Sie sind auf Seite 1von 7

SPE/IADC 85302

Fully Three-Dimensional Well Design Improves Margins in Critical Wells

Jan A. Aasen and Bernt S. Aadny, SPE, Stavanger University

Copyright 2003, SPE/IADC

This paper was prepared for presentation at the SPE/IADC Middle East Drilling Technology For burst calculations, the uniaxial Barlow equation1 is still
Conference & Exhibition held in Abu Dhabi, UAE, 20-22 October 2003.
quite popular because of its simplicity. This equation is
This paper was selected for presentation by an SPE/IADC Program Committee following
review of information contained in an abstract submitted by the author(s). Contents of the
derived assuming a thin-walled pipe with zero outside
paper, as presented, have not been reviewed by the Society of Petroleum Engineers or the pressure. The pipe will burst when the hoop stress reaches the
International Association of Drilling Contractors and are subject to correction by the author(s).
The material, as presented, does not necessarily reflect any position of the SPE, IADC, their yield strength. The major shortcoming of the Barlow formula
officers, or members. Electronic reproduction, distribution, or storage of any part of this paper
for commercial purposes without the written consent of the Society of Petroleum Engineers or
is the omission of axial load effects. For a pipe without axial
the International Association of Drilling Contractors is prohibited. Permission to reproduce in loads, the equation is fairly accurate for large diameter to
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 thickness ratios. It works well for a casing string, but is in
presented. Write Librarian, SPE, P.O. Box 833836, Richardson, TX 75083-3836, U.S.A.,
fax 01-972-952-9435.
error for small-diameter pipes.

By neglecting radial pipe stress, a two-dimensional (2D) load

Abstract diagram results with the shape of an ellipse. Studying this,
With the evolution of long, deep and hot wells, the design one observes that axial tension lowers the pipe collapse
margins have become very small. This motivated the resistance and axial compression makes the pipe weak in
presented study and prompted the investigation for a precise burst. Reversing the conditions, the opposite is true. In the
well design model for burst and collapse. Although most well biaxial approach, the axial effects are honored while the less
designs are based on one- and two-dimensional mechanics, important radial stress is neglected. Biaxial stress analysis of
more recent design packages include a three-dimensional well tubulars is an improvement compared to uniaxial theory.
version as well. It will be shown that these models are
accurate only for certain conditions. Triaxial design
Well tubulars may be subjected to a variety of loads during
This paper presents an exact solution to the three-dimensional installation and service. Tensile and compressive axial
well design problem. Field examples demonstrate the stresses are produced by axial loads and bending of the pipe.
accuracy of the new and conventional design models. The Pressure inside and outside of the tubular gives rise to radial
new model calculates design factors that are exact, thereby and hoop stresses. Pipe may also experience shear stresses if
offering a tool for optimized well design. The presented torque is applied. In this paper we consider axial loads and
theory is based on the von Mises yield criterion and the Lam pressure effects only. Tensile stress is positive and a negative
thick-walled solution for the pipe. sign indicates compression throughout the paper.
Introduction The axial stress a, the radial stress r and the hoop stress h
There are a wide variety of models and techniques available to are principal stresses. These stresses may be combined to a
estimate burst and collapse pressures of pipe used for drilling, single equivalent stress (VME) using the von Mises distortion
completion and well intervention. The results obtained using energy theorem2:
the different methods are not always easy to compare since the
underlying theory differs. In this study we present a three-
= ( a r ) + ( a h ) + ( r h )
2 2 2 2
dimensional (3D) model that is exact. 2 VME (1)

The early approach was to simplify the design equations for The yield strength of well tubulars is determined from simple
burst and collapse by neglecting the effects of radial stress and uniaxial tension tests. Two of the three principal stresses are
axial load. This is called uniaxial well design. Radial stress is zero for this case (r and h). The von Mises equivalent
usually small compared to hoop stress in tubulars with high (VME) stress for this situation is given as:
pressure-differentials and it is always compressive.
Neglecting axial load, on the other hand, could introduce large VME = y (2)
errors in the design.

