IADC/SPE 99074

A Re-examination of Drillpipe/Slip Mechanics

P. Paslay, Manatee; P.D. Pattillo, SPE, and P.D. Pattillo II, SPE, BP America; U.B. Sathuvalli, SPE, Blade Energy
Partners; and M.L. Payne, SPE, BP America

Copyright 2006, IADC/SPE Drilling Conference

tangential stress at the drillpipe inner diameter. Knowing the
This paper was prepared for presentation at the IADC/SPE Drilling Conference held in Miami, tangential and axial stresses on the inner diameter of the
Florida, U.S.A., 21–23 February 2006.
drillpipe (where the radial stress vanishes), the von Mises
This paper was selected for presentation by an IADC/SPE Program Committee following
review of information contained in a proposal submitted by the author(s). Contents of the
equivalent (VME) stress criterion is used to estimate the axial
paper, as presented, have not been reviewed by the International Association of Drilling load at which the drillpipe begins to yield. The Reinhold-Spiri
Contractors or 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 IADC, SPE, their formula for the slip crushing load is,
officers, or members. Electronic reproduction, distribution, or storage of any part of this paper
for commercial purposes without the written consent of the International Association of Drilling
2
Contractors and Society of Petroleum Engineers is prohibited. Permission to reproduce in print
is restricted to an abstract of not more than 300 words; illustrations may not be copied. The Fz , slip crush = Fzy 2 2
................(2)
abstract must contain conspicuous acknowledgment of where and by whom the paper was ⎛ K do ⎞ ⎛ K do ⎞
1 + ⎜1 + +
2 L ⎟⎠ ⎜⎝ 2 L ⎟⎠
presented. Write Librarian, SPE, P.O. Box 833836, Richardson, TX 75083-3836, U.S.A.,
fax 1.972.952.9435. ⎝

Abstract The predictions of Eq. (2) were compared with a limited series
Current models of slip crushing treat the rotary slip as an of meticulous tests on 5 in., 19.5 ppf, Grade E drill pipe
axisymmetric wedge that generates an axisymmetric radial loaded in standard and extended length manual slips2.
load on the drillpipe lateral surface. However, recent tests on Until recently, Eq. (2) has been used to calculate slip
strain gauged drillpipe specimens suggest that this model does crushing loads for drillpipe and casing hung in rotary slips,
not adequately capture the mechanical response of the and sometimes to calculate the loads exerted by packers on
drillpipe. The tests show that drillpipe response is a complex tubing. Eq. (2) is based on a statically determinate analysis of
and non-axisymmetric function of slip geometry, friction a conical wedge, a consequence of which is that the peak
between the slip and the bowl in the rotary table, and the stress is always on the drillpipe inner diameter in the vicinity
mechanics of load transfer between the slip and drillpipe. of the slip toe. An unstated consequence of the assumptions
This paper presents a new model of slip behavior that leading to Eq. (2) is that the drillpipe stress distribution is
represents the slip system as a series of line loads. Analysis of axisymmetric. This theory thus implies that the peak stress
the suspended tubular under these forces and axial tension region is a circle on the drillpipe inner diameter in the plane of
leads to a limit load that characterizes slip crushing. A the slip toe.
corollary of this analysis is the minimum slip length required Following the work by Reinhold and Spiri1 and Vreeland2.
to support a given axial tension. in the late 1950s, slip crushing analysis received little attention
until 1985. A 1985 paper by Hayatdavoudi discusses the
Background principles of slip insert design to reduce probability of
The failure of drillpipe in the region of contact between the yielding in the pipe3. The work includes representation of
drillpipe and slips was first addressed by Reinhold and Spiri in strain gage data from slip crush tests on 9-5/8 in., 53.5 ppf
19591. This paper recognized that drillpipe is subjected to bi- casing. Though the paper does not contain the details of
axial loading in the slip contact area. By treating the slip as an testing, the data presented in this paper indicates that the
immovable wedge between a rigid bowl and the hanging deformation of the casing inner diameter above the toe of the
drillpipe, a relation between the axial force on the drillpipe slip is larger than the deformation near the top of the slip.
and the transverse force exerted on it by the slips was derived. Focus on drillpipe slip crushing was renewed in the late
The ratio of the transverse force to the axial force known as 1990s with the rapid development of deepwater fields in the
the “K-factor” is given by Gulf of Mexico. Anecdotal evidence from several deepwater
operators and results of tests by equipment manufacturers
W 1 − μ BS tan α seemed to indicate that the Reinhold-Spiri formula may not be
K= = . .............................................. (1)
FzA μ BS + tan α a conservative estimate of drillpipe slip crushing capacity if
initial yielding is applied as the limit condition. In 2002
The angle of taper has been standardized by the API to 9o, 27’, Sathuvalli et al. reviewed all available test data and the current
45’’. state of the art4. Based on examination of best available data
The average radial pressure on the drill pipe outer diameter and limited heuristic modeling of slip-drillpipe interaction, the
and the axial stress in the pipe beneath the slip toe are paper concluded that yielding on the drillpipe inner diameter
estimated. This “radial pressure” is used to estimate the
2 IADC/SPE 99074

