SPE/IADC 105068

How Good is the Torque-Drag Model?

Robert F. Mitchell and Robello Samuel, Halliburton DE&DS

Copyright 2007, SPE/IADC Drilling Conference

force is attributed to friction generated by drillstring contact
This paper was prepared for presentation at the 2007 SPE/IADC Drilling Conference held in with the wellbore. When rotating, this same friction will
Amsterdam, The Netherlands, 2022 February 2007.
reduce the surface torque transmitted to the bit. It is useful to
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
be able to estimate the friction forces when planning a well or
paper, as presented, have not been reviewed by the Society of Petroleum Engineers or doing post-mortem analysis.
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
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
Analysis of these drillstring loads is done with drillstring
and International Association of Drilling Contractors is prohibited. Permission to reproduce in computer models, and there have been many drillstring models
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 developed over the last 30 years. By far the most common
was presented. Write Librarian, SPE, P.O. Box 833836, Richardson, TX 75083-3836 U.S.A.,
fax 1.972.952.9435.
method for drillstring analysis is the torque-drag model
originally developed by Dawson and Lane (Johancsik, 1973)
Abstract and put into differential equation form by Sheppard (1987).
Perhaps the only standard drillstring model in use today is Because of the simplicity and general availability of this
the torque-drag model originally developed by Johancsik model, it has been used extensively for planning and in the
(1973) in 1984, and put in a standard form by Sheppard field. If any drillstring model could be called standard, this
(1987). Because of the simplicity and general availability of would be the one.
this model, it has been used extensively for planning and in
the field. This model is known to be an approximation of real In the most basic terms, a drillstring model must determine the
drillstring behavior, in particular, that the bending stiffness is trajectory of an elastic rod constrained by a wellbore. The
neglected, so the torque-drag model is often called a soft- elastic rod will contact the wellbore in a number of places in
string model. There have been many stiff-string models either point contact or continuous contact, and the points of
developed, but there is no industry standard formulation. contact must be determined by the calculation process.
Field experience indicates that the friction coefficient that Furthermore, the geometric non-linearities associated with the
matches torque data does not match drag data. wellbore trajectory usually violate the basic assumptions used
in developing textbook elastic rod equations, so more general
Why does this model work well in some circumstances, yet elastic relations must be used [e.g. Love (1944) and Nordgren
not so well in other cases? Part of the problem may lie in the (1974)]. Friction further complicates the problem because the
description of the wellbore trajectory and the wellbore friction force direction is not known. One of the most well-
tortuosity. However, this study concentrates on the actual known formulations was developed by Walker and Friedman
model formulation. The torque-drag model formulation has (1977). A comprehensive paper on formulation, unfortunately
been reviewed in the context of a large displacement often overlooked, is the work of Ho (Ho, 1986). Ho followed
equilibrium analysis and it was determined that the model this paper with a simplification for soft string models (Ho,
satisfies all of the force equilibrium equations, but only one of 1988). Ho used a curvilinear coordinate system based on the
the moment equilibrium equations. The remaining two Serret-Frenet equations for a curve in space.
moment equations therefore provide objective criteria for
evaluating specific torque-drag problems. The torque-drag model formulation solves the wellbore
contact problem by assuming that the drillstring trajectory is
Several example problems with different inclinations and the same as the wellbore trajectory. Contact is further assumed
wellbore curvatures are studied to show how these criteria can to be continuous. This assumed trajectory is certainly within
be used to validate torque-drag models. Some general inches of the actual drillstring trajectory. Unfortunately, the
observations are made about torque-drag model applicability, most commonly used wellbore trajectory model (minimum
but the real value of this study is the test criteria themselves. curvature see Appendix B) implies that bending moments
are not continuous at survey points. The solution to this
Introduction problem is to assume that the drillstring is a single-force
Generally speaking, what do we mean by torque-drag beam, with the force assumed tangent to the trajectory.
modeling? Drag is the excess load compared to rotating Because the bending stiffness is neglected, the torque-drag
drillstring weight, which may be either positive when pulling model is often called a cable, chain, or soft-string
the drillstring or negative while sliding into the well. This drag model. Note that this assumption is not necessary if a different
2 SPE/IADC 105068

wellbore trajectory model is used that is smooth enough.

