Вы находитесь на странице: 1из 10

How Good Is the Torque/Drag Model?

Robert F. Mitchell and Robello Samuel, Halliburton

Summary equations, so more-general elastic relations must be used (Love


Perhaps the only “standard” drillstring model in use today is the 1944; Nordgren 1974). Friction complicates the problem further
torque/drag model originally developed by Johancsik et al. (1984) because the friction-force direction is not usually known. One of
and put in a standard form by Sheppard et al. (1987). Because of the most well-known formulations was developed by Walker and
the simplicity and general availability of this model, it has been Friedman (1977). A comprehensive paper on formulation, unfor-
used extensively for planning and in the field. Field experience tunately often overlooked, is the work of Ho (1986). He followed
indicates that the model usually works very well. This model is this paper with a simplification for soft-string models (Ho 1988).
thought to be an approximation of real drillstring behavior—in Ho used a curvilinear-coordinate system based on the Serret-
particular, the bending stiffness is neglected, so the torque/drag Frenet equations [see Zwillinger (1996)] for a curve in space.
model is often called a “soft-string” model. There have been The torque/drag-model formulation solves the wellbore-
many “stiff-string” models developed, but there is no “industry- contact problem by assuming that the drillstring trajectory is the
standard” formulation. same as the wellbore trajectory. Considering that survey data are
Why does this model work well in some circumstances, yet not measured inside the drillstring, this is an excellent assumption.
so well in other cases? This question is difficult to answer without Contact is further assumed to be continuous. This assumed trajec-
a comprehensive torque/drag model, so this study concentrated tory is certainly within inches of the actual drillstring trajectory.
on developing this model formulation. The standard torque/ Unfortunately, the most commonly used wellbore-trajectory
drag-model formulation was then reviewed in the context of model (minimum curvature—see Appendix B) implies that bend-
this analysis, and it was determined that the model satisfied all ing moments are not continuous at survey points. The solution to
of the force-equilibrium equations but only one of the moment- this problem is to assume that the drillstring is a single-force
equilibrium equations. Solving the remaining two equilibrium beam, with the force assumed to be tangent to the trajectory.
equations required that both torque and drag be zero. Fortunately, Because the bending stiffness is neglected, the torque/drag model
this condition is easy to resolve by assuming that the string has is often called a cable, chain, or soft-string model. Note that this
shear forces. While the standard torque/drag model is often termed assumption is not necessary if a different wellbore-trajectory
a soft-string model, in reality, it must have shear forces. model is used that is smooth enough.
Two field cases were studied. The first case was known to In this paper, we will take the first two basic assumptions and
match field data well; the second case was known to fail. Study derive all the consequences of mechanical equilibrium for the
of these two cases indicated that part of the problem may lie in the drillstring. The third assumption, the cable model, will be shown
description of the wellbore trajectory. This study provides approx- either to fail to satisfy equilibrium or to force unacceptable condi-
imation formulae to evaluate the effects of variable wellbore cur- tions on torque/drag modeling.
vature and wellbore torsion, and these formulae were able to show Note to the reader: There is not a consistent notation among
notable differences between the two sample problems. the pioneering papers on torque/drag modeling. The authors were
forced to choose a new set of notations that were complete enough
to describe the whole torque/drag problem. Notation for the geo-
Introduction metry of the torque/drag model is described in Appendices A and
Generally speaking, what do we mean by “torque-drag model- B, using common mathematical notation found in handbooks
ing?” Drag is the excess load compared to rotating drillstring (Zwillinger 1996). Notation for the mechanics of torque/drag uses
weight, which may be either positive when pulling the drillstring F for force and M for moment, as is the usual practice in mechan-
or negative while sliding into the well. (Note: there is no axial- ical engineering. See Appendices C and D for further details.
friction drag in a rotating drillstring, so rotating-drillstring weight
is the zero-drag reference point.) This drag force is attributed to Critique of the Standard Torque/Drag Model
friction generated by drillstring contact with the wellbore. When
To critique the standard torque/drag model, we need a more
rotating, this same friction will reduce the surface torque trans-
detailed model so that we can identify the simplifying assump-
mitted to the bit. It is useful to be able to estimate the friction
tions. In Appendix C, we identify all the loads associated with
forces when planning a well or performing post-drilling analysis.
Analysis of these drillstring loads is performed with drillstring friction, and in Appendix D, we derive the equilibrium equations
for pipe that has shear forces and axial forces, in effect a stiff-
computer models, and there have been many drillstring models
string version of the standard torque/drag model.
developed over the last 30 years. By far the most common method
for drillstring analysis is the torque/drag model originally devel-
oped by Johancsik et al. (1984) and put into differential-equation Torque Calculations. The equilibrium equations for torque cal-
form by Sheppard et al. (1987). Because of the simplicity and culations are given in Eqs. D-12 through D-17. We can immedi-
general availability of this model, it has been used extensively ately evaluate the magnitude of the shear forces from the moment
for planning and in the field. If any drillstring model could be equations:
called standard, this would be the one. Fn ¼ 0; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (1)
In the most basic terms, a drillstring model must determine the
trajectory of an elastic rod constrained by a wellbore. The elastic Fb ¼ kMt ; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (2)
rod will contact the wellbore in a number of places in either point
contact or continuous contact, and the points of contact must be where Fn is the shear force in the nY direction, Fb is the shear force
determined by the calculation process. Furthermore, the geometric in the bY direction, k is the wellbore curvature, and Mt is the axial
nonlinearities associated with the wellbore trajectory usually vio- torque. (The vectors and curvature are defined in Appendix A).
late the basic assumptions used in developing textbook elastic-rod Notice that Eq. 2 (moment equation) cannot be satisfied exactly
by the standard torque/drag model because there is no shear force
Fb. Even worse, the exact satisfaction of equilibrium requires that
Copyright ã 2009 Society of Petroleum Engineers
the torque be zero. Even so, we can see that for torque calcula-
This paper (SPE 105068) was accepted for presentation at the SPE/IADC Drilling Confer- tions, the assumption of a single-force beam is not too bad. First,
ence, Amsterdam, 20–22 February 2007, and revised for publication. Original manuscript
received for review 14 November 2006. Revised manuscript received for review 17 April
Fn is identically zero. Second, the curvature k is typically a small
2008. Paper peer approved 21 July 2008. number. The largest curvature, for a short-radius well, would be