initiated at values around 20% less than that predicted by Eq. condition of the mating surfaces at the slip-bowl interface, and
(2). point to the following general observations:
The “non-conservative” aspect of Eq. (2) posed a serious 1. Hoop stress in drillpipe is not axisymmetric.
constraint to the design of deepwater drillstrings, where hook 2. The axial and hoop stress do not always vary
loads in the range of 750 klbf to 1,000 klbf are not uncommon. monotonically across the length of the drillpipe in the
This prompted a closer examination of the theoretical basis of slip-contact region.
the Reinhold-Spiri equation4 and a program of full scale tests 3. The peak stresses in the drillpipe typically occur
to examine the performance of drillpipe and landing string along an axial plane, i.e. along a line on the drillpipe
joints (typically 5-7/8 in. and 6-5/8 in. outside diameter) inner diameter parallel to the drillpipe axis. This
loaded in different types of slip systems (both manual and plane usually coincides with a gap between slip insert
power slips). In February 2004, this led to full scale testing in carriers. The peak stress decays with circumferential
which the drillpipe was strain gaged extensively on the ID (see distance on either side of this line of peak stress.
Figs. 1 and 2)5. This effort was followed by a proprietary joint These peak stresses do not necessarily occur at the
industry project in which slip systems from various toe of the slip as predicted by the Reinhold-Spiri
manufacturers were tested. Those results are not being formula.
disclosed in this paper and the authors hope they can be the This series of tests has demonstrated discrepancies
subject of a future publication by the JIP participants. between current practice/theoretical understanding and the
industry’s predictive capability with respect to slip crushing
mechanics. The analysis of strain gage data reveals that the
peak stresses may not be confined to a single cross sectional
plane on the drillpipe inner diameter, as predicted by
conventional theory. Rather, yielding begins on the inner
diameter along a line parallel to the drillpipe axis. This line of
yielding corresponds to a gap between the slip insert carriers.
Though initial yield is indicated at loads below that predicted
by Eq. (2), the mechanics of slip crushing are such that the
yield failure mode is somewhat gradual and the system retains
reasonable stiffness and stability well after initial yield. The
post-yield behavior of the drillpipe under slip crushing has
provided a safety margin for the industry and prevented more
serious problems in borderline operations.

Slip System Model

The slip system considered here contains n (= 2, 3, 4, 5 or 6)
identical slips situated symmetrically in a bowl. An axial
downward load is applied to the tubular in the slips and
through the action of friction the slips slide downward and
inward in the bowl supporting the slips. Each slip is assumed
Fig. 1 – Typical Strain Gage Layout and Axial Load Transfer in a to apply the same radial load on the tubular. When the radial
Slip System loads are sufficiently large to support the axial load through
friction, the sliding stops and the slip system supports the axial
load.

14000

12000

10000 pL

8000
pL, lb/in

6000
TWO DIAMETRICALLY OPPOSED LOADS
NO AXIAL LOAD
4000 6
YOUNG'S MODULUS = 30x10 psi
POISSON'S RATIO = 0.3
YIELD POINT = 100,000 psi
2000 WALL THICKNESS = 0.5 in
MEAN RADIUS = 3.0 in
Fig.2 – Strain Gage Profiles in 6-5/8 in., 30.3 ppf, V-150 Drillpipe pL
Axially Loaded to 1,137 klbf 0
0 0.025 0.05 0.075 0.1 0.125 0.15 0.175 0.2 0.225 0.25

Preliminary analyses of results of tests described in w0, in.