modified by the term 1 + f . Where does this term come
2

In this paper, we will take these two basic assumptions and from? As shown in Figure 1, the pipe is trying to climb the
derive all the consequences of mechanical equilibrium for the wellbore wall because of the friction force generated by the
drillstring. rotation. As a result, the contact force will be slightly less than
if the pipe lay on the bottom of the wellbore. This term should
Mechanical Equilibrium of the Torque-Drag Model be applied to any torque calculation model. The second term
To critique the conventional torque-drag model, we need a in equation (4) is almost always unimportant. As already
more detailed model, so that we can identify the simplifying stated, the product rp will always be small compared to one.
assumptions. In appendix C, we identify all the loads The vertical component of the binormal vector is small except
associated with friction, and in Appendix D, we derive the for wells turning on a horizontal plane. The combination of a
equilibrium equations for pipe that has shear forces as well as short radius turn on a horizontal plane rarely occurs. Even so,
axial forces, in effect a stiff-string version of the torque-drag using worst case values, the second term will rarely be more
model. than 1% of the total.

Torque Calculations Drag Calculations

The equilibrium equations for torque calculations are given in The shear forces for drag calculations can be evaluated from
equations (D-7). We can immediately evaluate the magnitude equations (D-6)e,f:
of the shear forces from the moment equations: Fb = f rp w c sin
Fn = 0 . . . . . .(5)
. . . . . .(1) Fn = f rp w c cos
Fb = M t As shown in Appendix D, Fb is a constant, and:
v
where Fn is the
v shear force in the n direction, Fb is the shear Fb = f rp w bp b z . . . . . .(6)
force in the b direction, is the wellbore curvature, and Mt is
The total magnitude of the shear force is:
the axial torque. Notice that one moment equation cannot be
satisfied exactly by the torque-drag model. Even so, we can Fs = Fb2 + Fn2 = f rp w c . . . . . .(7)
see that for torque calculations, the assumption of a single
If we explicitly neglect dFn/ds, then the contact force wc can
force beam is not too bad. First, Fn is identically zero. Second,
be shown to be the conventional torque-drag contact force,
the curvature is typically a small number. About the largest
equation (3).
curvature, for a short radius well, would be approximately .02
ft-1. We can readily evaluate the goodness of the single force
How Good is the Torque-Drag Model?
assumption by using equation 1b.
The industry consensus is that the torque-drag model is pretty
good. For example, Lesso (1989) discusses an application of
The contact force wc is given by:
the torque-drag model to the Amaulijak Field. While this is a

wc =
(F + w
e bp n z ) + (w bp b z )
2 2

+
f rp w bp b z successful application of torque-drag models, note that the
friction coefficient needed to match drag results is very
1+ 2
f 1 + f2 different from the friction coefficient needed to match torque
. . . . . .(2) results.
where f is the coulomb friction factor, Fe is the effective axial
force in the pipe, wbp is the buoyant weight of the pipe, rp is In many other cases, the friction coefficient needed to model
the tool joint radius, nz is the vertical component of the normal running into the hole differs from the coefficient for pulling
vector, and bz is the vertical component of the binormal out of the hole. To further confuse the issue, the coefficient
vector. These terms are defined in the appendices. This may change from trip to trip. It is clear that the model is
equation was derived assuming rp is small compared to 1, useful, but caution in the use of the model is certainly
which is always true for real wellbore trajectories. The contact justified.
force derived by Sheppard (1987), wcs, in our notation, equals:
(w sin Fe ) + (Fe sin )
Is the model a soft string model? The magnitude of the shear
w cs =
2 2
bp force is readily calculated, using equation (1) for torque
calculations, and equation (7) for drag calculations. These
This can be shown to equal (see Appendix E):
terms cannot actually be zero, for equation (1) implies that the
w cs = (F + w
e bp n z ) + (w bp b z )
2 2
. . . . . .(3) torque would be zero in torque calculations, and equation (7)
implies that drag would be zero in drag calculations. One
The contact force, in terms of the torque-drag model contact could actually argue that torque-drag models have always had
force wcs, equals shear forces, just that nobody ever bothered to calculate them.
w cs f rp w bp b z That is, the torque-drag model is actually a stiff string model.
wc = + . . . . . .(4)
1 + f2 1 + f2 So where does the torque-drag model fail? We can fix up the
First, we note that the torque-drag contact force should be contact force for torque calculations easily, using equation (2).
But the real culprit is the trajectory model. We can see in
SPE/IADC 105068 3

