Journal of Petroleum Science and Engineering 32 (2001) 53 71

www.elsevier.com/locate/jpetscieng

Design of oil wells using analytical friction models

Bernt S. Aadny a,*, Ketil Andersen b
a

Stavanger University, P.O. Box 2557, 4091 Stavanger, Norway

b
Statoil, 4035 Stavanger, Norway
Received 12 March 2001; accepted 4 September 2001

Abstract
Recent wells have been drilled to more than 10 km from the platform, and companies are planning to extend these beyond
12 km. Well friction is one of the most important limiting factors in this process.
Torque and drag prognoses are developed today on in-house simulators. Although this is a good tool for planning,
improvements are made on a trial and error basis, and these simulators have limited availability. To provide more insight into
the frictional aspect, a larger study was undertaken. Explicit analytical equations are derived to model drill string tension for
hoisting or lowering of the drill string. The equations are developed for straight sections, build-up sections, drop-off sections
and side bends. Both constant curvature models and a new modified catenary model are derived. The new catenary model is
developed for arbitrary entry and exit inclinations. Equations to determine well friction in three-dimensional well profiles are
also given. In addition, expressions for torque and drag are developed based on the tension equations. Equations for combined
motion and drilling with a motor are also given. Using these equations, the total friction in a well is derived from the sum of the
contributions from each hole section.
Examples are provided to demonstrate the application for ordinary production wells, catenary wells, long-reach wells and
horizontal wells. Optimization criteria are developed to design the well for minimum friction. D 2001 Elsevier Science B.V. All
rights reserved.
Keywords: Torque; Drag; Analytic models; Long-reach; Catenary; Build-up; Drop-off; Build rate; Buckling

1. Introduction
The oil industry is in general producing the easiest
accessible oil first because they are more economical.
As existing fields are being produced, however, it
becomes important to drain these in an optimum way.
Drilling technology plays an important role here
because the horizontal reach has more than doubled
during the last decade.
*

Corresponding author. Hogskolen i Stavanger, P.O. Box 2557,

N-4091, Stavanger, Norway. Fax: +47-51-832250.
E-mail address: Bernt.Aadnoy@tn.his.no (B.S. Aadny).

It has become evident that well friction is a limiting

factor in extended-reach drilling. Sheppard et al.
(1987) showed that an undersection trajectory could
have reduced drag compared to a conventional tangent
section. Banks et al. (1992) summarized the extendedreach capabilities. McClendon and Anders (1985) studied the catenary well profile and demonstrated their
advantage over conventional methods.
In Norway, this development became very important not only to drain older fields more efficiently,
but to reduce the number of offshore platforms in
new development projects. Eek-Olsen et al. (1993,
1994) and Alfsen et al. (1993) demonstrate the evo-

0920-4105/01/$ - see front matter D 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 9 2 0 - 4 1 0 5 ( 0 1 ) 0 0 1 4 7 - 4

54

B.S. Aadny, K. Andersen / Journal of Petroleum Science and Engineering 32 (2001) 5371

lution from a 3-km horizontal reach to more than 7

km. Justad et al. (1995) show the planning and
drilling of a complex long reach designer horizontal
well to develop marginal satellite fields from an
existing platform, while Blikra et al. (1994) addresses both the achievements and the cost-benefits.
Benesch et al. (1996) discuss an extended reach well
in Japan.
Aarrestad and Blikra (1994) present a good review
of the various aspects of torque and drag problems
encountered in extended-reach drilling. A general review is given by Payne et al. (1994). Gou et al. (1993)
and Wiggins et al. (1992) define equations to calculate
the well trajectory. When the well trajectory has been
determined, the complete well must be designed. A
recent review over design considerations and potential
problems is presented by Guild et al. (1995), while
Aadny (1996) describes the complete well design
process. Sheppard et al. (1987) formulated the torque
and drag models that are implemented in most simulators. Analytical expressions for build, drop, hold and
side profiles are presented in this paper, and also a
new modified catenary profile. Using these equations,
friction analysis can be made without requiring a simulator.

2. Models for torque and drag

The models shown here are derived in detail by
Aadny and Andersen (1998).

2.1. Applications
The drag and torque solutions presented below are
derived independently. Their applications are as follows.
The drag equation applies when tripping in or out
of the well. This may also apply to coiled tubing,
logging, completion or workover operations. The
torque equation applies to pure rotation while drilling.
The equation governing combined motion, both axial
movement and rotation, is also given. Reaming, for
example, can therefore also be modeled. The drilling
phase of the well is analyzed, however, the equations
are valid for other phases such as well completions,
provided that a correct scenario is defined.
2.2. Drag and torque along straight sections
Before proceeding with various frictional models,
the basic principles for well friction are defined. All
equations that follow are based on the soft string
model. String stiffness is neglected because it contributes a negligible amount to the tension. Fig. 1a
defines the forces acting on an inclined drillstring.
The force required to pull a drillstring along an
inclined plane is: F = mgcos(a) + lmgsin(a). If the
drillpipe is lowered instead, the friction acts opposite
to the direction of motion, resulting in a top force of:
F = mgcos(a)
lmgsin(a). This is a Coulomb friction model. From a stationary position, increasing or
decreasing the load, an equal amount will lead to

Fig. 1. Forces and geometry in straight hole sections.

B.S. Aadny, K. Andersen / Journal of Petroleum Science and Engineering 32 (2001) 5371

55

upward or downward movement of the drillstring.

For a drillstring of weight mg( = wDs) and an inclination a, the axial weight and the drag force in a
straight section is (Fig. 1b).

The same principle applies for rotating friction, the

torque. The applied torque is equal to the normal
moment (wDsr) multiplied with the friction factor (l).
The torque becomes:

F2 F1 wDscosaFlsina

T lwDsrsina

The plus sign defines pulling of the pipe, whereas a

minus sign defines lowering of the pipe. The first term
inside the brackets defines the weight of the string, the
second term defines the additional frictional force
required to move the pipe. The change in force when
initiating motion either up or down is found by
subtracting the weight from the forces defined above.
The weight is:
wDscosa

Please observe that the axial weight of the drillstring

is always equal to the unit weight multiplied with the
projected height (Aadny et al., 1999), regardless of
borehole inclination.