62 March 2009 SPE Drilling & Completion


approximately .02 ft–1. We can readily evaluate the goodness of Note that the conventional torque/drag model would require Fb
the single-force assumption by using Eq. 2. and Fn to be zero; but worse, the exact satisfaction of equilibrium
The contact force wc is given by requires that the friction drag be zero. As shown in Appendix D,
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Fb is a constant, and
u   
u kFe þ wbp nz 2 þ wbp bz 2 mf krp wbp bz
wc ¼ t þ ; . . . . . . . . . (3) Fb ¼ mf rp wbp bz : . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (9)
1 þ m2f 1 þ m2f
The total magnitude of the shear force is
where mf is the Coulomb-friction factor, Fe is the effective axial qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
force in the pipe, wbp is the buoyant weight of the pipe, rp is the
Fs ¼ F2b þ F2n ¼ mf rp wc : . . . . . . . . . . . . . . . . . . . . . . . . . . (10)
tool-joint radius, nz is the vertical component of the normal vec-
tor, and bz is the vertical component of the binormal vector. These
If we neglect dFn/ds, then the contact force wc can be shown to be
terms are defined in Appendices A through F. Eq. 3 was derived
the conventional torque-drag contact force (Eq. 5).
assuming krp is small compared to one, which is always true for
real wellbore trajectories. The contact force derived by Sheppard
et al. (1987), wcs, in our notation, is given by How Good Is the Standard Torque/Drag Model? The industry
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2
consensus is that the standard torque-drag model is pretty good.
wcs ¼ wbp sin f  Fe f0 þðFe sin f#0 Þ : . . . . . . . . . . . . (4)
2 For example, Lesso et al. (1989) discusses an application of the
standard torque/drag model to the Amauligak field. While this is a
This can be shown to be equivalent to (see Appendix E) successful application of standard torque/drag models, note that
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi the friction coefficient that is needed to match drag results is very
 2  2 different from the friction coefficient needed to match torque
wcs ¼ kFe þ wbp nz þ wbp bz : . . . . . . . . . . . . . . . . . . . (5) results.
In many other cases, the friction coefficient that is needed to
The contact force, in terms of the torque/drag-model contact force model running into the hole differs from the coefficient for pulling
wcs, is given by out of the hole. To confuse the issue further, the coefficient may
wcs mf krp wbp bz change from trip to trip. It is clear that the model is useful, but
wc ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffi þ : . . . . . . . . . . . . . . . . . . . . . . . . . (6) caution in the use of the model is certainly justified.
1 þ mf 2 1 þ m2f
Is the standard model a soft-string model? The magnitude of
the shear force is readily calculated, using Eq. 1 for torque calcu-
First, we note thatqthe
ffiffiffiffiffiffiffiffiffiffiffiffiffi
torque/drag contact force should be mod- lations and Eq. 10 for drag calculations. These terms cannot actu-
ified by the term 1 þ m2f : Where does this term come from? As ally be zero because Eq. 1 implies that the torque would be zero in
shown in Fig. 1, the pipe is trying to climb the wellbore wall torque calculations and Eq. 10 implies that drag would be zero in
because of the friction force generated by the rotation. As a result, drag calculations. One could actually argue that torque/drag mod-
the contact force will be slightly less than that if the pipe is lying els have always had shear forces, just that nobody ever bothered
on the bottom of the wellbore. This term should be applied to any to calculate them. That is, the standard torque/drag model is actu-
torque-calculation model. The second term in Eq. 6 is almost ally a stiff-string model.
always unimportant. As already stated, the product krp will So where does the torque-drag model fail? We can fix up the
always be small compared to one. The vertical component of the contact force for torque calculations easily, using Eq. 3. But the
binormal vector is small except for wells turning on a horizontal real problem lies with the trajectory model. We can see in Appen-
plane. The combination of a short-radius turn on a horizontal dix D (Eq. D-4) that the constant-curvature assumption nullifies
plane rarely occurs, but such cases do exist (Grinde and Haugland the impact of the bending moment on the equilibrium equations.
2003). Even so, using worst-case values, the second term will This is because the equilibrium equations depend on the change in
rarely be more than 1% of the total. bending moment, and minimum curvature requires that the bend-
ing moment remain constant. Further, we see that the constant-
curvature model implies a discontinuity in the bending moment at
Drag Calculations. The shear forces for drag calculations can be survey points. In Appendix F, we have calculated this discontinu-
evaluated from Eqs. D-10 and D-11 (moment equations): ity, and we found, to the authors’ surprise, that it can be greater
than the bending moment itself. We have, in effect, neglected the
Fb ¼ mf rp wc sin y; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (7) contact loads that would have generated this nonnegligible change
in bending moment.
Fb ¼ mf rp wc cos y: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (8)
If we assume that the wellbore-trajectory model has a
variable curvature k(s) (i.e., not constant) and geometrical
torsion t (see Appendix A for more about k and t), then we can
write revised equilibrium equations for drag:

dFe dk
þ EIk þ wbp tz  mf wc ð1  kro cos yÞ ¼ 0; . . . . . . (11)
ds ds
Fe k þ t2 EIk þ wbp nz  wc ðcos y þ tmf ro sin yÞ ¼ 0; . . . . (12)

and

dk
wbp bz  tEI  wc ðsin y  tmf ro cos yÞ ¼ 0: . . . . . . . . . . (13)
ds
And for torque,

dFe dk
þ EIk þ wbp tz ¼ 0; . . . . . . . . . . . . . . . . . . . . . . . . . . (14)
ds ds
Fe kþwbp nz tkMt þt2 EIkwc cosymf wc siny ¼ 0; . . . (15)

Fig. 1—Rotation equilibrium position. and

March 2009 SPE Drilling & Completion 63


Fig. 2—Comparison of geometric torsion.

dk introduced by the constant-curvature assumption. The actual error will


wbp bz  ð2EIt  Mt Þ þ wc ðmf ro k  sin y þ mf cos yÞ ¼ 0: be greater because t and dk/ds may actually change more rapidly.
ds
To probe these ideas, we have looked at two drag example
: : : : : : : : : : : : : : : : : : : : : : : : : : : : : (16)
cases. Both cases are based on actual (proprietary) survey data.
The minimum torsion t needed to rotate the binormal vector bi to The first case showed good correlation with field data, while the
the binormal vector bi+1 is given by second case showed poor correlation.
We have speculated that the constant-curvature wellbore-
t ¼ cos1 ðbYi  bYtþ1 Þ=ðsiþ1  si Þ: . . . . . . . . . . . . . . . . . . . . . . (17) trajectory assumption has the potential to cause problems, so we
have compared the geometric torsion, as defined in Eq. 17, for the
The minimum value of dk/ds is given by
two cases in Fig. 2. In this bar chart, the logarithm (base 10) of the
dk kiþ1  ki geometric torsion was sorted into groups with width equal to ½, with
¼ : . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (18) magnitude increasing from left to right. Remember that we are using
ds siþ1  si
logarithmic values, so magnitudes increase by about three times as
We can now formulate two tests. We can use the approximate we move from bar to bar. The height of the bar is the fraction of the
values of t and dk/ds to estimate the contact force wc, and we can string length over which the torsion acts. We can see that Case 2
estimate the load terms in Eq. 11 or 14. Comparison with the same (dark bars) has significantly more of the higher torsion values
terms evaluated at t = 0, and dk/ds = 0, gives us the minimum error applied over a greater fraction of the string length.

