERIFICATION OF THE COMPUTER CODE
In this section, we discuss a number of deflection mea-

surements made on the frame and fabric of the model frame-
supported tents described in previous sections and compare
those measurements with computer code predictions. Initially,
we planned to make detailed measurements of stress in both
the frame and fabric, but the results described in the pre-
vious section discouraged us from this approach. In par-
ticular, the need for the inclusion of joint efficiencies
in the computer code to obtain adequate agreement of mea-
sured and predicted frame deflections indicated that detailed
measurements are not justified at this time.
7.1 Computer Models
The grids for the computer models of the slant-roof and
arch-roof test models are shown in Figs. 7.1 and 7.2, re-
spectively. Because the tents are symmetrical, only one-
half of each was modeled. The numbers at each dot are the
node designations, the numbers in the boxes are the beam
element designations, and the numbers in the circles are
the membrane (fabric) element designations. In both models,
fabric was attached to only a small portion of the frame.
Only one-half of the roof of the slant-roof frame was covered
with fabric, and only about 45° of the arch of the arch-
roof frame. In both cases, the fabric was assumed to be
attached to the centroids of the beams. In the tests
described below, the loading, which is applied vertically
(z direction) and is distributed over the fabric, is modeled
as a number of equal discrete vertical loads applied to all
fabric nodes.
In the slant-roof frame model, all joints are rigid with

the exceptio n of the feet of the model, nodes 18 and 19;
these are al lowed to freely rotate about the Y axis (pinned
joints). Al so the beam 4 connection to node 3 is taken as
pinned about the Y axis. The arch-roof frame model is simi-
lar, with pi nned joints about the Y axis at nodes 1 and 21.
The geometri c properties of the beams in both models are the
same as used in Sec. 6.2. The fabric is modeled as having
the biaxial stress/strain equation of state developed in
Sec. 4.2.
7.2 Model Tent Tests
To test the model tents, we attached fabric to the two
model frames, as described in Sec. 5. (Note that a number
of purlins were removed from the arch-roof frame, Fig. 7.2,
80
...... ii..,.„j
WS
FIG. 7.1. SLANT ROOF TENT COMPUTER MODEL
81
■ !. .«M—a—t ■JflliaäkmMfctaB:..^.;. ,.■;■■. -,...■. ,,- ..„;,. ■■ v, M^.a.^J, ■■-.^.-^-.j^â

• m»^mtîj.^m.wm9mt^mjm,K,t mm w^.«j»#,w-yn»fipj -'■«y '4n^.ni(iifl) »!»»■»«" flJV*H^wsj.' -tun*?""
FIG. 7.2. ARCH ROOF FRAME COMPUTER MODEL
82
!*«£&; J« va-.ii «i i*v*v.tiirtiÄü ^-.
to make it easier to measure frame and fabric deflections.)
The computer models illustrated Figs. 7.1* and 7.2 show
where the fabric was attached to each model. Each of the
four edges of the fabric was folded over and sewn to form a
narrow tube. Notches approximately 1 in. square were then
cut out of the corners of the fabric to facilitate attach-
ment at the corners and to minimize wrinkling. By disassemb-
ling the frame, we could insert the appropriate beams of the
frame into these tubes and then reassemble the frame. This
method of attaching fabric to frame resulted in the fabric
loading the beams approximately through their centroids.
The fabric in both frames was sized so that it would
lie slack in the frame. As a result, it was necessary, for
input to the computer program, to measure the position of
the fabric relative to the frame, as described in Sec. 6.3.
We determined the initial unloaded fabric position at each
of the fabric grid points shown in Figs. 7.1 and 7.2 after
loading the fabric a number of times to take out any slack.
The loading was accomplished, as in Sec. 6.3, by laying long
bags filled with lead shot on the fabric. The slant-roof
fabric initial unloaded position at, say, node 7 (see Fig.
7.1) was measured by laying a straight edge on the frame
along the line 6-7-8 and measuring, with a depth gauge at
node 7, perpendicular to the plane of the frame (the plane
defined by nodes 5-3-15). These measurements are given in
Table 7.1. A simple computer program transformed the mea-
surements to the global coordinates of Fig. 7.1. Similar
measurements were performed on the arch-roof frame, except
that we measured the distance to the fabric at the node of
interest in the radial direction (toward the center of the
circle that forms the arch), using a straight edge resting
on the two arches of the frame. These measurements are
given in Table 7.2. Again, a simple computer program trans-
formed these measurements to the global coordinates of Fig.
;-2. These data were smoothed to facilitate convergence of
;
computer code. The initial fabric position for the two
le
nts used in the code is given in Tables 7-3 and 7.4.
The frames were then each placed on the bed of a raili-
ng machine and the fabric was loaded. By attaching a
Dni f to the head of the milling machine, placing the
loaSa and°n the frame or fabric with no load, applying the
> moving the bed of the milling machine until the
7l1 Shows loads

loart
da
applied to nodes 3 through 17. The
at each node is the total load divided by 25.
83
■"T-wri^'r^îr-«: - ^W-^^rwr^^ä«-^.«^
nppHHppnppsc^^^^P^SF
TABLE 7.1. MEASURED UNLOADED INITIAL FABRIC SHAPE IN THE