Appendix D (equation D-4) that the constant curvature so magnitudes increase by about three times as we move from
assumption nullifies the impact of the bending moment on the bar to bar. The height of the bar is the fraction of the string
equilibrium equations. Further, we see that the constant length over which the torsion acts. We can see that Case 2
curvature model implies a discontinuity in the bending (dark bars) has much more of the higher torsions values
moment at survey points. We have, in effect, neglected the applied over greater fraction of the string length.
contact loads that would have generated this change in
bending moment. Next, we compared d/ds, as defined in equation 11, for the
two cases in Figure 3. In this bar chart, the logarithm of d/ds
If we assume that the wellbore trajectory model has a variable was sorted into groups with width equal to . Again, the
curvature (s) (i.e. not constant) and geometrical torsion , height of the bar is the fraction of the string length over which
then we can write revised equilibrium equations for drag: the torsion acts. Again, we can see that Case 2 has much more
dFe d of the higher values.
+ EI + w bp t z f w c (1 ro cos ) = 0
ds ds Using equation 8, we can derive the error produced by
Fe + EI + w bp n z w c (cos + f ro sin ) = 0
2
neglecting the torsion and d/ds terms for the contact force
and for the axial loads. The relative error for the contact force
d
w bp b z EI w c (sin f ro cos ) = 0 terms is given by:
ds wc w
c
. . . . . .(8) e1 = . . . . . .(12)
wc
and for torque: c is the modified contact force
where w

dFe d c =
+ EI + w bp t z = 0 w
ds ds
Fe + w bp n z M t + 2 EI w c cos f w c sin = 0 (Fe + 2 M b + w bp n z ) 2 + ( w bp b z EI d / ds) 2

d 1 + 2 f2 ro2
w bp b z (2EI M t ) + w c ( f ro sin + f cos ) = 0 . . . . . .(13)
ds
. . . . . .(9) We compared the contact force error terms in Figure 4, using
the logarithm of the error in the same way as Figures 2 and 3.
Again, we can see that Case 2 has much more of the higher
The minimum torsion needed to rotate the binormal vector bi
values.
to the binormal vector bi+1 is given by:
v v
= cos 1 (bi b t +1 ) /(s i+1 s i ) . . . . . .(10) For the axial load terms, we define the error:
The minimum value of d/ds is given by: wa w
a
d i+1 i e2 = . . . . . . .(14)
= . . . . . .(11) wa
ds s i+1 s i where
We can now formulate two tests. We can use the approximate w a = w bp t z f w c
values of and d/ds to estimate the contact force wc and we . . . . . .(15)
can estimate the load terms on the first equation in (8) or (9). a = M b d / ds + w bp t z f w
w c
Comparison with the same terms evaluated at = 0, and d/ds
= 0, give us the minimum error introduced by the constant We compared the axial load error terms in Figure 5, using the
curvature assumption. The actual error will be greater because logarithm of the error in the same way as Figures 2, 3 and 4.
and d/ds may actually change more rapidly. Again, we can see that Case 2 has much more of the higher
values.
To probe these ideas, we have looked at two drag example
cases. The first case shows good correlation with field data, While this study cannot be definitive, it seems that two criteria
while the second case is known to produce drastically different for torque-drag models are provided by error terms e1 and e2.
results when a stiff string model is used. Further work will be needed to refine these results.

