Society of Petroleum Engineers

SPE 26317

An Improved Design Equation for Tubular Collapse

J.A. Issa, * Exxon Production Research Co., and D.S. Crawford, Stanford U.
'SPE Member

Copyright 1993, Society of Petroleum Engineers, Inc.

This paper was prepared for presentation at the 68th Annual Technical Conference and Exhibition of the Society of Petroleum Engineers held in Houston, Texas, 3-6 October 1993.

This paper was selected for presentation by an SPE Program Committee following review of information contained in an abstract submitted by the author(s). Contents of the paper,
as presented, have not been reviewed by the Society of Petroleum Engineers and are sUbject to correction by the author(s). The material, as presented, does not necessarily reflect
any position of the Society of Petroleum Engineers, its officers, or members. Papers presented at SPE meetings are sUbject to publication review by Editorial Committees of the Society
of Petroleum Engineers. Permission to copy is restricted to an abstract of not more than 300 words. Illustrations may not be copied. The abstract should contain conspicuous aCknowledgment
of where and by whom the paper is presented. Write Librarian, SPE, P.O. Box 833836, Richardson, TX 75083-3836, U.S.A. Telex, 163245 SPEUT.

Abstract Introduction

This paper presents a new equation for predicting the collapse
of tubulars under external pressure. The development of the
equation is based on a large number of non-linear finite
element simulations of tubulars with different geometrical
tolerances and mechanical properties. The simulation results
for the collapse pressure and post-collapse geometry of the
tubulars were verified with full scale physical tests. A non-
linear regression analysis was then performed on the data
obtained from the simulations to optimize the parameters of the
collapse equation. The new equation accounts for variations in
the tubular diameter, wall thickness, ovality, eccentricity, and
material elastic-plastic behavior.
One of the most important considerations involved in the
design of casing and tubing for deep oil and gas wells, and
drilling risers in deep water, is the resistance to collapse under
external pressure. At the depths at which tubulars are now
commonly used, the collapse of tubulars largely determines the
wall thickness and material grade that is required for a given
application. In addition, there is a tremendous cost incentive to
design casing strings properly with respect to collapse. Over-
design leads to costly tubular purchases while under-design can
lead to failures and costly repair operations. For these reasons,
an accurate understanding of the collapse properties of tubulars
is extremely valuable.

This new collapse equation provides significant advantages
over existing formulas, mainly because it was developed based
on non-linear mechanics solutions of the collapse problem as
compared to equations derived from statistical analysis. In
addition, the equation gives the true collapse pressure of the
tubular, whether the tubular collapses elastically or after it has
yielded. The new equation could be used to design tubulars
based on manufacturing tolerances with respect to wall
thickness, ovality, and eccentricity as well as material
mechanical properties. This allows for design optimization
which could account for significant cost savings, especially
when designing expensive non-API tubulars such as corrosion
resistant alloy tubulars. The new collapse equation presented in
this paper provides accurate predictions of the collapse pressure
of tubulars and accounts for tolerances in the tubular geometry
and material elastic-plastic behavior. This new equation is
simple and could be used in optimizing tubular design..
The collapse problem presents difficulty for two reasons. First,
the collapse of tubulars is an instability type of failure and is
sensitive to imperfections such as ovality, eccentricity, and
localized wall reduction. The impact of these imperfections is
difficult to evaluate analytically. The second source of
difficulty arises because collapse can occur in two distinct
modes - the elastic mode in which failure occurs under elastic
deformation and the elastic-plastic mode in which failure is
preceded by permanent deformation of the tubular. Which
failure mode actually occurs depends on the ratio of the tube
diameter to wall thickness, the tube material yield strength, and
the elastic modulus. For tubulars that collapse elastically,
analytical solutions have succeeded in predicting collapse
pressures, but failed to account for the sensitivity of these
pressures to imperfections. For tubulars that collapse in the
elastic-plastic mode, analytical solutions are frustrated by
complex material non-Iinearities and a moving elastic-plastic