SLANT-ROOF FRAME.
Distance to the Fabric in Inches

Perpendicular to the Plane
Node Formed by Nodes (5-3-15) in
No. Fig. 7.1
7 0.52
8 0.51
10 0.73
11 0.69
13 0.62
14 0.64
TABLE 7.2 MEASURED UNLOADED INITIAL FABRIC SHAPE IN THE

ARCH-ROOF FRAME.
Distance to the Fabric in Inches in the

Node Radial Direction From the Cylindrical
No. Surface Formed by the Two Arches
84
HMtfi w. ■...- . .L-. .,.,-. .... ■ A v

■r^^-y^P^K^r''^^ '" ■* ■T-yir&r?::im+-**i vrrrr-°^-r~< w.ii*<?;->;^ j. :■. t,.-.-.™^ -*w :--T '.', ,>-*"r7 ■»!'!>'
oo o oo ooo oo oooo o oo oo o
1 1 ! 1 1 1 1 1 I 1 1 1 1 1 1 i 1 1 1 1
V)
+» oooo o oo\roHO (\IHSCJiO o o oo o

<o iriLTvininiA vo ON^r t^-t»- CM CO O00 IT\ inino o o
c M t— t~- C— IS- t— c—:=r .=r t»-m CO C*-- CO >- t-t^-o o o
t"— C-- 0\ <J\ <J\ vovovo coco CO o o o t— t-C—O Q O
j s- r-\r-{r^ r-i rt rt t-i <-\
o
o
o
4J oooo o ooo oo O OO OO oo oo o
c o o o mo omooir» ooinoo tno o o o
■r- >- O LHO C-Ln inoMno S1AO O O
o
a. o o\o^r o\ O^CTiOJ- C\ O^- CT\ O ^r a\o o o
< (0
a:
o o o o o o rori H t--H OOOOOHO oo oo o
z o o o o o ViD POOO C\J t^- CM c\tr>^3" o o o oo o
>e ooininin COVO IACM00 cointviaio o o o o in
o
o O O CO CO CO VOVOVO o CT\ ONOoroonc-- Nsosm
cc H H H rlHriNH rH CM CM CM CM CM CM CM H
i
l-
z
<
—i
O H O O H O H H O H H OH HO O H H H H
s
-
o >- O O O O O O H H O H H OH HO O O H H H
c
o
CO •r*
o 4-> X
•r- X O H O O rH O H H O H H OH H O OHHrlH
T3
C
o
cc o
GO M O O O O O O O O O O O OOOO O O H H rH
>»
J-
19
•o
c >- O H O C rH O O H O O H OO HO O H H H H
3
O
CQ
X O O O O O O O O O O O O OOO O O H rH H
CO
Node
No.
H c\j m-=r in vo c--co cr\o rH CM oo-=r in vo t>-ao o\o

H H H H H H riHHHW
CO
<
85
- *-• i mm Mnüg -—- - IM IHBHHUWtt ,:._r .-..-iÂS

«;•£>*-■;';■ -»■'---; ' :~V .- ■■■_--■ ■ ■- :—.■—.- - ... ■■■■■>-.:,.i*;-—-r ■>■■- ■ ;-,-. -.-"
OOOCOrlHO\OOOOHO\0-J OWO mClinOVDOOHHHHOOO oo o

oininnooo>c\i(\i wojvofl- e*-vo ONOO^Hoo^ôNOOOomifMrioo
Nodal P oint Coor dinates
o rvj CVJ ^r o o o^.=r -=r.=r-=rovo.3-mo\ in.=r -=rc\iooômoooo-3rc\ic\ioo

Z
OVOVOrHVOVOO«. HrHHHHOOO V\ ON C7\ ONCO t>- >-C- VO VO VO VO H«)\OQO

HHHHN(M<MNN(M(M(MHHHriHHHHHHHriH
oooooooooooooooooooooooooooooooo
Y
OOVOOOVOOOC\l^rVOOC\JJJ-VOOCM^J-VOOC\l^-<J30rVzTVOOOVOOO
oooino\o\<ôoootoaj(Mo\Ho\(Mcoinro moo HHHrtinoooo

oooH-=r-=rinc—N-t»-t*- invo CM o>co coc\joovor-it-o\a\OACr\c\i.=r:3-.=rt--
ocoo\inuiSHHHH-3-ooroMirimmmô o>oo t— t— c~-t-.=i- on ro on H
X
ooo on coco in in in inco OOOOOOHHHH^ÔI onvo vovovoc»oooin

HHHHHHHHC\jc\jajrvjojojnjojrvi'Aj'\jojc\irornrnrH
o
o
on
ZZ
t
X HOHOOHOOOOHOHHHOHHHOHrHHOOOHOOHHH
<J
0£
n Codes
<c
YY
HOOOOOOOOOOOHHHOHHHOHrHHOOOOOOOHH
o
•r—
•(- X HOHOOHOOOOHOr-iHHOHHHOHHrHOOOHOOHHH
CO •o
o
a.
c
o
o
o M HOOOOOOOOOOOOOOOOOOOOOOOOOOOOOi-IH
i—i >>
cc
CO (0
C >■ HOHOOHOOOOHOOOrHOOOHOOOrHOOOHOOHHH
o
CQ
X HOOOOOOOOOOOOOOOOOOOOOOOOOOOOOHH
Node
No.
H c\j o-).=r invo t- oo a\o H c\j m^j- invo t~-oo WOHOI m^r invo t-oo O\OHOI
HHHHHrHHHrHHrvirvjojc\i(\jcoc\jrvjojojmrnm
CO
86
1 MMMM -■ --'—■■
i^'V*^»r-,*?s*yr«s?K^^
pointer touched the point of interest again, we could deter-