We have speculated that the constant curvature wellbore Conclusions and Observations
trajectory assumption has the potential to cause problems, so Surprisingly, the torque-drag model cannot be a soft string
we have compared the geometric torsion, as defined in model and still satisfy the equilibrium equations. Non-zero
equation 10, for the two cases in Figure 2. In this bar chart, the shear forces must exist, otherwise, friction loads must vanish.
logarithm (base 10) of the geometric torsion was sorted into Fortunately, no changes to the formulation are necessary, and
groups with width equal to , with magnitude increasing from the shear forces can be evaluated with a side calculation. An
left to right. Remember that we are using logarithmic values, additional factor must be added to the contact force for torque
4 SPE/IADC 105068

calculations, because the rotating string will tend to climb Section Drillstring, Trans ASME, Journal of Pressure
the wellbore wall, reducing the contact force. Vessel Tech., (May), pp. 367-373.

The real weakness of the torque-drag model is the use of Zwillinger, Daniel (editor). 1996. CRC Standard
constant curvature wellbore trajectories. These trajectories Mathematical Tables and Formulae, 30th Edition, Boca
imply that drillstring bending moment does not vary smoothly Raton, Florida: CRC Press, pp 321-322.
at survey points, which means that some contact forces and
axial loads are missing from the model. Rough estimates of Nomenclature
wellbore torsion and change of curvature with depth, when
applied to the example cases, seem to identify wellbores that Ap = cross-sectional area of the pipe (in2)
are not good candidates for torque-drag modeling, but further
r
b = binormal vector
study will be necessary.
bz = z coordinate of the binormal vector
I = moment of inertia (in4)
References
E = Youngs elastic modulus (psi)
Fa = actual axial force in the pipe (lbf)
Ho, H.-S. 1986. General Formulation of Drill-string Under
Fe = the effective force (lbf)
Large Deformation and Its Use in BHA Analysis, paper
Fst = pressure-area force terms, the stream thrust (lbf)
SPE 115562, presented at the 60th Annual Technical
Mt = axial torque (lbf-in)
Conference and Exhibition of the SPE, New Orleans, r
(Oct.). n = normal vector
nz = z coordinate of normal vector
v
Ho, H.-S. 1988. An Improved Modeling Program for r = position vector (in)
Computing the Torque and Drag in Directional and Deep R = radius of curvature (in)
Wells, paper SPE 18047 presented at the 63rd Annual rc = radial clearance (in) )
Technical Conference and Exhibition of the SPE, ri = pipe inside radius (in)
(October 2-5). rp = pipe outside radius (in)
s = measured depth (in)
r
Johancsik, C.A., Dawson, R. and Friesen, D.B.1973. Torque t = tangent vector
and Drag in Directional Wells - Prediction and tz = z coordinate of the tangent vector
Measurement, paper SPE 11380 presented at the wa = axial distributed load (lbf/in)
IADC/SPE Drilling conf., New Orleans. wbp = buoyant weight of the pipe (lbf/in)
wst = gradient of the stream thrust (lbf/in)
Lesso, W.G., Mullens, E. and Daudey, J. 1989. Developing a wc = contact load (lbf/in)
Platform Strategy and Predicting Torque Losses for wcs = Sheppards contact force (lbf/in)
Modeled Directional Wells in the Amaulijak Field of the wd = friction drag load (lbf/in)
Beaufort Sea, Canada, SPE 19550 presented at the 64th
Annual Technical Conference and Exhibition, San wef = excess annular fluid loads (lbf/in)
Antonio, (Oct 8-11). = wellbore curvature (in-1)
f = dynamic friction coefficient
Love, A.E.H. 1944. A Treatise on the Mathematical Theory of = angle between survey tangent vectors
Elasticity, 4th Edition, New York City: Dover.
= wellbore trajectory inclination angle
= pipe angular deflection
Nordgren, R. P. 1974, On Computation of the Motion of
Elastic Rods, Trans. ASME, Journal of Applied = wellbore trajectory azimuth angle
Mechanics, (September), pp. 777-780.
Appendix A: Curvilinear Coordinates
v
Sawaryn, S.J. and Thorogood, J.L.2003. A Compendium of If the position of the drillstring is given as r (s) , where s is the
v
Directional Calculations Based on the Minimum arc length of the curve, then the unit tangent t (s ) to the
Curvature Method, SPE 84246 presented at the SPE v
curve r (s) is given by:
Annual Technical Conference and Exhibition, Denver, v
Colorado, (October 5-8). v d r (s) v
t (s) = = r (s) . . . . . .(A-1)
ds
Sheppard, M.C., Wick, C. and Burgess, T.M. 1986. Designing
where we have used to indicate derivative with respect to s.
Well Paths to Reduce Drag and Torque, paper SPE
The derivative of the tangent vector is:
115463, presented at the 61st Annual Technical
Conference and Exhibition of the SPE, New Orleans, v r
(Oct). t (s) = (s) n (s) . . . . . .(A-2)