For this case, the axial force has no influence. The

torque can be considered independent of the direction
of rotation. Later, combined rotation and pulling will
be addressed. The unit mass of the drillpipe, or the
weight always must be corrected for buoyancy. The
buoyancy factor is:
b1

qmud
qdrill pipe

and the buoyed unit mass is:

w bwdrill

pipe

Here, wdrill pipe refers to the unit weight of the pipe in

air. Please observe that the buoyancy correction de-

Fig. 2. Forces and geometries of various curved hole profiles.

56

B.S. Aadny, K. Andersen / Journal of Petroleum Science and Engineering 32 (2001) 5371

fined in Eq. (5) is valid only for cases of equal fluid

densities on both sides of the drillpipe. Aadny et al.
(1999) defines buoyancy factors for different densities.
Eq. (1) indicates that the following condition is
required for the drillstring to slide downward: cos
(a)  lsin(a). The following condition for the maximum sail angle of a well is therefore:



1 1
amaximum Vtan
6
l
2.3. Drag and torque for other well geometries
Fig. 2 shows a number of different well geometries. The inclinations at top and bottom are also
shown. Gu et al. (1993) derived some of the equations, whereas Aadny and Andersen (1998) presented the complete derivations. The equations for
the well geometries shown in Fig. 2 are given in Table
1 for the construction of the well geometry. All
equations required to compute the torque and drag
friction of each geometry are summarized in Table 2.
The equation for a straight section is given in Table
2a. The friction is additive, which implies that the
friction depends on the weight of the pipe itself and
not on the total axial force. The friction is independent
of whether a high or a low pulling force is applied at

the end of the pipe. Equations for a drop-off section

are given in Table 2b. Friction depends on two
elements; the weight of the pipe itself and, the bottom
pulling force multiplied by an exponential expression;
hence, friction is no longer additive, but can be
considered multiplicative. The side bends defined in
Table 2d and e has added complexity. The weight of
the pipe forces the pipe to its lowest position, whereas
the pulling force attempts to move it to the mid point
of the hole. The exact position along the sidebend
depends on the vector sum of these two forces.
The modified catenary profile is defined in Table
2f. Assume that a rope is hung between two points,
where the shape given by a cosine/hyperbolic function, or approximately as a parabole. If a well is
drilled with a shape identical to one half side of the
hanging rope and a drillstring with right weight and
tension is inserted, it may theoretically not touch the
borehole and hence have no friction. The idea is also
to minimize well friction by maintaining a minimum
of normal force between the drillstring and the well.
The drawback of using a catenary solution is that the
build-rate must be changed continuously. A catenary
profile will always have some inclination at the top
starting point. For a vertical well, Table 2g defines
equations to construct a build-up section, which is
continuous both in inclination and build-rate.

Table 1
Geometrical projections for various section profiles
Section profile

Section length Ds

Vertical projection Dz

Horizontal projection Dx

Horizontal projection Dy

(a) Straight inclined

(b) Drop-off
(c) Build-up
(d) Right side-bend
(e) Left side-bend

Ds
R(a2
a1)
R(a2
a1)
R(/2
/1)
R(/2
/1)
F1
sin a1 sinhfAg
cos a1

Dscos a
R(sin a2
sin a1)

R(sin a2
sin a1)
0
0
F1 sin a1
coshfAg
B

Dssin a

R(cos a2
cos a1)
R(cos a2
cos a1)

R(cos /2
cos /1)

R(cos /2
cos /1)

R(sin /2
sin /1)

R(sin /2
sin/1)

R * sin a2 *

R*(1
cos a2*)

(f) Modified catenary

(g) Entrance modified R * a2 *

catenary

Where:


wx
sinh
1 cot a1
B coshfsinh
1 cot a1 g
A
F1 sina1
R*

F0 wDs2 2wDsF0 cos a1

wF0 sin a1

tan a2 *

wDs F1 cos a1
F1 sin a1

Subscript 1 is the deepest position, subscript 2 is the highest.

Dx

Section profile

Static weight

Torque

Pulling force

Lowering force

(a) Straight inclined

(b) Drop-off
(c) Build-up

wDscos a
wR(sin a2
sin a1)

wR(sin a2
sin a1)

lwDsrsin a
lr{( F1 + C)Aa2
a1A
D}
lr{( F1 + C)Aa2
a1A + D}
lrA/2
/1A(H
F1)

(d) Right side-bend

(e) Left side-bend

lrA/2
/1A(H
F1)

F1 + wDs(cos a + lsin a)
F1el(a2
a1) + E
F1e
l(a2
a1)
G
"
#
1
wR2
l/2
/1
He


2
Hel/2
/1
"
#
1
wR2
Hel/2
/1

2
Hel/2
/1

F1 + wDs(cos a + lsin a)
F1e
l(a2
a1) + G
F1el(a2
a1)
E
"
#
1
H
wR2 l/2
/1
e


2 el/2
/1
H
"
#
1
H
wR2 l/2
/1
e


2 el/2
/1
H

(f) Modified catenary

wDz

lrDFtan
1I

J + lDFtan
1I

J
lDFtan
1I

(g) Entrance modified

catenary

wR*(sin a1)

lr{( F1 + C)a1 + 2wR*(1
cos a1)}

Where:
C wRsin a1 D 2Rwcos a2
cos a1
i
wR h
E
1
l2 sin a2
ela2
a1 sin a1
2lcos a2
ela2
a1 cos a1
2
1l
q
G wRbsin a2
e
la2
a1 sin a1 c H F1 F12 wR2


q
wDs F1 cos a1
J F12 wDs2 2wDsF1 cos a1
I
F1 sin a1
i
wR h
1
l2 sin a1 2lcos a1
K
2
1l
Subscript 1 is the deepest position, subscript 2 is the highest.

