BUCKLING MODELS OF THIN CIRCULAR PIPES ENCASED IN

RIGID CAVITY

By A. M. Omara,t L. K. Guice,z W. T. Straughan,) and F. A. Akl 4

ABSTRACT: Theoretical models for the structural design of cured-in-place plastic liners used in the rehabili-
tation process of deteriorated pipelines are reviewed. An introduction of the buckling theory of a free-standing
circular ring, a model that has been used widely for the design of liners, is presented. Approaches for analyzing
the buckling behavior of a thin circular ring encased in a rigid host pipe are discussed. A comparison between
different models and experimental results is presented. The paper concludes that the model proposed by Glock
for predicting the buckling pressure of a thin circular ring encased in a rigid cavity compares most favorably
Downloaded from ascelibrary.org by Florida International University on 10/02/13. Copyright ASCE. For personal use only; all rights reserved.

with the experimental results.

INTRODUCTION In his experimental work, Fairbairn concluded that the pipe

Structural design methodologies for pipeline rehabilitation length and the ratio of diameter to wall thickness are important
systems are still in their infancy when compared to those for parameters in determining the buckling pressure. Bresse
components in bridges, buildings, and other structures. Yet, (1866), using small deflection theory, studied the stability of
design approaches for these older structural systems are con- a thin free-standing (no outside constraint) circular ring under
tinuously evolving and improving as more research is con- external hydrostatic pressure (Fig. 1) and arrived at the fol-
ducted and more experience is gained. The pipeline rehabili- lowing expression:
tation design process is also evolving, and as new research 3£1
results are obtained, it is expected that enhancements in the Pcr = Ji3 (1)
design process will lead to more cost-effective and durable
designs. Tens of millions of feet of sewer pipelines have been where P cr = critical buckling pressure; E = modulus of elas-
rehabilitated with cured-in-place plastic (CIPP) pipe systems. ticity; I = moment of inertia of the ring's cross section; and R
CIPP liners are made of a thermosetting resin that is cured = mean radius of the ring.
inside the pipeline, forming a snug fitting condition within the Bryan (1888) used the minimum potential energy criterion
existing pipe. Some of these CIPP systems have proven that of stability to formulate a similar expression to (1) for the case
they can meet and exceed specified performance requirements. of an infinitely long free-standing pipe under hydrostatic ex-
However, with the tremendous needs faced by the United ternal pressure. His equation differs from (1) in the term E,
States in rehabilitating the underground infrastructure, it is im- which has been replaced by E/(1 - v 2) to account for the plane
perative that future systems be designed economically and strain condition of the infinitely long pipe. Considering this
with the highest quality to serve the public for many years. modification and substituting for I = bt 3/12, where b is the
The purpose of this paper is to provide an overview of re- width and in this case is unity, the resulting equation is
search that is applicable to the structural design of CIPP pipes.
Specifically, models that are related to the stability of flexible 2£
Pcr = 1 _ V 2
(t)3
D (2)
pipes formed inside stiff host pipes are considered, and the
effects of soil loading are not considered. The present paper where D = mean diameter of pipe; t = mean pipe wall thick-
reviews literature dealing with the basic theories for the buck- ness; and v = Poisson's ratio. This equation can be written in
ling phenomenon of thin rings and cylinders under various the following form:
loading forms, boundary conditions, and failure configura-
tions. It begins with the analysis of an unconstrained thin 2£ 1
Pcr = 1 - v 2 (SDR - 1)3
(3)
ring's buckling behavior, a relatively simple model but one
that has been used widely for the design of CIPP liners. The where SDR = standard dimension ratio = outside pipe diam-
paper also presents approaches for the buckling behavior of a eter/mean pipe wall thickness.
thin ring rigidly encased within a confining enclosure ring. The
theoretical approaches are compared with available experi-
mental results.
BASIC BUCKLING THEORY OF FREE RING
Research on cylindrical tubes under external pressure dates
back to the middle of the nineteenth century (Fairbairn 1858).
'Doctoral Candidate, Dept. of Civ. Engrg., Louisiana Tech Univ., Rus-
ton, LA 71272.
2Prof. of Civ. Engrg. and Academic Dir., Louisiana Tech Univ., Ruston,
LA.
'Asst. Prof. of Civ. Engrg., Louisiana Tech Univ., Ruston, LA.
'Prof. of Civ. Engrg. and Grad. Coordinator, Louisiana Tech Univ.,
Ruston, LA.
Note. Associate Editor: Raymond Plaut. Discussion open until May I,
1998. To extend the closing date one month, a written request must be
filed with the ASCE Manager of Journals. The manuscript for this paper
was submitted for review and possible publication on February 26, 1996.
This paper is part of the ]ourlUll of Engineering Mechanics, Vol. 123,
No. 12, December, 1997. ©ASCE, ISSN 0733-9399/97/0012-1294- FIG. 1. Buckling of Unconstrained Circular Ring Loaded with
1301/$4.00 + $.50 per page. Paper No. 12711. Uniform Hydrostatic Pressure

1294/ JOURNAL OF ENGINEERING MECHANICS / DECEMBER 1997

J. Eng. Mech. 1997.123:1294-1301.

 BUCKLED PORTION
Other models for the stability of a circular free ring under
different types of loading can be found in the literature; see
Boresi (1955), Bodner (1958), Wasserman (1961), Wempner
and Kesti (1962), Smith and Simitses (1969), and Lardner
(1980).
Eqs. (1) and (2) are frequently credited to Timoshenko be-
cause of the summary of ring stability presented in his classical
text (Timoshenko and Gere 1961). Timoshenko's equation has
served for many years as the basis for designing CIPP pipe
liner systems. Timoshenko's equation is primarily derived for
an unconstrained circular pipe loaded by an external hydro-
static pressure. The current practice in the design of CIPP uses
an enhancement factor for the buckling pressure to account for
Downloaded from ascelibrary.org by Florida International University on 10/02/13. Copyright ASCE. For personal use only; all rights reserved.