v
Walker, B.R. and Friedman, M.B. 1977. Three-Dimensional where (s) is the curvature, and n ( s ) is the unit normal to the
Force and Deflection Analysis of a Variable Cross-
SPE/IADC 105068 5

v
curve. The third coordinate is the binormal vector b ( s ) Sawaryn (2003). In this method, we connect two tangent
vectors with a circular arc, as illustrated in Figure B-2.
defined by:
v v v In this Figure we have a circular arc of radius R over angle ,
b(s) = t (s) n (s) . . . . . .(A-3) v
connecting the two tangent vectors t1 at measured depth s1,
v
where is the vector cross product. The triad and t 2 at measured depth s2. The arc length is R = s2-s1 =
v v v s. Notice that the angle is also the angle between the
{ t (s), n (s), b(s)} forms a moving coordinate system along v v
the drillstring trajectory. The last two derivatives we need to tangents t1 and t 2 . From this we can immediately determine
complete the definition of the coordinate system are: R:

v v v v v
n (s) = (s) t (s) + (s)b(s) R = s / = s / cos 1 ( t1 t 2 ) = 1 / . . . .(B-4)
v v . . . . . .(A-4)
b(s) = (s)n (s) The following equations define a circular arc:
v r v
where (s) is the torsion of the curve. The torsion of the curve r (s ) = t1 R sin[ ( s s 1 )] + n 1 R {1 cos[ ( s s 1 )]} + r1
is not to be mistaken for the mechanical torsion of the v r r
t (s ) = t1 cos[ ( s s 1 )] + n 1 sin[ ( s s 1 )]
drillstring. Instead, torsion is a measure of the helical nature of v v v
the curve. For instance, a constant pitch helix has constant n (s ) = t1 sin[ ( s s 1 )] + n 1 cos[ ( s s 1 )]
torsion, while a plane curve has zero torsion. r r r v
b (s ) = t1 n 1 = b 1
These equations are called the Serret-Frenet equations [see . . . . . .(B-5)
Zwillinger (1996)].
v v
The vector r1 is just the initial position at s = s1. The vector t1
Appendix B: Calculating the Wellbore Trajectory
v v
is the initial tangent vector. The vector n1 is the initial normal
The normal method for determining the well path r (s) is to
vector. If we evaluate equation (B-5)b at s = s2, we find:
use some type of surveying instrument to measure the
inclination and azimuth at various depths and then to calculate v v v v
the trajectory. t (s 2 ) = t1 cos s + n 1 sin s = t 2 . . . . . .(B-6)