( F1 + wR * sin a1)e

la1

F1 Ke
la1

2l
wR cos a1
1 l2

B.S. Aadny, K. Andersen / Journal of Petroleum Science and Engineering 32 (2001) 5371

Table 2
Torque and drag for various sections

57

58

B.S. Aadny, K. Andersen / Journal of Petroleum Science and Engineering 32 (2001) 5371

2.4. Combined friction for simultaneous hoisting and

rotation
During tripping, tight hole conditions may occur.
The remedy is typically to rotate the drillstring while
pulling, or to pump to provide hydraulic support
beneath the drill bit. Rotation will reduce the axial
drag as compared to a non-rotating pipe.
Fig. 3a shows a pipe section of weight wDs. If it is
pulled along a surface, the drag will be equal to the
normal force (weight) multiplied with the coefficient
of friction, lwDs. If the pipe is rotated instead, the
torque (T/r) ratio is also equal to lwDs. Thus, the
weight and friction coefficient results in the same
frictional resistance regardless of whether the movement is axial or rotational. If the pipe is subjected to
both motions simultaneously, the resultant friction is

still limited by the normal weight component. The

direction of the friction is determined by the resultant
of the velocities in the two directions (Fig. 3b). If the
pipe is only pulled, a drag of lwDs results in the axial
direction. If it is rotated only, a T/r ratio of lwDs
results in the tangential direction. If a combined
motion is applied, the same resultant frictional force
applies in some other direction. The torque and the
drag are related by the following equation (Fig. 3b):

2
T
F 2 lwDs2 Fcap
r

The term on the right side of Eq. (7) is produced by

the drag during no rotation, and may be defined as the
frictional capacity of the pipe. If combined axial and
rotational movement is applied, Eq. (7) can be used to

Fig. 3. Combined friction.

B.S. Aadny, K. Andersen / Journal of Petroleum Science and Engineering 32 (2001) 5371

compute the reduction both in torque and drag. The

angle g is defined by the velocities: g = tan
1(vt/va),
where vt is the tangential (peripheral) speed and va is
the axial speed of the drillpipe (Fig. 3b). The axial
drag is reduced to a minimum while reaming out of
the borehole. Two well operations that utilize this
effect are backreaming while tripping out, and, rotating the liner when placing it. In both cases, rotation
leads to reduced axial drag.
2.5. Drag and torque in combined bends
In some cases, the well is building or dropping and
turning sideways simultaneously, resulting in a threedimensional profile. The exact equations for the
friction are complex, because they involve forces
acting in two different planes. The resulting differential equation must be integrated through the well
path. Approximately one may compute the resultant
of the friction in the two planes as follows:
Pulling or lowering of the drill pipe:
q
F2 F22 build or drop F22 sidebend

Torque:
q
2
2
T Tbuild
or drop Tsidebend

ficient of friction. When bit load is applied, the string

is lowered resulting in a reduction of the hook load.
The tension in the drill string is reduced by the same
amount. Therefore, the bit load applied should be
subtracted when studying friction during drilling.
2.8. Torque when drilling with downhole motor
Using a downhole motor or turbine, the drill bit
rotates at a much higher speed than during rotary
drilling. The bit speed is the sum of the speed of the
motor, and the rotational speed of the drill string. The
string is often rotated at a low speed to reduce axial
friction. Experience shows that the bit torque is lower
when using a downhole motor (Guild et al., 1995).
The power source is downhole, and the reactive bit
torque must be balanced by friction in the borehole
and surface torque. Two examples are presented for
comparison. First, a conventional rotary drilling operation, and second, drilling with a downhole motor.
The power consumed at the drill bit for the two cases
can be expressed as:
Protary CTrotary nrotary
Pmotor CTmotor nmotor

2.6. Drag and torque in modified catenary sections

An ideal catenary profile results in no contact
between the hole and the borehole wall; hence, no
friction. During hoisting or lowering of the pipe,
however, the tension at the ends of the catenary
section changes due to friction elsewhere along the
drillstring. This deviation in tension results in friction,
which is expressed as:
DF F1 tripping out
F1 drilling

59

10

where C is a constant, P is the power, T is the torque

and n is the rotational speed of the drill bit. Assume
that the two cases consume the same power at the bit.
For this case, the bit torque when drilling with a motor
is expressed as:
nrotary
Tmotor
Trotary
11
nmotor
Thus, if the bit speed (when drilling with a motor) is
twice the rotary speed during conventional drilling, the
corresponding bit torque is just one half. The motor
actually acts as a torque reducer. Eq. (11) can be used to
normalize torque data for motor drilling, to make it
comparable to rotary drilling data. Also, using a downhole motor may be a means by which to minimize
torque when there are severe torque limits on the rig.

2.7. The total drag and torque

Any well can be analyzed by adding the contribution from each geometry starting from the bottom as
shown in the examples that follows. The result is
friction at surface due to contact friction along the
borehole. Current practice is to use an overall coef-

3. How to construct a well with the analytical

solutions
Table 1 defines the equations required to construct
various well geometries. One usually models a well

60

B.S. Aadny, K. Andersen / Journal of Petroleum Science and Engineering 32 (2001) 5371

constant radius to a sail angle, which is kept constant

to the bottom of the well. Tables 3 and 4 lists the
equations required to model the drag, torque and the
static weight for this case. The static weight is just the
unit weight multiplied by the projected height, regardless of the inclination of the well.
For this example, assume a unit weight of the
bottom-hole-assembly of 3 kN/m (200 m long), drillpipe of 0.3 kN/m (2000 m from BHA to end build
section), a 60 inclination, a build radius of 500 m, a
kick-off-depth of 1500 m and a mud weight of 1.56
s.g. which leads to a buoyancy factor (Eq. (4)) of 0.8.
The friction coefficient is 0.15. The torque is computed both with the bit off bottom, and with a bit
force of 150 kN, resulting in a bit torque of 6 kN/m.
With these numbers, the target is located at 3033 m
vertical depth, and the horizontal departure is 2155 m.
Inserting these numbers into the appropriate equations in Tables 3 and 4 (adding friction from bottom
to top):
Drag when pulling: F5 = 0 + 302 + 302 + 224 + 360
= 1188 kN;
Lowering: F5 = 0 + 178 + 178 + 18 + 360 = 734 kN;
Static weight: F5 = 0 + 240 + 240 + 104 + 360 = 944
kN;
Torque, bit off bottom: T5 = 0 + 6.24 + 6.24 + 10.94
+ 0 = 23.45 kN/m;
Torque, bit force 150 kN: T5 = 6 + 6.24 + 6.24 +
8.87 + 0 = 27.35 kN/m.