mine the X-Y-Z deflection of a given point unde?- load.
Each of the bags of lead shot weighed ~?.5 lb; four
were required to cover the surface of the slant roof frame
fabric, giving 10 lb per layer. At most, three layers were
used, for a total maximum load of 30 lb or approximately
16.5 lb/ft2.* Using the scaling laws of Sec. 3, we find
that this load corresponds to a full-scale load of
_^m Ss D
FS L
PS K
m m
Since the model is 1/8 scale and since the model fabric is
approximately 1/5 as stiff as cotton duck (see Sec. 4.1),
P
FS " 1/8 16.5 - 10 lb/ft:
For the arch-roof frame, three bags will cover the

fabric surface, giving 7.5 lb or "7.5 lb/ft2 per layer of
bags. Two layers of bags then correspond to a full-scale
load of
P
PS - 1/8 15 - 9.4 lb/ft:
7.3 Results
Computer model predictions of frame and fabric deflec-
tion of the slant-roof tent frame are compared with measure-
ments on the model tents in Figs. 7.3 to 7.10. Figures 7.3
to 7.6 show deflections of the fabric under the three load
increments in the X and Z directions. In general, agreement
of measurement and prediction is good, with the exceptions
of the-X deflection at node 8 in Fig. 7.3 and the Z deflec-
tion at node 10 in Fig. 7.4. We believe that these dis-
crepancies would be reduced (1) if a finer mesh were used in
the computer code, especially for modeling the fabric, and
(2) if a more accurate means of measuring the initial
*Based on the projected area of the slant roof on the X-Y

plane.
87
MMMHMMM 1—11
!i..'Ty?V™ .-■■■T-:-Ti-...--^,'-- -■>..:•■■ ' •;vn >7 .■»■«■?■;-. ■ :»..-JS ;
1 1 1
A MEASURED X DEFLECTION
A MEASURED Z DEFLECTION
PREDICTED X DEFLECTION
PREDICTED Z DEFLECTION
3 -
2
O
5*
LI
_l
U.
UJ
Q
A^-" *
1 — —
//A A
A
/^ 1 1 1
10 20 30 40
LOAD(LBS)
FIG. 7.3. COMPARISON OF MEASURED AND PREDICTED DEFLECTIONS

IN THE SLANT-ROOF TENT AT NODE 8.
88
«■te^riWi —•-■--■■■--■-. I-II in yfl
;.?•;..<-:■, :'■■ ..'.!•■ -- * -,-»".- *•*« -j - —>.. -*- - - l<3$£>p^r -",l-r=-~,':-:--.VV-^'.^;-, ;■■>,.?:,' ;j!£ ."■)'.•-.-: ■■.;-J:^-->..: ■,•7 .i;^.r—r--.-.',,-;r'■""J V-"""/' ."/(-■.■■^:»:rr-r'-^-':'--,V>;lv^^-^'--7::\v-JtT;-\-'- ■ -■..■.■.^*;:v;'--'r .^.r;.., . ---■- ■■-^
1 1 1
▲ MEASURED X DEFLECTION
5 -
6t
LÜ
-J
Ü.
tu
Q
A
1 - —
/A— ▲
A
yy
10 20 30 40
LOAD(LBS)

89
w^* ——-•■— - ■ ---■ UH ■ -' '-- :

— -■--'-■-+ - -i .TrTiiiTT 'i nl-.riMHiil
^^^^^^^-^•■-■'^.^"■■-':^rr^- :[~ -^■"-■•"^r^"';?-'** ^-.■-■r^r^ ■ ■•-...; J^V-ys-~,i^-^y.'.-i.-t. -;»■-»,-?■■-'■•-.'>f.iy ■-p *•-..■•■■..,\-/;i.):-■;■ ■-,^'- <■ ■■v'-.:.3vi;.;>"-'.-;fr i'r;-?~: _ ; -..T>'-j;*n"^r. : V'-: ■:■■■:.:_;■.■:' i- ":».' ■■-,-i--?:>':**'.;.■. .-.-y'Li".-":. *■ *;'■■' :^5
20 30 40
LOAD(LBS)

90
JMHtt ■Mb ■ iiiMiiiiriM—Tl—]

■ r.-iV.--?. ■ «o.:-.*w ■ -vr*T?i^\
1 1 1
▲ MEASURED X OEFLECTION
Z
O
O2 — —
LÜ
_J
U.
w
Q
£ *' "^"
/ s
/s
/s
/s
r i 1 1
20 30 40
LOAD(LBS)

91
Uftd ^M
jsppp^sr'"^ ■ **v*<w. ^^Bfj^ppgn^^ -■> *vi ^^pr^p*^^ /^m^w^F'";^^r,r:^r;''"^'';:^: -vv-" - T" ■'■"'■ ■■T^^^^^>..ffl«R
A MEASURED
PREDICTED
Z
ÜJ
5
LLI
Ü
<
Q.
(/)
FIG, 7.7. Z DEFLECTION IN THE SLANT-ROOF TENT ALONG NODES

