Int. J. Impact Engng Vol. 4, No. 1, pp. 41-56, 1986 0734-743X/86$3.00+0.

00
Printed in Great Britain PergamonJournalsLtd

THE INEXTENSIONAL COLLAPSE OF GROOVED

THIN-WALLED CYLINDERS OF PVC UNDER AXIAL LOADING

A. G. MAMALIS,* D. E. MANOLAKOS,* G. L. VIEGELAHN,t N. M. VAXEVANIDIS*

and W. JOHNSON$
*Department of Mechanical Engineering, National Technical University of Athens, Greece
?Department of Mechanical Engineering and Engineering Mechanics, Michigan Technological University,
Houghton, Michigan, U.S.A.
:~Ridge Hall, Chapel-en-le-Frith, via Stockport, Cheshire SK12 6UD, England, U.K.

(Received 28 November 1985; in revised form 13 January 1986)

Summary--An inextensional collapse mechanism is considered for the crumpling of thin-walled,

grooved tubes and tubes without any discontinuities (non-grooved), when subject to axial load.
Shortening of the tube (or shell) height is achieved by folding in a non-axisymmetric diamond
mode about stationary circumferential and inclined plastic hinges; collapse is initiated at the
thinnest section of the tube (groove) and spreads along the various rings and grooves of which the
tube consists, by passage as a travelling hinge. Expressions for the mean collapse load when the
shell material is rigid-perfectly plastic are developed.
Predicted loads based on the collapse modes encountered, when compared with those obtained
experimentally, using rigid PVC tubes of constant axial length and groove dimensions (depth and
length), are found to be in good agreement.

NOTATION

A cross-sectional area at a grooved region of the shell

c length of circumferential yield line
Dg minimum or root outside diameter of a grooved region of the shell
Di inside diameter of shell
Dr outside diameter at a ringed region of the shell
h height of lobe
k number of inclined hinge lines per lobe
lg axial length of a groove
l, axial length of a ring
L axial length of shell
m number of lobes in the axial direction
Mpg full plastic moment per unit length at grooved regions of shell
Mpr full plastic moment per unit length at ringed sections of shell
ng number of grooves per lobe
nT number of rings per lobe
P axial load
P mean collapse (post-buckling) load
P~nax peak load
P~ yield load
r radius of curvature
tg wall thickness at grooved regions of shell
tr wall thickness at ringed regions of shell
W plastic work dissipated
~" mean yield stress of material
z depth of a groove
e natural tensile strain
tr true tensile strain
+ angle

INTRODUCTION

The structural behaviour of relatively thin-walled components under static or dynamic

loading has been given much attention in crashworthiness studies as energy-absorbing
devices [1]. Tubular and conical components of circular and non-circular sections when
they collapse plastically under axial compression provide one of the best energy absorb-
ing devices (see the work reported in refs [2-6]). Crumpling mechanisms for such devices
using the concepts of stationary and travelling plastic hinges for calculating the buckling
load and the amount of energy absorbed have been proposed [4,6].

41
42 A. (; M..~M~11set a/

In this paper we report on theoretical and experimental investigations into the quasi-
static plastic collapse of a certain structural component, namely a thin-walled tube con-
taining a number of geometrical discontinuities in the form of a groove of constant depth
and axial length, with the aim of providing information about its behaviour as an energy-
absorbing device, e.g. details of the failure modes, load-deflection characteristics and
the influence of the plastic properties of the plastically collapsing material. An incx-
tensional collapse mechanism for folding the shell in a non-axisymmetrical diamond fold
mode, taking into account the concept of stationary, circumferential and inclined, travel-
ling plastic hinges, is considered. The plastically dissipated energy and the mean post-
buckling load necessary to perform the operation are estimated.
Experiments were carried out to verify the theoretical model developed. The shells,
made of rigid polyvinylchloride (PVC), were subjected to quasi-static compressive
loading; the modes of failure and the load-deflection characteristics were determined
and compared with the theoretical ones.
The work reported here is limited in as far as the results obtained are for PVC behav-
ing quasi-statically, and not dynamically, as would be the case for an impact energy
absorber.