Fig. 4. Build sail type well.

starting from the bottom. If the drill bit is off bottom,

the force there is zero. If drilling is taking place, the
bottom compressive force is equal to the force on the
drill bit.
3.1. Example: well with build-and-hold profile
The well under consideration is shown in Fig. 4. It
is vertical to the kick-off point, and build with a

This well is illustrated in Fig. 5. The static pipe

weight curve in Fig. 5a has a constant slope regardless

Table 3
Example of designing model for drag during pulling and lowering of drill string
Force

Equationspulling string

Equationslowering string

Bit force
Force top bottom-hole-assembly
Force top of sail section

F1 = 0
F2 = F1 + wBHALBHA(cos a + lsin a)
F3 = F2 + wDPLDP(cos a + lsin a)

Force at kick-off-position

F4=( F3 + wDPRsin a)ela

F1 = 0
F2 = F1 + wBHALBHA(cos a
lsin a)
F3 = F2 + wDPLDP(cos a
lsin a)



wDP R
2
F4 F3
1


l
sin
a


2lcos
a

e
la
1 l2
2lwDP R

1 l2

Force on top of well

F5 = F4 + wDPLKOP

F5 = F4 + wDPLKOP

B.S. Aadny, K. Andersen / Journal of Petroleum Science and Engineering 32 (2001) 5371

61

Table 4
Example of designing model for static weight and torque during drilling
Force/torque

Equationsstatic weight

Equationstorque

At bit
Top bottom-hole-assembly
Top of sail section
At kick-off-position
On top of well

F1 = 0
F2 = F1 + wBHALBHAcos a
F3 = F2 + wDPLDP cos a
F4 = ( F3 + wDPRsin a)
F5 = F4 + wDPLKOP

T1 = 0
T2 = T1 + lwBHALBHArsin a
T3 = T2 + lwDPLDPrsin a
T4 = T3 + lr[( F3 + wDPRsin a)a + 2wDPR(1
cos a)]
T5 = T4

of inclination as discussed above. It is clear that the

build-up bend contributes significantly to well friction. As discussed, any time a change in the direction
of the well takes place, the friction becomes a function
of the tension in the string, not only the weight itself.
This results in a multiplicative effect as seen. This
effect is also seen in the torque curve shown in Fig.
5b. When the bit is on the bottom, however, effective
string tension is lowered, resulting in less torque
through the bend. The surface torque change is therefore less than the bit torque. From this simple example, one may conclude that to minimize friction, a
minimum of direction changes should take place in
the well path.

3.2. Example: constructing a modified catenary well

profile
A catenary build profile can provide an absolute
minimum of well friction if it is properly designed.
The term modified catenary refers to the model, which
do not require a horizontal direction at the bottom, but
can be designed with any sail angle.
Fig. 6 shows a long-reach well. The well has a
horizontal reach of 6000 m, and maximum sail angle
is 80. It is planned to be drilled with a downhole
motor and no rotation of the drillstring. The string
must therefore be able to slide into the borehole. To
minimize friction, it is decided to construct a catenary

Fig. 5. Well friction for Example 2.

62

B.S. Aadny, K. Andersen / Journal of Petroleum Science and Engineering 32 (2001) 5371

Fig. 6. Catenary well profile for Example 2.

profile from 1000- to 3000-m vertical depth. During

this critical phase, tension at the bottom of the
catenary is 300 kN, the buoyed pipe weight is 0.25
kN/m, and the coefficient of friction is 0.15. Using
Table 1f, the following geometrical equations results:
n x
o
0:1754
B 1:0154
A
1182
Ds 1182sinh A
208

Dz 1182cosh A
1200

The well path is constructed by computing the vertical

height Dz for each value of x.
One starts at the bottom of the catenary section
and works upwards. The measured along-hole length
Ds is also computed. Fig. 6 shows that the catenary
ends at 80, but has a starting inclination of 21.7. A
short build-up section is required to go from vertical
to the starting value of the catenary. Assuming
continuity in the angle and the slope (derivative) at
the matching point between the two profiles would
require a build-up radius of 8665 m. To make this
transition shorter, a build-up radius of 1000 m was
chosen instead. The build rate is 1.72/30 m. The
build-rate for the catenary is computed using Eq.

(B1.6.) in Aadny and Andersen (1998) and the

build-rates for the various sections are shown in
Fig. 5b.
During drilling, the theoretical friction in the catenary section is zero, and the total well friction is due
to the short build-up section above the catenary, and in
the sail section below. This should clearly represent a
minimum; however, when pulling the drillstring more
friction develops. There are two obvious drawbacks
by using catenary profile: (1) the axial load must be
controlled accurately to minimize friction and (2) a
constantly changing build-rate is required. Therefore,
unless one is prepared for the increased complexity
and follow-up that is required to drill a catenary, it is
not worth the effort. Catenary profiles are not used
very much.
3.3. Example: comparing trajectories for long-reach
well
In the following friction, analysis will be conducted on a long-reach well, using the equations of
Tables 3 and 4. A target is located at a depth of 2950
mTVD (vertical depth). The total well depth is 3100

B.S. Aadny, K. Andersen / Journal of Petroleum Science and Engineering 32 (2001) 5371

mTVD with a horizontal reach of 7528 m. The rig has

a hoisting capacity of 4454 kN (1,000,000 lb), and a
top drive torque of 35 kN/m (25,800 ft lb). Assume
that the hoisting capacity is sufficient, but that the top
drive is a limiting factor. This exercise will determine
which well profile result in the lowest friction. A
comparison of the well trajectories is shown in Fig. 7.
All well paths are designed to build from vertical to a
sail angle, which is kept in the reservoir.
The maximum sail angle is taken from the friction
coefficient used in the previous well, about 0.15
average for open and cased hole. To ensure that the
drillstring will slide when orienting, the inclination
should not exceed tan
1(1/0.15) = 81.47 (Eq. (7)).
The maximum sail angle, as a function of well
friction, is shown in Fig. 8.
Table 5 shows the friction for the four well profiles.
The modified catenary profile gave the lowest torque.
The undersection profile gave a little higher torque,
but are better than the standard well profile and the
minimum dog-leg profile. Most of the torque is
generated in the long sail section.
The hook loads for the four well profiles are also
shown in Table 5. The hook loads are similar, but the
standard profile gives a higher pulling force than the
other profiles. The maximum load is still less than half
of the capacity of the drill pipe. Therefore, tension is
not the limiting factor.
The example shows that all four well profiles
developed similar amounts of friction. The choice of