5-8-11-14-17 WITH 30-L8 LOAD.
92
MHMUMMh. - —~———--*"—~ -■ ,....-...—im -^ T^J—-—a—ô....---.- ■ . -.i.-'.p-. .■-^..■-

:,3„,;,r,,,,.,: ■i .'.:■.->>.-- -v^7ViV"^';i-r'.-- •-■;::tT-;,-Tv ■,- rv-v. ■r,>-.-.--ilrv);---^'-'»pi-r;' * '. rif-^ji^'iv-
fr,--- '■'^•--■■.'^■■.î^^zrrw^^
1
▲ MEASURED
PREDICTED
Z
UJ
2
ÜJ
ü
<
-J
Q.
C/>
0.5
X/L
FIG. 7.8. X DEFLECTION WITH SLANT-ROOF TENT ALONG NODES

5-8-11-14-17 WITH 30-LB LOAD.
93
^^lg
I, .it''■.■^■n-rt^-'-r.^ "•■—-: ,->■.;:- -\'-yrr>.-:i zr-wffpf-r' ■ •rr^î--v~--:.y\-''f.^r^''-r'-r-- ■
▲ MEASURED
PREDICTEO
hi 1 - —
ÜJ
Ü
<
-J
Q.
W
L 1
0.5 1.0
X/L
FIG. 7.9. Z DEFLECTION IN THE SLANT-ROOF FRAME ALONG NODES

9-10-11 WITH 30-LB LOAD.
91
—*"——■— li^ftri^^.^^,,.^,,,.^, ^...A^^S^.^^^«^-/^^.^«;^^

M^pj^JgUfUy -M)fMKU4**itfc<^^p|^T^,i,ii^ iiiwiîi ■J-^^"v3CT7B,1J'~,f7^T^'-*7"r<'^'
:»"«'^s-
'•
A MEASURED
• PREDICTED
LÜ
1 —•
- —
5
ÜJ
ü
<
_l
Q.
CO
yi
O
i
0.5 1.0
X/L
FIG. 7.10. X DEFLECTION IN THE SLANT-ROOF FRAME ALONG

NODES y-10-11 WITH 30-LB LOAD.
95
waum mmm - —:'

.m. ui ••miwmmmmmm\%M%i\*n^ i ji.mn!wniMi|»ni|,'.Mj' '■»,.'-•" ■nam
f"
unloaded fabric shape could be found. A third source of

error is the modeling of the joints in the frame. Note that
the pinning of beam k about the Z axis at node 3 (see Sec.
7.1) is somewhat artificial, as the ridge pole of the tent
is simply bolted to the arch at the peak. Experimentation
with the frame suggested that we do this and. In fact,
if beam k is modeled as built-in rather than pinned, the
agreement is less satisfactory. Figures 7.7 through 7.10
show the deflection on the beams and fabric as a function of
position in the model. Figures 7.7 and 7.8 show the Z and X
deflections along the line defined by nodes 5-8-11-14-17*
and Figs. 7.9 and 7.10 give the same information along the
line defined by nodes 9-10-11+ for the full load of 30 lbs.
In all cases, deflection in the Y direction was quite small.
Figures 7.11 through 7.17 compare computer model pre-

dictions and measurements on the arch-roof tent model. Figures
7.11 through 7.13 show the X and Z deflections at points in-
terior to the fabric as a function of load. Agreement of
predictions and measurements is less satisfactory than was
obtained for the slant-roof frame, although at the higher
loads the agreement is generally pretty good. The discrep-
ancies in this case are due, in part, to the coarseness of
the grid in the computer model, especially in the fabric.
This result was particularly noticeable when we measured the
unloaded initial shape of the fabric. Because of the curved
geometry, this initial shape is too complicated to be ade-
quately modeled by the roarse grid of Fig. 7.2. In addi-
tion, some minor discrepancies between predicted and mea-
sured frame deflections are probably traceable to inade-
quately modeled joints. Errors in frame deflection are
particularly bad because they are amplified in the fabric.
This problem is discussed in some detail in the appendix,
where it is shown that a small horizontal deflection in one
of the end supports of a fabric strip leads to large verti-
cal deflections at the center of the fabric.
Figures 1.1k through 7.17 show the X and Z deflections
of the fabric and frame as a function of position at the
full load. Figures 7.Ik and 7.15 show the X and Z deflec-
tion along the line of nodes 11-15-19-23-27, where X/L = 0
corresponds to node 11, X/L =1.0 corresponds to node 27,
*In these figures, the position X/L = 0 corresponds to node

5, X/L = 0.5 corresponds to node 11, X/L = 1.0 corresponds
to node 17, etc.
As in the previous figures, X/L = 0 corresponds to node 9,
X/L =0.5 corresponds to node 11, etc.
96
:
mm MÜMH «a»"—*»^- Mfi njjjjljigiatMmfrmgiaMH
W-PWWJJJIRWJPUI^ API" ¥mm /J..U.4,... u- vfui.. j—^r~- tyLWfnwmwr"**T!<-*li>m:v>t>v*Tnr*- ' •»■«*."• •■' «"■ ■ .
20 30 40
LOAO(LBS)