EXPERIMENTAL OBSERVATIONS OF COLLAPSE MODES

The axial compression of thin-walled grooved cylindrical tubes between the parallel
steel platens of an Instron testing machine was carried out. The test material used was
commercial rigid polyvinylchloride PVC [4], the stress-strain curve of which, as obtained
from a quasi-static tension test, is given in Fig. 1. From this curve, modelled as a rigid
perfectly plastic material, the initial yield stress was estimated to be 0.0480 kN mm
The tests were carried out at a crosshead speed of 10 mm rain -1 or an overall com-
pression strain rate of 10 -3 s
Details relating to the specimen dimensions are presented in Table 1 and refer to the
geometrical representation of Fig. 2a. The initial axial length L of the specimens and the
length and depth of groove were kept constant whilst the ring length lr varied in the range
O<-lr/L<-l, which corresponds to a number of rings (with the two extreme cases being
tubes of different thickness) (see Table 1). The specimens were machined from com-
mercial tubes of constant wall thickness with the end faces machined square. All
specimens were frictionlessly axially compressed in a 'dry' condition.
A series of photographs of buckling modes were taken during the various crumpling
stages, which, along with terminal top and bottom views of the buckled specimens, are
shown in Fig. 3(a)-7(a). After collapsing, the PVC tubes were immersed in boiling water
for a few seconds and they immediately and completely recovered their original form,

0.05 j ~ e r f e c t l y ptostic materiot

~E
E 0-04
z
m"
m 0.03
o
2m
0.02
E

0.01
@

0,05 0,10 0.15 0 20

Natural t e n s i t e strQin
FIG. 1. Tensile stress-strain curve for rigid PVC.
 TABLE la.

"-" =r' ~. Buckling load (kN) P/Py

~, ~',
~ ~ ~, O
~ < ~'~
Inside Outside Axial Ring Length Pmax
='~ ~ ,.~ ~,. diameter, diameter length length No. of last Elastic Initial Y=--
~ ~ ~ ~ ,~"
~-- ,~.m''~ ~ • ~'~ Specimen Di Dr Dg L lr of ring recovery peak Mean fi A
No. Shape (mm) ( m m ) ( m m ) (ram) (mm) rings (mm) (mm) Pmax Expt. Theoret. (kN/mm 2) Expt. Theoret.

1 Cylinder 41.7 48 -- 127 127.0 1 0.0 7.3 26.6 10.2 10.7 0.060 0.371 0.389
Grooved 21.6 6.6 6.8 0.065 0.320 0.330
2 41.7 48 46.5 127 25.4 4 12.7 11.3
~ ~ ~" Cylinder
3 " 41.5 48 47.0 127 19.3 5 16.0 7.1 22.4 10.7 10.3 0.059 0.451 0.435
4 " 41.6 48 46.7 127 12.7 7 15.4 9.5 22.5 7.7 7.5 0.064 0.351 0.342 O
5 " 41.7 48 46.7 127 6.35 13 2.8 7.9 22.1 9.2 8.6 0.064 0.427 0.400
" o ~ = ~ ~ 6 " 41.3 48 46.9 127 3.175 19 5.3 9.7 22.6 7.3 7.2 0.058 0.304 0.299
~ ; ~ ' ~ ~
7 Cylinder 41.6 48 46.2 127 0.0 -- 0.0 12.5 20.0 6.5 6.3 0.063 0.331 0.320
O
Ycomp. =0.062 kN mm 2. A=~(D2g-D2i)/4. Axial length of groove: /g=3.175 mm. O
Yt.... = 0.048 kN/mm w2. Pv = YcompA. Depth of groove : z = 0.75 mm. g.
~.. r~ O"t~ O ~" O
=,
TABLE l b .