At each station, inclination angle and azimuth angle are v

which we can solve for n1 :
measured, as well as the course length s between stations.
These angles have been corrected to true north, if a magnetic
survey, or for drift, if a gyroscopic survey. The survey angles v r
v v t t cos( s) v v
define the tangent t to the trajectory at each station, where n1 = 2 1 = t 2 csc( s) t1 cot( s)
the tangent vector is defined in terms of inclination and sin( s)
azimuth in the following formula: . . . . . .(B-7)
v v v
t = { cos() sin() , sin() sin() , cos() } (B-1) Notice that equation (B-7) fails if t1 = t 2 . For this case, we
use equation (B-3) for the straight wellbore.
The survey angles and the tangent vector are shown in Figure
B-1. If we knew how the angles and varied between The vertical components of equation (B-5) are used in this
stations, or equivalently, if we knew how the tangent vectors text:
varied between stations, then we could determine the
trajectory by integrating the tangent vector from equation A-1: t z ( s ) = t 1 z cos[ ( s s 1 )] + n 1 z sin[ ( s s 1 )]
s v
n z ( s ) = t 1 z sin[ ( s s 1 )] + n 1 z cos[ ( s s 1 )]
v v v v
r (s) = r0 + t (s)ds . . . . . .(B-2) t 1 z = t 1 iz
s0
v v
n 1 z = n 1 iz
The easiest case would be a constant tangent vector between r r v v v
stations, which integrates into a straight line: b z ( s ) = ( t 1 n 1 ) iz = b 1 i z
v v v t 2z + n 2z + b 2z = 1
r (s) = r0 + t (s s 0 ) . . . . . .(B-3) . . . . . .(B-8)

The method most commonly used to define a well trajectory is

called the Minimum Curvature method, as, for example in
6 SPE/IADC 105068

Appendix C: The Loads on a Drillstring direction to the motion, and has the magnitude of the product
v
The change in the drillstring force F due to applied load of the contact force and the dynamic coefficient for
v friction f . The Coulomb friction relationship is shown in
vector w is given by the following equation:
Figure C-2.
v
dF v v
+w=0 . . . . . .(C-1) If the drillstring is sliding, there will be a friction drag force
ds wd tangent to the wellbore, and pointing in the direction
opposite to the sliding, as shown in Figure C-3. The friction
v force may act in either direction, depending on whether the
where w is force per length of the drillstring. The change in
v v pipe is being run into or out of the hole.
moment M due to applied moment vector m and pipe force
v For sliding friction:
F is given by the following equation:
v v
v w d = f w c t . . . . . .(C-7)
dM r r v v
+t F +m =0 . . . . . .(C-2)
ds where the choice of depends on the direction of sliding. If
the string is sliding into the hole, the negative sign holds,
v while pulling out of the hole uses the positive sign. The
The total drillstring load vector w is:
applied couple per unit length associated with this drag force
v v v v r v is given by:
w = wbp + wst + wc + wd + wef . . . . . .(C-3)
v v v v
v m = f w c t ro ( cos n sin b)
We define the buoyant weight wbp of the pipe: v v . . . . . .(C-8)
= f ro w c (sin n cos b)
v v
wbp = [ w p + ( i Ai o Ao ) g ] iz . . . . . .(C-4)
When the drillstring is rotated, the friction force is no longer
oriented axially, but is now applied opposite the direction of
v r v
The next term, w st , is the gradient of the pressure-area forces. rotation, in the n - b plane, as shown in Figure C-4:
The pressure-area forces, when fluid momentum is added, are
known as the stream thrust terms, Fst , are given by: For clockwise pipe rotation:
v v v
v v wd = f wc (sin n cos b ) . . . . . .(C-9)
Fst = [( po + o vo2 ) Ao ( pi + i vi2 ) Ai ] t
v
v dFst . . . . . .(C-5) If mt is the applied torque per unit length in the tangential
wst = direction, then the applied torque can be inferred from Figure
ds v
v C-3 ( t points into the paper):
The term wef is due to complex flow patterns in the annulus.
For many cases of interest, this term is zero, particularly for m t = f w c ro . . . . . .(C-10)
static fluid and for narrow annuli without pipe rotation.
Because of the advanced nature of the computation of this
term, this term will be neglected for the remaining discussion. Appendix D: Mechanical Response of a Drillstring
We will model the drillstring as an elastic solid material. Since
v
The remaining terms are the mechanical force terms. If the a solid material can develop shear stresses, we formulate F in
drillstring contacts the wellbore, there is a contact force wc the following way:
perpendicular to the wellbore, as shown in Figure C-1.
v v v v
r v F = Fa t + Fn n + Fb b . . . . . .(D-1)
Note that wc lies in the n - b plane at angle with respect to
r
the n vector. There is no contact force in the tangent vector where Fa is the axial force, Fn is the shear force in the normal
direction, since the contact force is perpendicular to the direction, and Fb is the shear force in the binormal direction. If
wellbore: we consider equation (D-1) with the equilibrium equation (C-
v 1), we can group the stream thrust terms with the axial force to
v v
w c = w c (cos n + sin b) . . . . . .(C-6) define the effective force Fe :