IN THE ARCH-ROOF TENT AT NODE 15.
97
m "~- ■• - ■■ —-.—---»■*■*-.- iwniWtiirinifif-'■"-■"■--'•-■•■'- ■ -■■■■■' ■■-"•-- .-■■-»•■^-•--^

9pi>wr^HA'^w«lH".«*,-"w<i.i"»^^»-,f <** ^w,s^v«^^^~-y^,y'-^?^^<^"r"f^*^"'^
mmmm ■ pgBgpwgPiUSP * ^^-^'-' t^y^ggwwww^ jgwswsy
TTSF**/£
10 20 30 40
LOAD(LBS)

98
ntMMWMMi ■ - -—
ttl—M-il »i- -■.-... ■■-■■«.. :i.i^:îi.t:iJii..iiaia
war^gîFjf-i ■:_ «irr,wt,I!i'?W
^^^^rrîîrm^^m^^.^
1 1 1
A MEASURED X DEFLECTION
3h-
UJ
_J
U.
UJ
/ s A
1 1
10 20 30 40
LOAD(LBS)

99
■UM^MMMI ■ iM
HMIMMMaM.
W.« HUMP .JWUl P»p«pJ»!»w»i'W w naPîmWF -•».«««■■MUPIMJIWI *WI«J«JW»."WSf». »JW «UMIUPIi.l." I HV!» ._Mm»-:<<<>l>*M . . ^»..•»»'W^W..'«^^
▲ MEASURED
PREDICTED
ÜJ
o
<
_l
en
FIG. 7.14. Z DEFLECTION IN THE ARCH-ROOF TENT ALONG

NODES 11-15-19-23-27 WITH 21-LB LOAD.
100
:
dm~.ru. i-T I r I'irf
IfPJP^^PWpjPQPFmmppBniVilMpBlt' 9fBBW " """^ w ^MHWHW .^,w..ii>i,w'wim Mi.pwujwi.umi' i ^n-mm»'...^' JIMI« w^,.-.* *"*W.i 11.' i 'Hi,, , jpii, 9, y* Fi
A MEASURED
PREDICTED
UJ
1 -
5
ÜJ
ü
<
-i
OL
A ▲
s^* X
FIG. 7.15. X DEFLECTION IN THE ARCH-ROOF TENT ALONG

NODES 11-15-19-23-27 WITH 21-LB LOAD.
101
mmmmmmmmm — - r—1M i l«M«M—

Ijp ^MMv&jgHHMNfi-uiifiw&i ww)i».):irtJ»<vii|»»..tfr'»'-'-J';,»'HMr
▲ MEASURED
PREDICTED
LU
s
ÜJ
O
<
_l
Q.
CO
0.5 .0
X/L
FIG. 7.16, Z DEFLECTION IN THE ARCH-ROOF TENT ALONG

NODES 16-17-18-19 WITH 21-LB LOAD.
102
-"- ■ ■ ... -~:-:.J--| ,,- „ MMJ

wmmmmwmf!mmm*m. tm^m^mmmmmmw ■' " ipiii!iiwii!t-Niiijjii)iiMjjifr^'"f^jiMawi'wp
▲ MEASURED
PREDICTED
Z
LU
2
UJ
ü
<
Q.
(/)
Q
0.5
X/L
FIG. 7.17. X DEFLECTION IN THE ARCH-ROOF TENT ALONG

NODES 16-17-18-19 WITH 21-LB LOAD.
10"-;
KM
JJ a I^^P^p^g^lIIPWPIIJ.IlllU.1. >" . ' IK-MKU^ÄJMWWIWJPW^W'H' - •'^'^w*«:«»!fW«WUlMLliWA.W!i",Wi.W!i«iJ' j;i(iiWWfBiiii„,.j.i.i.p»piiiLi(.-w5^»;
etc. Figures 7.16 End 7.17 present the same information for
the line of nod^s 16-17-18-19, where X/L ■ 0 corresponds to
node 16, X/L »0.5 corresponds to node 19, etc.
104
Ml« ^^^^^^^^ ^__^-.

.... ^- ^ ...IT-..- . 11...■, , , |M| |.l
8. CONCLUSIONS
It has been demonstrated during this program that the

stress strain behavior of fabrics can be measured and used
to predict the stresses and deflections in these fabrics
when they are used as membrane structural components. Mea-
sured uniaxial properties of a dacron fabric were used
successfully to predict the stresses and deflections in a
strip of that fabric when it was loaded normal to its plane,
The measured biaxial stress strain properties of cotton
typewriter ribbon cloth were used with some success in a
computer code to predict the stresses and deflections in a
rectangular piece of that fabric mounted in a rigid frame
and loaded normal to its plane. These same biaxial proper-
ties were again used successfully in the same computer code
to predict the deflections in the fabric when it is mounted
in small scale metal frames designed to simulate the slant-
roof and arch-roof frame-supported tents used by the Army.
Discrepancies between computer code predictions and mea-
surements on the model tents are traceable partly to too
coarse a mesh in the computer model but primarily to inade-
quate modeling of the joints between members in the model
frames. Further efforts should concentrate on building
into the computer code a better capability for modeling
these frame joints.
105
^^m^r^^^^'-^^^mm^^'-^w^ ft^Srnji^'^î^f^rT™^^^
APPENDIX A: SIMPLIFIED MEMBRANE ANALYSIS
To evaluate results obtained from the computer program