Buckling modes at the various

~-~ ~ o ~. ~.~ Specimen levels of folds
~=B.o =o., No. a b c d e Remarks:
o
1 A-2D 2D 2D 2D All patterns : Ellipse (Axes : 70/63, 63/73, 73/68, 68/53)
2 2D 2D 2D 2D All patterns : Ellipse (Axes : 52/70, 70/56, 56/66, 66/66)
i ° 3 2D 2D 2D 2D All patterns : Ellipse (Axes : 57/63, 63/67, 67/ /74)
4 2D 2D 2D 2D All patterns : Ellipse (Axes : 28/74, 74/62, 62/701 70/69)
5 2D 2D 2D 2D 2D All patterns : Ellipse (Axes : 55/64, 64/53, 53/70, 70/69, 69/54)
6 2D 2D 2D 2D All patterns : Ellipse
7 A-2D 2D 2D 2D All patterns : Ellipse (Axes: 54/69, 69/65, 65/60, 60/65)

A: axisymmetric ring. Note: Dimensions are in mm.

2D: 2-diamond (ellipse).
44 A. (i. M.,',MALISt'l al.

i
ll p
Or -- irlc[ined
J xij ~ hinge l i n e s

fEJlg

grooves(

IP
(o) b
i I I z'z
I I ..Jr"

~.J hi

(c)
Fro, 2. (a) A schematic diagram of grooved tubes in axial loading. (h) Assumed deformation
mode for inextensional collapse. (c) Simplified travelling hinge, showing formation of stationary
horizontal hinges in inextensional collapse.

(a) Deformation modes

In almost all the cases of the grooved tubes examined some common deformation
characteristics are apparent (see Figs. 3-7).
(i) At the first maximum (or peak) load the material yielded and a buckle (a circum-
ferential plastic hinge line) into a two-lobe type of failure started to form in a groove (at
the thinnest section of the tube) at a distance from the top end of the tube adjacent to the
platens of between 1/2 and 4/5 of its axial length (see 2 in Figs 3a-7a).
(ii) After the initial peak in load corresponding to the formation of the elliptical fold
described in (i), the load increased only slightly, in a manner similar to that of column-
buckling (see Fig. 9 for specimens 2-6).
(iii) The development of inclined plastic hinge lines was observed at each side of a
groove where deformation had been initiated, and led to the plastic collapse of the
adjacent ring-sections of the tube with the formation of a two-lobe (elliptical) type of
failure. Note that the major axis of the ellipse at the one level is orthogonal to that of the
ellipse at the next level and then parallel to that of the ellipse at the next level, i.e. level
by level rotations of ~r/2 radians. The major and the minor axes are almost equal for each
ellipse.
(iv) As a result of the compression process the lower part of the grooved specimen 4
ends with its lowest circumference adjacent to the platen becoming an ellipse; lifting of
the material from the bottom platen was also observed and is believed to be due to
tension (see, for example, 3 in Figs 3a and 7a). The top end of the tube remains circular
and in contact with the press platen.
The experimentally obtained deformation patterns for the water-immersed PVC
grooved specimens are shown mapped in Fig. 3(b)-7(b). After the deformation stages
(i)-(iv) as described above for specimen 2 (with lJL=0.20), progressive collapse was
 Collapse of grooved thin-walled cylinders 45

1 2 3 4 5

100
Hinge
tines
50

0
mrn 6 7 8 9
Z~L

10 , , 11
2¢m

Ca)
FIG. 3(a). Buckling modes for spec. 2, see Table 1 for details, 1-8, views of progressive collapse;
9, the 'as recovered' side view, 10, a top view; 11, a bottom view.

X Top end
ring 1
groove 1
ring 2
groove2
/
ring3 \
groove3
ring/-,
groove4
/
ring 5
Bottom end

(b)

FIG. 3(b). Assumed deformation mode for spec. 2.