We define the design factor (DF) as the ratio of allowable Insert these equations into Eq. 3:
stress to working stress. The allowable stress is the yield
strength of the pipe and the working stress is the VME stress. 1 y
We need to select DF greater than one in our designs. It DF = = (10)
x xy + y 2 VME

2 y If we assign z = DF, this equation describes a surface that

DF = (3) totally represents the loads caused by axial stress, inside
( a r ) + ( a h ) + ( r h )
2 2 2
pressure and outside pressure seen in relation to the yield limit
of the pipe. Furthermore, Eq. 10 is an exact solution that is
quite simple. Instead of a pressure differential, one uses the
actual outside and inside pressures as defined by Eqs. 8 and 9.
Conventional triaxial model
Current triaxial design is based on methods developed by
Johnson, Jellison and Klementich3 and Kastor4. These This 3D yield surface is shown in Fig. 1. If z is equal to one,
solutions are incorporated in most triaxial design packages this combination of a, r and h gives a loading on the pipe
used in the oil industry. Simplifications were introduced to that is equal to its yield strength. The part of the surface that
obtain a practical design method. For burst calculations, zero lies below z equal to one identifies loading in excess of yield.
outside pressure is assumed. Similarly, collapse calculations If the z-coordinate is greater than one, the calculated VME
are based on zero inside pressure. In applications they apply a stress is lower than the yield strength of the pipe and the
pressure differential to this model. design is in order. We notice that that the pipe is infinitely
strong at x = y = 0 since the denominator in Eq. 10 is zero at
For extremely thin-walled pipes this approach is in order. In this point. This condition implies that a = -pi and pi = po. This
the general case, however, the difference in wall surface area is a hydrostatic stress state.
causes a considerably larger force on the outside pipe wall
than on the inside of the pipe. This leads to an error that is In Fig. 2 we look at a 2D plot obtained by intersecting the 3D
magnified for large outside pressure in case of burst. yield surface by horizontal planes through different values of
z. In this plot the envelopes for burst and collapse become
The new 3D model smaller as the design factor increases. Any load path
Various approaches were attempted to solve this rather describing the service loads on casing or tubing over time
complex nonlinear problem. Using dimensional analysis we needs to remain inside the ellipse. It is also necessary to plot
found a solution that was rather simple. In the following the VME stresses associated with the occurrence of possible
derivation we assume no torque or bending. leak situations and emergency scenarios. The upper half of the
ellipse corresponds to burst, while the bottom half describes
Appendix A presents the Lam solution for radial and hoop collapse pressures. All needed parameters to describe a pipe
stresses in a thick-walled pipe5. Investigation of the equations (pi, po, a, do/t and y) are included in this 2D dimensionless
revealed that burst or collapse failures will initiate on the inner plot. It is not necessary to normalize the equations by
surface of the pipe. Introduce the following geometry factor6: selecting pi = 0 (collapse) or po = 0 (burst), as is the case in
conventional triaxial well design3.
2ro2 (do t )2 It follows from Eq. 10:
= = (4)
ro2 ri2 2(d o t 1)
x 1 3 2
The radial and hoop stress formulations at the inside pipe wall y= x (11)
2 DF 2 4
The plus sign is selected for burst calculations, while the
r = pi (5) negative sign is used for calculating collapse pressure. Define
h = ( pi po ) pi (6) VME = y/DF in the following. Substitute Eqs. 8 and 9 for x
and y in this equation and obtain the collapse pressure:
In absence of bending, the axial stress is calculated as:
pi (2 1) a + 4 VME 3( pi + a ) 2
F po = (12)
a = (7) 2
The burst pressure may be calculated as:
Introduce the following dimensionless variables:
x = ( pi + a ) / y (8) po (2 1) + a ( 2) + 4 VME ( 2 + 1) 3 2 ( po + a ) 2
pi =
2( 2 + 1)
y = ( pi po ) / y (9)

The presented theory for burst and collapse is based on the uniaxial models use burst or collapse pressure to define the
assumption that the pipe fails by yielding at the inner wall. design factor. The results obtained from the four example
There are situations where this may not be true. If the pipe is cases are plotted in Fig. 2.
bent, either by buckling or in curved sections of the well, it is
necessary to add the bending stress to the axial load picture. Case 1: Burst of gas-filled casing
Bending stresses are greatest at the outside wall and the We consider burst of the production casing just below the
resulting VME stress needs to be calculated on both the inside wellhead. This scenario could happen both during drilling and
and outside pipe walls to determine which surface fails first. production. During the drilling phase, the worst case is that
Structural instability is another issue. Collapse experiments on the entire casing fills up with produced natural gas. After the
oil-country tubulars7 show that yield failure, as described in well has been completed with the production tubing, a leak in
this study, is valid for pipe with small and moderate do/t ratios. the tubing below the wellhead could cause the same burst
As a rule of thumb1, the limiting do/t value for yield collapse is pressure. In both cases the greatest differential between inside
14. The presented theory is valid for collapse of drillpipe, and outside pressures are located in the very top of the casing.
snubbing pipe, production tubing and small casing strings. For Inside the casing at this depth the pressure is equal to the
large casing strings other collapse formulas must be employed. reservoir pressure minus the hydrostatic head of the gas
The burst theory, on the other hand, is valid regardless of pipe column. Outside the production casing is the hydrostatic
size. If torque is added to the pipe, as in drilling and some pressure equal to 360 m (1180 ft) of seawater in this subsea
well intervention operations, the shear stress must be included well. We consider in the following a leak during production
in the von Mises formulation. This is not considered in the and derate the yield strength of the C110 casing to 6750 bar
present study. Finally, pipe connectors may be pressure rated (97 900 psi) because of elevated temperature. Thermal
less than the pipe body8 and this is especially true in expansion of the casing changes the axial load in the casing at
compression. In these cases a second performance ellipse for the wellhead from tension to 104 ton (229 000 lbs) of
the connection may be generated on top of the pipe-body compression. The resulting axial stress is -634 bar (-9200 psi)
envelope in Fig. 2. as shown in Table 1.

Comparing models using HP/HT well cases From Eq. 10, the general 3D solution, calculate VME = 71 500
Previous models can be compared with the new exact 3D psi and DF = 1.37. This is deemed an acceptable design. The
model as a reference. In conventional triaxial analysis3 the conventional triaxial procedure gives results in close
design is performed using pressure-differentials. In certain agreement with the general solution. In this case it follows
cases this approach gives exact results, but in other cases an that VME = 71 700 psi and DF = 1.36. The reason for this
error is introduced using the conventional triaxial design. good match is that the outside pressure is relatively low.
Biaxial stress analysis has been used for many years for both
burst calculations4 and collapse calculations9. In the biaxial In the biaxial case, based on the given compressive stress in
method the radial stress is neglected, an assumption that in the casing, we calculate the allowable hoop stress equal to
many cases does not introduce a large error. In uniaxial 6410 bar (93 000 psi). The resulting burst pressure calculated
design each stress component is used separately and compared from the Barlow formula is 985 bar (14 300 psi), which gives
to the yield strength of the pipe material. In uniaxial burst DF = 1.37 based on pressure. In the uniaxial case, the
design, the hoop stress is assessed without considering the allowable hoop stress is the adjusted yield strength equal to
effects of axial load. This uniaxial approach often causes large 6750 bar (97 900 psi). The uniaxial model calculates an
design errors. optimistic DF = 1.44 since the effect of compression is
We consider four design examples from HP/HT wells located
offshore Norway. In the three first cases the concern is the Case 2: Burst of casing caused by leaking tubing
pressure integrity of the 10 x 9 production casing in a In this scenario we look at burst at the top of the production
subsea well. We look at two burst cases and one collapse packer during production. The production tubing leaks just
case9. In the last example, we look at collapse of a snubbing below the wellhead and wellhead pressure enters the
pipe during live well intervention from a fixed platform. Input production casing on top of the packer fluid. This is the same
data for the examples are summarized in Table 1. Reservoir situation described in the previous case, the difference being
temperature is 170 C (338 F) and wellhead temperature at that we now look at burst at the bottom of the well. The
full production is 155 C (311 F). Yield strength values are proposed design calls for a 1.18 SG (9.85 ppg) packer fluid.
derated according to prevailing temperature. The adjusted There is a significant compressive load on the production
yield strength values obtained from pipe manufacturers and casing caused by thermal expansion during production. Once
published material are named y* in the table. In the examples the wellhead pressure enters the tubing/casing annulus, this
we calculate the design factors based on the new exact 3D compression is relived somewhat by wellhead growth. The
solution presented in this study. These results are then resulting axial stress at the bottom of the casing is 2140 bar
compared to the design factors obtained using conventional (31 100 psi) as shown in Table 1. Outside the production
triaxial design, biaxial design and uniaxial design. This casing we have seawater gradient, while inside the casing the
comparison is shown in Table 2. Note that the design factors pressure is equal to wellhead pressure plus the hydrostatic
calculated using the new 3D model and the conventional head of the packer fluid column.
triaxial model are based on VME stress, while the biaxial and

The new 3D model calculates VME = 7100 bar (103 000 psi) compares favorably with wellhead pressure equal to 683 bar
and DF = 1.10. If this is not considered an acceptable design, (9900 psi). We also calculate VME = 3420 bar (49 600 psi)
a solution is to select a lighter packer fluid. The triaxial and DF = 1.69 based on the derated yield strength, which is
procedure gives VME = 7390 bar (107 100 psi) and DF = 1.06. acceptable for this critical well intervention operation. We
In this case the outside pressure is large and the conventional note that the pipe is moderately compressed and therefore very
triaxial theory overestimates the VME stress by 4%. In critical resistant to collapse. The conventional triaxial method gives
well design this error is significant. It is therefore the same results since the outside pressure is zero.
recommended to use the exact 3D theory presented in this
study in well situations were the margins are small. Biaxial The biaxial approach gives a collapse pressure identical to that
burst analysis gives DF = 1.04, while the calculated uniaxial of the 3D model. Since the inside pressure is zero, the radial
burst pressure yields DF = 1.17. stress is also zero. In this case the biaxial model for collapse
is exact. The biaxial design factor based on collapse pressure
Case 3: Casing collapse caused by mud loss to thief zone is equal to 1.64. This compares to DF = 1.69 when calculated
In this scenario the production casing has been cemented in in terms of VME stress.
place. Outside the casing there is original 1.86 SG (15.5 ppg)
drilling mud causing a collapse pressure. Inside the casing the Table 2 summarizes this field case. It is observed that the new
current drilling fluid has leaked into a permeable zone exact model provides an improvement in the design margins
penetrated during drilling. The new mud level in the well is for certain scenarios. The cases where the uniaxial design give
assumed in equilibrium with the normal pore pressure in the highest design factor, is not a correct solution as two stress
area. Inside the casing at the new mud level the pressure is components are neglected.
zero. Pertinent data is given in Table 1.
The do/t ratio for this Q125 casing is 15.8 and the temperature- The conventional triaxial approach for burst and collapse has
adjusted yield strength is 8280 bar (120 000 psi). According an inherent error due to simplifications. An exact 3D tubing
to API Bulletin 5C37, the yield collapse formula for this high- mechanics model is therefore developed in this paper. By
grade casing is valid for do/t less than 12.2. This range does introducing dimensionless parameters, the theory is presented
not cover the casing under investigation. Instead we need to as a three-parameter model that is simple to use.
use the API plastic collapse formulas, which are valid for do/t
between 12.2 and 19.9. In spite of this, we will carry on with Field cases from two HP/HT wells offshore Norway
the yield collapse analysis. demonstrate the new model. Typical errors resulting from
using simplified models are shown. It is seen that an
The full 3D model produces DF = 2.14. This is an optimistic improvement in design margins is achieved for certain cases.
value since we assume yield collapse. However, the design
factor is in excess of two and we conclude that the casing will The model can be used for cases including bending and
resist the collapse pressure caused by the drilling mud. The torsion, but must be checked for failure position. It is
triaxial model, also based on yield failure at the inside pipe recommended to use this new exact model for all tubing and
wall, gives the same result as the 3D model since the inside casing calculations.
pressure is zero.
The biaxial collapse analysis is performed using the API As Cross-sectional area of pipe, m2 (in2)
formulations7. The tension-adjusted yield strength is 7420 bar DF Design factor defined by Eq. 10
(107 500 psi) and the calculated collapse pressure is 699 bar do Outside pipe diameter, m (inch)
(10 100 psi). This gives DF = 2.04 based on pressure. There F Axial pipe force, N (lbs)
is no uniaxial model available for collapse calculations and pi Fluid pressure inside the pipe, Pa (psi)
therefore no results are obtained for the uniaxial case. po Fluid pressure outside the pipe, Pa (psi)
r Pipe radius, m (inch)
Case 4: Live well entry with snubbing ri Inside pipe radius, m (inch)
A 460k snubbing jack is used to deploy 5 premium tubing in ro Outside pipe radius, m (inch)
this well with 683 bar (9900 psi) wellhead pressure. The t Pipe wall thickness, m (inch)
maximum snub force is experienced while pushing the first x Dimensionless parameter defined by Eq. 8
joint into the well. The concern is that the wellhead pressure y Dimensionless parameter defined by Eq. 9
will collapse the pipe, which is run closed-ended and empty in z Dimensionless parameter defined by Eq. 10
the well. The compressive force working on the pipe below
the stripper rams is equal to the wellhead pressure multiplied Geometry factor defined by Eq. 4
by the cross-sectional area of the closed pipe. The resulting a Axial stress, Pa (psi)
compressive stress on the pipe is 1970 bar (28 600 psi) as h Hoop (tangential) stress, Pa (psi)
shown in Table 1. r Radial stress, Pa (psi)
VME Equivalent von Mises stress, Pa (psi)
For a design factor equal to 1.25, calculate the collapse y Yield strength, Pa (psi)
pressure po = 911 bar (13 200 psi) from Eq. 12, which y* Temperature-adjusted yield strength, Pa (psi)

References For the burst and collapse conditions assumed in this study,
1. Craft, B.C., Holden, W.R. and Graves, E.D., Jr.: Well the pipe will always fail at the inside wall. The Lam
Design: Drilling and Production, Prentice-Hall, equations expressed at the inside pipe wall are:
Englewood Cliffs, New Jersey (1962).
r = pi (A3)
2. Bourgoyne, A.T., Jr., Millheim, K.K, Chenevert, M.E.
and Young, F.S., Jr.: Applied Drilling Engineering, 2nd
printing, Society of Petroleum Engineers, Richardson, 2ro2 ( pi po )
h = pi (A4)
Texas (1991). ro2 ri2

3. Johnson, R., Jellison, M.J. and Klementich, E.F.:

Triaxial-Load-Capacity Diagrams Provide a New
Approach to Casing and Tubing Design Analysis, SPE
Drilling Engineering (September 1987) 268-274.

4. Kastor, R.L.: Triaxial Casing Design for Burst, paper

SPE 14727 presented at the IADC/SPE Drilling Case Pipe do/t pi po a y*
Conference, Dallas, Texas, 10-12 February 1986. bar bar bar bar
(psi) (psi) (psi) (psi)
5. Timoshenko, S.P. and Goodier, J.N.: Theory of Elasticity,
3rd edition, McGraw-Hill Book Company, New York 1 10 13.51 7.292 717 36 -634 6750
(1970). 85.3# (10400) (520) (-9200) (97900)
6. Pattillo, P.D.: How to Apply the New API Axial Load
2 9 15.80 8.434 1230 451 -2140 7850
Adjustment Part 1, Petroleum Engineer International 62.8# (17900) (6540) (-31100) (113800)
(March 1985) 40-50. Q125

7. Bulletin on Formulas and Calculations for Casing, 3 9 15.80 8.434 0 342 1510 8280
Tubing, Drill Pipe, and Line Pipe Properties, Bull. 5C3, 62.8# (0) (4960) (21900) (120000)
6th edition, American Petroleum Institute, Washington, Q125
D.C. (October 1994).
4 5 10.46 5.783 0 683 -1970 5770
8. Jellison, M.J. and Brock, J.N.: The Impact of 23.2# (0) (9900) (-28600) (83600)
Compression Forces on Casing-String Designs and
Connectors, SPE Drilling and Completion (December
2000) 241-248. Table 1: Data for HP/HT example wells

9. Aadny, B.S: Modern Well Design, 2nd printing, A.A.

Balkema, Rotterdam (1999).
Design factor (DF)

Case New 3D Conventional Biaxial Uniaxial

model triaxial model model
Appendix A: The Lam equations (exact) model
Axial stress is caused by axial load and bending. This stress is
directed along the pipe axis. Fluid pressure inside and outside
1 1.37 1.36 1.37 1.44
the pipe causes radial and hoop stresses. Both radial and hoop
stresses depend upon radial location. In the following 2 1.10 1.06 1.04 1.17
equations, subscript i refers to inside pipe wall and subscript o
denotes the outside wall. The general equations for radial and 3 (2.14) (2.14) 2.04 ---
hoop stress are5:
4 1.69 1.69 1.64 ---
ri2 ro2 ( po pi ) 1 pi ri po ro2
r = + (A1)
ro2 ri2 r2 ro2 ri2 Table 2: Design factors obtained for HP/HT example wells

ri2 ro2 ( po pi ) 1 pi ri2 po ro2

h = + (A2)
ro2 ri2 r2 ro2 ri2

z = DF

y = ( pi po ) / y

x = ( pi + a ) / y

Figure 1: Three-dimensional yield surface


DF = 1.00
DF = 1.10
DF = 1.20
DF = 1.30


y = ( pi po ) / y


Case 1
Case 2
S Case 3
Case 4

-1,5 -1,0 -0,5 0,0 0,5 1,0 1,5

x = ( pi + a ) / y

Figure 2: Two-dimensional design factors