INTRODUCTION

Filament winding is an efficient and often used means of fabricating

pres sure vessels, partly because it permits the designer to optimize the
structure by placing only enough fibers along given directions to withstand
the stresses along those directions. If homogeneous materials are used to
fabricate closed end pressure vessels, the longitudinal stresses are only half
the hoop stresses, so that material is effectively wasted in the longitudinal
direction. Because
of the efficiencies offered by filament winding, this technique has been used
for a wide variety of pressure vessel applications in the years since World War
II; these include motor cases for nearly all U.S. strategic missile systems, and
re cently they have been considered for light man-portable antitank missile
launchers.

The increase in efficiency offered by filament winding is accompanied by an

increase in the complexity of the techniques which must be employed in vessel
design and analysis of vessel performance. The elementary equations of strength
of materials are not valid, since the assumptions of homogeneity and isotropy
used in their derivation are not met. There is also a greater number of
material variables which must be considered, e.g., the number of helical and
hoop layers, and the angle of helical winding. The purpose of this note is to
present an ana lytical technique applicable to filament-wound pressure vessels
which can predict the internal stresses arising from a given combination of
internal pressure and choice of materials and processing parameters.

Netting analysis1,2 is perhaps the simplest of the analytical techniques

used in rationalizing the behavior of fiber-reinforced composite materials, but
it is well suited for filament-wound pressure vessels. Netting analyses assume
that all loads are supported by the fibers only, neglecting any contribution by
the matrix and any interaction between the fibers. These assumptions do not
accrue in general - they predict, for instance, zero transverse strength in a
unidirec tionally reinforced plate - but they are consistent with pressure
vessels, in which the fibers are loaded in tension and the shearing stresses are
small in comparison.

DEVELOPMENT OF GOVERNING
EQUATIONS
Closed-End Bottles
Closed-end refers to pressure vessels constructed so that internal
pressure sets up both hoop and longitudinal stresses, as opposed to open-tube
configura tions in which the longitudinal loads are taken by the next fixture.

Consider a cylindrical section of a chamber of radius R containing an

inter nal pressure P, with denoting the longitudinal (axial) and 6 the
circumferen tial (hoop) directions. The strain Ea at an angle a from the
longitudinal is given in terms of the axial hoop strains E ,Ee by the
transformation equation:3

1
 σα = εϕ cos2 α+ εθ sin2a + γϕθ sinθ cosθ (1)

where the term involving the shear strain γϕθ will be

neglected.

This development will assume the presence of two sets of helical

windings oriented at ± α1 and ± α2, and one set of hoop windings. The
stresses in these fibers are given directly from Equation 1 as

σ = E ε (2)
h θ

σ
α1
= E(εθ sin2a1 + ε
ϕ
cos2 α1) (3)

σ
α2
= E(εθ sin2a2 + ε
ϕ
cos2 α2) (4)

where σ denotes stress and E is the fiber modulus.

Denote the total cro5s-sectional area of fibers in a unit length of hoop or

the two helical windings as Ah, Aα1, and Aα2. (These areas can be computed in
several ways for actual vessels; an example will be given later.) Then the area
of helical fibers intersecting a unit length of surface along the hoop and axial
directions are Aα sinα and Aα cosα. The resolved components of force from
the helical fibers in the hoop and axial directions are then

hoop= (Aα sinα)σα sinα

axial= (Aα cosα) σα cosα

It now remains to establish force balances in the hoop and axial

directions.
First the hoop:

Unit
Leng

F F

(5)
2
 Now the longitudinal:

(6)
The stresses may now be written in terms of strains by means of Equations
2 to 4; Equations 5 and 6 then become:

(7)

(8)

which can be written as:

(9)

(10)

where:
The K's constitute a geometrical description of the vessel. Equations 9 and 10
are solved easily to give:

3
The magnitudes of the fiber stresses are then given directly by using the
results of Equations 14 and 15 in Equations 2 to 4.

Open-End Tubes
For the case in which the longitudinal loads are reacted by the test
fixture rather than the test specimen, the left hand side of Equation 6 is set to
zero.
Equations 9 and 10 then become

which yields (16)

(17)

(18)

(19)

As before, the fiber stresses can be obtained by using these values in Equations
2 to 4.

A BAS1C-LANGUAGE PROGRAM FOR CLOSED-END

CHAMBERS
General
In order to facilitate rapid and error-free implementation of the foregoing
analysis, a small BASIC-language program has been implemented for the Hewlett
Packard 9830 calculator. This program outputs values of hoop and axial strains,
and hoop and helical fiber stresses, given values of layer areas, helix angle,
chamber radius, fiber modulus, and internal pressure. A program listing is pre
sented in the Appendix. Conversion of the program to other computational systems
would not be difficult.

Definition of Program Variables

Input Variables

T$ E
p
Zl, 22
R
Al, A2
A3
Title s, inches
(Alphanumer Helical layer areas, square
ic, but inches Hoop layer area, square
avoid inches Fiber modulus, mpsi
quotation Internal pressure, psig
marks)
80
character 4
s maximum
H
e
l
i
c
a
l

w
i
n
d
i
n
g

a
n
g
l
e
s
,

d
e
g
r
e
e
s

C
h
a
m
b
e
r

r
a
d
i
u
 Internal Variables

KI, K2, K3 As defined in Equations 11 to

13 Output Variables

El Hoop strain, percent

E2 Axial strain,
percent
Sl Fiber stress in #1 helical layer,
ksi S2 Fiber stress in #2 helical
layer, ksi S3 Fiber stress in hoop
layers, ksi

User Instructions
1. Insert program cassette, rewind to start
2. Press LOAD, EXECUTE (program is at File #0)
3. Rewind cassette
4. Press RUN, EXECUTE
5. Enter input parameters as requested by calculator
6. After each problem is solved the user has the option of inputting
a new pressure while leaving the chamber geometry unchanged, or
starting a totally new problem.

EXAMPLE PROBLEMS
Analysis of Burst Test Data
As part of an Organic Materials Laboratory program investigating
filament wound Kevlar/epoxy as a possible material for interceptor missile
motor cases, a series of 6-inch-diameter, 14-inch-long bottles have been
fabricated using an XOXOO winding pattern (X denotes helical windings, 0
denotes hoop windings) with ±25° helix angle. These bottles are gaged to
permit measurement of hoop and axial strains, then tested by hydrostatic
pressurization. Once the burst pressure is known experimentally, the netting
analysis can be used to calculate the fiber stress at failure. This delivered
fiber strength is an important parameter in chamber design.

Before the netting analysis can be performed, it is necessary to calculate

the cross-sectional fiber areas Ah, Aal• and Aa2 of the hoop and helical layers.
Accurate values of these areas could be obtained by quantitative microscopy of
suitably sectioned chambers, but acceptable values can be calculated more con
veniently from values of filament denier and winding bandwidth as follows:

The denier (grams weight per 9000 meters length) of the 12-end roving used
in winding is 4560, or 5.067 x 10-3 g cm-1. Kevlar's density is 1.45 g cm-3,
so
the area of the roving is (5.067 x 10-3/1.45) = 3.49 x 10-4 cm2 = 5.42 x 10-4
in.2.
The width of a single band is measured during winding as 0.135 to 0.145 in.; tak
ing the bandwidth as 0.140 in., there are (5.42 x 10-4/0.140) = 3.87 x 10-3
in.2
of fiber area in each inch of pass length. In the XOXOO winding pattern, there
are two helical and three hoop layers, each consisting of two winding passes.
The total areas are then:

5
 Ah= (3) (2) (3.87 x 10-3) = 0.0232 in.2/in.

Aa.l = (2) (2) (3.87 x 10-3) = 0.0155 in.2/in.

Aa.2 = 0.
A general formula, using mixed but commonly used tmits is:

A= (1.722 x 10-7) (DEN) N/p B

where A is the resulting area in square inches, DEN is the roving denier, N is
the number of passes, pis the fiber density in grams per cubic centimeter, and B
is the bandwidth in inches.

Test bottle 4 experienced catastrophic failure at 3042 psig internal pres

sure. Output from the netting analysis program for this pressure and the above
layer areas is shown in Figure 1. These predictions can be compared with the
strain gage readings for axial and hoop strains, and the fiber stresses calcu
lated from the gage readings and Equations 2 to 4. These values are:

From Netting Differ-

Gages Analysis ence
(%)
Hoop Strain, Percent 1.25 1. 84 32
Axial Strain, Percent 0.69 1.89 63
Hoop Stress, ksi 238 351 62
Helical Stress, ksi 160 358 58

Agreement is poor. Netting analysis is generally expected to yield predic

tions approximately 10% higher than those of more sophisticated finite-element
analyses,2 but the large disagreements above are ascribed primarily to gaging

1!
! I, 1
1

i.'

iii i
iI i '.I' !.,: i I
1·H.<!>!.1l..l.. l'·, i!..

l.. i ... :.:.11::. ···. l"F: 1·: .\ i·i:..: :-'-: ,::!. :::·;
:: i I i 1... I !;: :i
' ,

'!'"I.'../ !I! H: .i J !. I iL
i! ii: Hiil[i'

Figure 1. Program output for six-inch Kevlar bottle at 3042 psig.

6
errors. Reasons for suspecting the gage readings are as follows: (a) The
XOXOO pattern was designed by Hercules, Inc., to be balanced, i.e., have equal
hoop and helical fiber stresses. This equality obtains in the netting analysis,
but not in the gage values. (b) A delivered fiber strength of approximately 350
ksi was measured in an intensive Hercules test program using identical bottles.
The stresses implied by the gage readings appear much too low. (Note that the 350
ksi strength is much less than either the mean (525 ksi) or the minus 3cr (447
ksi) strength measured in impregnated Kevlar strand tests. This loss in
delivered fiber strength is likely due to still unexplained failure mechanisms in
the filament-wound bottles, and is a serious problem in motor case design.)

Parametric Design Studies

Although the netting analysis is valuable in analyzing burst test results
and checking the validity of gage readings, it is perhaps even more useful as a
means of designing pressure vessels. In this capacity the program is run
repeat edly using various values of layer areas and helix angles on a trial-and-
error basis until the desired fiber stresses are obtained. To reduce the number
of trials it may be convenient to construct plots showing the influence of
various parameters; the desired values may then be located easily.
As an example, Figures 2 and 3 show the influence of internal pressure and
helix angle on the fiber stresses of 6-inch XOXOO Kevlar bottles having layer
areas Ah= 0.0216, Aa1 = 0.0144, and Aa2 = 0. The helix angle in Figure 2 is
±25°. Figure 1 demonstrates the linear relation between pressure and fiber
stress, while Figure 3 illustrates that a helix angle near 25° yields a balanced
design (equal helical and hoop stresses).

500.---------------------,

400 600r----------------------,
500
... HOOP

j 200

100 300
HELi CAL

o-._........ 10'----'---20 .... 4_._o ,. 6.0....

1000 2000 3000
Pressure, psig Helix Angle, deg

Figure 2. Influence of internal pressure on helical and Figure 3. Influence of helix angle on helical and
hoop fiber stresses for six-inch Kevlar bottles. hoop fiber stresses for six-inch Kevlar bottles.

7
 APPENDIX. PROGRAM LISTING
J j' I I? F. l'"l -:•:• ,:1:, ·:I:' -:1:• -:,:, -:1:, ·:I:' •'.I;· -:1:, ,:1:, ,:1:, ·\· :1· -1:- :1: -:1:- ·:l;· ,;I:' :1- :,:, ,:,:- :1:· , 1:- •!i'• ·:I: ·f ·I:• ·i!: ·:I:" •!I:• -:1:, :,:, •:1:- ·:!:' • :- :1:, -:1:- -:1:, -:1:• ·'.(· -:1:, -:1: ·f - :· -:I:" - :- ,:1:- -:1:- -:1;, :1:' -:1:· -:1:- ·:I:· -:,:- -:1:· ,:1:, :1;, -:,: -:i:, -:1:, -:1:- -:1:, ·:I:' :1; -:1:• -:i:, ;1:• -:1:
2U 1;:F:i"I t·•IETT I 111; HIIHL..··,':; J '.::; l'F:L::;::UF:[ ',,'[:::,::::EL.. FI t:EP •::TF:E::;:::::
;11 1::: E i"I I? u·,·L.. H1•11..:: E ...I uI... ?!j [: 1·111 F' R i', ,:; i?Hi·I :,.,:. 0" :::: !j ') 0•:::;
411 REl"l***********1f ***f***+****\***** ********i*****0*****i*************•**
i;u n1r·1 rrt::?uJ
1.11 F·l?ll·IT L.. Il··I•:;?,
lli 1·1:::: J l·!T rr:11:::: 1? :: ., 1·1:::E:;:::;1,11;:1:. ',..'[·::::::i: L. 1:· 1B[F: :: 1·kf::i::::: 1.11::·rr I 1.11:; Hl•IHl \·:::i J :::: ..
::O f•RJl·IT Lil'l(l 1
·:,1:1 F'F:Il·!T THF:i:,:•; ··uA'·•'F: 1;:0·1"1.. Al·I ..!UL.\' ?'.:,··
11'11:i F•1;:: I Ill L. I II ( 1 I

118 REl"l********* i***********************f************l** ******************"*

121'1 REl"I El'ITER PROBL.Ei"I PARAl"IETEF:S
JO REl"l**************f********* i********* ****i************ *******0******** *
411 DI :::::P .. IIITl:k Tl Tl. E .. ;
liU I l··•IF:•ur T::f::
CU Pkll'IT T$,LIN(l
!° IJ {I1::1:;
;,,:;U Dl'.::il·:• ·'t:,.l'lli:J•: 11[!.. J:: l·'ll'F"Ll:::.3 I.Hi..JhHL,,HL.F'HH/.1 :;
I. )0 I IIF·U 1· :: l ,, z::
2011 f I ::t::::D l
2 I 1:1 F•fii I HT .. Hl:L.. l :,: 11111;1.... 1: :; :: HL.l·:•HH l: ., :: :'.I,,., RL.. F'HHi:'.''' ., :: z,:
2i'.U DI:::F· "1,:l·•IT[R \.'[:;::[L. f:::HDIU::.,, Il•I"::
::SU [ l•lf:-U'l I?
2411 f•t::: I l·IT ,. '•/1:::::::::EL 1:·1•1iJ I U::::::::: ,. :: I?; .. I 1·1 ..
i::l',;:.1 II l ::i:· .. [l·ITFI? L A\·11::: ·f1Rl:::1-1:: ,, :::::) i. :.1 '· r:i :I ,, H::· ,, 'iS I •• ::

i::,::u ll··IF·u·!· Hl,,H2,i::1:::

:'./11 r:1:,,:EJ.1 6
;:·.::,1 1::•r::::Il·•IT "LR\.[I? flf:i.:.I! ,.::u 11·1::,:: HI "::r:i:t,,"H;::::,,,,":;11;::,,"HJ:.. ":,H:J
:: .1u 1,: 1,,,,,fl 1.,,,., :::: 11·.1z 1 :, cu:::::? 1 .' t:::::+1:1:::::i• ,: ::• .t 1·.1z2::, t2·,1,,: 1:.:u:::z::::: ::, t:::::
auu
1 ::· ,: '·

K2 H1*rcosz1::,1 +R2+1.cosz2::,t4
310 K:J=H8+H11rs1HZl)t +H2tlSil'IZ2 t4
.:20 DJ :::iF' .. El·IT[I:: FI !:LI? MODUL..1.1:::: ,, 1·1i•::. I,.::
, :i:i 1Ht·:•ur 1
::: IC'I F 1·.,:FTI J
:<511 l·'I:: J 1·-IT "I J [:[],: l·h!UUl... 1..1' · ':: [ .:: PH'::::: I"" L l H,: , l
,::; U [ ::: [ .,,:. 11:i +06
;;-11 111:::i:· ··t:1·1r1·1::: 1•1; ::::URE,, 1·:::11:;
:?C:CI l l·•IF"·UT F·
::·)1"1 FI :,,:ED 0
.::i,u;:, i'k J t·rr Tl!H.::::: :; "F'l?1:::::< l,ik[:::·:. :: F' :: ,. F:::. JC"
41U
42U
.::i-:::1::i
1. :.1.i:J [ l :: I '· F· :,.i::;: I/ ( ::: ,[ I< I.•····.::: Il·.. 2 '/,.I< l ti::·.. l<:?·1:1<3 :. :,
i :, .,,,., (

•l i:::; ,J [;?::::: (, 1: ·' '·R:, / ,. :? , ,·, :, ( :::> i:t•: l ··l<:3 , · ,: ,, 1. l:?· · ·1<2 '·1<:::, :,
1 1

.,1c:u FI ::1:11
4,"U 1:•1:;,: Il··IT THI:: :I/',:. Hi.. Uf' :;[ F,:H J 1.•I::.. :; l ·1'· JUD
l'F'Jt.lT f'HBl/:: ··:-;, 11:<JHL ·:TF:HIII ,., :: 1:.::1:1.00
.1, ') :: l :::[ •I , [ , l '· : ; 11i.:: J • l 2 1 ,: CO:::;;::: l::, "I;::.,
) E
: ;1: :11::,1 S2=E IE1i:(SIN! )l2+E'+(COSZ2)12:,
:; I :;J::::·[ ,[ 1
U FJ::Ell l
!j;?l'::i PRil'IT THF:10; .. :R
•Lo" r"·•
'I:,, I, I. I •
1:::i:::::c1 I 1::: H;:,: ::: U TH[ 1··1 I'",,::: 0
I',
IU
:.·1 i:::, 1:.:i 1•1; 11rr IHl:Ju::. ·:Ti•:[:. Ill l:i..: !-IL.i... 1;1-11 :; :• 1.U:111:: I< : I "
:·:i(; t·i::: I l···IT ·1i::it::: J II:.: ::::; 11;:1::::::: r l·•i HU1:II•·:.: :: .. :i :::, J i:1uu ::
l;J

88 kl:1"1 8kHl'ICH OPTlOl'IS

:::; '':i 1J F: F: 1·,·1-:1:- ,:(, -:1:• -:t:- -:,:, ·:I:" ·f ·:I:' ·:!:· :i: ):, ·:i: :i:- ·:I:' ·:I:' -:1:. ·:I:' ·:I:' ·:l. :1: -1: ·:':' :(, ._i: .I: "i. \ :i:, -:1: ·:!:' -: • .:,:, •:1:, ·:Ii· •,1: :1:• -;i I• ·/ :i: <
1.1:111 DJ::::F·· "f.:JITE::r::: I i ;JI? IIF::1-1 1:::•1:::1::::..::111;:i::: ..
(.11:1 l 1·.1F'I.JI I)
(,.,U 1,0TO U UF. ::,:?IJ ,, .I UIJ
.:.:::H ::rur::•
6•1 I 1:1{[1
9