the failure of Timoshenko's model to consider the existence

of the rigid constraints enclosing the plastic liner. It is referred
FIG. 2. Snap-Through Buckling of Thin Ring
to as an enhancement factor because it considers the enhance-
ment in the pipe strength due to the existence of confining
compression load as a function of the height of the imperfec-
constraints. Despite this effort to modify Timoshenko's equa-
tion, its suitability for CIPP design purposes has been ques- tion. Chan and McMinn (1966) studied the stability of a thin
steel lining installed inside concrete cylinders. Their results
tioned. Models that consider the stability of a circular pipe
show that for the shrink buckling problem there is no critical
encased in a rigid cavity resemble the problem of the plastic
liner used in the rehabilitation process more closely. These load as defined by classical theory. However, a state of unsta-
models are discussed in more detail in the following sections. ble equilibrium could be reached only by a small displacement
from the unbuckled position, which means that an initial im-
perfection is needed to invoke buckling. Bucciarelli and Pian
BUCKLING OF THIN RING ENCASED IN RIGID CAVITY
(1967) studied the effect of three kindS of initial imperfections
Several approaches for the analysis of the instability of a on the resulting shrink buckling critical load.
thin circular ring encased in a rigid cavity can be found in the Chicurel (1968) considered the shrink buckling of a thin
literature. In this type of problem the inner ring is restrained elastic circular ring compressed by being inserted into a cir-
by an outer wall so that the ring can deflect internally and cular cavity of smaller diameter than the outside diameter of
circumferentially but not externally. The amount of the cir- the free ring. Using TImoshenko's formulation for the deflec-
cumferential deformation the ring can experience during load- tion of an axially compressed curved beam, Chicurel was able
ing is dependent on the degree of friction developed on the to develop equations for the prebuckling hoop compression
interface between the outer cavity and the inner ring. Because force for the extreme cases of no slip and no friction. The case
of the existence of the outer wall, the inner ring could collapse of no slip was found to provide a lower bound of the buckling
under the effect of loading by snap-through buckling (Fig. 2), pressure.
as opposed to the symmetrical buckling assumed in the deri- EI-Bayoumy (1972) approached the shrink buckling prob-
vation of Timoshenko's equation (Fig. I). lem of an encased thin ring using the fundamental principles
The types of loading that can lead to the instability of the of the calculus of variations. The differential equations of the
inner ring include thermal expansion and hydrostatic pressure. problem and the boundary conditions were derived and solved
When the ring fails by its thermal expansion, the resultant analytically. Although EI-Bayoumy (1972) used a completely
buckling is referred to as "shrink buckling." Shrink buckling different approach, he reached the same conclusions as Chan
is also initiated when the outer cavity shrinks inward and and McMinn (1966). DaDeppo and Schmidt (1970) provided
forces the inner ring to buckle. Similarly the term "hydrostatic a bibliography for buckling of confined rings. Sun et al. (1994,
buckling" is used hereafter to refer to the failure of the inner 1995a,b,c) presented four different experimental and analytical
ring under the effect of hydrostatic pressure induced in the models for the shrink buckling of a thin ring encased in a rigid
interface between the inner ring and the outer wall. cavity. In these models the buckling stress induced in the inner
The shrink buckling phenomenon is not the same as the ring wall is obtained in terms of the ring geometry and stiff-
hydrostatic buckling phenomenon. For shrink buckling, the ness, the imperfection height, and the interface coefficient of
hoop compressive force is relieved immediately after buckling friction. Their results show that for a geometrically perfect
occurs. For a ring under hydrostatic pressure, the load is sus- system, there is no limit point, a conclusion that is in agree-
tained after buckling is initiated and the energy associated with ment with the results of Chan and McMinn (1966) and El-
the sustained pressure must be considered. Shrink buckling is Bayoumy (1972).
common in applications that involve sleeving the inside of Soong and Choi (1985) studied the shrink buckling of a thin
cylinders, pumps, or other types of pressure components or ring encased in elliptical rigid boundaries. Solving the resul-
storage containers. Sleeving is provided to improve acoustic, tant nonlinear differential equations, the critical end compres-
thermal, electrical, or radiation insulation of the outer cavity sive force was obtained.
or to prevent corrosion and any other mechanical damage. Hy- While these efforts contributed to the understanding of the
drostatic buckling is most common in applications that involve behavior of constrained pipes, the resulting models are not
lining the inside of deteriorated pipelines in order to stop leak- particularly applicable to the problem of hydrostatic buckling.
age, improve the flow hydraulic characteristics, or provide ad- As mentioned before, shrink buckling is a different phenom-
ditional strength for the outer pipe. enon from hydrostatic buckling. Other constrained models that
Research on the shrink buckling phenomenon of a thin ring are more relevant to this specific problem are summarized in
within a rigid cavity began with the analysis of a confined ring the following.
subjected to thermal expansion by Lo et al. (1962). They The buckling strength of an encased ring or a cylindrical
treated the buckled portion of the ring as a curved beam with shell under external pressure has been investigated mainly in
initial curvature. Hsu et al. (1964) approached the same prob- relation to steel liners of waterway tunnels subjected to exter-
lem by considering a small imperfection on the interface be- nal pore pressure. Amstutz (1950, 1953, 1969), a pioneer in
tween the ring and the confining cavity. They defined the critical this field, proposed a two-dimensional theory for the collapse
JOURNAL OF ENGINEERING MECHANICS / DECEMBER 1997/1295

J. Eng. Mech. 1997.123:1294-1301.

 of a long shell assuming that it deflects in proportion to a BUCI:LI!D PORTION !-d;:
I ,
sinusoidal pattern. The Amstutz formula for critical buckling
pressure has been commonly used for design purposes, al-
though its validity has not been proven on a sound theoretical
basis (Yamamoto and Matsubara 1982).
Zagustin and Herrmann (1967) investigated a thin elastic
ring constrained in a rigid circular surface under a uniformly
distributed parallel loading. An analytical solution for the re-
lation between the load and the associated deformed shape of x
the ring was found.
Yamamoto and Matsubara (1977) studied the buckling of a
short restrained circular shell provided with stiffeners at both
ends. The problem was formulated using the finite strip
Downloaded from ascelibrary.org by Florida International University on 10/02/13. Copyright ASCE. For personal use only; all rights reserved.

method. Results show that the existence of any initial imper-

fection decreases the buckling pressure dramatically. However,
the perfect circular case has showed a bifurcation limit load FIG. 3. Cheney's Model of Buckling of Thin Ring Loaded with
similar to those cases with imperfection. Uniform External Pressure
Yamamoto and Matsubara (1982) presented a finite strip for-
mulation for the case of a restrained long cylindrical shell. they defined the reduction induced in the ring buckling resis-
They built their model assuming that in the first stage of load- tance because of the ovalness of the outer host pipe.
ing, deflections are almost uniform and are the same as those In the next sections, we focus our discussion on two models
for a corresponding free cylindrical shell. Buckling in two developed by Cheney (1971) and Glock (1977) for the analysis
lobes occurs as soon as the pressure arrives at the critical value of hydrostatic buckling of a rigidly encased thin ring. These
for the free cylindrical shell. At that moment the shell crosses two models and Timoshenko's equation are more commonly
the gap spacing it from the outer cavity and comes into contact used by engineers in the analysis and design of pipes encased
with the cavity wall. Once the contact occurs, any further in a rigid cavity.
growth of the shell deflection stops and the shell capability
increases to carry additional loads. Based on the numerical Cheney's Model
results, they proposed an empirical formula to predict the
buckling load for a certain gap size and standard dimension Cheney (1971) used small-deflection linear theory to study
ratio range. the hydrostatic buckling of a rigidly encased thin ring. The
Kyriakides and Youn (1984) presented a study for the large- ring is envisioned to be made up of two parts: an upper part
deflection collapse of thin-walled confined circular rings under that buckles inward from the rigid circular wall, and a lower
external pressure. The ring is assumed to be inextensional and part that bears tightly against the rigid circular ring wall, as
shown in Fig. 3. In the upper part of the ring, the ring is free
to have an initial imperfection. Using the nonlinear formula-
to take any shape compatible with the differential equation and
tion of strains, the governing differential equations were de-
the boundary conditions at e = ::!:</>. The lower part, however,
veloped and solved numerically. The study shows that the ring
can only have deflection in the circumferential direction; radial
collapse pressure is a function of the initial imperfection and,
deflections are not admitted. He further assumed that the walls
for a perfect ring with no initial imperfection, the ring response
of the cavity move inward with the ring resisting outward
does not exhibit a bifurcation point.
movement but not inward movement. Based on these assump-
Kyriakides (1986) conducted an experimental program on
tions, Cheney derived the following equation for the critical
long cylindrical confined tubes loaded by constant external
pressure in a rigidly encased thin ring:
pressure. The purpose of the study was to establish conditions
under which local buckling could propagate and destroy the (k~r - 1)£/
whole shell. Li and Kyriakides (1991) studied the response Pcr = R3 (4)
and stability of two concentric, smoothly contracting, thin,
elastic rings under external pressure. The outer ring is assumed where kcr = 1.57(Rli)7J5; and i = radius of gyration. For rela-
to be perfectly circular, but the inner one has a small initial tively thin infinitely long pipes (e.g., SDR > 30), the critical
localized imperfection. The pressure is applied hydrostatically pressure can be expressed as
to the outer ring as well as to the cavity formed by the im- 11/5

perfection. This study differs from the aforementioned works _ 2.55£ !... (5)
cr
P - 1- v 2 ( D)
in that it considers the confining structure to be a deformable
ring. The study shows that the presence of the outer ring has
in general a stiffening effect on the response of the structure
Glock's Model
and leads to higher instability pressures. If the outer ring is
relatively thin, the two rings deform in a doubly symmetric Glock (1977) analyzed the stability of a rigidly encased thin
fashion similar to the one followed by an unconfined ring un- ring under external hydrostatic pressure as well as under ther-
der external pressure. When the relative stiffness of the outer mal load. Glock assumed that there is no friction between the
ring is increased, the response of the composite structure (the ring and the rigid cavity, but unlike Cheney's model, Glock's
inner and the outer ring together) becomes stiffer, and the na- model does not require the cavity wall to move inward with
ture of the instability changes to a limit load condition. That the ring. Glock used nonlinear deformation theory to develop
limiting load is dependent on the relative stiffness of the two his model. However, the derivation of Glock's model is not
rings and the geometry of the initial imperfection. For higher fully documented and the theoretical bases upon which he built
outer-ring stiffness, the mode of deformation switches to the his model are not cited. For the benefit of future research, the
snap-through buckling mode. complete derivation of this model is presented, and the theo-
Li and Guice (1995) studied the stability of a thin ring en- retical bases are cited in the following section.
cased in elliptical boundaries and loaded by a hydrostatic pres- Consider the system of polar coordinates and deformations
sure. Using a derivation similar to Chicurel's approach (1968), shown in Fig. 4, where u is the radial displacement and v is
1296 / JOURNAL OF ENGINEERING MECHANICS / DECEMBER 1997

J. Eng. Mech. 1997.123:1294-1301.

 where the approximation p «R and (1 - p/R) .... 1.0 are used
to reach the above expression; and R = ring mean radius. Sub-
stituting (15) into (14), the final forms for u(r, 6) and v(r, 6)
in terms of a and v are

u(r, e) = -u(e); v(r, e) = v + ~ u' (16, 17)

Using (16) and (17), the expression for the nonlinear cir-
cumferential strain, (12), becomes

£88 =e +
88 pI<. (18)
where
Downloaded from ascelibrary.org by Florida International University on 10/02/13. Copyright ASCE. For personal use only; all rights reserved.

e =1.R (v'
88
- u) + _1_2 (U')2
2R (19)

and

(20)
FIG. 4. Displacements In Polar Coordinates
Eq. (18) is the general form for the nonlinear circumferen-
the circumferential displacement. The following assumptions tial strain of an elastic ring in terms of its central axis dis-
are used for the case of a thin ring (El-Bayoumy 1972; Kerr placements. This expression encompasses two parts. The first
and Soifer 1969): e
part, 88 , represents the axial extension of the ring's centroidal
axis, and the second part, PK, represents the change of cur-
= 0; (J'rr«
(J'zz (J'89 (6a,b) vature at the point under consideration. Many researchers such
as Novozhilov (1953), Kerr and Soifer (1969), and EI-Bay-
ezr = eze = 0; eM! = 0 (7,8)
oumy (1972) used (18) to study several problems of circular
With the preceding assumptions, the strain energy expres- rings and arches.
sion for a thin circular ring reduces to In a similar approach to Cheney's model (1971), Glock
(1977) envisioned the ring to be made up of two regions, as
(9) shown in Fig. 5. Region I includes the buckled part of the
ring, where there is no contact between the ring and the ex-
ternal rigid pipe. Region II is the unbuckled portion of the
where V = volume of the ring. Assuming the validity of
ring, and it is in contact with the rigid host pipe. Region I
Hooke's law, (9) can be rewritten as follows:
carries both a hoop force and a bending moment. But in region
II, because no change of curvature is induced in this part of
(10) the ring, the bending moment vanishes and only the hoop force
needs to be considered.
where r = radius of the point under consideration, as shown Substituting (19) and (20) into (11), the strain energy ex-
in Fig. 4. The general expression for the nonlinear circumfer- pressions in regions I and II can be written as follows:
ential strain is (EI-Bayoumy 1972; Kerr and Soifer 1969)

e =-1r (av
-ae + u ) + -2r1 (au
-ae - v )2 + -2r1 ( u + -av)2 (11)
VI =L~ ER(Aei8 + Il<.i) de (21)
88
2 2
ae
For thin rings it is possible to assume that u, v, v' « u'
(22)
(EI-Bayoumy 1972), which reduces (12) to

e 88 =-1
r
, 1
(v + u) + - 2 (u')
2r
2
(12) where
BljCKLED PORTION
where ( )' == [d/(d6)]( ). I (REGION \)
Denoting the displacement of the ring axis by a and v, the I

displacement components u(r, 6) and v (r, 6) may be expressed

in terms of the displacement components of the ring axis as
follows:
u(r, e) = -u(e) (13) a.
\) d~,
u+u'd8
v(r, e) = V(e) + p/(e) (14)
-~~
where p = radial distance from the mean radius of the liner to
the point under consideration (Fig. 4). The unknown function
1(6) is determined from the linearized form of the condition
£r8 = 0; therefore

1(e) =R-1 -+v

ae
(au ') (15) FIG. 5. Glock's Buckling Model of Thin Ring Loaded with Uni-
form Hydrostatic Pressure

JOURNAL OF ENGINEERING MECHANICS / DECEMBER 1997/1297

J. Eng. Mech. 1997.123:1294-1301.

 where aD = radial deflection at e = O. Substituting the preced-
A
£10 =Ii1 (A'
VI -
A)
UI + 2R1 Z (.UI')Z ; (23a,b) ing approximate form of radial displacements into the expres-
sion of NavI and performing the integration results in
(24a,b)
N _ EA<\J [ _
aVI - 2R7T aD 8R <\J
u; (!!:)Z] (32)
The terms al and az denote the radial displacements in
regions I and II, respectively, and VI and Vz are the correspond- It is clear from (32) that N OVI = a constant value, and not a
ing circumferential displacements. Since region II is attached function of e. By substituting (31) into the expression of the
to the host pipe, az and Kz are assumed zero in the expression total potential energy, (30)
of the strain energy, (22). The total potential energy of the
system is II = E/3 L'" (-a
R 0 0
~
2<\Jz
cos 7Te )Z de + L" N;vI R de
2<\J EA 0
(25)

-f
Downloaded from ascelibrary.org by Florida International University on 10/02/13. Copyright ASCE. For personal use only; all rights reserved.

where W = work potential of the external hydrostatic pressure.

By substituting (21) and (22) into the expression of the total
paD cos Z ( ; ; ) R de (33)
potential energy, (25), and collecting similar terms
By performing the integration, (33) reduces to

n = E/ f K~R de + EA {f doR de + L" £ieR de} II = ~.l.

3
16R 't'
(!!:)4
<\J
.z + N;vI R7T _ PR
2 EA
U
o
2 <\Ja o (34)

Using the principle of minimum total potential energy gives

(26)
an. an d.l. = 0
The first term in (26) represents the potential energy due to dII = -
aao du + -a<\J0 't'
(35)
the change of curvature, and the second term represents the
potential energy due to the axial strains of the ring centerline. all all
-=0' -=0 (36a,b)
Substituting (23) and (24) into (26), the total potential energy aa o 'a<\J
in terms of the centerline displacements can be stated as
which results in the following expression:
n = E/3 ('" a';z de + EA { ('" [.!.. (v; - a + ~ a:z]Z R de
)
!!: . 4
2R Jo Jo R
l
2R all _
aao -
.E!...
8R 3 <\J ( <\J ) Uo
+
NavI
aNavI R'TT _ PR _
aa o EA 2 <\J - 0 (37)

+ L" [~Zr R de} - 2 L'" poalR de (27) but since

Another simplification for the preceding expression of the

total potential energy was suggested by Glock (1977). This
~= EA ~
aa o R7T 2
[1 _.!. (!!:)Z aD]
4 <\J R
(38)
modification entails the assumption of
then (37) becomes
aVI avz
-=0' -=0 (28)
ae 'ae
which results in the elimination of the term Uz, the strain en·
ergy in region II. Of course this is not the actual situation in Similarly
that part of the ring. To account for this discrepancy, Glock
(1977) averaged the strain energy density due to the exten- an = _~ E/3 (!!:)4 UZ + N. aNavI R'TT _ PR a _0 (40)
sional strains over the ring's circumference. The reason behind a<\J 16 R <\J 0 aVI a<\J EA 2 0-

that, as shown later, is to get the expression of the total po-

tential energy, (27), i:l terms of only one unknown displace- and
ment, al • Averaging the strain energy due to hoop forces car-
ries the assumption of having a constant hoop force along the ~
a<\J
=1.2 R7T
EA [a 0
+1.8 a; (!!:)Z]
R <\J (41)
entire ring. The average value for this hoop force is calculated
by summing the hoop force in region I and dispensing it over With the substitution of (41) into (40), we obtain
the ring circumference as follows:

NavI = ;~ (L'" ar de - L'" ~~ de) (29)

a
o
[-316 R 3 <\J
E/ (!!:)4 + !is (!!:)Z] =PR2 _
16R <\J
~
2 (42)

By replacing the hoop strain potentials in (27) with equiv- Eqs. (39) and (42) are two equations with two unknowns,
alent terms of the average hoop force, then the total potential aD and $. Solving (42) and (39), we get
energy reduces to z
5 E/ 7T

II = ;~ L'" ar z de + ;A f N;vI R de - L" paiR de (30)

Equating (43) and (32) results in
NavI = 3" R Z ( ~ )
(43)

Eq. (30) is the total potential energy form used in Glock's

solution (1977). An approximate form for the radial deflection 1. (~) a _.!.. (!!:) a; _ ~ E/ .!.. (!!:)Z = 0 (44)
in region I can be assumed as follows: 2 7T 0 16 <\J R 3 EA R <\J
al = aD cosz (;;) 0:5 e :5 <\J (31) Eq. (44) is a quadratic equation in terms of aD with the two
roots
1298/ JOURNAL OF ENGINEERING MECHANICS / DECEMBER 1997

J. Eng. Mech. 1997.123:1294-1301.

 Uo = 4R (~r ± ~16R2 (~r
Similarly, with the substitution of (43) into (39), we get
- :~ (~) 80
3
(45)
TABLE 1.
Model
(1)
Buckling Equation Parameters
Coefficient TI
(2)
Exponent
(3)
13

Timoshenko unconstrained 2 3

-uo = 6~~4 (~r + lOR (~r

Cheney's encased ring 2.55 2.2
(46) Glock's encased ring I 2.2

Equating (45) with (46) lutions with an exponent of 2.2. However, the coefficients can
= (~)2
3 vary significantly depending on the types of assumptions
PR
EI <\>
[1 ±.!.
6
I _ 80EI
'Y 16
-..!.- (.:!:!:)5]
3EA R <\> 2
(47) made. Each of the models presented has merits and under cer-
tain conditions might be the most appropriate one for consid-
Assuming that 1T/<I> = "( and 0: = (PR 3 )/EI then (47) can be eration. However, each of the models also has limitations, and
Downloaded from ascelibrary.org by Florida International University on 10/02/13. Copyright ASCE. For personal use only; all rights reserved.

simplified to even for the most complex of the models some factors have
not been included.
0:=,,( 2 [ 1±(i1 16 _ 80EI -..!.-
3EA R 2 "(
5] (48)
The stability analysis of restrained pipe is complicated by
nonlinear geometries, various interface and boundary condi-
tions, and nonlinear material behavior. Other factors, which
To find the critical pressure Pen we can use must ultimately be considered by a model, include viscoelastic
effects, host pipe geometry (such as ovality), and defects as
ap =0 ::::) ao: =0 well as other anomalies that may occur.
(49)
a"( a"(
EVALUATION OF EQUATIONS THROUGH
or EXPERIMENTAL RESULTS
The application of any analytical model into practice re-
quires validation and calibration through experimental re-
search. Until analytical models have been developed to include
the necessary conditions, the resulting equations must be tuned
400 EI "(4
by experiment, and safety factors must be included to address
+- "(2 3EAlf =0 (50)
various uncertainties.
12 ~ 80 EI 1 5
The equations derived by the preceding approaches may be
16 -"3 EA R 2 "( evaluated through comparisons with available experimental
data. The mathematical tools of linear regression and error
analysis may be used to develop the value of the coefficient
with the root and exponent based on the experimental results.
115
EA 2 Limited published experimental data is available for eval-
'Yer = 0.856 ( EI R ) (51) uating the theoretical models of CIPP pipes. Some organiza-
tions are known to have conducted proprietary experimental
Substituting the value of "(er into (48), we get work related to this study, but most have not been published
215 in the public domain. Three sets of data used in this paper
0.9690 EA 2 include work done by Aggarwal and Cooper at Coventry
O:er = [ 0.4964 ] ( EI R ) (52)
Polytechnic in 1984 (unpublished), Lo et al. at Utah State Uni-
versity (1993), and Guice et al. at Louisiana Tech University
or (1994).
PerR 3 = 0.969 ( EA R 2 )215 (53) Aggarwal and Cooper Tests
EI EI
Aggarwal and Cooper conducted external pressure tests of
Assuming that E is constant for both flexural and axial stiff- Insituform liners at Coventry Polytechnic, U.K. in 1984.. In
ness, and accounting for the plane strain condition of long these tests the liners were inserted in steel pipes. The pressure
pipes, the preceding expression for the critical pressure of thin was then applied and increased between the liner and casing
rings can be modified to a form similar to the case of thin in increments of approximately 1/1 Oth of the expected failure
pipes encased in rigid cavity as follows: pressure until failure. Internal observation was carried out to
2.2 determine when bulging occurred. The experimental failure
Per =1 : v2 (~ ) (54) pressure was found to be much larger than the theoretical
buckling pressure obtained by (3).
An "enhancement factor" was defined by Aggarwal and
CONSIDERATION OF MODELS Cooper as K = Ptest/Ptheory' The enhancement factor reflects the
The models reviewed result in equations for the critical difference between the results by experiment and theory. Ag-
buckling pressure that are similar in form. The basic form of garwal and Cooper indicated that the values of the enhance-
these equations is ment fact varied from 6.5 to 25.8 with a range of pipe SDR
from approximately 30 to 90. Aggarwal and Cooper indicated
P_~ I that 46 of the 49 tests gave a value of K greater than 7. The
(55) term "enhancement" was used because the buckling resistance
er - 1 - v 2 (SDR - 1)~
of the liners appeared to be significantly enhanced by the con-
where TJ = coefficient; and 13 = exponent. straining effects of the host pipe.
The resulting coefficients and exponents for each model are Aggarwal and Cooper's tests included data for 49 specimens
summarized in Table 1. It should be noted that there are con- with a relatively large range of SDR (from 29.86 to 90.25),
sistencies in the exponents for the different models. Those and a variety of material properties (modulus of elasticity from
models that impose a constraint around the surface yield so- 895,700 kPa to 2,521,740 kPa).
JOURNAL OF ENGINEERING MECHANICS / DECEMBER 1997/1299

J. Eng. Mech. 1997.123:1294-1301.

 Shell Development Company Tests TABLE 2. Evaluation of Analytical Models with Aggarwal and
Cooper Data
Shell Development Co. conducted an experimental program
at Utah State University to evaluate the collapse resistance of Equation 'l'I ~ log K K s
(1 ) (2) (3) (4) (5) (6)
CIPP lin~rs made with various epoxy resins (Lo et al. 1993).
The specimens had a constant outside diameter and different Timoshenko 2 3 1.075 11.89 0.139
thicknesses. The results of these tests were also analyzed to Glock 1 2.2 0.Q78 1.20 0.107
Best-fitting regression 1.07 2.17 0.000 1.00 0.107
determine the enhancement factor K. They found the values
of the factor ranged from 9.66 to 15.1.
TABLE 3 Analysis of Data
Louisiana Tech University Tests -
Timoshenko Glock
Louisiana Tech University conducted research on the long- Data log K K s log K K s
term effects of hydrostatic pressures on CIPP and FFP liners (1 ) (2) (3) (4) (5) (6) (7)
Downloaded from ascelibrary.org by Florida International University on 10/02/13. Copyright ASCE. For personal use only; all rights reserved.

under the Corps of Engineers Construction Productivity Ad-

vancement Research program (Guice et al. 1994). Seven dif- Aggarwal and Cooper 1.075 11.89 0.14 0.08 1.20 0.11
Shell Development 1.126 13.38 0.06 0.04 1.10 0.06
ferent products from five companies were evaluated. Several Louisiana Tech University 0.99 9.81 0.11 -0.05 0.90 0.08
short-term tests for each product were also conducted. Test
specimens were 12 in. in diameter and SDRs ranged from 30
to 60. results of the best-fitting regression analysis of Aggarwal and
Cooper's data are also presented in Table 2.
Analysis of Data Since the other analytical models presented are similar in
form and vary only by a constant with either Timoshenko's or
If the term P test is defined as the experimental buckling pres-
Glock's equations, the results for the other models were not
sure and the enhancement factor K is defined as the ratio of
included for comparison in Table 2. The average deviations
P test to Pen the following equation can be written:
between the theoretical and experimental buckling pressures
are smaller for Glock's equation than Timoshenko's equation.
P - KP - K [TJE 2 1 ] (56) Further, a best-fitting regression analysis results in coefficients
test - er - (1 - v ) (SDR - I)~
and exponents very similar to those for Glock's model.
and The test results from Shell Development Co. and Louisiana
Tech University were analyzed for comparison with the Ti-
log [E/(::" V2)] = log TJ - [3 10g(SDR - I) + log K (57) moshe~ko and Glock equations. The data were analyzed to
determme average values of statistical deviations based on
If the subscript i is used to designate individual test results each of the two models. Those results are summarized in Table
(i = I, 2, 3, ... , n), where n is the number of the tests in the 3. The smaller deviations between experiment and theory are
sample, (57) can be expressed in terms of subscript i as fol- again noted for Glock's equation.
lows:
CONCLUSIONS
p,.",J
1og [ E;I(1 _ v2) ] = log TJ - [3 log (SDR i - I) + log K J + Ei The development of several analytical models for CIPP
were presented and compared with experimental data. Results
(58) show that the coefficient and exponent suggested by Glock's
The values of 1) and [3 may be determined by the particular model may be the most appropriate values for use in conjunc-
analytical equation used for the critical pressure Per. The term tion with the experimental results obtained to date. Further-
P test• i is the test result of the ith specimen, whereas E i and SDR i more, it is clear that an enhancement factor K of 7, which is
are the corresponding modulus of elasticity and standard di- in common use in industry to compensate strictly for the de-
mension ratio, respectively. The term KJ reflects the deviation viation between experimental results and the theoretical un-
between the theoretical and experimental buckling pressures constrained pipe critical buckling pressure, is conservative.
of a specific test specimen, while EJ is the associated statistical This factor, of course is not all encompassing, as it does not
error. If the results of a series of short-term buckling tests are address conditions such as host pipe ovality and stiffness,
used to verify a theoretical equation with the corresponding visco-elastic effects, and uncertainties about loading. Other
values of constant log 1) and [3, a sample log K i (i = 1, 2, factors must be used to address these issues in the model.
... , n) can be obtained, and the log K and standard deviation
The analysis and results presented in this paper show clearly
that the development of a new design expression for a pipe
s of the test sample can be determined.
When evaluating different equations, the theoretical formula liner encased in a host pipe is a justifiable and attainable ob-
jective. However, more research is needed before any new
is first selected and the error related to the deviation between
the theoretical and experimental buckling pressures is deter- equation can be recommended as a design expression. An ideal
mined. On the other hand, it would be interesting to determine objective in the development of a new design model would be
to totally eliminate the use of the enhancement factor, which
the coefficients and exponents that can best predict the buck-
ling pressures. To do this, best-fitting regression is employed shows such variability in adjusting the theoretical uncon-
strained pipe critical buckling pressure to actual experimental
using the method of least squares.
results. This can be accomplished through the use of an 1)
coefficient in Glock's model based on a best-fitting linear re-
Results
gression analysis of additional test data.
If the coefficients and exponents presented in Table 1 are Since the bulk of the experimental data is based on 12 in.
substituted into (58), the data may be analyzed to yield esti- diameter host pipes and SDR of 50, more short-term buckling
mates of the mean error log K and the enhancement factor K tests need to be conducted on liners installed in smaller di-
corresponding to the error for each theoretical equation. The ameter and larger diameter host pipes. Liner specimens of both
results of this analysis for Timoshenko's unconstrained model smaller and larger SDR ratios should be tested in both pipe
and Glock's constrained model are presented in Table 2. The diameters. Moreover the CIPP should be tested in host pipes
1300 I JOURNAL OF ENGINEERING MECHANICS I DECEMBER 1997

J. Eng. Mech. 1997.123:1294-1301.

 of known ovality to facilitate the necessary adjustment of the Kyriakides, S. (1986). "Propagating buckles in long confined cylindrical
design expression to consider ovality effects. Finally, a long- shells," Int. J. Solids Struct., 22(12), 1579-1597.
Kyriakides, S., and Youn, S. K. (1984). "On the collapse of circular
term test program of all specimens should be enjoined to ad- confined rings under external pressure," Int. J. Solids Struct., 20(7),
just the design expression for visco-elastic effects. 699-713.
Langhaar, H. L. (1962). Energy methods in applied mechanics. John Wi-
APPENDIX. REFERENCES ley & Sons, New York, N.Y.
Amstutz, E. (1950). "Das Einbeulen von Schacht-und Stollenpanzerun- Lardner, T. J. (1980). "Brief note on the nonbuckling of a circular ring
gen." Schweizerische Bauzeitung, 68, 102-105 (in German). under a wrapping load," J. Appl. Mech., 47, 973-974.
Amstutz, E. (1953). "Das Einbeulen von vorgespannten Schacht-und Li, F. S., and Kyriakides, S. (1991). "On the response and stability of
Stollenpanzerungen." Schweizerische Bauzeitung, 71, 229-231 (in two concentric rings under external pressure," Int. J. Solids Struct.,
German). 27(1), 1-14.
Amstutz, E. (1969). "Das Einbeulen von Schacht-und Stollenpanzerun- Li, J. Y., and Guice, L. K. (1995). "Buckling of encased elliptic thin
gen." Schweizerische Bauzeitung, 87; U.S. Dept. of the Interior, Trans- ring." J. Engrg. Mech., ASCE, 121(12), 1325-1329.
lation No. 826, 541-549. Lo, H., Bogdanoff, J. L., Goldberg, J. E., and Crawford, R. F. (1962).
"A buckling problem of a circular ring," Proc., 4th U.S. Nat. Congr.
Downloaded from ascelibrary.org by Florida International University on 10/02/13. Copyright ASCE. For personal use only; all rights reserved.

Bodner, S. R. (1958). "On the conservativeness of various distributed

force systems." J. Aerone. Sci., 25(2), 152-158. of Appl. Mech., Am. Soc. Mech. Engrs., New York, N.Y., Vol. I,
Boresi, A. P. (1955). "A refinement of the theory of buckling of rings 691-695.
under uniform pressure." J. Appl. Mech., 22(1), 95 -102. Lo, K. H., Chang, B. T. A., Zhang, Q., and Wright, W. J. (1993). "Col-
Bresse, M. (1866). Cours de Mechanique Appliquee, 2nd Ed., Paris, lapse resistance of cured-in-place pipes." Proc., North Am. No-Dig '93,
France. NASTT, San Jose, Calif., E2:1-13.
Bryan, G. H. (1888). "Application of the energy test to the collapse of Novozhilov, V. V. (1953). Foundation ofthe nonlinear theory ofelasticity.
a long pipe under external pressure." Proc., Cambridge Philosophical Graylock Press, Rochester, N.J.
Soc., Vol. 6, Cambridge, U.K. Smith Jr., C. v., and Simitses, G. J. (1969). "Effect of shear and load
Bucciarelli Jr., L. L., and Pian, T. H. H. (1967). "Effect of initial imper- behavior on ring stability," J. Engrg. Mech., ASCE, 95(3), 559-569.
fections on the instability of a ring confined in an imperfect rigid Soong, T. C., and Choi, I. (1985). "Buckling of an elastic elliptical ring
boundary." J. Appl. Mech., 34(4),979-984. inside a rigid boundary." J. Appl. Mech., 52(3), 523-528.
Chan, H. C., and McMinn, S. J. (1966). "The stability of uniformly Sun, C., Shaw, W. J. D., and Vinogradov, A. M. (1994). "Buckling of
compressed ring surrounded by a rigid circular surface." Int. J. Mech. circular rings confined within rigid boundaries," Proc., 17th Annu. En-
Sci., Pergamon Press Ltd., U.K., 8, 433-442. ergy-Source Technol. Conf. and Exhibition, T. J. Kozik, ed., Am. Soc.
Cheney, J. A. (1971). "Pressure buckling of ring encased in cavity," J. Mech. Engrs., New York, N.Y., 62, 219-224.
Engrg. Mech., 97(2), 333-342. Sun, C., Shaw, W. J. D., and Vinogradov, A. M. (1995a). One-way buck-
Chicurel, R. (1968). "Shrink buckling of thin circular rings." J. Appl. ling of circular rings confined within a rigid boundary." J. Pressure
Mech., 35(3),608-610. Vessel Technol., 117(2), 162-169.
DaDeppo, D. A., and Schmidt, R. (1970). "Stability of an arch under Sun, C., Shaw, W. J. D., and Vinogradov, A. M. (I 995b). "Instability of
combined loads; Bibliography on stability of arches." Indust. Math., confined rings: An experimental approach." Experimental Mech.,
20(2),71-89. 25(2),97-103.
EI-Bayoumy, L. (1972). "Buckling of a circular elastic ring confined to Sun, C., Shaw, W. J. D., and Vinogradov, A. M. (1995c). "Discrete-
a uniformly contracting circular boundary," J. Appl. Mech., 39(3), element model for buckling analysis of thin ring confined within rigid
758-766. boundary," J. Engrg. Mech., ASCE, 121(1), 71-79.
Fairbairn, W. (1858). "On the resistance of tubes to collapse." Phil. Timoshenko, S. P., and Gere, J. M. (1961). Theory of elastic stability.
Trans. Royal Soc., London, U.K., Vol. 148. McGraw-Hili, New York, N.Y.
Glock, D. (1977). "Uberkritisches Verhalten eines starr ummantelten Wasserman, E. (1961). "The effect of the behavior of the load on the
Kreisrohres bei Wasserdruck von aussen und Temperaturdehnung." frequency of free vibration of a ring," NASA Rep. IT F-52, Nat.
(Post-critical behavior of a rigidly encased circular pipe subject to ex- Aeronautics and Space Admin., Washington, D.C.
ternal water pressure and thermal extension) Der Stahlbau, Wilhelm & Wempner, G., and Kesti, N. (1962). "On the buckling of circular arches
Sohn, Berlin, Germany, 7, 212-217 (in German). and rings," Proc., 4th U.S. Nat. Congr. ofAppl. Mech., Am. Soc. Mech.
Guice, L. K., Straughan, W. T., Norris, C. R., and Bennett, R. D. (1994). Engrs., New York, N.Y., Vol. 2, 843-852.
"Long-term structural behavior of pipeline rehabilitation systems," Yamamoto, Y., and Matsubara, N. (1977). "Buckling of a cylindrical shell
Tech. Rep. #302, Trenchless Techno!. Ctr., Louisiana Tech Univ., Rus- restrained by an outer rigid wall." Proc.• 27th Japan Nat. Congr. of
ton, La. Appl. Mech., Univ. of Tokyo Press, Tokyo, Japan, 115-123.
Hsu, P. T., Elkon, J., and Pian, T. H. H. (1964). "Note on the instability Yamamoto, Y., and Matsubara, N. (1982). "Buckling of a cylindrical shell
of circular rings confined to a rigid boundary," J. Appl. Mech., 31(3), under external pressure restrained by an outer rigid wall," Proc.• Col-
559-562. lapse and Buckling, Struct. Theory and Practice Symp., Cambridge
Kerr, A. D., and Soifer, M. T. (1969). "The linearization of the pre- University Press, London, U.K., 493-504.
buckling static and its effect on the determined buckling loads," J. Zagustin, E. A., and Herrmann, G. (1967). "Stability of an elastic ring
Appl. Mech., 36(4), 775 - 783. in a rigid cavity," J. Appl. Mech., 34(2), 263-270.

JOURNAL OF ENGINEERING MECHANICS / DECEMBER 1997/1301

J. Eng. Mech. 1997.123:1294-1301.