and fabric tests and to develop some insight into the geo-
metrically nonlinear deflections of membranes, we derived
an approximate solution of a one-dimensional membrane strip.
The following sections contain brief derivations of a mem-
brane (string) with and without initial displacements.
A.l Analysis of an Initially Flat Membrane
The membrane is assumed to be initially flat with linear

elastic material properties. Figure A.l shows the initial
and deformed configurations of a portion of the membrane.
Summing forces in the x-direction will produce the x-
coordinate equilibrium equation as
■—: (T cos<t>)dx = 0 (A.I:

dX
where T is the total membrane tension and <|> is the angle be-
tween the x coordinate and the deformed membrane. Similarly,
the y-coordinate equilibrium equation becomes
■—
ox (T sin<j))dx + PdSvx = 0 (A.2)
where P is the distributed loading in the global y-direction

and dS is the projection of the deformed elemental portion
of the membrane onto the x-coordinate. The dS quantity is a
measure of strain and may be written as (see Fig. A.2)
dS = Xdx = [(1+u J2 + w2 1* dx (A.3)
where u, w are the displacements in the x and y directions,

respectively, the comma represents differentiation, and X
becomes the extension parameter. Also,
dSvx - (1+u , j )dx , (A.4)
which may be used in Eq. A.2 to produce
106
mmtm
&'■?>?*$?•■*:.'■■" >■'-■■ v:*f;;sw^n!*,Y»*'--r'^ ,. .--i>,_y.:-rv:-r- ;-'- -tf*".- — ■ 'T- ' .' —j:
$^ T+dT
FIG. A.l. COORDINATE SYSTEM FOR INITIALLY FLAT MEMBRANE

STRIP.
r dx
w+dw = w + w,- dx
FIG. A.2. DISPLACEMENTS OF ELEMENTAL PORTION OF MEMBRANE
107
■tefeü^ ---- —■ - rt—jSI

Wr^jRJiTTr3TW. ""-«?■«:«?* ■'£ ■T.'*«*.'"»
(T sin<fr) v + P(l+u _) - 0 . (A.5)
From Fig. A.2, it can be shown that the sine and cosine
functions of * may be expressed in terms of X and displace-
ment derivatives as
w
tin* = -jp (A.6)
1+u
JA
COSlJ) = (A.7)
The equilibrium equations may now be written as
•P^ = 0 (A.8)
4H + p(i+U}X) = 0 (A.9)
with boundary conditions as
w(0) = w(L) » 0
u(0) = u(L/2) = u(L) = 0 (A.10)
The load strain relationship for the membrane is
T z Ke. (A.11)
where K is the stiffness parameter with units in pounds and

e is the strain in the x-direction. The strain can be ex-
pressed as
e Adx-dx = A, - 1, .
= —-= (A.12)
x dx
3 08
mm aniaaaaaigijaiüüa -■■- ■ m Mt, - .^J>i~.,~.j>~^,â

*W"<™««5T-';TFr71s««wl^."wi-»» ' u-'..i»r»"'--^J . u MMW.UNwmmigr T^7»^5^.>w.(*iTrP7?WM,**W>-.yw
If we assume that the components of strain are small com-

pared to unity, then
X
= * + u
,x +
I w!x (A.13)
and Eq. A.11 becomes
+
- 4,* 1 "!*) (A.14)
To obtain an approximate solution of Eqs. A.9 and A.10,

we assume that
\ z 1 ; 1 + u = 1
in Eq. A.9 to give
T ■ K(u + w w ) = 0 (A.15)
The above relation implies
P = constant , (A.16)
which may be used in Eq. A.10 to give
Tw + P = 0 (A.17)
Integrating Eq. A.17 twice and applying the appropriate

boundary conditions produces
w = || (L-x) , (A.18)
where L is the distance between membrane supports. Note

that the vertical displacement is a parabolic function of
position.
109
ilttîiiiin- - "—MiflMill i a,.\j.-:\L: .*-^jLî^V:W,„i,

PTî»!^j«Bjj5j^5P^5TjK^r ja^s^«™-**TT-^^v:^r'r-^'TTTaê«H»"'«*fW*,T!7; ■,-^r^f ^ >*,M^;-;^,»^fl1s----.\?>p77;"'< ■.'-,--Wi-sn^-i,-5ÄV=i- -
The u displacement is obtained by using Eq. A.18 in