Fig. 3—Comparison of variable curvature.

64 March 2009 SPE Drilling & Completion


Fig. 4—Contact-load error.

Next, we compared dk/ds, as defined in Eq. 18, for the two cases We compared the contact-force error terms in Fig. 4, using the
(Fig. 3). In this bar chart, the logarithm of dk/ds was sorted into logarithm of the error in the same way as Figs. 2 and 3. Again, we
groups with width equal to ½. Again, the height of the bar is the can see that Case 2 has significantly more of the higher values.
fraction of the string length over which dk/ds acts. Again, we can For the axial-load terms, we define the error as
see that Case 2 has significantly more of the higher values.
Using Eqs. 11 through 13, we can derive the error produced by ^a j
jw a  w
neglecting the torsion and dk/ds terms for the contact force and e2 ¼ ; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (21)
wa
for the axial loads. The relative error for the contact-force terms is
given by where

jwc  w^c j wa ¼ wbp tz  mf wc ; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (22)


e1 ¼ ; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (19)
wc
^a ¼ Mb dk=ds þ wbp tz  mf w
w ^c : . . . . . . . . . . . . . . . . . . . . . . (23)
^c is the modified contact force,
where w
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi We compared the axial-load error terms in Fig. 5, using the
ðFe k þ t2 mb þ wbp nz Þ2 þ ðwbp bz  EIt dk=dsÞ2 logarithm of the error in the same way as Figs. 2, 3, and 4. Again,
^c ¼
w qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi we can see that Case 2 has significanly more of the higher values.
1 þ t2 m2f ro2 While this study cannot be definitive, it seems that two criteria
for torque/drag models are provided by error terms e1 and e2.
: : : : : : : . . . . . . . . . . . . . . . . . . . . . . . . : (20) Further work will be needed to refine these results.

Fig. 5—Comparison of axial-load error.