A second force, friction, is associated with the contact force. Fe = Fa + Fst

The Coulomb friction model is particularly simple in concept. . . . (D-2)
If two surfaces in contact with normal force FN are sliding = Fa + (p o + o v o2 )A o (p i + i v i2 )A i
relative to each other, the friction force points in the opposite
SPE/IADC 105068 7

dFe
The casing moments for a circular pipe are given by: Fn + w bp t z f w c = 0
ds
v v r
M = EI b + M t t . . . . . .(D-3) d
Fn + Fe + w bp n z w c cos = 0
ds
where EI is the bending stiffness and M t is the axial torque. d . . . .. .(D-6)
Fb + w bp b z w c sin = 0
The first term in equation (D-3) may seem peculiar. The ds
situation is illustrated in Figure D-1: Mt = 0
For a round pipe, the bending moment points in the direction Fb + f ro w c sin = 0
v v
perpendicular to the t n plane, and is proportional to the Fn f ro w c cos = 0
curvature of the pipe. In the figure we can see that the pipe
v v
curves in the t n plane, so that the steel on the inside of the For the torque calculations, we will assume that the pipe is
curve is compressed, while the steel on the outside of the being rotated. If we adopt the friction equation for pipe
curve is extended. The forces associated with these rotation, the equilibrium equations now have the form:
displacements generate a moment that is proportional to the dFe
curvature (1/R) of the pipe and points perpendicular to the Fn + w bp t z = 0
plane. ds
dFn
The moment equilibrium equation (C-2) gives, using + Fe + w bp n z w c cos f w c sin = 0
equations (D-1)-(D-3): ds
d v v dM t r dFb
+ w bp b z w c sin + f w c cos = 0
EI( b n ) + t
ds ds . . . . . .(D-4) ds
v v v Fn = 0
+ (M t Fb )n + Fn b + m = 0
Fb = M t
where we have used the Frenet equations (Appendix A).
d
For a constant curvature wellbore, the first term in equation M t w c r0 = 0
(D-3) is constant, which means that the bending moment term ds
vanishes: . . . . . .(D-7)
Because bz is a constant for the minimum curvature trajectory,
dM t r v v v equations (D-7)c,e have the analytic solution:
t + (M t Fb )n + Fn b + m = 0 . . . . . .(D-5) s
ds Fb = Fb0 exp( ) + r0 f w bp b z . . . . . .(D-8)
r0 f
Because r0 f is typically small relative to measured depth s,
For a constant curvature trajectory, the bending moment
defined in equation (D-3) may be discontinuous at survey
locations. Further, the bending moment does not even appear the exponential term in equation (D-8) either grows very large
in the equilibrium equations (D-5). These considerations (running in) or very small (pulling out changes sign of f ).
suggest that the minimum curvature trajectory is not an exact 0
representation of a real drillstring configuration. The most reasonable choice for Fb is zero. This will result in
a discontinuity in Fb at survey points. We have already
For the drag calculations, we will assume that the pipe is being observed that the bending moment may also be discontinuous
run into the hole. If we assume linear Coulomb friction to at survey points, so this is another negative point for the
v
model the drag force on the pipe, then wd is given by minimum curvature trajectory.
equation (C-7). For this assumption, the drag force will be
pointing in the negative s direction. To modify these equations Appendix E: Equivalent Forms of the Contact Force
for pulling out of the hole, the only change needed will be to Sheppard (1987) expresses the contact force as:
change the sign of the friction coefficient f . The equations wc = (w bp sin Fe ) + (Fe sin )
2 2
. . . .(E-1)
modeling drag calculations are the following:
We wish to show that:
wc = (w bp n z + Fe ) + (w bp b z )
2 2
. . . . . . (E-2)

is an equivalent expression in terms of a minimum curvature

formulation. First, we need an expression for in terms of
8 SPE/IADC 105068

inclination and azimuth. The Frenet equations give: 0.300

d v r
t = n . . . . . .(E-3) 0.250 Case 1
ds
v Case 2

and since n is the unit normal vector: 0.200

d r d r

fraction od string length

r r
t t = 2n n = 2 . . . . . .(E-4)
ds ds 0.150

we can evaluate by differentiating equation (B-1) and

0.100
evaluating equation (E-4). We get:
2 = 2 + sin 2 2
. . . . . .(E-5) 0.050

If we expand (E-2), substituting (E-3) and (E-5), we get:

w c = Fe2 2 + 2Fe n z + w 2bp n 2z + w 2bp b 2z
0.000
log dK/ds

Figure 3: Comparison of Variable Curvature

= Fe2 (2 + sin 2 2 + 2Fe tz + w 2bp (1 t 2z )

. . . . . .(E-6) 0.180

From equation (B-1), the z component of the tangent vector

r 0.160

t is cos, so: 0.140

1 t z2 = 1 cos 2 = sin 2 0.120

fraction string length

. . . . . .(E-7)

t z = (cos ) = sin
0.100 Case 1
Case 2
0.080

Substitute (E-7) into (E-6) yields:

(w sin Fe ) + (Fe sin ) . . . .(E-8)
0.060

wc =
2 2
bp 0.040

0.020

0.000
log load
error
wellbore Figure 4: Contact Load Error

0.180

0.160

0.140
rotation
friction force
0.120
fraction of string length

Case 1
0.100
formation Case 2
contact force
0.080

gravity force
0.060

Figure 1: Rotating Equilibrium Position 0.040

0.020

0.350
0.000
log relative
error
0.300
Figure 5: Comparison of Axial Load Error
Case 1

0.250 Case 2
fraction of string length

0.200

0.150

0.100

0.050

0.000
log torsion

Figure 2: Comparison of Geometric Torsion

SPE/IADC 105068 9

iN

FN

Motion

iE
t

Force = fFN

Figure C-2: The Coulomb Friction Force

iZ

Figure B-1: Inclination and Azimuth

ro

t2 wd = mwc

m=mwcro
t1
C-3: Sliding Generates Loads and Distributed Moment

n
R = 1/
wd wc
R

rotation
t2
b
s
wellbore

t1 formation

Figure B-2: Circular Arc

Figure C-4: Friction Force Due to Rotation

wc

Mb


R
drillstring
b
n
wellbore

formation
t

Figure C-1: Contact Force Angle Figure D-1: Bending Moment Perpendicular to t-n Plane