Eq. A.15, integrating twice and applying boundary conditions
u--GF[4-^%] (A.19)
Performing the appropriate differentiation of u and w

and substituting into Eq. A.14 gives
-m■ (A.20)
This equation can be used to express the displacements as a

function of the loading:
w(x)
L - W (!) (i - !) (A.21)
u(x)
L ■ -! wT (!) m - m* rt • (A.22)
Equation A.21 csn be modified to account for any ini-

tial tention TQ in the membrane
2T.2\%
T - T0 + (*§££)' (A.23)
and Eq. A.21 becomes
w (»- i) ■
w 3PL/K
v^m
L (A.24)
Equations A.23 and A.24 are used to solve for the ten-
sion and displacement of the initially flat membrane with a
pretension.
110
■**"*iA'^'-*-"'" mmmämtmm ■ - ■
Bjffiüi iigijmaa^âaaB ■■-■ >frr; jaümMtefc -A-a**»^
A.2 Analysis of a Membrane With an Initial Deflection
The development of the deflection equations of an ini-

tially deflected membrane is similar to the derivation per-
formed in Sec. A.l, except that the strain relations must be
modified to account for the initial displacement.
From Pigs. A.3 and A.4 we see that the X parameter is
X = [(1+u Y + u°)2 + (w Y + w°.)2] (A.25)

,A ,A ,A jA
where the superscript zero on the displacements refers to

the initial displacement. The parameter X may be simplified
by expanding Eq. A.25 and assuming the strains to be small
compared to unity:
1 + u „ + u° + % (w „ + w°J2 . (A.26)
* A • A C- m A % A
The load-displacement relationship as described in the pre-

vious section becomes
T = K u + u
[ ,x :x + i<\x + w
:x>2] (A.27)
The equilibrium equations become
T = 0 (A.28)
Tw»jXX = -P (A.29)
w' w + w (A.30)
where w' is the total displacement of a point on the membrane

relative to the flat unstrained, undeflected position of the
membrane and w is the incremental displacement.
The vertical free-hanging shape of the membrane is as-

sumed to be parabolic as
o _
w (w
(t) M). (A.3D
ill
— ----- ■ yîmmmm -•■ lüagnüi

I ■■■■BBPWt 8 ■--V.^'-SJj-l'^'4».!^*v^>*WL^lV^»**><W.V^-'^gTWUItl.»-' , l.',,'HWil —r y.^1-,»ji,,».ji.ji^w,wj-<Jw^vij>»ii^^>.i.iji|^^ »fWrt^W.W*!»'"**'«'
x,u
INITIAL SHAPE-
ZERO TENSION
u°{L)
FIG. A.3. COORDINATE SYSTEM FOR INITIALLY DISPLACED

MEMBRANE STRIP.
w°+w£ dx
NJ°+U °dx
w+w,jJx
\u+u,xdx
FIG. A.4. DISPLACEMENTS OF MEMBRANE WITH INITIAL

DEFLECTION.
112
mmtm HMMi II ■!«■■ lahgafuWiiMii ii ■'11in ■ *ii it ü liii i

P*^W»r~^>~'^*i'^T^~r:^''T?7^«<^^ • '■-' Tnr^^lw^TCTTSTSpT
where wfl is the initial center displacement of the membrane.

Using Eqs. A.6 and A.7 in Eq. A.5 and integrating as before
produces
w
L
1
7 fc-^MM- (A.32)
The corresponding initial displacement in the x direc-

tion is obtained from Eq. A.27 when zero load has been
applied to the system. Therefore,
U
U - - 7 *'J ■ (A.33)
Integrating and applying appropriate boundary conditions

gives
•• ■ - ¥ to h - a) 1 m ■ + (A.34)
The amount of extra material used to create the initial dis-

placement of the membrane (see Pig. A.3) is determined by
substituting the length of membrane into Eq. A.3**. Thera-
fore
u"(L) = - 3 — •
ft K (A.35)
The actual initial distance between membrane supports when

there Is no initial deflection can be approximated by
L = 1
■[-!(£)']. (A.36)
113
M«M NOtftfMMiMittlM fiMflÜ hii^^.,,.-,^».^^^

wgf^r^K'^f*^*^; "r^:^. '■"■*■'*V*?^ "'
mm
where L' is the new distance between end supports that re-
sults in the initial deflection w° in Eq. A.31.*
The u-displacement function is obtained by using Eqs.

A.28 and A.33:
u ,xx = -[w « A
(w « A, A.+w k A. Ai ) + w°, X w , XX 1 (A.37)
The u displacement as a function of membrane parameters is

determined by substituting the appropriate terms in Eq. A.37
using Eqs. A.31 and A.32, integrating twice, and applying
the boundary conditions to give
* - - * [W - (tJJ [(tf -1 (SF * ®\ ■ <*•»> +
The relationship between the tension and the applied

load is obtained from Eq. A.27 using Eqs. A.31, A.32, A.31*,
and A.38:
2
i^N -W]- (A.39)
Using this equation in the equation for the membrane incre-

mental displacement, we obtain
w 12T/K x /, x\ (A.ltO)
L
as an alternative expression for Eq. A.32.
*Note that in the model the initial deflection is envisioned

as being produced by moving the membrane end support at
x = L a distance u°(L) (see Eq. A.33) closer to the other
membrane support. Thud, the distance between end supports
that one measures in an actual experiment with initial de-
flection w° is L».
114
»in
:if^»^^..?t>mi1îs^-:n;7^^-^;,J'75^^T^"*"'™;L1^^?,5l"
Equations A.39 and A.40 are used to determine the ten-

sions and incremental displacements of an initially dis-
placed membrane.
115
iwnii--i" a- .Arr------ nVi * fa. .i..i..baa

LIST OF SYMBOLS
B B B B Material constants in the fabric stress

WW» WF» FW» PP strain model each of which is a func-
tion of the fabric tension.
C
W» C
F Warp and fill biaxial stress strain
model constants respectively.
E Young's modulus of the frame material.
f Load per unit area.
Ow(o) Gp(o) Warp and fill biaxial stress strain

model functions, respectively.
I Frame beam moment of inertia.
K Fabric stiffness constant (in Appendix).
L Characteristic length, distance between

supports.
P
W> P
F Exponents in the biaxial fabric stress
strain model.
P Load per unit length (in Appendix).