March 2009 SPE Drilling & Completion 65


Conclusions and Observations rp = pipe outside radius (in.)
For the first time, a comprehensive analysis of torque/drag model- R = radius of curvature (in.)
ing is available. There is considerable value in having this analy- s = measured depth (in.)
sis. For example, there are several commonly held assumptions ~
t = tangent vector
about torque/drag modeling that are not valid. tz = z-coordinate of the tangent vector
For instance, many believe that the torque/drag model is a wa = axial distributed load (lbf/in.)
single-force model with no bending resistance. Surprisingly, the ^a
w = modified axial load (lbf/in.)
torque/drag model cannot be a soft-string model and still satisfy
wbp = buoyant weight of the pipe (lbf/in.)
the equilibrium equations. Nonzero shear forces must exist, other-
wise, friction loads must vanish. Fortunately, no changes to the wc = contact load (lbf/in.)
formulation are necessary, and the shear forces can be evaluated ^c
w = modified contact force (lbf/in.)
with a simple side calculation. An additional factor must be wcs = contact force of Sheppard et al. (1987) (lbf/in.)
added to the contact force for torque calculations because the wd = friction-drag load (lbf/in.)
rotating string will tend to climb the wellbore wall, reducing Dwef = excess annular-fluid loads (lbf/in.)
the contact force. wst = gradient of the stream thrust (lbf/in.)
The lack of bending resistence is also illusory because the # = wellbore-trajectory azimuth angle
wellbore trajectory commonly used, minimum curvature, reduces y = pipe angular deflection
the bending effects in the equilibrium equations to zero. What is k = wellbore curvature (in.–1)
worse, the bending moment is discontinuous at survey points. This
mf = dynamic coefficient of friction
discontinuity may be greater than the bending moment itself.
Ultimately, the real weakness of the torque/drag model is the t = geometric torsion of a curve (in.–1)
use of the minimum-curvature wellbore trajectory. These trajec- f = wellbore-trajectory inclination angle
tories imply that drillstring bending moment does not vary smooth- C = angle between survey tangent vectors
ly at survey points, which means that some contact forces and
axial loads are missing from the model. Missing loads and forces
References
imply that the torque and drag values may be underestimated.
What can be done to quantify these potential errors? Rough Grinde, J. and Haugland, T. 2003. Short Radius TTRD Well with Rig
estimates of wellbore torsion and change of curvature with depth Assisted Snubbing on the Veslefrikk Field. Paper SPE 85328 presented
have been determined, along with estimates of their impact on at the SPE/IADC Middle East Drilling Technology Conference and
contact load and axial force. When these criteria are applied to Exhibition, Abu Dhabi, UAE, 20–22 October. DOI: 10.2118/85328-MS.
the example cases, they seem to identify wellbores that are good Ho, H.-S. 1986. General Formulation of Drillstring Under Large Deforma-
candidates for classic torque/drag modeling and wellbores that tion and Its Use in BHA Analysis. Paper SPE 15562 presented at the
may be troublesome. Further study and better wellbore-trajectory SPE Annual Technical Conference and Exhibition, New Orleans, 5–8
models will likely be necessary. October. DOI: 10.2118/15562-MS.
When should the torque-drag model be used with care? The Ho, H.-S. 1988. An Improved Modeling Program for Computing the
first case would be actual survey data, as opposed to a planned Torque and Drag in Directional and Deep Wells. Paper SPE 18047
well trajectory. Planned well trajectories will not usually include presented at the SPE Annual Technical Conference and Exhibition,
variable radius of curvature and 3D torsional effects. Survey data Houston, 2–5 October. DOI: 10.2118/18047-MS.
almost always contain some of this because real wells are not 2D Johancsik, C.A., Friesen, D.B., and Dawson, R. 1984. Torque and
circular arcs. Short-radius wells are also candidates for poor clas- Drag in Directional Wells—Prediction and Measurement. JPT 36 (6):
sic-torque/drag-model performance. Finally, complex, 3D well 987–992. SPE-11380-PA. DOI: 10.2118/11380-PA.
plans are much more common today, and these plans have poten- Lesso, W.G., Mullens, E., and Daudey, J. 1989. Developing a Platform
tial to produce poor torque-drag predictions. Strategy and Predicting Torque Losses for Modeled Directional Wells
in the Amauligak Field of the Beaufort Sea, Canada. Paper SPE 19550
presented at the SPE Annual Technical Conference and Exhibition,
Nomenclature
San Antonio, Texas, USA, 8–11 October. DOI: 10.2118/19550-MS.
Ap = cross-sectional area of the pipe (in.2) Love, A.E.H. 1944. A Treatise on the Mathematical Theory of Elasticity,
b~ = binormal vector fourth edition. New York: Dover Publications.
bi = binormal vector at survey point i Nordgren, R.P. 1974. On Computation of the Motion of Elastic Rods.
bz = z-coordinate of the binormal vector Journal of Applied Mechanics 41 (September): 777–780.
e1 = relative error in the contact force Sawaryn, S.J. and Thorogood, J.L. 2005. A Compendium of Directional
e2 = relative error in the contact force Calculations Based on the Minimum Curvature Method. SPEDC 20
(1): 24–36. SPE-84246-PA. DOI: 10.2118/84246-PA.
E = Young’s elastic modulus (psi)
Sheppard, M.C., Wick, C., and Burgess, T.M. 1987. Designing Well Paths
Fa = actual axial force in the pipe (lbf)
to Reduce Drag and Torque. SPEDE 2 (4): 344–350. SPE-15463-PA.
Fb = shear force in the b~ direction DOI: 10.2118/15463-PA.
Fe = the effective force (lbf) Walker, B.R. and Friedman, M.B. 1977. Three-Dimensional Force and
Fn = shear force in the ~ n direction Deflection Analysis of a Variable Cross-Section Drillstring. Journal
Fs = total shear force (lbf) of Pressure Vessel Tech. (May 1977): 367–373.
Fst = pressure-area force terms, the stream thrust (lbf) Zwillinger, D. ed. 1996. CRC Standard Mathematical Tables and Formu-
I = moment of inertia (in.4) lae, 30th edition, 321–322. Boca Raton, Florida: CRC-Press.
mb = applied distributed bending couple (lbf-ft /ft)
mt = applied distributed torsional couple (lbf-ft /ft) Appendix A—Curvilinear Coordinates
Mb = bending moment Y
If the position of the drillstring is given as rðsÞ; where s is the arc
Mt = axial torque (lbf-in.) length of the curve, then the unit tangent tYðsÞ to the curve rðsÞ
Y is
~
n = normal vector given by
nz = z-coordinate of normal vector
Y
drðsÞ
rY = position vector (in.) Y ¼
tðsÞ ¼ rY0ðsÞ; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A-1)
rc = radial clearance (in.) ds
ri = pipe inside radius (in.) where we have used 0 to indicate derivative with respect to s. The
ro = outside radius of the pipe (in.) derivative of the tangent vector is