Reference 5 indicates that the mechanical behavior of the Fig. 3 – Line Load vs. Radial Displacement, Elastic, Perfectly
drillpipe is a complex function of slip geometry and the Plastic Material
IADC/SPE 99074 3

The loads applied to the tubular by the slips are modeled as corresponding inward radial displacement with increasing
n line loads on the surface of the tubular parallel to the axial load.
tubular axis. The load per unit length on each slip is The influence of tubular material work hardening is not
designated by pL . In practical applications the radial loads expected to be large for the ductile materials commonly used.
can cause yielding of the tubular material. The inset in Fig. 3 Figure 5 shows results for a bilinear work hardening material
shows a case for n = 2. The curve in this figure has been for the same tubular dimensions as Fig. 4 and n = 6. This
figure confirms that for the usual tubular materials the
determined for pL being constant along the axial direction
influence of work hardening is small.
and the axial extent of the loading is large compared to the Clearly, the load per unit length, pL , varies along the
tubular diameter. The curve was found using the theory
described in the appendix and shows an important property of radially loaded length of the slip. In the analysis presented
the solution for an elastic, perfectly plastic material. As the here the tubular radial displacement at a point along the loaded
inward radial displacement increases, the value of pL reaches length is determined from the value of pL at that point. The
a maximum value and then decreases. Under load controlled plausibility of this procedure is justified by the well-known
conditions the maximum point on the curve is an unstable shell theory solution shown in Fig. 6 where a uniform external
point and the displacement will continue to grow beyond this pressure has been applied over a finite length. The radial
point. displacement is seen to decay rapidly with distance away from
the loaded region.
24000
AXIAL TENSILE LOAD = 300,000 lb
AXIAL TENSILE LOAD = 100,000 lb 0.0007
22000
20000 AXIAL TENSILE LOAD = 500,000 lb 0.0006
18000
AXIAL TENSILE LOAD = 700,000 lb 0.0005
16000
14000 0.0004
pL, lb/in

6
YOUNG'S MODULUS =30x10 psi
AXIAL TENSILE LOAD = 900,000 lb
12000 POISSON'S RATIO = 0.30
w, in.

MEAN RADIUS = 3.00 in

0.0003 WALL THICKNESS = 0.5 in
10000 NUMBER OF LINE LOADS = 3
6 LOADED LENGTH = 15 in
YOUNG'S MODULUS = 30x10 psi
8000 POISSON'S RATIO = 0.3 0.0002
YIELD POINT = 100,000 psi
6000 WALL THICKNESS = 0.5 in
MEAN RADIUS = 3.0 in 0.0001
4000
2000 AXIAL TENSILE LOAD FOR PIPE YIELD (WITH pL = 0) = 942,500 lb 0.0000
0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 -0.0001
w0, in. -5 -2.5 0 2.5 5 7.5 10 12.5 15 17.5 20
z, in.
Fig. 4 - pL vs. w0 for Selected Values of Axial Load
Fig. 6 – Inward Radial Displacement vs. Axial Position along
Tubular, Tubular Loaded with 1000 psi External Pressure from z =
0 to z = 15 in.
55000
160,000
50000
140,000 Application – Minimum Theoretical Slip Length
45000 120,000
Under the above assumptions it is possible to establish a
40000 UTS = 100,000 psi
limiting condition called the minimum theoretical slip length,
35000 NUMBER OF SLIPS = 6
TOTAL AXIAL FORCE ON PIPE = 600,000 lb Lmin . In order to derive the minimum slip length condition an
pL, lb/in

6
30000 YOUNG'S MODULUS = 30x10 psi
POISSON'S RATIO = 0.3
YIELD POINT = 100,000 psi
axial force equilibrium equation is introduced for incipient slip
25000
ULTIMATE TENSILE STRENGTH = variable
PLASTIC STRAIN AT UTS = 0.094
in the form,
20000 WALL THICKNESS = 0.50 in
MEAN RADIUS OF PIPE = 3.0 in
dFz
= − μTS npL . ............................................................(3)
15000 ΔwO = 5x10 in
-5

10000 dz
5000 AXIAL TENSILE LOAD FOR PIPE YIELD (WITH pL = 0) = 942,500 lb
The maximum inward radial loading on the tubular, pLMax ,
0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 is a function of Fz that is determined numerically. Fig. 7
w0, in.
shows, for the slip system and tubular defined in the figure,
Fig. 5 - pL vs. w0 for Selected Values of Ultimate Tensile the relation of pLMax to Fz . The maximum value of Fz at z
Strength = 0 is its pipe body yield strength (tubular cross-sectional area
multiplied by the yield stress). When this value is used as an
The loaded tubular also has an axial load that lessens its initial condition at z = 0 and the above differential equation is
ability to withstand inward radial loading. The derivation in integrated numerically to the point where Fz = 0, the dashed
the appendix includes the equations necessary to determine
curve in Figure 8 is obtained. The curve shows that Fz
solutions including axial loads. Fig. 4 shows results for the
same tubular with n = 3 and several different axial loads. vanishes when z = 10.23 in. This value of z is the shortest
Note the reduction of both the maximum load and the length of slip that can apply an axial load sufficient to yield
the entire cross section of the tubular in the plane z = 0.
4 IADC/SPE 99074

100,000 For this case Eq. (2) becomes,

90,000
Fz 0 1 + μ BS cot α
= = μTS ...........................................(6)
80,000 P cot α − μ BS
70,000

60,000
so that,
pLMax, lb/in

μTS cot α − 1
NUMBER OF SLIPS = 3
6
50,000
μ BS =
YOUNG'S MODULUS = 30x10 psi
POISSON'S RATIO = 0.30 . ......................................................(7)
40,000 YIELD POINT = 135,000 psi
MEAN RADIUS = 2.125 in
cot α + μTS
WALL THICKNESS = 0.75 in
30,000
The relation between μTS and μ BS is a result of regarding
20,000

10,000 AXIAL TENSILE LOAD FOR PIPE YIELD (WITH pL = 0) = 1,352,000 lb

the slip as an axisymmetric wedge. For an ideal wedge, Eqs.
(6) and (7) imply that μ BS , the coefficient of friction between
0
0 200000 400000 600000 800000 1000000 1200000 1400000 the slip and the bowl, should be equal to or less than the value
FzA, lb given by Eq. (7).
Fig. 7 - pLMax vs. FzA , Elastic, Perfectly Plastic Material If μ BS is greater than the value given above, the slip
cannot grip the pipe, since the limiting condition for the
minimum slip length assumes incipient slip at the drillpipe-
slip interface. On the other hand, if μ BS is less than the value
1.60E+06

NUMBER OF SLIPS = 3
1.40E+06 6
YOUNG'S MODULUS = 30x10 psi given above, the radial force on the tubular might crush the
POISSON'S RATIO = 0.30

1.20E+06
YIELD POINT = 135,000 psi
MEAN RADIUS = 2.125 in
pipe. The design of the slip system is based on ensuring
WALL THICKNESS = 0.75 in
FRICTION COEFFICIENT = 0.5
“better than incipient slip” at the drillpipe-slip interface and
1.00E+06
simultaneously preventing radial crushing of the pipe. Since
the value of μTS to prevent incipient slip is governed by the
Fz , lb

10.23 - 6.19 = 4.04 in.

8.00E+05

relative magnitudes of Fz 0 and P , and since the slips cannot

FzA = 900,000 lb
6.00E+05 Fz0 = YIELD LIMIT

be treated as axisymmetric wedges, it is, very unlikely that Eq.

4.00E+05
Lmin = 6.19 in.
z = 10.23 in. (7) will be satisfied in a real design.
2.00E+05 The design goal should be to have the value of P as small
0.00E+00
as practically possible. The average line load, pLAve , is,
0 2 4 6 8 10 12
z, in. pLine = P ..................................................................(8)
nL
Fig. 8 - Fz Distribution along Slip for FzA = 900,000 lb, Elastic,
where L is the length of each slip. The value of pLAve should
Perfectly Plastic Material
be compared to curves corresponding to Fig. 4 to ensure the
Normally, a design value for the slip system capacity load, tubular will not be too damaged by the slips when the design
Fz 0 , is specified. In this case, the Fz versus z relation is capacity of the slip system is reached.
obtained by shifting the dashed curve in Fig. 8 parallel to the
Application – Slip/Tubular Axial Load Distribution
z -axis until the ordinate of the shifted curve at z = 0 equals
In addition to the determination of the minimum theoretical
Fz 0 . The case for Fz 0 = 900,000 lbf is shown as the solid slip length, it is possible to estimate the axial load distribution
curve in Fig. 8. The dashed curve had to be shifted along the between the slips and the tubular. That is, still considering the
z -axis by 4.04 inches so that Fz ( 0 ) = 900,000 lbf. The solid loads to be line loads, the intensity variation with distance
curve meets the z -axis is at z = 6.19 in. and this value is from the bottom of the slip is estimated. This analysis
defined as the minimum theoretical slip length when Fz 0 = accounts for the possible misalignment of the slips with
respect to the tubular. An angular misalignment, ξ , is
900,000 lbf.
The limiting condition assumed for the calculation of the introduced. When ξ > 0, the bottom of the slip contacts the
minimum theoretical slip length assumes that at every point of tubular before the top of the slip. Results obtained by varying
contact there is incipient slip between the tubular and the slips. this angle should help determine desirable limits on tolerance
Let the total radial force on the tubular from the slip be, P , in in the design of the slip system. The axial load distribution is
this limiting condition, so that found for cases where ξ is held constant while the slip inward
Lmin radial displacement is increased in steps. For each step the
P=n ∫
0
pL dz .............................................................. (4) maximum axial load is found as μTS n ∫ pL dz along the contact
length. The value for pL is found from curves similar to Fig.
and 4.
Fig. 9 shows the result of calculations based on the above
μTS = Fz 0 P .................................................................. (5) paragraph. In this figure the maximum axial load at the
bottom of the slip is plotted against the inward radial
IADC/SPE 99074 5

displacement. The different curves correspond to different 9 101,000

lengths of slips. The ordinates of the curves are assumed to 10 30,000
become constant once the maximum value is reached.
Obviously, the shorter slips can carry less maximum load than Curves like those in Fig. 10 should be useful for setting
the longer slips. some of the manufacturing tolerances in the design of the slip
system.
800000
Nomenclature
MAXIMUM AXIAL LOAD AT SLIP BOTTOM, lb

700000
25"
do outside diameter of tubular, [L], in.
Young's modulus, [ML-1T-2], lb/in2
NUMBERS ON CURVES GIVE THE LENGTHS OF THE SLIPS 20
600000 E
ANG = 0.500 deg
NUMBER OF LINE LOADS = 3 ez axial strain, [L/L], in./in.
500000 YOUNG'S MODULUS = 30E6 psi 15"
POISSON'S RATIO = 0.3
YIELD POINT = 100,000 psi
eθ circumferential strain, [L/L], in./in.
400000
tensile load on tubular at position z , [MLT-2], lb
WALL THICKNESS = 0.500 in
MEAN RADIUS = 3.000 in Fz
10"
300000
FzA applied tensile load on tubular, [MLT-2], lb
200000
5"
Fz , slip crush Reinhold-Spiri slip crushing load, [MLT-2],
100000 lb
0
Fzy pipe body yield strength, [MLT-2], lb
0.00 0.05 0.10 0.15 0.20 0.25
Fz 0 design value of axial tensile load, [MLT-2], lb
w 0, in

Fig. 9 – Maximum Load on Tubular vs. Slip Inward Radial

h thickness of shell, [L], in.
Displacement for Different Slip Lengths K ratio of W to FzA
Kθ circumferential curvature, [L-1], 1/in.
In Fig. 10 a fixed length of slip is considered and the load
distribution along the slip is shown. Each curve is for a Lmin minimum theoretical slip length, [L], in.
different inward radial displacement. The numbers on the L slip length, [L], in.
curves may be used to determine the maximum axial load N displacement amplitude, [L], in.
from Table 1 given below. Note that the abscissa of this curve n number of line loads
is distance measured from the top if the slip. Since the value P total radial force on tubular from slip, [MLT-2], lb
of ξ > 0 for this case, the initial contact between the slip and Pe perimeter, [L], in.
the tubular is at the bottom of the slip. pL line load magnitude, [MT-2], lb/in.
pLAve average value of line load magnitude, [MT-2], lb/in.
pLMax maximum value of line load magnitude, [MT-2], lb/in.
25000
EACH CURVE NUMBER CORRESPONDS TO AN AXIAL LOAD GIVEN IN TABLE 1
1

2 R radius of shell midsurface, [L], in.

20000
r radial coordinate, [L], in.
3
v circumferential displacement, [L], in.
15000 W transverse load on tubular from slips, [MLT-2], lb
pL, lb/in

w radial displacement, [L], in.

10000 w0 amplitude term in expression for radial displacement,
[L], in.
5000 z axial position coordinate measured from the bottom
of the slip, [L], in.
4

0
5 6 7 8 9 10
α slip taper angle, [-], rad
0 2 4 6 8 10 12 14 16 18 εr radial midsurface strain, [L/L], in./in.
DISTANCE FROM TOP OF SLIPS, in.
εz axial midsurface strain, [L/L], in./in.
Fig. 10 - pL vs. Distance from the Top of the Slips, 3 Slips, Same εθ circumferential midsurface strain, [L/L], in./in.
Data as Fig. 9, Except Length of Slips is 16.59 in. Δλ plasticity parameter
μ Poisson's ratio
Table 1 – Maximum Axial Load for Curves in Fig. 10
Curve Number Axial Load, lb μ BS coefficient of friction between bowl and slip
1 537,000 μTS coefficient of friction between tubular and slip
2 537,000
3 531,000 θ circumferential coordinate, [-], rad
4 491,000 σr radial stress, [ML-1T-2], lb/in2
5 415,000
6 338,000 σz axial stress, [ML-1T-2], lb/in2
7 259,000 σθ circumferential stress, [ML-1T-2], lb/in2
8 180,000
6 IADC/SPE 99074

ξ misalignment angle, [-], rad 1

2 (
v + v,θθ ) ................................................. (A-3)
2
εθ =
ζ distance from midsurface, [L], in. 2R
Integrating Eq. (A-3) around the circumference
References yields a non-zero strain which must be eliminated
1. Reinhold, W. B.and Spiri, W. H.: “Why does drill pipe fail by introducing a corrective or scaling strain that
in the slip area?,” World Oil (October 1959) 100. renders εθ zero in a global sense.
2. Vreeland, T., Jr.: "Deformation of Drill Pipe Held in
Rotary Slips,” ASME paper No. 61-PET-20, presented at Inextension with an Assumed Displacement. Assume the
Petroleum Mechanical Engineering Conference, Kansas tangential displacement is of the form
City, Sept. 24-27, 1961.
3. Hayatdavoudi, A.: “Elastic Yield of Casing Due to v = N sin nθ ............................................................. (A-4)
Elevator/Spider System,” paper SPE 13449 presented at the
where, as elsewhere, n is the number of line loads and N is a
1985 SPE/IADC Conference, New Orleans, March 6-8.
4. Sathuvalli, U. B., Payne, M. L, Shepard, Suryanarayana, P. constant. Then, from Eq. (A-2)
V., and Shepard, J.: “Advanced Slip Crushing w = −nN cos nθ ....................................................... (A-5)
Considerations for Deepwater Drilling,” SPEDC (2002)
210. and the hoop strain is given from Eq. (A-1) as
5. Payne, M. L, Pattillo, P.D., Driscoll, P. M., and Sathuvalli,
N 2 sin 2 ( nθ )
U. B.: “Experimental Investigation of Drillpipe Loaded in
Slips,” paper WTC 2005-63324, Proceedings of the ASME
εθ =
2R 2 (1 − n ) 2 2
...................................... (A-6)
World Tribology Congress III, Washington, D. C., Sept.
12-16, 2005. Integrating the circumferential strain around the
6. Huang, N. C. and Pattillo, P. D.: "The Effect of circumference, the change in perimeter not accounted for by
Length:Diameter Ratio on Collapse of Casing," Journal of the linear inextension equation is
Pressure Vessel Technology (May 1984) 160. 2π 2π
N 2 sin 2 ( nθ )
7. Sanders, J. L.: "Nonlinear Theories for Thin Shells,"
Quarterly of Applied Mathematics (1963) 21.
ΔPe = ∫ εθ Rdθ = ∫ 2R 2 (1 − n )
2 2
Rdθ
0 0
8. Timoshenko, S. F.: Theory of Elastic Stability, Second ....... (A-7)
π N 2 (1 − n 2 )
2
Edition, McGraw-Hill Book Company, London (1961) 445.
=
Appendix – Governing Equations 2R
That is, under an assumed displacement of the form (A-4),
Infinite Shell with Axial Line Loads. Consider a circular and the local linear inextension constraint of Eq. (A-2), a
cylindrical shell of infinite length subjected to axial line loads circular cross section, as it deforms, will undergo an increase
equally distributed at n locations around its circumference. in perimenter given by Eq. (A-7).
Neglecting rotations about the normal to the midsurface of the This additional extension is removed via the following
shell, the circumferential midsurface strain can be written as6,7 reasoning. Given the initial circular cross section, the change
in radius necessary to effect the change in perimeter given by
v,θ + w 1
+ 2 ( − w,θ + v ) ................................. (A-1)
2
εθ = Eq. (A-7) is
R 2R
N 2 (1 − n 2 )
2

where a comma denotes differentiation with respect to the ΔP

ΔR = − e = − ................................... (A-8)
subscripted quantity, that is, ( ) ,θ = ∂ ( ) ∂θ . The primary 2π 4R
interest here is with bending deformation. We shall therefore Defining the Displacement Field. With the above results,
assume an inextensional shell. Ideally, inextension would be the radial displacement may be written as the sum of two
imposed locally by setting the circuferential strain, εθ , equal components – a component (see Eq. (A-5)) compatible with
to zero at each location on the circumference. Unfortunately, the assumed circumferential displacement, v , and linear
this leads to a nonlinear differential equation. As an inextension, and a component (see Eq. (A-8)) sufficient to
alternative, inextension will be invoked in two steps: render the cross section nonlinearly inextensible, at least in a
• Locally, each position on the circumference is global sense,
required to adhere to the classic linear
N 2 (1 − n 2 )
2
inextensionality constraint8 which requires the w = −nN cos ( nθ ) − ............................. (A-9)
linear terms in Eq. (A-1) to vanish, 4R
v,θ + w = 0 ................................................................ (A-2) Defining w0 = − nN , the outward radial displacement is
• Globally (e.g. for the entire circumference), the given by

w0 2 (1 − n 2 )
change in circumferential midsurface length 2

obtained by integrating Eq. (A-1) with constraint w = w0 cos ( nθ ) − .............................. (A-10)

(A-2) around the circumference is required to 4n 2 R
vanish. Substituting Eq. (A-2) into Eq. (A-1), The circumferential curvature is defined as
IADC/SPE 99074 7

1 1 ( 2σ r − σ θ − σ z ) Δσ r + ( 2σ θ − σ z − σ r ) Δσ θ
2 (
Kθ = − w,θ +v ) ,θ = − 2 ( w,θθ + w ) .................. (A-11) .......... (A-20)
R R + ( 2σ z − σ r − σ θ ) Δσ z = 0
where the last step uses Eq. (A-2). Substituting Eq. (A-10)
where the cylindrical coordinates are assumed to be aligned
into Eq. (A-11),
with the principal directions of the stresses, and Δλ is

Kθ =
(n 2
− 1) w0 ⎡
⎢ θ +
(1 − 1 n2 ) w0 ⎤⎥ ............... (A-12) positive. When the above equations yield a non-positive value
cos n for Δλ the increment has no plastic deformation and then Δλ
R2 ⎢⎣ 4R ⎥⎦ must be set to zero and the last equation deleted.
In the above, Eqs. (A-17) – (A-19) state that the change of
with the last term in brackets being smaller than the first by the plastic strain component is proportional to the current
the order of w0 R . This latter term will be ignored in future component of the corresponding deviatoric stress component.
expressions involving the circumferential curvature, Equation (A-20) constrains (to first order terms in the stress
changes) the von Mises stress to be unchanged during an
Kθ =
(n 2
− 1) w0
cos nθ .......................................... (A-13) incremental change which has non-trivial plastic flow. As
R2 written, Eq. (A-20) specifies perfectly plastic behavior, but a
similar development could be used to analyze a work
With the midsurface strain defined by Eq. (A-3), the
hardening material.
circumferential strain at any position through the wall
In keeping with the conventional treatment of thin shells,
thickness a distance ζ from the midsurface is given by
the radial stress is assumed to vanish everywhere. The
increment of tangential strain for each element follows from
eθ = (1 + ζ R ) (εθ + ζ Kθ ) ≅ (εθ + ζ Kθ ) ............. (A-14)
−1

Eq. (A-16). The algorithm developed here uses a search

or, using Eq. (A-4) and Eq. (A-10) in Eq. (A-1) procedure to find the increment in axial strain. An assumed
value for the increment of the axial strain is chosen and the
⎡ w 2 ⎛ n2 − 1 ⎞2
eθ = ⎢ − 0 2 ⎜ + ζ
( n2 − 1) w0 cos nθ ⎤⎥ ........ (A-15) assumed value is treated as a known quantity along with the
⎟ increment of the tangential strain. The incremental equations
⎢⎣ 4 R ⎝ n ⎠ R2 ⎥⎦ may be written as
Now consider a displacement controlled experiment where ⎡
(σ θ + σ z ) ⎤⎥
1
the radial displacement is incremented by incrementing w0 . ⎢E μ μ
3
⎢ ⎥ ⎡ Δer ⎤
The increment in circumferential strain is from Eq. (A-15) 1
⎢0 −1 μ − ( 2σ θ − σ z ) ⎥ ⎢⎢ Δσ θ ⎥⎥
⎢ 3 ⎥ =
⎡ 1 ⎛ n2 − 1 ⎞2 ⎤ Δw ⎢ ⎥ ⎢ Δσ z ⎥
Δeθ = ⎢ − ⎜ ⎟ w0 + ζ ( n − 1) cos nθ ⎥ 2 ... (A-16)
2 0
1
⎢⎣ 2 ⎝ n ⎠ ⎥⎦ R
⎢0 μ −1 − ( 2σ z − σ θ ) ⎥ ⎢ Δλ ⎥
⎢ 3 ⎥⎣ ⎦
⎢0 ⎥
This expression is important in the incremental solution ⎣ ( 2σ θ − σ z ) ( 2σ z − σ θ ) 0 ⎦
procedure to be described below. ⎡ 0 ⎤
⎢ − E Δe ⎥
Numerical Solution for Uniformly Loaded Element. ⎢ θ ⎥

Consider a small element bounded by r and r + Δr , θ and ⎢ − E Δez ⎥

θ + Δθ , and z and z + Δz , where Δr and Δθ are small ⎢ ⎥
⎣ 0 ⎦
enough so the stress state in the element may be considered to ................................................................. (A-21)
be uniform. Using the von Mises yield function, the
governing stress-strain equations for the element during The solution proceeds as follows:
nontrivial plastic flow are, 1. Starting with the results of the previous step, w0 is
E Δer − Δσ r + μΔσ θ + μΔσ z = known. Assume a value for Δw0 .
1 .............................. (A-17) 2. For a given calculation step with w0 and Δw0 fixed,
( 2σ r − σ θ − σ z ) Δλ assume a trial value for Δε z .
3
3. For each calculation cell (through the wall thickness
E Δeθ − Δσ θ + μΔσ z + μΔσ r = and around the circumference) compute Δeθ with
1 .............................. (A-18) Eq. (A-16).
( 2σ θ − σ z − σ r ) Δλ 4. Using Eq. (A-21) compute Δσ θ , Δσ z and Δλ .
3
Repeat with Δλ = 0 if the first calculation results in a
E Δez − Δσ z + μΔσ r + μΔσ θ = negative value of Δλ .
1 .............................. (A-19) 5. Compute the increments in axial force and bending
( 2σ z − σ r − σ θ ) Δλ moment on the cross section.
3
6. The search routine uses the increment in axial force
as the error and adjusts Δε z (e.g. return to Step 2) in
8 IADC/SPE 99074

order to reduce the absolute value of the error below

a small value. When this has been accomplished, the
work done through second order terms during the
change Δw0 owing to bending is calculated as the
sum for all the sections of the product of the moment
and the change of curvature. The work done on the
pipe by the axial external force and the line loads
must equal the work done by bending.
7. This completes the calculations for one increment in
Δw0 . Following this calculation, the stresses and
moments must be updated by adding the changes to
the values at the beginning of the preceding step.
Then calculation for another step can be commenced.