T T Warp and fill tension/unit length,
respectively.
T Tension/unit length (in text).
T Total tension in a fabric strip (in

Appendix).
Initial total tension (in Appendix).
x and y fabric strip deflections (in

Appendix).
u, wp In plane and out of plane fabric de-

flections .
wB Frame beam deflections.
TV Initial center deflection of the fabric
strip (in Appendix).
116
HM ■■--■'-—
■iiiüMMIaj*aa"^*" MÜli
LIST OF SYMBOLS (Cont'd.)
w° Initial free hanging fabric strip shape

(in Appendix).
w' Total fabric strip deflection initial

plus incremental (In Appendix).
X, Y, Z Global coordinates.
a Ratio of warp to fill stress.

Fabric strain, warp and fill strain,
respectively.
Fabric stiffness — slope of the uni- ■;
axial stress strain curve.
Fabric elongation including rotational

effects (in Appendix).
v Poisson's ratio.
Rotation of the fabric strip (in Ap-
pendix) .
3
( )
a2
c > ,xx
.. —2
ax"
( )m Refers to scale model.
( ) FS Refers to full scale.
( ;w Refers to fabric warp direction.
( )B Refers to fabric fill direction.
117
gggjf —"—-^»»*-^—mtm a, n ■*■« ■ ■* '-*-' »- -

PDF Bicycle

Uploaded by

Copyright:

Available Formats

PDF Bicycle

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

PDF Bicycle

Uploaded by

Copyright:

Available Formats

A 001 012.

ANALYSIS OF STRESSES AND

Bolt, Beranek and Newman Inc.

UNITED STATES ARMY

mimt !F STSESSES HNS

Contract No. DAAG?7-73~C-0*O7

ANALYSIS OF STRESSES AND DEFLECTIONS

Bolt Beranek and Newman Inc.

Contract No. DAAG17-73-C-0107

This report was prepared by Bolt Beranek and Newman

2.3 Program Input and Output 4

2.3.1 Input requirements 4

3. SCALING LAWS FOR FRAME-SUPPORTED TENTS 8

3.1 Fabric Scaling 8

3.2 Frame Scaling 10

4.2.1 Biaxial testing apparatus 24

4.3 Fabric Force Sensor 38

TABLE OF CONTENTS (Cont'd.)

5. MODEL TENT FRAMES 43

5.1 Slant Roof Frame 43

6. PRELIMINARY VERIFICATION OF THE COMPUTER CODE 52

6.1 Fabric Strip Membrane 52

6.3.1 Testing procedure 69

7.2 Model Tent Tests 80

2 1 Flow Chart of Program 5

4 .15 Comparison of Uniaxial Data With the Biaxial

LIST OF FIGURES (Cont'd.)

4 16 Load Link in the Fabric Force Sensor 39

4 17 Fabric Force Sensor Drawn Actual Size 40

4 18 Fabric Force Gauge Calibration Curve 41

5 1 Tent Maintenance Shelter Beam Cross Section ... 14

5 2 One-Eighth Scale Approximate Model of TMS 45

5 4 Fritche Shelter Frame Members 48

5 5 One-Eighth Scale Approximate Mode of the FrS .. 50

6 .6 Strip Deflection With an Initial Deflection

6.7 Comparison of Theoretical and Measured Fabric

6.9 Arch Roof Frame Computer Model 65

6.10 Slant Roof Frame Deflections 67

6.12 Fabric Frame Test Arrangement 70

LIST OF FIGURES (Cont'd.)

7.1 Slant Roof Tent Computer Model 81

7.3 Comparison of Measured and Predicted Deflec-

7.6 Comparison of Measured and Predicted Deflec-

7.8 X Deflection With Slant-Roof Tent Along

7.9 Z Deflection in the Slant-Roof Frame Along

7.10 X Deflection in the Slant-Roof Frame Along

7.12 Comparison of Measured and Predicted Deflec-

LIST OF FIGURES (Cont'd.)

.15 X Deflection in the Arch-Roof Tent Along Nodes

.17 X Deflection in the Arch-Roof T^nt Along Nodes

A.l Coordinate System for Initially Flat Membrane

A.3 Coordinate System for Initially Displaced

A finite element computer code has been developed for

«■■t ■MHM „__-, ■ttHMM

SECURITY CLASSIFICATION OF THIS PACE (Whtn Dim Entered; /fPMnnzi 07.<b

Analysis of Stresses and Deflections in Final

16. DISTRIBUTION STATEMENT (of «lie Report;

Approved for public release; distribution unlimited,

is. SUPPLEMENTARY NOTES ^ ^ ^3^^ by the Technical Information Service

Frame Supported Tents

""* ■ - 1 mitt^ätmmmlttmmmmiimm SSE

0Callahan, J.C., "NONFESA - Nonlinear Finite Element Struc-