66 March 2009 SPE Drilling & Completion


tY0 ðsÞ ¼ kðsÞ ~
nðsÞ; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A-2) equivalently, if we knew how the tangent vectors varied between
stations, then we could determine the trajectory by integrating the
where k(s) is the curvature and nðsÞ
Y is the unit vector normal to the tangent vector from Eq. A-1:
Z s
curve. The third coordinate is the binormal vector bðsÞ defined by
Y ¼ rY0 þ
rðsÞ tYðsÞds: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (B-2)
Y ¼ tYðsÞ  nðsÞ;
bðsÞ Y . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A-3) s0

The easiest case would be a constant tangent vector between


where  is the vector cross product. The triad ftðsÞ; Y nðsÞ; Y Y
bðsÞg stations, which integrates into a straight line:
forms a moving coordinate system along the drillstring trajectory.
The last two derivatives that are needed to complete the definition Y ¼ rY0 þ tYðs  s0 Þ: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (B-3)
rðsÞ
of the coordinate system are
The method most commonly used to define a well trajectory is
nY 0ðsÞ ¼ kðsÞ tYðsÞ þ tðsÞbðsÞ
Y called the minimum-curvature method, as, for example, in
; . . . . . . . . . . . . . . . . . . . . . (A-4)
bY 0ðsÞ ¼ tðsÞnðsÞ
Y Sawaryn and Thorogood (2005). In this method, we connect two
tangent vectors with a circular arc, as illustrated in Fig. B-2.
where tðsÞ is the torsion of the curve. The torsion of the curve is In Fig. B-2 we have a circular arc of radius R over angle c,
not to be mistaken for the mechanical torsion of the drillstring. connecting the two tangent vectors tY1 at measured depth s1 and tY2
Instead, torsion is a measure of the helical nature of the curve. at measured depth s2. The arc length is Rc = s2–s1 = Ds. Notice
For instance, a constant-pitch helix has constant torsion, while a that the angle c is also the angle between the tangents tY1 and tY2 :
plane curve has zero torsion. From this, we can determine R immediately:
These equations are called the Serret-Frenet equations [see
Zwillinger (1996)]. R ¼ Ds=c ¼ Ds= cos1 ðtY1  tY2 Þ ¼ 1=k . . . . . . . . . . . . . . . . (B-4)
The following equations define a circular arc:
Appendix B—Calculating the Wellbore Trajectory
The normal method for determining the well path rðsÞ Y is to rY ðsÞ ¼ ~ n1 Rf1  cos½kðs  s1 Þg þ rY1 ;
t1 R sin½kðs  s1 Þ þ ~
use some type of surveying instrument to measure the inclination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (B-5)
and azimuth at various depths and then to calculate the trajectory.
At each station, inclination angle f and azimuth angle # are tYðsÞ ¼ ~ n1 sin½kðs  s1 Þ ; . . . . . . . . . . . (B-6)
t1 cos½kðs  s1 Þ þ ~
measured, along with the course length Ds between stations.
Fig. B-1 shows the relationship of these angles to the coordinate Y ¼ tY1 sin½kðs  s1 Þ þ nY1 cos½kðs  s1 Þ ; . . . . . . . . . (B-7)
nðsÞ
directions iN pointing north, iE pointing east, and iZ pointing
downward. These angles have been corrected to true north, if a and
magnetic survey, or for drift, if a gyroscopic survey. The survey
angles define the tangent tY to the trajectory at each station, where ~ ¼~
bðsÞ n1 ¼ bY1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (B-8)
t1  ~
the tangent vector is defined in terms of inclination f and azimuth
# in the following formula: The vector rY1 is simply the initial position at s = s1. The vector
tY1 is the initial tangent vector. The vector nY1 is the initial normal
tY ¼ f cosð#Þ sinðfÞ; sinð#Þ sinðfÞ; cosðfÞ g: . . . . . . . . . . (B-1) vector. If we evaluate Eq. B-6 at s = s2, we find:

The survey angles and the tangent vector are shown in Fig. B-1. If Y 2 Þ ¼ tY1 cos kDs þ nY1 sin kDs ¼ tY2 ; . . . . . . . . . . . . . . . . . (B-9)
tðs
we knew how the angles f and # varied between stations, or
which we can solve for nY1 by

Fig. B-1—Inclination  and azimuth #. Fig. B-2—Circular arc.