4~ A. (i, MAMAHSetal.

1 2 3 4 5

100

Hinge
tines
0
mm 6 7 8 9
Ak

10 11
¢m

(a)
Fl(~. 4(a).Buckling modes for spec. 3, see Table 1 for details. 1-8, views of progressive collapse
9, the 'as recovered' side view; 10, a top view; 11, a bottom view.

Topend
ringl ~ ~ ]
groove1 ~
ring2
groove 2
ring3
gr°°ve3~start of ~ ./
ring 4 %.¢,~ttctpse/ "~ /
groove4
ring5 / 5 ~ ~
groove5_ ~5
. . . . . . . Bottomend

(b)

Fl(;. 4(b). Assumed deformation mode for spec. 3.

 Collapse of grooved thin-walled cylinders 47

2 3 4 5

100
nge
50 nes

0
mm 6 ? 8 9
AL

10 ;;cm, 11

(a)
FIG. 5(a). Buckling modes for spec. 4, see Table 1 for details. 1-8, views of progressive collapse;
9, the 'as recovered' side view; 10, a top view; 11, a bottom view.

Top end
ring 1
groove 1
ring 2
groove 2
ring 3
groove 3
ring4
groove 4
ring 5
groove 5
ring 6
groove 6
ring 7
groove 7

Bottom end

(b)

FIG. 5(b). Assumed deformation mode for spec. 4.

4~ A . (;. M..xrq:,t.is ct al.

II

1 2 3 4 5

100

50

0
mm 6 7 8 .,
Hinge
AL lines

, 11
10 2 cm

(a)
FIG. 6(a). Buckling m o d e s for spec. 5, see Table 1 for details. 1-8, views of progressive collapse;
9, the "as recovered" side view: 10, a top view; 11, a bottom view.

Topend[ / ~. / \ Iring 1

2
3
4
5
6
7
8
9
10

11
, 11
12
, 12 . . . . . . . . . . .
13
B i,o1~
end

(b)

FIG. 6(b). A s s u m e d deformation mode for spec. 5.

 Collapse of grooved thin-walled cylinders 49

1 2 3 4 5 6

I00

Hinge
lines
0
mm 7 8 9 10
&L

11 , , 12
2cm

(a)
Fto. 7(a). Buckling modes for spec. 6, see Table 1 for details. 1-9, views of progressive collapse;
10, the 'as recovered' side view; ll, a top view; 12, a bottom view.

groov.
2 ,, Top e n d

11
12
13
14
15 S t a r t of
16 _ _ _ cottapse
17
18
19
Bottom
end

(b)

FIG, 7(b). Assumed deformation mode for spec. 6.

5(i A. (;. M,,X~A[ Is et al.

~
Top e n d
First
lobe
(2D)
Second
lobe
(2D)

I
~ Third
lobe
(2D)

Fourth
lobe
(2D)
Bottom end

(a)

FIG. 8(a). Assumed deformation mode for spec. 1.

Top e n d
First
lobe
(2D)
Second
lobe
(2D)
Third
lobe
(20)
Fourth
Lobe
(2D)

Bottom end

(b)

FK;. 8(b). Assumed deformation mode for spec. 7.

developed in groove 2 with the formation of an ellipse orthogonal to the previous one of
groove 3 (see 5 in Fig. 3a); further, the ellipse in groove 1 is set at an angle of ~r/2 to the
one in groove 2 (see 6 in Fig. 3a), there having been simultaneous spreading of the in-
clined plastic hinges in rings 1 and 2. We observe that in the final stages of deformation the
top circular end of a specimen is terminally distorted to, essentially, an ellipse, also losing
contact with the upper steel platen along its major axis (see 6-8 in Fig. 3a). Moreover,
the levels of the ellipses in rings 3 and 4 did not remain horizontal but moved downwards
until the ends touched the bottom platen (see 5 in Fig. 3a). This resulted in an increase of
buckling load indicated in these stages of deformation by Fig. 9. Fracture was found
along most of the circumference of the first buckle in groove 3, probably due to the sharp
corner of the groove acting as a stress raiser.
We must point out that some elastic recovery is found in all collapsed specimens after
unloading.
The covering of ring 5 takes place by the initial elliptical collapse of groove 3 moving
downwards (see 4 in Fig. 4a); where inclined hinge lines spread into rings 5 and 6 there is
the formation of an ellipse along the circumference of groove 2 at an angle of ~r/2 relative
to the previous one. Grooves 4 and 5 in Fig. 4(a) are seen at characteristic stages of
deformation after the initiation of collapse (stages i-iv) in specimen 3 which has a ratio
l,/L=O. 15. Final collapse stages are clearly indicated in 6-8 of the same figure along with
the deformation patterns shown in Fig. 4(b). Note that the end of the grooved specimen
after collapsing is an ellipse and that no fracture occurred.
Stages of deformation subsequent to the initial stages (i)-(iv), for specimen 4
(l,/L=O.lO), specimen 5 (l,=0.05) and specimen 6 (l,/L=0.025) show the progressive
deformation of grooves into ellipses with the inclined hinge lines spreading along the
adjacent rings, each lobe have rotated through "rr/2radians relative to its previous one.
 Collapse of grooved thin-walled cylinders 51

Hinge
lines

Specimen 1 Specimen 7
I I

20mm

(c)
FIG. 8(c). Recovered tubes (specimens 1 and 7) being immersed in boiling water showing
deformation zones.

Material was seen to move downwards and touch the platens, the top end of the grooved
specimen being finally distorted and losing contact with the steel platen along its major
axis. This is well shown in Figs 4(a)-7(a) where the obtained deformation patterns for the
immersed specimen are clearly indicated in Figs 4(b)-7(b).
As would be expected, the buckling wavelength is fully restricted by the geometrical
discontinuities caused by machining the specimens. Thus, for the tubes with four and five
rings (specimens 2 and 3, respectively) the lobes are formed after each ring and for the
tube with seven rings (specimen 4), the wavelength consists of two adjacent rings. For
the specimens with more rings than seven (specimens 5 and 6) structural behaviour
approaches that of a non-grooved specimen, with equal wavelengths.
For comparison purposes two tubes without any geometrical discontinuities, of the
same initial length and inside diameter as the grooved specimens, were caused to
undergo collapse in the same manner as the grooved specimens. The two tubes have
different wall thicknesses corresponding to (i) Ir/L=l, specimen 1, i.e. with an outside
diameter Dr and (ii) lr/L=O, specimen 7, with an outside diameter Dg. They both showed
the well-known structural behaviour in which the initial plastic and axisymmetric (ring)
buckle system develops into a non-symmetric two-diamond (elliptic) pattern with a ~r/2
radians level-by-level rotation as subsequent loading occurred (see Table 1). The in-
extensional progressive collapse modes with the formation of circumferential and
inclined stationary hinges, and travelling ones as given in ref. [6] and as experimentally
obtained in the present work, are shown in Figs. 8(a) and (b).
52 A.G. MA~,I,XHSet aL

"%\ / ..... -

"<" I--#: 6.6kN

- _ t i _ i i i •
0 0.1 0.2 0.3 0.4 0.5 0.6
Deflection / L
(a)

5 3°I
201
o

A
c~
0f"k / A 093kN
J ~ . . . . . • L___ I . . . . . .

0 0.I 0.2 0.3 0.4 0.5 0.6

Deflectionl L

(b)
FIG. 9(a) Load-deflectioncurves for specimens 1-4; - - spec. 1, - - - - - spec. 2 , - - - -
spec. 3, -..- spec. 4. (b) Load-deflectioncurves for specimens 5-7:- spec. 5, -- - -- spec.
6, - - spec. 7.

(b) Buckling load

The load-deflection characteristics for the various modes of collapse for the grooved
specimens 2-6 and the two tubes without geometrical discontinuities, 1 and 7, are pre-
sented in Fig. 9. Initially, the shell behaves elastically, and the testing machine load rises
at a steady rate. Collapse similarly begins for all the specimens examined (see Fig. 9)
with the rate of change of load thereafter falling similarly. After the maximum value, a
fold is developed, which then extends to its maximum degree. For all specimens the
initial part of the load-deflection diagram clearly has about the same shape.
The grooved specimens show features characteristic of buckling; the load rapidly
increases to reach a maximum and later the post-buckling phase is extended with suc-
cessive rises and falls in load level as the various plastic regions form and develop. The
secondary peaks of the buckling load in the final stages of deformation occur with the
touching of the bottom platen by the deformed material (as ellipses).
Note that the maximum load, i.e. initial peak, for the grooved specimens is almost
constant since yielding occurs at the thinnest section of the specimen which is the same
for all the specimens examined (see Table 1 and Fig. 9).
The well-known initial elastic behaviour of a tube with a rise of load at a steady rate
and with the same slope as earlier described, the appearance of the peak load, the rapid
falling off of the load until an axisymmetric ring is formed and the change into the 2-
diamond fold with a series of fluctuations about a mean post-buckling load for the sub-
sequent 2-diamond lobes developed are observations characteristic of collapsing tubes
which did not have geometrical discontinuities, e.g. specimens 1 and 7. For experi-
mentally obtained results for peak load, see Table 1.
 Collapse of grooved thin-walled cylinders 53

THEORETICAL MODEL OF PLASTIC COLLAPSE

Based on the experimental observations of the collapse modes outlined above (see
Figs 3b-7b), a theoretical model of inextensional plastic collapse has been developed.
Figure 2(b) shows a generalized, developed shape of the shell surface after a non-
symmetric deformation mode. In the analysis the assumptions made for the inextensional
collapse of thin-walled tubes [6] and conical shells [4] are:
(i) The material is rigid-perfectly plastic with infinite ductility;
(ii) The middle surface of the shell remains unextended and its position does not
depend on the deflections which have taken place: this is to say that membrane
stretching is neglected, among other things;
(iii) The compression is frictionless and both ends of the shell are free to deform or
change shape as they choose;
(iv) The developed theoretical model is based upon the terminal collapse mode and
not upon the history of plastic deformation;
(v) Strains due to direct and shear stresses are neglected and only the vertical action
of the load P is considered.
A collapsing shell contains two sorts of plastic hinges, horizontal ones around the
circumference of a groove and inclined hinges crossing between the rings.
The various stages of plastic compression during the formation of one lobe are:
(i) Flattening each layer from its initial radius of curvature to infinity.
Each layer (see Fig. 2b and also Figs 3b-7b) consists of a number of curvilinear
trapeziums which make up the whole surface of the shell and after full compression these
become fiat trapeziums separated by a horizontal line along which acute bending has
taken place (see also similar remarks in refs [4] and [6]).
The plastic work done in completely flattening and removing the curvature of the
layer, say number j, consisting of nr rings and ng grooves, is, following the notation
given above:

Wl =2"rr (nrlrMpr+nglgMpg)p (1)

where Mpr=Y~/4 is the full plastic moment per unit length in ringed-regions of the shell
and Mpg= y t 2 / 4 is the full plastic moment per unit length in a grooved region of the shell.
(ii) Bending along the stationary inclined hinge lines crossing the above-mentioned
rings and grooves. Plastic work dissipated during a -rr-radians rotation of the shell-portion
at stationary inclined hinge lines is

W2 = ~rkj( nrlrMpr + nglgMpgVCOS~bj, (2)

where kj is the number of inclined hinge lines of lobe j and

~bj=tan-'{(xo-xsj)/h~} (see Fig. 2b).

(iii) Bending along the stationary horizontal circumferential hinge line, number j, the
plastic work dissipated being

W3='~c Mpg. (3)

(iv) The mechanism of a travelling hinge for the formation of the stationary horizontal
hinges is now considered. According to the mechanism described in refs [6] and [4],
collapse does not occur simultaneously along the whole of a hinge line, but is 'laid down'
in a progressive manner with travel along the specimen in circumferential bending and
unbending hinges; with the mid-surface inextensibility condition maintained, this
mechanism describes the progress of the travelling hinge - - how it reverses direction and
travels outwards again until the limit of its travel is reached when two more stationary
bends are formed. This is illustrated in Fig. 2(c). The plastic work done in the travelling
hinge around the entire effective circumference of a groove is
5.1 \. (i M,\~lxJls ~'lal'

(l,, wrl I.,

4i
F Trr

The work done by the mean axial load Pi is

W , = P I It, 2r). ~'~1

If the plastic work done by the axial load is completely dissipated at the plastic hinges
then by equating the external work to the above calculated internal work dissipated,

- lg
Pj( h i - 2r)= WI + W2 + W3 + 2W3( - - - 1),
~TF

Or

~5i= W I + W 2 - W s 2WJg
+ ~ (6)
hl- 2r wr(hi-2r)
Differentiating equation (6) with respect to r gives that value of ~ which makes fii a
minimum. It is obtained as

-B+ ~/BZ-4AC
ri* -- 2A /7)

where A = W 1 --1-W 2 - W 3 , B=4W3lg/~ and C= - W3hjlg/~.

Thus, the mean post buckling load P for the collapse of the grooved tube is
m

m
(s)
~,' hi
is

where m denotes the total number of formed layers.

Values of P for the grooved specimens 2-6 obtained from the theoretical model
outlined above and for the tubes without discontinuities, specimens 1 and 7, using the
inextensional collapse analysis of ref. [6], are given in Table 1. It will be convenient to
compare the mean post-buckling load/5 with the end load Pv that would cause simple
yielding at the groove, i.e. the thinnest section of the shell where plastic collapse is
initiated in direct compression.
If Y is the initial uniaxial yield stress of the material (see Fig. 1), then

P,,= '7~(D~ -O1 )/a. (,~)

Values of fi/Pv calculated from the theoretical model for inextensional collapse and
experimentally predicted from the present work (see Table 1) are plotted against ljL
(length of the ring-section of the grooved specimen/its initial axial length) in Fig. 10.

DISCUSSION AND CONCLUSIONS

The deformation mechanism described, which is based on experimental observations,

accounts for the crumpling of thin-walled grooved tubes subjected to axial loading.
For the geometries examined the shells collapse by inextensional folding into non-
symmetric diamond-shaped lobes (of elliptical form) first yielding at a circumferential
plastic hinge which forms at a groove. The proposed deformation mechanism for this
case assumes inextensibility of the shell mid-surface with the deformation being accom-
plished by bending, mainly concentrated at plastic hinge lines. The possible form of
stationary circumferential and inclined hinges and travelling hinge regions of constant
radius which generate the stationary hinges were examined.
 Collapse of grooved thin-walled cylinders 55

Experimental tests on rigid PVC specimens provided basic deformation patterns and
the simple failure mechanisms proposed. Collapsed PVC shells immersed in boiling
water immediately recovered (almost) their initial form showing clearly the various
regions of heavily concentrated bending.
To check for circumferential inextensibility of the mid-surface of the shell, measure-
ments of the deformed circumferential lobes at any level were made and allowed the
diameter of the circumscribed and inscribed circles of the non-symmetric diamond folds
to be calculated (see Table 1). No significant increase in the circumference was observed.
Examination of the specimens indicated that the travelling hinge is probably of con-
stant radius around the circumference and throughout the deformation process; similar
observations were reported in ref. [4] for the collapse of conical shells. It has to be noted,
as mentioned above, that some elastic recovery takes place by unloading.
Plastic analysis based on the proposed mechanism of inextensional plastic collapse,
and assuming rigid-perfectly plastic material, allowed the prediction of the mean post-
buckling load and the energy absorbed by the shells. Comparison of those with the
present experimental results for rigid PVC grooved tubes and plain tubes without dis-
continuities (see Fig. 10 and Table 1) indicates that for the geometries examined,
experimental and theoretical values of mean post-buckling load are in good agreement
and in the range of +5%.
Shell geometry and the groove geometrical discontinuities involve factors affecting the
deformation modes and the buckling loads. The initial peak load corresponding to the
elasto-plastic behaviour of the shell and the post-buckling load seem to remain almost
constant for all ratios of lr/L (length of the ring-section of the grooved specimen/total
axial length of the shell). Dividing the peak load by the cross-sectional area of the
groove, where yielding of the material was initiated, gives the initial compressive yield
stress of the material at 0.062 KN mm -2, which is a reasonable value. The initial tensile
yield stress of rigid PVC taken at 3% of strain is 0.048 KN mm -2 (see Fig. 1). The dif-
ference is due to the yield behaviour of the PVC material; it is slightly sensitive to the
hydrostatic component of stress. Note also that the experimentally obtained value from a
quasi-static test compared to the initial compressive yield stress of similar PVC material
was 0.0658 KN mm -2 at 3% of strain used for the work reported in ref. [4].
Calculated values of P/Pr using the estimated compressive yield stres of the material
(see Table 1) were plotted in Fig. 10. Taking into account all the assumptions made the
theoretical (and experimental) results for PVC fit well the equation,

=0-03 Ir +0.356. (10)

Py L

0.5

0.4

>-
0.3 ~, py=0.03 Elr .o.3s6

0.2

0.1

I L I __

0 0.1 0,2 1.0

lrl L

Ft6. 10. Variation of P/Pr with the ratio lJL: see Table 1 . 0 : Theoretical. X: Experimental.
5~ A. (;. M.~xM,xJis cta/.

Acknowledgements--The authors are grateful to Professor P. S. Theocaris of the National lcchnical Univcr~,it~
of Athens for providing the laboratory facilities for the present experimental work, Professor W. Johnson ~s
grateful for the support of the Levcrhulme Trust. The authors are also very indebted to Mrs R. Tsopela t(}r
typing their manuscript. The suggestions for improving the initial draft of this paper, made by Profcsst~r
Norman Jones and Professor S. R. Reid. were much appreciated.

REFERENCES

1. W. JOHNSON and A. G. MAMALIS, Crashworthiness ~)/' Vehicles. Mechanical Engineering Publications.

London (1978).
2, A. G, MAMALISand W. JOHNSON~The quasi-static crumpling of thin-walled circular cylinders and frusta
under axial compression. Int. J. Mech. Sci. 25, 713 (1983).
3. A. G. MAMALlS,W. JOHNSON and G. L. VIEGELAHN,The crumpling of steel thin-walled tubes and frusta
under axial compression at elevated strain-rates: some experimental results. Int. J. Mech. Sci. 26, 537
(1984).
4. A. G. MAMALIS,W. JOHNSON, D. E. MANOLAKOS,G. L. VIEGELAHNand N. M. VAXEVAN|DIS,On the axial
collapse of thin PVC conical shells (in press).
5. P. D. SODEN, S. T. S. AL-HASSANIand W. JOHNSON,The crumpling of polyvinylchloride tubes under static
and dynamic axial loads. In Mechanical Properties at High Rates of Strain (Edited by HARDING J.),
pp. 327-338. Institute of Physics, London, Conf. Series No. 21 (1974).
6. W. JOHNSON, P. D. SODEN and S. T, S. AL-HASSANk Inextensional collapse of thin-walled tubes under
axial compression. J. Strain Anal. 12, 4 (1977).