7
References and illustrations at end of paper.
2 AN IMPROVED DESIGN EQUAnON FOR TUBULAR COLLAPSE
interface.. Solutions for elastic-plastic collapse are limited to
substituting the "reduced modulus" for the elastic modulus in
the elastic collapse solutions. These solutions are, again,
insensitive to the geometric imperfections of the tubular.
Because of these difficulties, the collapse pressure of tubulars
with imperfections are investigated using numerical
simulations.
A series of numerical simulations were conducted using non-
linear finite element models to predict the coIlapse pressure of
tubulars with different sizes and strengths. The predictions
were verified with fuIl-scaie physical test data for the coIlapse
pressure and post-coIlapse geometry. The numerical simulation
results were then used to develop a generalized coIlapse
equation.
This paper presents the formulation, results, and verification of
the methodology used in developing an improved design
equation for tubular collapse. The dependence of coIlapse
pressures on the tubular diameter to waIl thickness mtio and
material yield strength is fuIly investigated for commercial
steel tubulars. In addition, the sensitivity of coIlapse pressure
to imperfections in the geometry is modeled. A single equation
that accumtely predicts the coIlapse pressure of tubulars of any
diameter to wall thickness mtio, yield stress, ovality, and
eccentricity is presented.
Numerical Simulation of Tubular Collapse
The numerical simulation of tubular coIlapse was based on the
development of non-linear finite element models. A distinct
advantage of the finite element method is the capability of
solving non-linear governing equations. A collapsing tubular is
a complicated physical system governed by highly non-linear
equations for the stresses, the strains and displacements due to
large deformations and material non-linearity.
The finite element modeling was performed using ABAQUS I , a
general purpose, non-linear finite element progmm. The
program allows accurate modeling of material non-Iinearities
(plasticity) and geometric non-Iinearities (large deformations).
The incremental loading of the tubular is controIled
automaticaIly by ABAQUS. The automatic incrementation
allows the progmm to follow the load-deflection curve for the
tubular to the maximum pressure (the coIlapse pressure).
FoIlowing the maximum, the program incrementaIly reduces
the pressure so that the collapse simulation can be continued.
In this way, the post-coIlapse behavior of the tubular can be
modeled.
Description of the Finite Element Model
A finite element model of the cross section of a tubular is
8
SPE26317
shown in Figure 1. The elements used in this model are second
order, isoparametric, generalized plane strain continuum
elements. This element type is an eight-noded quadrilateral.
modified by adding two nodes to model the out-of-plane loads
such as axial and bending loads. Although it is not plainly
visible, the finite element models were constructed with a small
initial ovality. This imperfection was introduced to allow the
coIlapse process to occur. Without this ovality, hydrostatic
pressure could be applied indefinitely without collapse. The
ovality was introduced through the equation:
r =R(I+Scos29) ...(1)
where r is the imperfect tube radius, R is the nominal perfect
tubular radius, S is the ovality ratio, and 9 is the angular
coordinate measured from the center of the tubular as shown in
Figure 2.. An alternative measure of ovality, the ovality factor,
N is also used in this report. N can be expressed as l/S. Also
shown in Figure 2 is the chosen measure of tubular eccentricity.
The numerical simulation is performed by applying a uniform
pressure to the faces of the elements on the outside diameter of
the tubular. Since the pressure is increased incrementally, and
since the pressure is applied along the normal to the element
surfaces, the line of action of the pressure changes as the
tubular deforms. This is an accumte modeling of the action of
pressure which always acts normal to a sutface.
Material Constitutive Relations
The material constitutive relations were based on a strain
hardening elasto-plastic material model using the Von Mises
yield criterion. The generic shape of the material stress-strain
relations was taken from tensile tests of the axial properties of
commercial steel tubulars. However, it was designed to model
the coIlapse of tubulars constructed of materials of various
yield strengths. In order to insure consistency of material yield
behavior, each point on the generic curve was shifted up or
down pamllel to the elastic modulus as required to create stress-
strain relations for materials with any desired yield strength. A
stress-strain relation for an 80 ksi yield material is shown in
Figure 3.
Comparison with Physical Test Data
FuIl-scale physical test data were used to verify the accumcy of
corresponding finite element models using two criteria:
coIlapse pressure and post-coIlapse geometry. The physical test
data used in the verification were obtained from collapse tests
of corrosion resistant alloy (CRA) tubulars and from the
collapse test data used in deriving the plastic collapse equation
of API Bulletin 5C32 .
SPE 26317 1. A. ISSA AND D. S. CRAWFORD
In order to demonstrate the ability of the finite element model
to simulate collapse, each of the specimens was modeled
independently and the model predictions were compared with
the experimental collapse pressures. Table 1 lists the results of
this comparison. All of the collapse pressures were predicted
within 6%. Such an error level is expected because of several
assumptions concerning the material properties. The success of
any collapse simulation using a finite element model is
determined largely by the extent to which the circumferential
material properties of the tubular are accurately represented by
the material model used. In highly cold-worked tubulars,
however, significant directionality of material properties may
be introduced. For these simulations, the axial material
properties of the tubulars were used since they were most
readily available. Some error may therefore be expected. In
addition, the model assumes that the material properties are
uniform throughout the tubular. Variations in the quenching
rate and extent of cold work around the circumference and
along the length of the tubular, however, can introduce
significant variations in the material properties.
In addition to predicting collapse pressures, it is important that
the finite element model is capable of predicting the post-
collapse deformation of tubulars. Figure 4 shows the finite
element model's prediction of the progress of collapse. Figure
5 shows the pressure-deformation response from the collapse
simulation. As shown in this figure, the pressure increases until
the maximum pressure (collapse pressure) is reached, and then
a decrease in the pressure is observed with a large increase in
the deformation. Figure 6 compares a photograph of a cross-
section of the collapsed tubular with the finite element model's
prediction for the same tubular. The excellent agreement
between the collapse test data and the predictions of the finite
element model justify a high degree of confidence in the
model's ability to accurately simulate the collapse of tubulars.
The accuracy of the results stems from the fact that no
approximations were made in formulating the non-linear
equations that govern the collapse phenomenon other than the
intrinsic approximations involved with numerical solutions to a
physical problem.
To further demonstrate the ability of the finite element model
to simulate collapse, the collapse database that form the basis
of the plastic collapse equations of API Bulletin 5C3 was
analyzed. Included in the database are records of the actual
diameter, wall thickness, and the measured yield strength for
each collapsed specimen. The database does not, however,
include any measure of ovalities or eccentricities. As a result,
specific test cases could not be modeled. Instead, various D/t
ratios were modeled with an average yield strength and an
assumed oVality ratio of S = 0.002. A comparison of the model
predictions and the physical data is shown for the three tested
materials, J-55, N-80, and P-110 in Figures 7, 8, and 9. The
predicted collapse values are an excellent predictor of the
9
3
central tendency of the API data. Also shown in these figures
are predicted upper and lower bounds for the API data. These
bounds are based on best case and worst case geometry and
material properties as prescribed by the API 5CT
Specifications3 • The upper and lower bound curves effectively
contain all of the API test data with the exception of a few
widely scattered data points.
Development of a Generalized Collapse Equation
Once a reliable finite element model is developed, it is
desirable to generalize the predictions of the model in the form
of a collapse equation. Such an equation would be capable of
predicting collapse pressures based on all of the important
parameters affecting collapse. The parameters considered here
are: the ratio of the diameter to the wall thickness D/t, the
material yield strength Y and modulus of elasticity E, the
tubular "out-of-roundness" or ovality S, and the tubular
eccentricity e.
The problem of deriving the generalized collapse equation was
first divided into two tasks. The first task involved fully
investigating the relationship between the collapse pressure P,
the ratio of the outside diameter to wall thickness D/t, the
material yield strength Y, and modulus of elasticity E, for
tubulars without geometric imperfections. The second task
consisted of modeling the dependence of the tubular collapse
pressure on the geometric imperfections. In essence, the
following functional form was hypothesized:
D D
P = Po("t' Y, E) g (S,"t) h(e) ...(2)
where P is the tubular collapse pressure, Po(D/t, Y, E) is the
collapse function for near perfect tubulars, and g (S,D/t) and
h(e) are reduction functions that account for tubular ovality and
eccentricity respectively.
Near Perfect Tubular Collapse Function
Over 200 finite element models were used to investigate the
near perfect tubular collapse function. The data from these
models are shown in Figure 10. The data are plotted against
tID rather than D/t on the ordinate since they are more easily
viewed and curves are more easily fit to the data in this format.
For low tID ratios, the collapse pressures are independent of
material yield strength as expected and correspond closely to
the elastic instability equation. For higher tID ratios, the
collapse pressure depends strongly on material yield strength.
In the region between the elastic collapse and the elastic-plastic
collapse, each of the materials undergoes a transition at the
onset of plasticity. This can be seen in the data in Figure 10 as
a sudden change of slope.
4 AN IMPROVED DESIGN EQUAnON FOR TUBULAR COLLAPSE SPE26317

After conducting detailed non-linear regression analyses, the ovalities and has been used to predict the reduction in collapse
best function that fits the data in Figure 10 with an error less strength of tubulars with ovalities as high as S = 0.1 with
than 2% has the following form: reasonable accuracy.

...(3)
Eccentricity Reduction Function

The eccentricity reduction function was derived by studying the

where; impact of eccentricity on the collapse strength in the range of e
from 0 to 0.5. The relationship was then fit with a simple
polynomial using non-linear regression analyses. This
polynomial has the form:

...(5)

where;

Cl =0.8123. C2 = -1.1272

Figure 12 shows the finite element predictions and the

and;
Collapse Imperfection Sensitivity
Once the collapse of tubulars without geometric imperfections
has been described, it is necessary to account for the impact of
these imperfections. The remaining work was aimed at
deriving the reduction functions that account for the tubular
ovality and eccentricity.
Ovality Reduction Function
reduction function approximations. All of the errors are less
than 6% and most are less than 2%. Although the eccentricity
reduction function is based on finite element models of
eccentricities in the range of e from 0 to 0.5, the chosen
polynomial form predicts a collapse pressure of zero for e = 1.
This corresponds to the physical observation that the tubular's
collapse resistance approaches zero as its inside diameter
approaches its outside diameter.
The Generalized Collapse EQuation
The generalized collapse equation is summarized below:

The ovality relation was derived by constructing a series of

models in the range of ovality ratios allowed by the API 5CT
Specifications. The API outside diameter limitations may be
where;
translated through Equation 1 to limits on the ovality variable
S. When this is done, S is found to be constrained between S=O
and S=0.OO2. Figure 11 shows the finite element predictions
and the reduction function approximation. The reduction
function obtained with less than 2% error has the form:

...(4)

where;

Although the ovality reduction function is based on finite

element models of ovality ratios within the API limits, the
chosen hyperbolic function form behaves well at higher

10
SPE26317 1. A. ISSA AND D. S. CRAWFORD 5

and the constants of the equation are defined; r = radius at an angular location

Al = 7.0333, A2 = 0.1295, A3 = 12.3298 R =nominal radius

BI = 0.1648, B2 = 0.5972, B3 = 0.7618
CI = 0.8123, C2 = -1.1272 S = ovality ratio

Conclusions t =wall thickness

1. A general equation for predicting the collapse pressure of Y =yield stress
tubulars has been developed. The equation was derived from
non-linear finite element simulations of the collapse of tubulars v =Poisson's ratio
with varied diameter to wall thickness ratio, material yield
strength, ovality, and eccentricity. e =angular position
2. The collapse simulation data were verified with collapse Acknowledgments
pressures from physical tests. Excellent agreement between
physical test and simulation collapse pressures was achieved in The authors thank Exxon Production Research Company for
every case. permission to publish this work.

3. Significant reduction in the collapse pressure of a tubular References

occurs when a small ovality or eccentricity is introduced.
1. ABAQUS Users Manual, Version 4.8.1, Hibbit, Karlsson,
4. The general equation can be used to evaluate the effect of and Sorenson, Inc., Providence, Rhode Island, February
manufacturing tolerances on the collapse pressure of tubulars. 1989.

5. The new equation could be used in optimizing tubular
design. Significant cost savings could be realized, especially
when designing non-API tubulars such as corrosion resistant
alloys.
Nomenclature
2. API Bulletin 5C3, "Formulas and Calculations for Casing,
Tubing, Drill Pipe, and Line Pipe Properties", Fourth
Edition, February I, 1985.
3. API Specification 5CT, "Specification for Casing and
Tubing", Second Edition, November I, 1989.

AI, A2, A3 = Constants of the elastic-plastic collapse function

BI, B2, B3 =Constants of the ovality reduction function TABLE 1

COLLAPSE MODEL VERIFICATION DATA
CI, C2 =Constants of the eccentricity reduction function TUB£ OVAL AXIAL TEST MODB MODEU
SPEC GRADE 00 WALL OIt S LOAD P P TEST
IN! !PSI !Psn
D =tubular nominal outside diameter 110-02
'lD~
110
110
"N!
4025
'025
0""
0225
17'
.7.
0001
0001
"BSl
0
'25Ollll
9900
"'0
'''0
8Bl0
0 ..
0"

e =eccentricity ,105
1f4.Q2 110 0385 DODOS 0 =00 098
114-00 110
'025
4025 0385 O,OOos 20‫סס‬oo
"""
2<HlOO 20000 100
114-08 110 4025 0385 10 Oooos ~OOOOO 251S<l _0 0 ..

E =Young's modulus l1G~ 140 403 0304 133 DODOS 0 ,.... "0000 '06

121-02 '40 401G 0378 10' DODOS 0 23550 2.,.0 103

P = collapse pressure 126-03 '10 '02 022. 179 0001 0 11500 11200 0'7

152-02 110 4013 0178 225 DODOS 0 6500 ...0 '02

Po = collapse pressure for near perfect tubular 15NI4 110 4, 13 0219 183 DODOS 0 '2000 11500 0 ..

1 .... 125 40' om '06 DODOS 0 28500 8300 0 ..

158.{14 125 401 om 106 00005 00000
2"'" 30500 104

PE =elastic collapse pressure rl-1 130 289" 02444 118 00012 ·:'19000 21DtlO 2200 '06

PEP =elastic-plastic collapse pressure

11
 0\

OVALITY ECCENTRICITY

~
~
• r = R(I +Scos(28» e=~t
~
.... ~
~
FIgure 1. Finite Dement Model for CoDapse Simulation FIgure 2. Measures 01 ImperfectlODS
I\)

~
110
TRUE STRESS VS TRUE STRAIN

-L----I-----=r
X_. + -I
*
80 KSI YlBLD MATBRIAL ~
g

T--
100 I I -e Z
_
~
I ,

:zr~~--L ~~+ -
90

~
80

----t-----~

-i=I:~~-~~F_--- -:f==l ~
':;;'~ 70
-'II
~~ 60
II _
•
II ~
III 0
lUI
f<f<
50
jJ n
~
II'" 40

30 +-1I ' - - - - - --.

-l--. - ------; :--------1 _+I
: ----1 i
20

10

0
0
I ---=r==J-----L----
0.02 O.lM
--

STRAIN
tn

FIgure 3. Stre8l-StralD Curve for 10 ksI Yield Materlal FIgure 4. Progressive Stages of Deformation FoHowing Tubular Collapse
t8
-ei
- .J
 PLOT OF LOAD VS. DISPLACEMENT
11,100 PSI EXTERNAL til
~
c~

~e ... ~
· .
.
-~
-..!

PHYSICAL TEST SPECIMEN

~7 - ~ - ~.. ~ .

I
~
~6

~5
..................... _-

••••••••••••••• _ •••••••••••• _

:
·
:
,
.

•••••••• 4.

.
:
_

Z C.4 - - ············r···--··--···1·········· -1" -_ - .

~
~3 ...... .- _ -
· -~ _
,
~ _-_.~
. ~. /':¢;~~lf-l:: ..
·;,:'.,,'ff~'(···· -
~tjjHJ1j.:t\TIDJ1:ti-J'..rtl;r~'.
· . .
..... - ~ -.. _ ~ -~ (;<", .•. ~~
~2

.:
---

.
- -_ ..
FINITE ELEMENT MODEL PREDICTION 00 hlj

··~'i.%'itEmn~±I±ll~li~~~v.'
,
~I .... -...... ._--_ ;_ -': - _-

: : ;
0 i i i
0 0.05 0.1 0.15 0.2 0.25 c.]
OJ SPLACEMENT (I NGHES I :-<
~
...... Figure 5. Pressure·Deftection Response for Tubular CoDapse FIgure 6. Comparison of Model Prediction F;j
c.> til
and Physical Specimen Post-CoDapse Geometry. >
~
~
til

1‫סס‬oo
20000
I
~
I
UPPER BOUND
9000 18000

iw 8000 ! 16000
7000 ~ 14000
a:
j
j
1Il 8000 gj 12000
1Il
W
a: 5000 ~ 10000
lL lL
w 4000 UPPER BOUND W 8000
1Il 1Il
lL

~ j
3000 6000

0 2000 LOWER BOUND

oJ
0 4000 .L -=---- =- •
U u
1000 2000
0 o I I I I I I I I
15 17 19 21 23 25 27 29 11 13 15 17 19 21 23 25
0/1 Olt

FIgure 7. Comparison of Model Predictions Figure 8. Comparison of Model Predictions

with J·55 Plastic CoBapse Data with N-80 Plastic CoDapse Data

-..!
 00

25000
UPPER BOUND

'iii 20000
S ~
35000 T
w
a:
:l s'" 30000 t
en 15000 W I
en
w a:
:l
25000 t,
a: en
Do en
~ 10000
Do
:s
...J
o 5000
w
a:
a..
w
en
a.
:::1
10000
~

~
<l:
U ...J
...J
0 5000
u
o I I I I I I t Ii , • I I I I

11 13 15 17
Dit
19 21 23 -0.01 0.01 0.03 0.05
tiD
0.07 0.09 0.11
oo
m
CI:l
5
......
Figure 9. Comparison of Model Predictions
with P·110 Plastic Collapse Data
Figure 10. Comparison of Generalized Collapse Equation z
~
with Simulation Data
.j:>.
c::::
~
o
z
~
~
iii
S
20000
18000
+
--...0!1=10 s'iii ~~~:IT
16000
?
~
16000 w
w !5 14000 n
a:
:l
14000
:::I
o
en
en 12000 I •
• .D!I=13 l1.i
12000
b
w
a:
a.
w
en
a.
<l:
10000
8000
6000
"~'- --- ------ ------------ ------
iii _,
0/\;15.5

F"O.. .t=21
a: 10000

~~ 8000
6000
t ~tTl
...J j 4000
...J
0
U
4000
2000 ••• J'iI Il Ii .O/t:;:.30
o
U 2000 + Ii ;I
.!II 11 Ii
• ~
0/t-30
~

~
a i I I I I I I I I j

-0.0005 0.0005 0.0015 0.0025 0.0035 0.0045 0.0055 o 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5

OVALITY RATIO (S) ECCENTRICTY (e)

CI:l

Figure 11. Reduction of the Collapse Pressure Due to Ovality Figure 12. Reduction of the Collapse Pressure Due to Eccentricity ~
N

-
0'\
'..»
- .J