March 2009 SPE Drilling & Completion 67


tY2  ~
t1 cosðkDsÞ FY st ¼ ½ðpo þ ro v2o ÞAo  ðpi þ ri v2i ÞAi  tY
nY1 ¼ ¼ tY2 cscðkDsÞ  tY1 cotðkDsÞ: . . . (B-10)
sinðkDsÞ dFY : . . . . . . . . . . . . . (C-5)
wY st ¼ st
Notice that Eq. B-10 fails if tY1 = tY2 : For this case, we use Eq. B-3 ds
for the straight wellbore. The term Dw Y ef is because of complex flow patterns in the
The vertical components of Eqs. B-5 through B-8 are used in annulus. For many cases of interest, this term is zero, particularly
this text: for static fluid and for narrow annuli without pipe rotation.
tz ðsÞ ¼ t1z cos½kðs  s1 Þ þ n1z sin½kðs  s1 Þ Because of the advanced nature of the computation of this term,
this term will be neglected for the remaining discussion.
nz ðsÞ ¼ t1z sin½kðs  s1 Þ þ n1z cos½kðs  s1 Þ
The remaining terms are the mechanical-force terms. If the
t1z ¼ tY1  iYz drillstring contacts the wellbore, there is a contact force wc per-
: . . . . . (B-11)
n1z ¼ nY1  iYz pendicular to the wellbore, as shown in Fig. C-1.
Note that wc lies in the ~ b plane at angle y with respect to the
n–Y
bz ðsÞ ¼ ð~t1  ~ n1 Þ  iYz ¼ bY1  iYz
~
n vector. There is no contact force in the tangent-vector direction
t2z þ n2z þ b2z ¼ 1 because the contact force is perpendicular to the wellbore:
Y c ¼ wc ðcos y nY þ sin y bY Þ: . . . . . . . . . . . . . . . . . . . . . . . . . (C-6)
w
Appendix C—The Loads on a Drillstring A second force, friction, is associated with the contact force.
The change in the drillstring force FY because of applied-load The Coulomb-friction model is particularly simple in concept.
Y is given by the following:
vector w If two surfaces in contact with normal force FN are sliding relative
to each other, the friction force points in the direction opposite to
d FY
Y ¼ 0; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (C-1)
þw that of the motion and has the magnitude of the product of the
ds contact force and the dynamic coefficient for friction mf : The
where w Y is force per length of the drillstring. The change in Coulomb-friction relationship is shown in Fig. C-2.
If the drillstring is sliding, there will be a friction-drag force wd
moment MY because of applied moment vector m Y and pipe force FY
tangent to the wellbore and pointing in the direction opposite to
is given by the following equation:
that of the sliding, as shown in Fig. C-3. The friction force may
Y
dM act in either direction, depending on whether the pipe is being run
þ tY FY þ m
Y ¼ 0Y . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (C-2) into or out of the hole.
ds
For sliding friction,
Y is
The total drillstring-load vector w
Y d ¼ mf wct;Y . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (C-7)
w
Y ¼w
w Y st þ w
Y bp þ w wef : . . . . . . . . . . . . . . . . . . . (C-3)
~d þ DY
Yc þ w
where the choice of  depends on the direction of sliding. If
Y bp as
We define the buoyant weight of the pipe w the string is sliding into the hole, the negative sign holds,
while pulling out of the hole uses the positive sign. The applied
Y bp ¼ ½wp þ ðri Ai  ro Ao Þg iYz : . . . . . . . . . . . . . . . . . . . . . (C-4)
w bending couple per unit length associated with this drag force is
given by
Y st ; is the gradient of the pressure-area forces.
The next term, w Y b ¼ mf wctY ro ð cos y nY  sin y bYÞ
m
The pressure-area forces, when fluid momentum is added, are : . . . . . . . . . . . . . (C-8)
known as the stream-thrust terms, Fst ; and are given by ¼ mf ro wc ðsin y nY  cos y bYÞ

When the drillstring is rotated, the friction force is no longer


oriented axially, but it is now applied opposite the direction of
rotation, in the ~
n–Y
b plane, as shown in Fig. C-4.
For clockwise pipe rotation

Y d ¼ mf wc ðsin y nY  cos y bY Þ: . . . . . . . . . . . . . . . . . . . . . . . (C-9)


w

If mt is the applied torque per unit length in the tangential direc-


tion, then the applied torque can be inferred from Fig. C-4
(~
t points into the paper):
mt ¼ - mf wc ro : . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (C-10)

Fig. C-1—Contact-force angle. Fig. C-2—The Coulomb-friction force.

