SPE 46007

Catastrophic Buckling of Coiled Tubing in the Injector

Ken Newman, SPE, CTES, L.C. and Jan Aasen, SPE, OSCA Inc.

Copyright 1998, Society of Petroleum Engineers, Inc.

Buckling Theory
This paper was prepared for presentation at the 1998 SPE/ICoTA Coiled Tubing Roundtable Column buckling depends on the slenderness ratio of the
held in Houston, Texas, 15–16 April 1998.
column. To define slenderness ratio, first the radius of
This paper was selected for presentation by an SPE Program Committee following review of gyration must be defined:
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 rg = 1
2 ro2 + ri2 Eq. 1
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper Now the slenderness ratio ξ is defined as:
for commercial purposes without the written consent of the 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 abstract must contain conspicuous
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O.
ξ = Le rg Eq. 2
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.
Le depends on the end conditions of the column as follows:
Abstract
A series of tests were performed to determine the compressive Le End Conditions
load at which CT will catastrophically buckle between the L both ends free
bottom of the chains and the top of the stripper in the injector. 2L one end free, one end fixed
This paper documents these tests and discusses analytical L/2 both ends fixed
calculations of the buckling load. An equation for calculating 0.7L one end fixed, one end guided
the buckling load is given.
Classical elastic buckling theory based on the Euler
Introduction equation does not apply to columns with small slenderness
Coiled Tubing (CT) is often used in live well applications in ratios (less than 80). The CT problem we are addressing
which the CT must be snubbed into the well against the typically has small slenderness ratios. The authors chose an
wellhead pressure. If the wellhead pressure is high, the empirical formula instead to study this problem.
compressive load in the CT above the stripper can become The Gordon-Rankine empirical formula presented by
quite high. There is an unsupported length of CT between the Timoshenko1 can be written as:
bottom of the chains and the top of the stripper. If the
Pb 1
compressive load in the CT becomes too high the CT may = Eq. 3
catastrophically buckle in this unsupported interval. Py 1 + β 2 ξ 2
A series of tests were performed to measure the buckling
load of an unsupported length of CT. Three types of tests Where Py is the full body yield load given by:
were performed: Py = σ y A Eq. 4
1. New pieces of CT which had never been bent were
tested in a lab using a hydraulic press to provide the If the effect of internal pressure is being included, σ’y
compressive load. should be used in the above equation in place of σy. The value
2. Used pieces of CT with residual curvature were tested of β is determined experimentally. For this paper the
in the lab using the hydraulic press. multiplier to convert from L to Le is considered to be part of β,
causing Le to be equal to L.
3. Used pieces of CT were tested in an injector with a
load cell mounted below the stripper
The theory for the effects of the residual bending moment Yield Stress
and the internal pressure were reviewed. An empirical Used CT samples of 80,000 psi grade CT were tested by the
solution to this buckling problem was developed based on the manufacturer to determine the uniaxial yield stress. A portion
test data. Recommendations for using this empirical solution of the stress-strain curve from one of these tests is shown in
to determine the buckling load were made. Figure 1. This curve continued to a strain of 30% and a stress
2 KEN NEWMAN AND JAN AASEN SPE 46007

of 85,000 psi. In this buckling type problem the CT is seeing

large strains, up to 2% before buckling occurs.
σ yc = 1
2 bσ + σ g −
r h b
σ 2y − 43 σ r − σ h g 2
Eq. 10

Figure 1 shows that the yield stress could be between

73,000 psi and 78,000 psi depending on the strain. This
When both values are needed at the same time in an
difference in yield stress causes a relatively small change in
equation an approximation of the yield stress adjusted for
the buckling load compared to the other variations in
internal pressure is:
chain/stripper alignment, unsupported length, etc. For
simplicity it was decided to use the nominal yield stress for
calculation purposes. b
σ ′y = σ 2y − 43 σ r − σ h g 2
Eq. 11

Effect of Internal Pressure

Some of the tests were performed with internal pressure. These adjusted values of the axial yield stress are used to
Analytically the yield stress was adjusted to compensate for account for the internal pressure. In typical CT situations
internal pressure. The Von Mises yield criteria for a tri-axial these adjustments are relatively small. For example, 1.25” OD
stress state involving radial, axial and hoop stresses is: CT with a .109” wall, 80,000 psi yield and 5,000 psi internal
pressure, the adjusted axial yield stress to account for the

σy = 1
2 bσ −σ g +bσ −σ g +bσ −σ g
a h
2
a r
2
r h
2

Eq. 5
internal pressure is 75,308 psi. This difference in yield stress
due to internal pressure causes a relatively small change in the
buckling load compared to the other variations in
chain/stripper alignment, unsupported length, etc.

Effect of Residual Bending Moment

σa = 1
2 bσ + σ g ±
r h σ − 2
y
3
4 bσ − σ g
r h
2
Once CT has been bent around a reel or over a guide arch and
straightened again, it has large residual stresses left in the
Solving this equation for the axial stress yields: material. Figure 2, taken from Reference 2, shows the stress
across the CT cross-section when the CT is bent, and when it
is straightened. This residual stress causes a residual bending
Eq. 6 moment in the CT which must be resisted by the chains and
Assuming an average radial stress through the wall stripper to hold the CT column in a straight position.
Note from Figure 2 that when the CT is bent nearly the
entire cross section is at the yield stress either in tension or in
σ r = − 21 Pi compression. Assuming the entire cross section is at the yield
thickness: stress, the bending moment caused by the residual stress can
Eq. 7 be calculated as:

Pi ro
4
c
M b′ = σ ′y ro3 − ri 3
3
h Eq. 12

σh =
b ro − ri g
Using the thin wall assumption:

Eq. 8 Yang3 showed that this approximation of Mb’ is accurate

when compared to an exact solution with less than 0.3% error.
Once the CT is straightened there is a small area about the
neutral axis in which the stress returns to zero, as can be seen
Using these values of the hoop and radial stress, two in Figure 2. Thus the residual bending moment in the CT
values of the axial stress can be obtained, one for yielding in when it is straight is somewhat less than when it is bent. An
compression and the other for yielding in tension. These approximation for the residual bending moment when the CT
adjusted axial yield values are written as: is straight developed by the authors is:

σ yt = 1
2 bσ + σ g +
r h b
σ 2y − 43 σ r − σ h g 2
4
M b = σ ′y 35
3
b g c
. ro − ri ry2 − ro3 − ri 3 h
Eq. 9 Eq. 13

Where ry is the distance from the neutral axis at which the

pipe yields plastically when bent to the radius Rb which is the
SPE 46007 CATASTROPHIC BUCKLING OF COILED TUBING IN THE INJECTOR 3

smallest radius of bending the CT has been bent to, usually at Lab Testing
the reel: New straight CT samples and curved used CT samples, of
various lengths, wall thickness and yield stresses, were
Note that the sign of Mb is opposite to the sign of Mb’ due buckled in a hydraulic press. The ends of the CT were rigidly
σ ′y Rb supported in the test fixture. The applied force was measured
ry = Eq. 14 both from the hydraulic pressure applied to the press and by an
E electronic load cell.
to bending in the opposite direction. There were some problems with the end supports in the
As a compressive axial load is applied to the column, the test fixture causing the ends to become misaligned, especially
portion of the CT already at the yield stress in compression with the curved CT samples. The resulting data was probably
yields, while the rest of the cross section absorbs the more scattered due to these problems.
compressive load. This yielding and shifting of the stress Figure 4 shows the data from these tests. The Gordon-
curve reduces the bending moment. Rankine equation was used to model the buckling load. For
A numerical plasticity program called Plastic described in the straight CT samples, a beta value of 0.015 appeared to give
Reference 2 was used to simulate this situation. A sample of a fair fit to the data. For the curved used CT samples, a beta
CT which had been bent and straightened was simulated. value of 0.03 was used.
Axial loads, Fa, were then applied. The sample was then bent, As expected, the curved used CT samples had a lower
in the plane of its original curvature, to various radii of buckling load than the straight new CT samples.
curvature, with Fa still applied, and the bending moment due
to the stress profile in the CT was recorded. The results of Injector Testing
these simulations are given in Figure 3. In one case in Figure Three HR 440 injectors were used to buckle CT between the
2 the sample was bent 90 degrees out of plane with Fa=0. chains and the top of the stripper. The unsupported length in
Only one side of this curve is shown since the curve is each injector was different from the others. Used pipe (20%
symmetric about the vertical axis, which means the maximum of service life) was taken from 1.0” and 1.25” CT working
moment is the same for bending either direction from the reels, and cut in 20’ lengths. These samples were fed into the
initial plane of curvature. injector with the curvature in the CT pointing towards the reel.
Assuming the unsupported column is initially straight (0 A load cell was placed underneath the injector, and the
curvature in Figure 3) and an axial compressive load is compressive load recorded using a computer. A grid was
applied, the residual bending moment in the sample would be placed behind the unsupported CT column and a video was
as is shown along the vertical axis in Figure 3. There are then used to record the deflection of the column as it buckled.
three possible bending scenarios which have been considered: Photo 1 shows a sample of CT after it was buckled.
1. Bending in plane in the positive curvature direction – Table 1 and Figures 5 and 6 show the results of this
when this bending occurs the bending moment testing. Figure 5 shows the CT diameter and two lengths for
decreases quickly. Thus it would appear that this the unsupported column, “long” and “short”. Figure 6 shows
would be the preferred direction of bending. the same data designated by the internal pressure that was in
2. Bending in plane in the negative curvature direction – the CT during the test.
when this bending occurs the internal bending It was impossible to determine the length of the
moment basically remains constant, increasing only unsupported column. The lower support at the stripper was
slightly. rigid and easy to measure, but the upper support at the chains
3. Bending 90 degrees out of plane either way – when was not well defined. The chains had 3.75” long segments
this bending occurs the moment decreases slightly. which gripped the CT sample. The chain passes around a
With more severe bending the moment will increase sprocket at the bottom of its travel. The bars that force the
again. segments against the CT sample (known as skates) are above
In none of these 3 cases was the moment required to increase the sprockets. The sprockets are a fixed distance apart. When
significantly for bending to occur. Thus when eccentric the segment comes off of the skate above the sprocket, it no
loading is applied, the CT can bend in any of these directions longer has any lateral force pushing it against the CT, but it
depending on the direction of the eccentric load. Later when may still be in contact with the CT. As it travels around the
the injector buckling testing is discussed, it will be noted that sprocket the segment rotates away from the CT. Thus, as the
bending in all of these directions did indeed occur. chain moves the point at which the column is supported
From this discussion of the residual bending moment the moves. By visual inspection it was impossible to tell where
following conclusions were reached: the chain actually supported the column. Thus the column
1. The residual bending moment decreases as length, one of the most important parameters in the buckling
compressive axial load is applied. calculation, was basically unknown.
2. The direction of the eccentricity of axial loading, not In several cases the column would bend (buckle) slightly,
the residual moment, determines which direction the bringing it into contact with a lower segment of the chain.
column will buckle. Once in contact with the lower segment, the support point
would be lower, making the column shorter so it could bear
4 KEN NEWMAN AND JAN AASEN SPE 46007

more load. This would result in another higher buckling load L= unsupported length of CT
after the column had already been buckled. Le= equivalent unsupported length of CT
In cases where the column bent slightly and the Mb= residual bending moment in CT when straight
compressive force was released, the column remained bent, Mb’= residual bending moment in CT when bent
proving that the bending was indeed plastic. Pb= catastrophic buckling load of CT
As was stated earlier, one would expect the sample to Pi= internal pressure
buckle in the plane of curvature of the CT sample, with the Py= full body yield load of CT
preferred direction being away from the reel. Most of the rg= radius of gyration of the CT
buckling did indeed occur in the plane of curvature, though ri= inside radius of the CT
not all. Some samples buckled perpendicular to the plane of ro= outside radius of the CT
curvature in both directions. Those that did buckle in the ry= distance from neutral axis to point CT yields
plane of curvature buckled in both directions, toward and Rb= minimum radius of bending CT has been bent to
away from the reel. β= calibration factor
σy= uni-axial yield stress
Recommended Calculation σ’y= axial yield stress adjusted for internal pressure
The buckling load for the unsupported length of CT, Pb, can be σyc= compressive axial yield stress adjusted for
calculated using the Gordon-Rankine Equation 3 with β equal internal pressure
to 0.03. The length used to calculate the slenderness ratio, ξ, σyt= tensile axial yield stress adjusted for internal
should be the vertical length from the top of the stripper to the pressure
centerline of the lower chain sprockets. ξ= slenderness ratio
Once the buckling load is obtained, a safety factor must be
applied. Because of variations in chain/stripper alignment, Acknowledgements
chain wear, etc. the authors suggest a safety factor of 50% The authors appreciate the support of OSCA, Inc. (a Great
should be used. Lakes Chemical Corporation company) and CTES, L.C. (A
Table 2 gives calculated values of Pb/σy for common CT Tuboscope affiliated company) in this work.
sizes. Multiply the value from Table 2 by the yield stress of
the CT to obtain an estimate of the safe compressive load References
limit. This table includes a 50% safety factor.
1. Timoshenko, S.: Strength of Materials – Part I, Third Edition,
Conclusions Krieger Publishing Company, 1958
The following conclusions were drawn from this work:
2. K. Newman, U.B. Sathuvali, S. Wolhart; “Elongation of Coiled
• The buckling load for the unsupported section of CT Tubing During its Life,” paper SPE 38408, presented at the 2nd
between the chains and the stripper is difficult to North American ICoTA/SPE Roundtable, April 1997
calculate accurately because of uncertainty of the
3. Y. Yang; “Collapse and Burst Pressure of Coiled Tubing Under
location of the upper support, variations in
Axial Load, and Bending Torque and Strain Energy in Spooling
chain/stripper alignment, chain wear, etc. CT,” SPE 36338, presented at the 1st North American
• Intuitively, one would expect the CT to buckle in the ICoTA/SPE Roundtable, February 1996
direction of its residual curvature. In fact, the
direction the CT will buckle is random.
• There is a large variation in the stress while yielding,
making it difficult to choose a specific “yield stress.”
The nominal uniaxial yield stress of the material was
choosen for calculation purposes.
• The variation in the buckling load due to internal
pressure in the CT is small compared to the other
variations.
• An empirical calculation was developed to use in
estimating the buckling load.
• Due to the variations mentioned above, a large safety
factor should be applied to the empirical calculation.
• A table of calculated buckling loads for typical CT
sizes was presented.

Nomenclature Photo 1 – Buckled CT

A= cross sectional area of CT = π(ro2-ri2)
E= modulus of elasticity for CT material
 SPE 46007 CATASTROPHIC BUCKLING OF COILED TUBING IN THE INJECTOR 5
1.2
85,000
83,000 1
81,000
0.8
Stress (psi)

Load Ratio
79,000
77,000 0.6
75,000
Gordon-Rankine (beta = 0.015)
73,000 0.4
Straight pipe data
71,000
0.2 Gordon-Rankine (beta = 0.03)
69,000
67,000 Curved pipe data
0
65,000 0 10 20 30 40
Slenderness Ratio
0% 1% 2% 3%
Strain

Figure 1 – Used CT Uniaxial Stress Strain Curve Figure 4 – CT Buckling Using Hydraulic Press

1 1.2

0.75 1
Bent
Straightened
(Distance from neutral axis /

0.5
Load Ratio 0.8

0.25
outer radius)

0.6
0
-70 -50 -30 -10 10 30 50 70 0.4
- 0 .25 Gordon-Rankine (beta = 0 .03)

0.2 1.25-inch long

- 0 .5 1-inch long
1.25-inch short
0
- 0 .75
0 10 20 30 40
S l e n d e r n e s s R a tio
Stress,
-1 ksi

Figure 2 - Stress Profiles for Bent and Straight CT Figure 5 – Injector Buckling Tests

1500 1.2

1000 1
Moment (ft lb)

0.8
Load Ratio

1.25" OD CT 500
0.109" wall, 80Ksi
0 0.6

-20 -10 0 10 20 0.4

-500 Fa = 0
Fa = -5000 0.2 Gordon-Rankine (beta = 0.03)
-1000 Fa = -15000 Zero pump pressure
Fa = -20000 0 5,000 psi pump pressure
-1500 Fa = 0@90deg 0 10 20 30 40
Curvature = 1/R (1/in)*1000 Slenderness Ratio

Figure 3 – Residual Moment Versus Curvature Figure 6 – Injector Buckling Tests

6 KEN NEWMAN AND JAN AASEN SPE 46007

Open Buckling Outside CT Wall Yield Internal

Length Load Diameter Stress Pressure
(in) (lbs) (in) (in) (psi) (psi)
12 18,000 1.241 0.111 80,000 -
12 17,400 1.241 0.111 80,000 -
12 18,200 1.241 0.111 80,000 -
12 19,200 1.241 0.111 80,000 -
12 15,400 1.252 0.100 80,000 -
12 16,400 1.252 0.100 80,000 -
12 14,700 1.252 0.100 80,000 -
12 16,600 1.252 0.100 80,000 -
12 17,500 1.241 0.111 75,300 5,100
12 17,500 1.241 0.111 75,500 5,000
12 15,100 1.241 0.111 75,300 5,100
12 16,000 1.241 0.111 75,000 5,300
8.5 12,800 1.0 0.095 80,000 -
11 10,000 1.0 0.095 80,000 -
10.5 9,100 1.0 0.095 80,000 -
10 10,400 1.0 0.095 80,000 -
11 11,500 1.0 0.095 80,000 -
11 12,600 1.0 0.095 80,000 -
10 13,600 1.0 0.095 80,000 -
10 13,200 1.0 0.095 80,000 -
10 12,700 1.0 0.095 76,600 4,600
10 14,100 1.0 0.095 75,700 5,200
11 10,300 1.0 0.095 75,000 5,600
11 11,800 1.0 0.095 75,700 5,200
8 24,000 1.25 0.109 80,000 -
8 20,100 1.25 0.095 80,000 -
8 18,800 1.25 0.095 80,000 -
8 23,800 1.25 0.109 80,000 -
8 18,800 1.25 0.095 79,800 950
8 17,900 1.25 0.095 73,900 5,000
8 18,300 1.25 0.095 73,900 5,000

Table 1 – Data From Injector Head Buckling Tests

SPE 46007 CATASTROPHIC BUCKLING OF COILED TUBING IN THE INJECTOR 7

OD (in) Wall (in) Unsupported Length (in)

2 4 6 8 10 12 14 16
1.00 0.075 0.1054 0.0961 0.0838 0.0710 0.0594 0.0494 0.0413 0.0347
1.00 0.080 0.1118 0.1019 0.0887 0.0751 0.0627 0.0522 0.0435 0.0366
1.00 0.087 0.1206 0.1097 0.0954 0.0806 0.0672 0.0559 0.0466 0.0391
1.00 0.095 0.1305 0.1186 0.1029 0.0868 0.0722 0.0600 0.0499 0.0419
1.00 0.102 0.1390 0.1261 0.1092 0.0920 0.0765 0.0634 0.0527 0.0442
1.250 0.087 0.1556 0.1465 0.1335 0.1187 0.1039 0.0902 0.0780 0.0675
1.250 0.095 0.1687 0.1587 0.1445 0.1283 0.1122 0.0973 0.0840 0.0727
1.250 0.102 0.1800 0.1693 0.1539 0.1366 0.1193 0.1033 0.0892 0.0770
1.250 0.109 0.1912 0.1796 0.1632 0.1446 0.1262 0.1092 0.0942 0.0813
1.250 0.125 0.2160 0.2027 0.1837 0.1625 0.1414 0.1221 0.1051 0.0906
1.250 0.134 0.2297 0.2153 0.1949 0.1721 0.1496 0.1290 0.1110 0.0955
1.500 0.095 0.2067 0.1982 0.1854 0.1701 0.1538 0.1377 0.1225 0.1087
1.500 0.102 0.2208 0.2116 0.1979 0.1814 0.1639 0.1466 0.1304 0.1156
1.500 0.109 0.2347 0.2249 0.2102 0.1926 0.1739 0.1554 0.1381 0.1223
1.500 0.125 0.2660 0.2546 0.2377 0.2174 0.1960 0.1749 0.1551 0.1373
1.500 0.134 0.2832 0.2710 0.2527 0.2310 0.2080 0.1855 0.1644 0.1453
1.500 0.156 0.3242 0.3098 0.2885 0.2631 0.2364 0.2103 0.1860 0.1641
1.750 0.109 0.2780 0.2695 0.2564 0.2401 0.2219 0.2031 0.1846 0.1671
1.750 0.125 0.3156 0.3058 0.2907 0.2719 0.2510 0.2295 0.2084 0.1884
1.750 0.134 0.3365 0.3259 0.3096 0.2894 0.2670 0.2440 0.2213 0.2000
1.750 0.156 0.3863 0.3738 0.3548 0.3311 0.3050 0.2782 0.2520 0.2273
1.750 0.175 0.4280 0.4140 0.3924 0.3658 0.3365 0.3064 0.2772 0.2497
1.750 0.188 0.4560 0.4408 0.4175 0.3889 0.3573 0.3251 0.2938 0.2644
2.000 0.109 0.3212 0.3137 0.3020 0.2869 0.2697 0.2512 0.2324 0.2139
2.000 0.125 0.3652 0.3565 0.3430 0.3257 0.3058 0.2846 0.2630 0.2419
2.000 0.134 0.3896 0.3803 0.3657 0.3471 0.3258 0.3030 0.2799 0.2573
2.000 0.156 0.4481 0.4372 0.4201 0.3983 0.3734 0.3469 0.3200 0.2938
2.000 0.175 0.4974 0.4850 0.4658 0.4412 0.4132 0.3834 0.3533 0.3240
2.000 0.188 0.5305 0.5171 0.4963 0.4699 0.4397 0.4077 0.3754 0.3440
2.375 0.109 0.3858 0.3795 0.3694 0.3561 0.3404 0.3229 0.3045 0.2857
2.375 0.125 0.4393 0.4320 0.4203 0.4050 0.3869 0.3669 0.3457 0.3241
2.375 0.134 0.4690 0.4612 0.4486 0.4322 0.4127 0.3912 0.3685 0.3454
2.375 0.156 0.5406 0.5314 0.5167 0.4974 0.4747 0.4496 0.4231 0.3962
2.375 0.175 0.6012 0.5908 0.5742 0.5525 0.5269 0.4986 0.4689 0.4387
2.375 0.188 0.6420 0.6308 0.6129 0.5895 0.5619 0.5315 0.4995 0.4671
2.875 0.125 0.5379 0.5319 0.5221 0.5090 0.4931 0.4750 0.4552 0.4343
2.875 0.134 0.5747 0.5683 0.5577 0.5437 0.5266 0.5071 0.4859 0.4635
2.875 0.156 0.6637 0.6561 0.6438 0.6273 0.6073 0.5846 0.5598 0.5337
2.875 0.175 0.7393 0.7307 0.7168 0.6982 0.6757 0.6501 0.6222 0.5929
2.875 0.188 0.7904 0.7811 0.7661 0.7461 0.7219 0.6943 0.6643 0.6327
2.875 0.203 0.8486 0.8386 0.8223 0.8006 0.7744 0.7445 0.7121 0.6780

Table 2 – Buckling Load Divided by Yield Stress for Common CT Sizes

(includes 50% safety factor)