63

Fig. 8. Critical sail angle vs. friction coefficient.

well profile was not significant. Of course, this conclusion is only valid for the case considered. One
advantage the modified catenary profile has, compared to the minimum dog-leg profile, is the fact that
the build-up friction is generated over a shorter length
(975 vs. 2758 m). Friction reduction subs may therefore be applied over a much shorter length, resulting
in less cost. A significant part of the friction of the
modified catenary is due to the build-up before the
catenary starts. If a slant rig could be used, this
friction could be reduced to a minimum.

Fig. 7. Well trajections for Example 3.

64

B.S. Aadny, K. Andersen / Journal of Petroleum Science and Engineering 32 (2001) 5371

Table 5
Drag and torque for Example 3
Well Profile

Pulling
force (kN)

Lowering
force (kN)

Torque
(kN/m)

Modified catenary
Minimum dog-leg
Under-section
Standard

1360
1332
1321
1350

593
609
568
543

28.6
30.9
29.5
30.4

3.4. Example: ultra-long reach well design

The record long-reach wells today have more than
10-km horizontal reach. It is fully feasible, however,
to extend this towards, and even beyond, 12 km by a
well-planned design and a operational follow-up.
The well profile should be as simple as possible
consisting of a vertical section, a build section and a
sail angle towards the target. Avoid drop-off into the
reservoir, if possible, to minimize friction.
The sail angle should in general be as high as
possible, to reduce axial tension and hence friction in
the curved hole sections. Usually the reservoir depth is
such that the target is within reach only if a very high
sail angle is used. The maximum sail angle is given by
the friction coefficient. Therefore, another requirement
is low friction. This can be obtained by using oil muds
or friction reducers. To limit the load on the drillpipe, a
high buoyancy can be beneficial. The disadvantage of
a dense fluid is the compromise between buoyancy and
friction. Usually more particles in the mud will
increase the friction (Aadny, 1996).

In long wells, hydraulic friction may limit the flow

rate, thus leading to poor hole cleaning. Increased pipe
size will reduce this problem. Since increased pipe
size leads to increased pipe weight, drill pipes of
alternative materials may be required. Today drill
pipes are available both in aluminum, titanium and
composites.
In the following, an example of extending the
well from Section 3.3 from a reach of 7528 m to 12
km will be shown. Fig. 9 shows the resulting well
path. Maintaining the same sail angle, this well will
reach the target at a depth of 3767 mTVD (from
RKB). The build-up section chosen is the minimum
dog-leg profile.
An analysis of the hydraulics and hole cleaning
problems of such a long well resulted in the conclusion
that the drill string should have an outer diameter of at
least 5.5 in. The analysis is therefore assuming a 6-5/8
in. drillpipe in the upper 2755 m, and 5.5-in. drillpipe
down to the bottom-hole-assembly. Due to the fact that
the vertical depth of the well is not deep, the hook load
is reasonably low. Table 6 shows the results. Using steel
drillpipe, the hook load is about a third of the hoisting
capacity on the rig, and less than the strength of the
pipe. The hook load is therefore not a limiting factor
during drilling of the well. Table 6 also shows the
torque for the case of using 5.5-in. drillpipe. The total
predicted torque is 53.9 kN/m. Although a few drilling
rigs can handle torque of this magnitude, it is too high
for many drilling rigs. Using an all steel drillstring the
well would probably not reach the target.

Fig. 9. Extending the well of Example 3 to a horizontal reach of 12 km.

B.S. Aadny, K. Andersen / Journal of Petroleum Science and Engineering 32 (2001) 5371
Table 6
Comparison of torque and drag in 12 km horizontal reach well using
5.5-in. drillpipes of various materials in the sail section
Drillpipe

Hook load (kN)

Steel
Titanium
Composite

Torque (kN/m)

Pulling

Static

Lowering

Buildup

Sail
section

Total

1790
1330
1020

1750
1300
1010

600
600
600

10.2
6.2
3.7

43.7
26.9
16.0

53.9
33.1
19.7

Assume 6-5/8 in. steel pipe from surface to start of sail section.

Minimum lowering force:

dF5
0 ! tana
da
2lwBHA LBHA wDP LDP
wDP R
lwDP R
1
l2 wBHA LBHA wDP LDP
13
Minimum torque:
dT
0 ! tana
da

It was decided to try with a light drillpipe. Most of

the torque is generated from the tension from the sail
section, which is converted into torque in the build
section. The sail section was therefore first modeled
with a drill pipe made of titanium. As seen in Table
6, the torque drops to 33.1 kN/m. Since composite
drill pipes now are available (Hareland et al., 1997)
(8.72 lb/ft), the sail section was also modeled with
this as well. The cumulative torque was now down to
19.7 kN/m. Bit torque must be added to all these
values.
From this study, it is concluded that it is fully
possible to drill extended reach well to 12 km and
even longer. Actually, by using lighter drill pipes as
shown, it should be possible to drill several kilometers
beyond that. The most important limitation is believed
to be the hydraulic system; one probably needs a
higher pressure mud pump.

4. Well path optimization

The example of Section 3.1 will be used to investigate the optimum path of the well. Specifically, the
depth to the kick-off point that gives a minimum well
friction will be determined. Minimum drag and torque
can be found from Tables 3 and 4:
Minimum pull force:
dF5
0 ! tana
da

2wBHA LBHA wDP LDP wDP R

1
l2 wBHA LBHA wDP LDP
lwDP R
12

65

2wBHA LBHA wDP LDP awDP R

awBHA LBHA wDP LDP
3wDP R

14

4.1. Example: determine optimal kick-off-depth and

well inclination
Using Example 1, and inserting the data into Eqs.
(12), (13) and (14), the following optimal inclinations
are obtained:
Pulling: a = 68
Lowering: a = 0
Torque: a = 90
Maximum sail angle (Eq. (6)): a = 81.5.
From torque considerations, and also from pull
considerations, a deep KOP is preferred; however, to
minimize friction during lowering, a shallow KOP is
best. This demonstrates that an exact optimum does
not exist, but one has to select an optimization criterion. The drillstring is heavily loaded during rotation
and during pulling; therefore, these are selected to be
optimized. Lowering is usually not critical, and will be
neglected; however, for long horizontal wells lowering
can be critical. Different optimization criteria may
apply for different well types.
From this simple geometry studied the following
observations can be drawn. The build-up section
contributes most to well friction, and minimum friction can be obtained by placing the KOP deep. The
assumption is that lowering friction is considered
unimportant. The sail angle should have an inclination
between 68 and 81.5. The kick-off point should be
therefore between 1740 and 2173 m.
A kick-off-depth of 2100 m, resulting in a sail
angle of 76 as a compromise between the pull force
and minimum torque will be used. If the real coef-

66

B.S. Aadny, K. Andersen / Journal of Petroleum Science and Engineering 32 (2001) 5371

Fig. 10. Well path and drilling configuration for long horizontal well.

ficient of friction should increase slightly (e.g., by fill

in hole), the drillstring will still slide.
Equations can be derived for more complex geometries, and a solution obtained as shown above. For
general three-dimensional geometries, Rudolf et al.
(1998) and Suryanarayana and McCann (1998) have
derived non-linear optimization procedures.

of differences compared to an ordinary wells, which

we will address in the following.
Using Table 2, equations for the case of a horizontal well are given in Tables 7 and 8. For simplicity, we
have assumed drill collars covering the whole buildup bend. Equations can be derived for any drillstring
geometry.
An example is presented to demonstrate the properties of this well type.

5. Long horizontal wells and buckling

5.1. Example: well friction in a horizontal well
The well types covered in the previous analysis all
assume that the drillstring slides into the borehole by
gravity, as defined by Eq. (6). There is another class of
wells where the drill string must be forced into the
borehole. This is long horizontal wells (Fig. 10). Here,
the drillcollars are placed in the vertical or in the
build-up section to provide weight. The complete
drillstring in the horizontal section is placed in compression during drilling. This case leads to a number

With reference to Fig. 10, a horizontal well has the

following data:
Mud weight: 1.33 s.g. which leads to a buoyancy
factor of 0.83.
Weight BHA 1 kN/m, length 60 m.
Weight drill pipe 0.4 kN/m, length horizontal
section 1360 m.

Table 7
Example of designing model for drag during pulling and lowering of drill string, for horizontal well of Fig. 10
Force

Equationspulling string

Equationslowering string

Bit force
Force end build-up

F1 = 0
F2 = F1 + l(wBHAL1 + wDPL2)

Force start build-up

F3 = (F2 + wDCR)el2

F1 = 0
F2 = F1 + l(wBHAL1 + wDPL2)



1
l2
l p
2l
F3 = F2 wDC R
wDC R
e 2

1 l2
1 l2

Force top drill collar

Force on top of well

F4 = F3 + wDCL4
F5 = F4 + wDPL5

F4 = F3 + wDCL4
F5 = F4 + wDPL5

B.S. Aadny, K. Andersen / Journal of Petroleum Science and Engineering 32 (2001) 5371

67

Table 8
Example of designing model for static weight and torque during drilling, for horizontal well of Fig. 10
Force/torque

Equationsstatic weight

Equationstorque

At bit
Bottom build-up

F1 = 0
F2 = F1

Top build-up

F3 = F2 + wDCR

Top drillcollar

F4 = F3 + wDCL4

T1
T2 = T1 + lr (wDCL1 + wDPL2)

p
T3 = T2 lrRwDC 2
2
T4 = T3

On top of well

F5 = F4 + wDCL5

T5 = T4

Weight drill collars 1 kN/m, height 40 m above

kick-off point.
280 m drillpipe to surface.
Radius build-section 300 m.
Radial clearance of rotating pipe 0.18 m.
These data are inserted in Tables 7 and 8, and all
forces computed. The results are shown in Fig. 11.
The static string weight are shown in Fig. 11a. This
is zero at 620-m depth, as the well is horizontal at this
point; however, during pulling, the horizontal section
give friction. The build-up section provides a significant portion of friction and the lowering curve appears

completely different from an ordinary well. In the

horizontal section, the pipes must be pushed in since
there is no gravitational effect here. The complete
bottom portion of the well is in compression, and one
must consider pipe buckling.
During drilling most of the axial friction disappear due to string rotation as explained earlier. For
fast rotation, the hookload approaches the static
string weight. As shown later, a buckling force of
109 kN is computed for this example. Let us assume
that we apply a bit force of 109 kN. During a fast
rotation this will be the compressive force throughout the horizontal section (Fig. 11a). For this exam-

Fig. 11. (a) Drag forces for horizontal well of Example 6. (b) Torque for horizontal well of Example 6.

68

B.S. Aadny, K. Andersen / Journal of Petroleum Science and Engineering 32 (2001) 5371

ple, the critical buckling load is exceeded during

downward sliding of the drill string, but not during
drilling.
Fig. 11b shows the torque through the well. Again
the build-up section creates most of the friction.
Assuming a bit force of 109 kN and a bit torque of

5 kN/m, the torque during drilling is shown. It is

interesting to observe that a 5 kN/m bit torque at
bottom results in only 2 kN/m additional torque at
surface. The reason is that the reduced axial load
when the bit force is applied results in less torque in
the build-up bend.

6. Neutral point and buckling

The lower part of the BHA is usually in a compressive state whereas the drillstring is usually in tension. The
transition from tension to compression is called the neutral point. If a pipe is loaded axially beyond a critical point,
it will buckle and assume a sinus shape or some other shape that deviates from a straight pipe. Loading the pipe
beyond the buckling point may lead to increased well friction, ultimately a stuck pipe, or pipe failure. Obviously, it
is important for practical well operations not to exceed the critical buckling load.
Buckling limit should be checked for any component put into a compressive load. In addition, this may be the
case at the bottom of the drillstring for completion and production equipment as well.
The simplest case of buckling is the sinusoidal buckling of the drillpipe. Increasing the load further leads to
helical buckling, the string assumes the shape of a coil. These are valid for straight pipes and wellbores. If the
wellbore is curved, this actually stabilizes the pipe, and a higher critical load is required to initiate buckling. All
these models are given by He and Kyllingstad (1995) and Kyllingstad (1995). Reference is also given to Mitchell
(1996) and Dawson and Paslay (1994).
The starting point for any buckling analysis is the so-called Euler equation. From classical mechanics, the
critical load to initiate buckling is given by the following equation:
Fcr

p2 EI
L2

If other end condition applies like fixed ends, a constant must be multiplied to the above equation. This equation
is at best valid for vertical boreholes. For the general case with deviated boreholes, the following equation defines
the critical load to initiate buckling:
r
bFN EI
Fcr
r

15

Here, FN is the unit normal force on the pipe, and r is the radial clearance between the drillpipe and the
borehole.
The scaling factor b is defined by Kyllingstad (1995) as follows:
b = 4 for sinusoidal buckling in straight wellbores
b = 8.4 for helical buckling in curved boreholes.
In other words, a curvature in the wellbore actually stabilizes the pipe. He et al. (1995) also showed that by
simultaneously applying torque to the pipe only lowered the critical buckling force by a few percent, suggesting
that this effect may be neglected.
Measuring the hook load during a drilling operation, the coefficient of friction can be calculated using the
models presented in this paper. Normally this coefficient of friction should be similar when hoisting or lowering
the drillstring. In some cases, a higher coefficient of friction is found when lowering the drillstring. This can be due

B.S. Aadny, K. Andersen / Journal of Petroleum Science and Engineering 32 (2001) 5371

69

to buckling in the horizontal section, but it could also be due to cuttings in the borehole. When hoisting, the
drillstring may move along the high side of hole since it is in total tension. As shown in Fig. 11a, during lowering
of the string may be in compression resulting in sliding along the low side of the hole. Cuttings may accumulate on
the low side of the hole resulting in higher friction. This demonstrates the necessity of proper hole cleaning, and
may also warrant use of lubricants. In general, buckling may be eliminated by applying rotation, but if drilling in
an oriented mode by sliding the string, hole cleaning must be adequate to minimize friction.
6.1. Example: buckling in horizontal well
For the drillstring of Example 6, the critical buckling force defined by Eq. (15) will be computed.
The moment of inertia for the drill string in the horizontal section is:
I

p
p 4
D4
d 4
5
44 18:11 in4 7:54  10
6 m4
64
64

The normal force on the drillstring is the buoyed weight of the string itself, since the wellbore is horizontal.
Inserting the data into Eq. (15), the critical buckling force becomes:
s
4  0:83  0:4 kN=m  215  106 kN=m2  7:54  10
6
Fcr
0:18 m
The critical buckling force becomes: Fcr = 109 kN. This represents the maximum bit force that can be applied,
with reference to Fig. 11a. While sliding the drillstring downwards, friction results in buckling at the start of the
horizontal section; however, if the string is rotated, axial friction decreases and the axial load reduces.
Buckling causes constraint in the drilling operation. Repeating the buckling analysis with a heavy-weight
drillpipe (5  3 in.), the critical buckling load increased to 133 kN. This is not the best solution. Inspection of the
equation for the moment of inertia, one observes that increasing the outer diameter of the pipe has a more
significant effect. Using a 6-5/8 in. thin-walled drillpipe with inner diameter of 5.96 in. results in a buckling force
of 166 kN. This pipe has the same unit weight as the initial 5-in. drillpipe.
Another issue is the weight required to provide sufficient bit force. Assuming a bit force of 166 kN, and a
buoyed weight of drillcollars, a vertical depth of drillcollars of 201 m is required. Since the drillcollars are in the
build-up section, a minimum length is 210 m. In summary, the following conditions are observed on a long
horizontal well.
. Buckling may occur at the start of the horizontal section. Use large diameter thin-walled pipe to increase pipe
stiffness, and to minimize pipe weight. Small clearance between hole and drillstring also reduces buckling.
. Maximum bit force is given by this critical buckling force. During drilling, the force will be constant
throughout the horizontal section.
. Weight of drill collars required is also defined by the buckling force. As a minimum, let the vertical height of
drillcollars times the buoyed weight equal the buckling force. The buckling force is the major controlling factor (or
limitation) and is the design parameter for bit force, and drill collar weight. To reduce axial friction when buckling
occur, always rotate pipe. Rotation has negligible effect on buckling.

7. Summary
Friction models for a number of different well
geometries are presented. Explicit equations are given

to model both the rotary torque and the drag forces

associated with hoisting or lowering of the drill string.
Equations to compute well friction for a fully threedimensional well path are also given. Equations for

70

B.S. Aadny, K. Andersen / Journal of Petroleum Science and Engineering 32 (2001) 5371

combined motion and for drilling with a motor are

also given. These equations can be applied to any type
of well using, e.g. a spreadsheet.
Several examples demonstrate the use of the
analytical models. It is shown that any change of
direction in the well paths contributes to increased
friction. In a straight well section, the friction is
additive, while in bends it is defined as multiplicative,
that it depends on the pipe tension locally. One
example compares various well design, whereas
another example shows a way to design ultra-long
wells by using light-weight pipes.
An optimization analysis is performed with the
models, and it is shown that for the build-sail well
geometry it is generally preferable to place the kickoff-point (KOP) deep.
Horizontal wells often require using the drillstring
in compression. An example demonstrates the design,
and also defines relevant buckling criteria.
Nomenclature
x, y
coordinates in the horizontal plane
z
vertical coordinate
Ds
measured length along hole section
Dx, Dy, Dz projected distances
w
unit weight of drillpipe (kg/m)
F
force along drillstring (kN)
F1
force at bottom of section
F2
force at top of section
T
torque along drillstring (kN/m)
b
buoyancy factor
P
power (kW)
n
rotary speed (rpm)
C
conversion constant
a
inclination of string from vertical (rad)
/
azimuth of the section (rad)
l
coefficient of friction
DF
additional force applied to catenary
WOB weight-on-bit
BHA
bottom-hole-assembly
R
radius of bend (m)
r
radius of tool joint (m)
L
length of pipe (m)
TVD
true vertical depth (m)
RKB
drill floor reference
DL
dog-leg; angular change in well path
DLS
dog-leg severity; rate of dog-leg (= 30 (m) 
57.3(/rad)/R (m))