68 March 2009 SPE Drilling & Completion


Fig. C-3—Sliding generates loads and distributed moment.

Appendix D—Mechanical Response of a


Drillstring
We will model the drillstring as an elastic-solid material. Because
a solid material can develop shear stresses, we formulate F ~ in the
following way:

FY ¼ FatYþ FnnY þ Fbb;


Y . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (D-1)

where Fa is the axial force, Fn is the shear force in the normal


direction, and Fb is the shear force in the binormal direction. If we
consider Eq. D-1 with the equilibrium equation Eq. C-1, we can
group the stream-thrust terms with the axial force to define the Fig. C-4—Friction force caused by rotation.
effective force Fe :
Fe ¼ Fa þ Fst Y d is given by Eq. C-7. For this
the drag force on the pipe, then w
: . . . . . . . . . . . (D-2) assumption, the drag force will be pointing in the negative s
¼ Fa þ ðpo þ ro v2o ÞAo  ðpi þ ri v2i ÞAi
direction. To modify these equations for pulling out of the hole,
The moment for a circular pipe is given by the only change needed will be changing the sign of the friction
coefficient mf : The equations modeling drag calculations are the
MY ¼ EIk bY þ Mt~
t; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (D-3) following:
dFe
where EI is the bending stiffness and Mt is the axial torque.  kFn þ wbp tz  mf wc ¼ 0; . . . . . . . . . . . . . . . . . . . . (D-6)
ds
The first term in Eq. D-3 may seem peculiar. The situation is
d
illustrated in Fig. D-1. Fn þ Fe k þ wbp nz  wc cos y ¼ 0; . . . . . . . . . . . . . . . . (D-7)
For a round pipe, the bending moment points in the direction ds
perpendicular to the tY nY plane and is proportional to the d
curvature of the pipe. In Fig. D-1, we can see that the pipe curves Fb þ wbp bz  wc sin y ¼ 0; . . . . . . . . . . . . . . . . . . . . . . (D-8)
ds
in the tY nY plane, so that the steel on the inside of the curve is
compressed, while the steel on the outside of the curve is exten- Mt ¼ 0; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (D-9)
ded. The forces associated with these displacements generate a
 Fb þ mf ro wc sin y ¼ 0; . . . . . . . . . . . . . . . . . . . . . . . . . (D-10)
moment that is proportional to the curvature (1/R) of the pipe and
points perpendicular to the plane. and
Eq. C-2 (moment-equilibrium equation) gives, using Eqs. D-1
and D-3 Fn  mf ro wc cos y ¼ 0: . . . . . . . . . . . . . . . . . . . . . . . . . . . (D-11)
 
dk dMt
EI bY  kt nY þ ~
t For the torque calculations, we will assume that the pipe is being
ds ds ; . . . . . . . . . . . . . . . . . (D-4) rotated. If we adopt the friction equation for pipe rotation, the
þ ðMt k  Fb ÞnY þ FnbY þ m
Y ¼0 equilibrium equations now have the form
dFe
where we have used the Serret-Frenet equations (Appendix A).  kFn þ wbp tz ¼ 0; . . . . . . . . . . . . . . . . . . . . . . . . . . (D-12)
Notice that for a constant-curvature trajectory, the bending ds
moment given in Eq. D-3, EIkY b; is also constant. Because the
equilibrium equations depend on the change in bending moment,
this means that the bending-moment term vanishes:
dMt
~
t þ ðMt k  Fb ÞnY þ FnbY þ m
Y ¼ 0: . . . . . . . . . . . . . . . . (D-5)
ds
For a constant-curvature trajectory, the bending moment
defined in Eq. D-3 may be discontinuous at survey locations, and
this discontinuity is not negligible (see Appendix F). Further, the
bending moment does not even appear in Eq. D-5 (equilibrium
equation). These considerations suggest that the minimum-
curvature trajectory is not an exact representation of a real drill-
string configuration.
For the drag calculations, we will assume that the pipe is being
run into the hole. If we assume linear Coulomb friction to model Fig. D-1—Bending moment perpendicular to ~
t–~
n plane.

March 2009 SPE Drilling & Completion 69


dFn Appendix F—Bending-Stress Discontinuity in
þ Fe k þ wbp nz  wc cos y  mf wc sin y ¼ 0; . . . . . (D-13)
ds Conventional Torque/Drag Analysis
dFb In this paper, it was stated that bending stresses were discontin-
þ wbp bz  wc sin y þ mf wc cos y ