Buckling Analysis in Deviated Wells:

A Practical Method
R.F. Mitchell, SPE, Landmark Drilling and Well Services

Summary and dogleg angle are developed. An application problem was

Current helical buckling models are valid for vertical wells, but solved and the effects of well deviation on stability, length
provide only approximate solutions for horizontal wells. Solutions change, and maximum bending stress were examined. Well devia-
of the nonlinear buckling equations for arbitrary well deviation tion is shown to have significant impact on buckling results and
have been developed, but are too complex for practical use. This tubing stress analysis.
paper presents a set of correlations that match the exact solutions At the end of this paper is a complete nomenclature and refer-
extremely well, but are simple to use. These correlations show the ence list.
effects of well deviation on buckling shape, tubing length change,
contact force and bending stress. Buckling Models for Deviated Wells
Analysis of Helical Buckling. The theoretical basis for the analy-
sis of buckling in deviated wellbores is described in this section.
Introduction The first generally useful buckling solution was published by
The most generally accepted method for the analysis of buckling, Lubinski.1 In Lubinski’s analysis, the wellbore is assumed to be
tubing movement, and packer selection is the method developed vertical and the tubing buckled shape was modeled as a helix with
by Lubinski et al. in Ref. 1. Analyses following Lubinski’s basic variable pitch. Mitchell4 showed that Lubinski’s solution was an
approach have been developed for more complicated tubing con- approximate solution to the beam-column equations6 with dis-
figurations, e.g., tapered strings.2,3 Woods, in the Appendix to placements constrained to a cylinder. In this formulation of the
Lubinski et al.,1 developed a mechanical model of well buckling beam-column equations, the lateral displacements, shown in Fig.
behavior that predicted the buckled configuration as a function of 1, are given by
well loads. This model featured: u 1 5r cos u , ~1!
1. slender beam theory is used to relate bending moment to
curvature, u 2 5r sin u , ~2!
2. the tubing is assumed to buckle into a helical shape,
where u is the helix angle, and r is the tubing/casing radial clear-
3. the wellbore is assumed to be straight and vertical,
ance. The differential equation for the helix angle u is given by
4. the pitch of the helix is related to the buckling load through
the principle of virtual work, @ 2EI u - 12EI ~ u 8 ! 3 2F u 8 # 8 1 ~ w l /r ! sin u 50, ~3!
5. friction between the tubing and casing is neglected.
where w l is the lateral tubing weight per unit length, EI is the
Mitchell developed a more general approach that replaced the bending stiffness, F is the axial buckling force, and 8 denotes
virtual work relations with the full set of beam-column equations d/dz. The axial buckling force F and the lateral tubing weight are
constrained to be in contact with the casing.4 Helical buckling in a
deviated well, in this formulation, is described by a fourth order
nonlinear differential equation. For a vertical well, the solution to
this equation can be accurately approximated by the simple alge-
braic equation discovered by Lubinski and Woods. This solution
is not valid for deviated or horizontal wells because of the lateral
gravity forces. The full deviated well equation was solved by
Mitchell using numerical methods.5 The purpose of this paper is
to put these results in a more usable form.
Accurate solution of the buckling equations is important for
several reasons. Bending stresses due to tubing buckling will be
overestimated for deviated wells using Lubinski’s formula. How-
ever, Lubinski’s solution applied to deviated wells will also over-
predict tubing movement. For a fixed packer, this solution will
overestimate tubing compliance, which may greatly underestimate
the axial loads, resulting in a nonconservative design. For a free
packer or PBR, exaggerated tubing motion will require excessive
seal length. Further, because tubing incremental motion will con-
trol the friction load direction, errors in overall tubing displace-
ment will generate further errors in friction loads.
This paper presents correlations to the numerical solution of the
buckling differential equation. Calculation of results, including
buckling length change, tubing contact forces, bending stresses,

Copyright © 1999 Society of Petroleum Engineers

This paper (SPE 55039) was revised for publication from paper SPE 36761, first presented
at the 1996 SPE Annual Technical Conference and Exhibition held in Denver, Colorado,
6–9 October. Original manuscript received for review 24 October 1996. Revised manu-
script received 28 October 1998. Paper peer approved 4 November 1998. Fig. 1–Coordinate system for buckling analysis.

SPE Drill. & Completion 14 ~1!, March 1999 1064-6671/99/14~1!/11/10/$3.5010.15 11

 Fig. 2–Buckling in deviated wells.

both strongly influenced by fluid pressures and must be formu- Correlations for Maximum Buckling Dogleg. The correlation
lated accordingly.7 Note also that F is the value of the axial force developed in Ref. 5 for the maximum value of u 8 for lateral
in the buckled state. Lubinski’s helical pitch solution: buckling can be expressed

u 8 56 A F
2EI
~4! 8 5
u max
1.1227
A2EI
F 0.04~ F2F p ! 0.46

is an approximate solution to Eq. ~3! for w l equal 0, but not the for
only possible solution. An approximate solution to Eq. ~3! for
inclined wells, with F constant, is 2.8F p .F.F p . ~7!

u 5 a sin~ b z ! , u a u !1, ~5! The corresponding helical buckling correlation is

which requires
u 8 56 A F
2EI
for F.2.8F p . ~8!
F> A4w l EI/r5F p ~6!
The region 2.8F p .F.1.4F p may be either helical or lateral,
for b to be real. This is the Paslay–Dawson equation,8 with F p however, 2.8F p is believed to be the lateral buckling limit on
defined as the Paslay critical buckling load. loading, while 1.4F p is believed to be the helical buckling limit on
General solutions to Eq. ~3! are not easy to obtain. The solution unloading from a helical buckled state.5,9 The lower limit, 1.4F p ,
technique developed in Ref. ~5! used the Galerkin method with is the minimum pipe strain energy needed to develop a helix, but
cubic interpolation functions to calculate u as a function of z. This this minimum energy is insufficient to cause the spontaneous de-
method produced a system of nonlinear equations that required velopment of a helix from a laterally buckled pipe. The value of
numerical solution techniques. This method is too complex and the upper limit is still under study, but in the particular numerical
too slow for use in casing and tubing design, and is obviously method used to develop these results, 2.8F p was the maximum
unsuited for hand calculations. Examination of the results of the limit under which a stable solution of lateral buckling could be
numerical analysis, however, suggested that simple formulas developed. In practice, pipe will transition from lateral to helical
might match the numerical results fairly well. In fact, these simple buckling at some intermediate value between 1.4F p and 2.8F p
formulas matched the numerical results perfectly. There is prob- because of irregularities in the actual geometry, and will fall out
ably a theoretical reason for this good match, but to date, no one of the helix for values of F less than 1.4F p . Experimental studies
has discovered it. However, we can benefit from this fortuitous have verified these results, at least qualitatively, and further ex-
match by developing simple, useful correlations for buckling perimental studies are continuing to investigate this behavior.10
analysis. For a vertical well, F p is zero, so only Eq. ~8! applies, which is

12 R.F. Mitchell: Buckling Analysis in Deviated Wells SPE Drill. & Completion, Vol. 14, No. 1, March 1999
 Fig. 3–Bending moment correlations.

the buckling solution determined by Lubinski and Woods.1 An The equation for dogleg curvature for a helix is
important distinction between Eqs. ~7! and ~8! is that Eq. ~7! is the
k 5r ~ u 8 ! 2 . ~9!
maximum value of u 8 while Eq. ~8! is the actual value of u 8 . As
8 over the lateral
can be seen in Fig. 2, u 8 varies between 6 u max The dogleg units for Eq. ~9! is radians per inch. To convert to the
buckling interval. This distinction will be used when developing conventional unit of degrees per 100 feet, multiply the result by
buckling length change results. 68,755.

Fig. 4–Bending stress correlations.

R.F. Mitchell: Buckling Analysis in Deviated Wells SPE Drill. & Completion, Vol. 14, No. 1, March 1999 13
 Fig. 5–Typical lateral buckling
cycle.

Correlations for Bending Moment and Bending Stress. Given M d o Ed o r ~ u 8 ! 2

the tubing curvature, the bending moment is determined: s b5 5 , ~11!
2I 2
M 5EI k 5EIr ~ u 8 ! 2 . ~10!
where d o is the outside diameter of the pipe. The following cor-
The corresponding maximum bending stress is relations can be derived using Eqs. ~7! and ~8!:

Fig. 6–Buckling strain comparison.

14 R.F. Mitchell: Buckling Analysis in Deviated Wells SPE Drill. & Completion, Vol. 14, No. 1, March 1999
 Fig. 7–Buckling length change comparison.

M 50 for F,F p , s b 50 for F,F p ,

M 50.6302rF .08~ F2F p ! 0.92 for 2.8F p .F.F p , d o r 0.08

s b 50.3151 F ~ F2F p ! 0.92 for 2.8F p .F.F p ,
I
M 50.5000rF for F.2.8F p , ~12!
d or
and s b 50.2500 F for F.2.8F p . ~13!
I

Fig. 8–Contact force correlation.

R.F. Mitchell: Buckling Analysis in Deviated Wells SPE Drill. & Completion, Vol. 14, No. 1, March 1999 15
 TABLE 1– SQUEEZE CEMENTING OPERATION

Tubing-Casing Properties

Type o.d. (in.) i.d. (in.) Weight (ppf)

Tubing 2–7/8 2.441 6.5

Casing 7 6.094 32
Fluid Properties

Fluid Location Fluid Type Density (psi/in)

Tubing 15 ppg cement 0.0649

Annulus 30 degree API crude 0.0317

Miscellaneous Properties:
w 50.6396 cos f lbf/in.
w l 50.6396 sin f lbf/in.
E 5303106 psi
I 51.611 in.4
r 51.610 in.
Packer Bore area58.296 in.2

The dependence of bending moment and bending stress on buck- Fig. 9–Example problem wellbore geometry.
ling force is shown in Fig. 3 and 4. Bending moment and bending
stress are expressed in dimensionless form on these two figures,
and since bending moment and stress are proportional, the figures
have the same form. Conventional analysis shows the dimension- e b 52 21 ~ r u 8 ! 2 . ~14!
less moment to be constant ~this means the bending moment is
proportional to the buckling force!, while the correlation predicts For the case of lateral buckling, we have only the maximum value
no moment below the Paslay buckling force, and lower bending of u 8 , so we must determine the average strain in terms of the
moment below the helical limit. Similar observations apply to the maximum strain. The actual shape of the u 8 curve ~Fig. 5! was
bending stress. integrated numerically to determine the following relationship:

Correlations for Buckling Strain and Length Change. The r 2 0.08

buckling ‘‘strain,’’ in the sense of Lubinski,1 is the buckling e b 520.7285 F ~ F2F p ! 92 2.8F p .F.F p , ~15!
4EI
length change per unit length. The buckling strain is given by the
following relationship: which compares to the helical buckling strain:

Fig. 10–Buckling forces.

16 R.F. Mitchell: Buckling Analysis in Deviated Wells SPE Drill. & Completion, Vol. 14, No. 1, March 1999
 Fig. 11–Tubing buckled length.

r2 To determine the buckling length change DL b , we need to

e b 52 F, F.2.8F p . ~16! integrate Eqs. ~15! and ~16! over the appropriate length interval:
4El
The buckling strain is illustrated in Fig. 6. No buckling strain is
shown for buckling force less than the Paslay Buckling force F p . DL b 5 E
z1
z2
e b dz ~17!
The buckling strain is roughly half the conventional helical buck-
ling strain up to the lateral buckling limit, where the tubing goes where z1 and z2 are defined by the distribution of the buckling
into helical buckling. force F. For the general case of arbitrary variation of F over the

Fig. 12–Buckling length change.

R.F. Mitchell: Buckling Analysis in Deviated Wells SPE Drill. & Completion, Vol. 14, No. 1, March 1999 17
 Fig. 13–Bending stresses.

interval DL5z22z1, Eq. ~17! must be numerically integrated. W n 5rF 2 /4EI1w l . ~24!
However, there are two special cases that are commonly used. For
that case of constant force F, such as in a horizontal well, Eq. ~17! The contact force dependence on the buckling force is shown in
is easily integrated: Fig. 8. When the buckling mode changes from lateral to helical,
the contact force increases substantially.

Ez1
z2
e b dz5 e b DL, ~18!
An Application Problem
where e b is defined by either Eqs. ~15! or ~16!. The second special The sample application presented in this section is based on the
case is for a linear variation of F over the interval: example used by Lubinski.1 The basic parameters for this problem
are summarized in Table I. The principle difference in the present
F ~ z ! 5wz1c. ~19! calculation from Lubinski is the addition of a kick-off point to
We can now evaluate Eq. ~18! by change of variables: make this a deviated well. The deviation angle f is used as a

E E
variable to show the relative importance of well deviation on the
z2 F2
e b dz5w 21 e b dF. ~20! buckling results. The well deviation geometry is shown in Fig. 9.
z1 F1 The buckling force at bottom hole is shown in Fig. 10. Because
Eq. ~15! cannot be integrated in closed form, but a good approxi- the force is generated by the hydrostatic pressure, the force de-
mation discussed in the Appendix can be integrated to give the creases as the deviation angle increases. The critical buckling
following result: force ~Paslay force! increases with the square root of the sine of
the deviation angle, and the critical helical buckling force is 2.8
2r 2 times the Paslay force. Fig. 10 shows that helical buckling is
DL b 5 ~ F 2F p !@ 0.3771F 2 20.3668F p #
4EIw 2 surpressed for angles above about 54 degrees, but that lateral
buckling occurs throughout the range shown.
for 2.8F p .F 2 .F p . ~21!
Fig. 11 shows the total buckled length of the tubing. The Lu-
The integration of Eq. ~16! gives the familiar Lubinski result: binski theory indicates that buckling length increases as the devia-
tion angle increases. Since hydrostatic pressures control buckling
r2
DL b 52 @ F 2 2F 21 # for F.2.8F p . ~22! in this application, the neutral point of the tubing corresponds to a
8EIw 2 fixed true vertical depth. The measured depth corresponding to a
Buckling length change is illustrated in Fig. 7. The buckling fixed true vertical depth increases as the deviation angle increases,
length change predicted is much less than the conventional model resulting in an increasing buckled length. The buckled length for
when buckling forces are less than about four times the Paslay the deviated well correlation shows reduced buckled length as
buckling force. deviation increases, which is qualitatively more reasonable than
the Lubinski prediction. The helical buckled length decreases
Correlations for Contact Force. From equilibrium consider- even more rapidly, since lateral loads tend to prevent helical buck-
ations only, the average contact force for lateral buckling is ling.
The buckling length change is shown in Fig. 12. The difference
W n 5w l . ~23!
in buckled length change between the correlation and the Lubinski
The average contact force for the helically buckled section is model, for small deviation angles, is surprisingly large, with about

18 R.F. Mitchell: Buckling Analysis in Deviated Wells SPE Drill. & Completion, Vol. 14, No. 1, March 1999
 Fig. A-1–Correlation of lateral buckling strain.

a 20% reduction in length change for a deviation angle of 10 will be overestimated for deviated wells using Lubinski’s formula.
degrees. The difference in buckled length for large angles is sub- However, Lubinski’s solution applied to deviated wells will also
stantial. For angles in excess of 60 degrees, the correlation buck- overpredict tubing movement. For a fixed packer, this solution
led length change is negligible, while the Lubinski theory predicts will overestimate tubing compliance, which may greatly underes-
80% of the vertical well length change. timate the axial loads, resulting in a nonconservative design. For a
The bending stress induced by buckling is shown in Fig. 13. As free packer or PBR, exaggerated tubing motion will require ex-
long as helical buckling is predicted, the correlation and the Lu- cessive seal length.
binski model coincide. At about 50 degrees, helical buckling is The analysis in this paper indicates the following
surpressed, and the bending stress is reduced by more than 20%.
1. For a vertical well, helical buckling can be accurately ap-
For inclination above 80 degrees, bending stress predicted by the
proximated by a simple algebraic equation discovered by Lubinski
correlation is near zero, while the Lubinski model predicts over
and Woods. This solution is not valid for deviated or horizontal
5,000 psi.
wells because of the lateral gravity forces.
Results, Observations, and Conclusions 2. Buckling in deviated wells is of two forms, a lateral ‘‘snake-
Accurate analysis of buckling is important for several reasons. like’’ buckling and helical buckling. The buckling force must ex-
First, buckling generates bending stresses not present in the origi- ceed the Paslay force to initiate lateral buckling, and must exceed
nal configuration. If the stresses in the original configuration were 1.4 times the Paslay force for helical buckling. Only helical buck-
near yield, this additional stress could produce failure, including ling occurs for buckling forces exceeding 2.8 times the Paslay
permanent plastic deformation called ‘‘corkscrewing.’’ Second, force.
buckling causes tubing movement. It can be seen that a coiled 3. Simple correlations for the analysis of buckling in deviated
tubing is shorter than straight tubing, and this is an important wells have been developed from numerical solutions of the buck-
consideration if the tubing is not fixed in the packer. Free packers ling equation. Contact forces, buckling length changes, tubing
and PBRs have seals that are designed to accommodate this tubing bending stress and maximum dogleg angles are determined.
movement, so an important design consideration is the length of 4. The Lubinski buckling model predicts an increase in the
these seals. Third, tubing buckling causes the relief of compres- length of buckled tubing for an increase in deviation angle, which
sive axial loads when the packer is fixed. This effect is not as is not considered realistic. The correlation presented in this paper
recognized as the first two buckling effects, but is equally impor- predicts a reduction in buckled tubing length, and a substantial
tant. The axial compliance of buckled tubing may be much less reduction in helically buckled length.
than the compliance of straight tubing. Tubing movement due to 5. Buckling length change predicted by the correlation shows
thermal expansion or ballooning can be accommodated with a
considerable reduction compared to the Lubinski model. This re-
lower increase in axial load for a buckled tubing.
duction is primarily due to the reduction of helical buckling by
The accuracy and comprehensiveness of the buckling model is
lateral forces in the deviated well.
important for designing tubing. For example, the most commonly
used buckling model was developed by Lubinski and Woods in Clearly, a comprehensive model for buckling is desirable. The
the 1950’s. This model is accurate for vertical wells but needs correlations presented in this paper make the comprehensive
modification for deviated. Tubing bending stress due to buckling analysis of buckling in deviated wells practical.

R.F. Mitchell: Buckling Analysis in Deviated Wells SPE Drill. & Completion, Vol. 14, No. 1, March 1999 19
Nomenclature 7. Mitchell, R.F.: ‘‘Forces on Curved Tubulars Due to Fluid Flow,’’
SPE Prod. Facil. 30 ~February 1996!.
do 5 tubing outside diameter, L, in. 8. Dawson, R. and Paslay, P.R.: ‘‘Drillpipe Buckling in Inclined
E 5 Young’s modulus, m/Lt 2 , psi Holes,’’ J. Pet. Technol., 1734 ~October 1984!.
I 5 moment of inertia of tubing, L 4 , in4 9. Chen, Y.-C., Lin, Y.-H., and Cheatham, J.B.: ‘‘Tubing and Casing
F 5 the buckling force, mL/t 2 , lbf Buckling in Horizontal Wells,’’ J. Pet. Technol. 140 ~February 1990!.
Fo 5 the packer to tubing force, mL/t 2 , lbf 10. Suryanarayana, P.V.R. and McCann, R.C.: ‘‘An Experimental Study
Fl 5 the Paslay buckling force, mL/t 2 , lbf of Buckling and Post-Buckling of Laterally Constrained Rods,’’ J.
L 5 the tubular string length, L, ft. Energy Resources Technology, Trans. ASME 117 ~June 1995!.
M 5 bending moment, mL 2 /t 2 , ft-lbf
r 5 the tubing-casing radial clearance, L, in. Appendix—Approximations to the Lateral Buckling
w 5 axial distributed load in the tubing, m/t 2 , lbf/ft. Strain
wl 5 lateral distributed load in the tubing, m/t 2 , lbf/ft. The average lateral buckling strain is given by Eq. ~15!:
wn 5 the contact force between the tubing and casing, m/t 2 ,
r 2 0.08
avg520.7285 2.8F p .F.F p .
lbf/ft. eb F ~ F2F p ! 0.92
z 5 measured depth, L, ft. 4EI
eb 5 buckling strain ~A-1!
sb 5 bending stress, m/Lt 2 , psi The quantity F 0.08(F2F p ) 0.92 cannot be integrated in closed form.
u 5 angle between the pipe center location and the x coor- We define the dimensionless function E as a function of F/F p :
dinate axis, radians
E ~ j ! 5 j 0.08~ j 21 ! 0.92, ~A-2!
u8 5 derivative of u with respect to z, L 21 , radians/ft.
8
u MAX 5 lateral buckling amplitude correlation, L 21 , radians/ft. where j 5F/F p . Function E is plotted in Fig. A-1. Function E
f 5 deviation angle measured from the vertical, radians appears to be very nearly linear. This function can be approxi-
j 5 F/F p mated over the range j P @ 1,2.8 # by the following least-squares fit:
DL b 5 length change due to buckling, L, ft.
E ~ j ! 521.021211.0352j . ~A-3!
References The closeness of fit to Eq. ~A-2! is demonstrated in Fig. A-1. The
1. Lubinski, A., Althouse, W.S., and Logan, J.L.: ‘‘Helical Buckling of use of Eq. ~A-3! in Eq. ~A-1! gives
Tubing Sealed in Packers,’’ J. Pet. Technol. ~June 1962!.
2. Hammerlindl, D.J.: ‘‘Movement, Forces, and Stresses Associated r2
With Combination Tubing Strings Sealed in Packers,’’ J. Pet. Tech- eb avg520.7285 ~ 1.0352F21.0212F p ! 2.8F p
4EI
nol. ~February 1977!.
3. Hammerlindl, D.J.: ‘‘Packer-to-Tubing Forces for Intermediate Pack- .F.F p . ~A-4!
ers,’’ J. Pet. Technol. 195–208, 515–527 ~March 1980!.
4. Mitchell, R.F.: ‘‘New Concepts for Helical Buckling,’’ SPEDE, 303– Robert F. Mitchell is a senior technical advisor with the Drill-
310 ~September 1988!. ing and Well Services division of Landmark Graphics. e-mail:
5. Mitchell, R.F.: ‘‘Effects of Well Deviation on Helical Buckling,’’ [email protected]. He was Vice President of Enertech Engi-
SPE Drilling Completion ~March 1997!. neering and Research Co. from 1980 to 1996. He holds BA,
6. An Introduction to the Mechanics of Solids, Crandall, S.H. and Dahl, MS, and PhD degrees from Rice U. Mitchell, a member of
N.C. ~eds.! McGraw–Hill Book Inc., New York City ~1959! 360– the Twenty-Five Year Club, was a 1994–97 member of the
384. Editorial Review Committee.

20 R.F. Mitchell: Buckling Analysis in Deviated Wells SPE Drill. & Completion, Vol. 14, No. 1, March 1999