Fcap
g

frictional capacity of the drill string

tangential speed/axial speed of the drillpipe

References
Aadny, B.S., 1996. Modern Well Design, 1st edn. Balkema, Rotterdam, Netherlands, 240 pp.
Aadny, B.S., Andersen, K., 1998. Friction analysis for long-reach
wells. IADC/SPE 39391, 1998 IADC/SPE Drilling Conference,
Dallas, TX, 3 6 March.
Aadny, B.S., Larsen, K., Berg, P.C., 1999. Analysis of stuck pipe
in deviated boreholes. SPE 56628 1999 Annual Technical Conference and Exhibition, Houston, TX, 3 6 October.
Aarrestad, T.V., Blikra, H., 1994. Torque and dragtwo factors in
extended-reach drilling. J. Pet. Technol., 800 803, Sep.
Alfsen, T.E., Heggen, S., Blikra, H., Tjoetta, H., 1993. Pushing the
limits for extended reach drilling: World record from platform
C, well C2. SPE Drilling and Completion 10 (2), 71 76.
Banks, S.M., Hogg, T.W., Thorogood, J.L., 1992. Increasing extended-reach capabilities through wellbore profile optimization.
IADC/SPE 23850, 1992 IADC/SPE Drilling Conference, New
Orleans, Feb. 18 21.
Benesch, J.M., Camacho, G., Matsuzawa, S., Dawson, C.R., 1996.
Planning a record extended-reach well in Japan. Pet. Eng. Int.,
59 67 April.
Blikra, H., Drevdal, K.E., Aarrestad, T.V., 1994. Extended reach,
horizontal and complex design wells: challenges, achievements
and cost-benefits. 14th World Petroleum Congress, Stavanger
May 29 June 1. In Proceedings: WPC, London, vol. 2, pp.
191 201.
Dawson, R., Paslay, P.R., 1994. Drillpipe buckling in inclined holes.
J. Pet. Technol., 1734 1738, Oct.
Eek-Olsen, J., Sletten, H., Reynolds, J.T., Samuell, J.G., 1993.
North Sea advances in extended reach drilling. SPE/IADC
25750, 1993 SPE/IADC Drilling Conference, Amsterdam, Feb.
23 25.
Eek-Olsen, J., Drevdal, K.E., Samuell, J., Reynolds, J., 1994. Design directional drilling to increase total recovery and production rates. SPE/IADC 27461, 1994 SPE/IADC Drilling Conference, Dallas, Texas, Feb. 15 18.
Gou, R., Lee, R.L., Miska, S., 1993. Constant curvature equations improve design of 3-D well trajectory. Oil Gas J., 38 47, Apr. 19.
Gu, H., Newman, K.R., Haugland, L.F., 1993. Analysis of slack-off
force transmitted downhole in coiled tubing operations. SPE
26511, 68th Ann. Tech. Conf. and Exhibition of the SPE, Houston, TX, Oct. 3 6.
Guild, G.J., Hill, T.H., Summers, M.A., 1995. Designing and drilling extended reach wells. Pet. Eng. Int., 35 41, Jan.
Hareland, G., Lyons, W.C., Baldwin, D.D., Briggs, G., Bratli, R.K.,
1997. Extended reach composite materials drill pipe. SPE/IADC
37646, 1997 SPE/IADC Drilling Conference, Amsterdam,
March 4 6.
., 1995. Helical buckling and lock-up conHe, X., Kyllingstad, A
ditions for coiled tubing in curved wells. SPE Drill. Completion
10 (1), March 10 15.

B.S. Aadny, K. Andersen / Journal of Petroleum Science and Engineering 32 (2001) 5371
., 1995. Interactions between
He, X., Halsey, G.W., Kyllingstad, A
torque and helical buckling in drilling. SPE 30521, 1995 SPE
Ann. Tech. Conf., Dallas, TX, 20 25, Oct.
Justad, T., Jacobsen, B., Blikra, H., Gaskin, G., Clarke, C., Ritchie,
A., 1995. Extending barriers to develop a marginal satellite field
from an existing platform. SPE/IADC 28294, 69th Ann. Tech.
Conf. of the SPE, New Orleans, Sept. 25 28.
., 1995. Buckling of tubular strings in curved wells. J.
Kyllingstad, A
Pet. Sci. Eng. 12, 209 218.
McClendon, R.T., Anders, E.O., 1985. Directional drilling using the
catenary method. SPE/IADC 13478, 1985 SPE/IADC Drilling
Conference, New Orleans, La., March 6 8.
Mitchell, R.F., 1996. Buckling analysis in deviated wells: a practical
method. SPE 36761, 1996 SPE Annual Technical Conference
and Exhibition, Denver, Colorado, 6 9, October.

71

Payne, M.L., Cocking, D.A., Hatch, A.J., 1994. Critical technologies for success in extended reach drilling. SPE 28293, SPE 69th
Ann. Tech. Conf. and Exhibition, New Orleans, LA, Sep. 25
28.
Rudolf, R.L., Suryanarayana, P.V.R., McCann, R., Rupani, R.A.,
1998. An algorithm and program to plan optimal horizontal well
paths. ETCE 98-4532, Energy Sources Tech. Conf., ASME.
Sheppard, M.C., Wick, C., Burgess, T., 1987. Designing well paths
to reduce drag and torque. SPE Drill. Eng., 344 350, Dec.
Suryanarayana, P.V.R., McCann, R.C., 1998. Horizontal well-path
planning and correction using optimization techniques. ETCE
98-4540, Energy Sources Tech. Conf., ASME.
Wiggins, M.L., Choe, J., Juvkam-Vold, H.C., 1992. Single equation
simplifies horizontal, directional drilling plans. Oil Gas J., 74
77, Nov. 2.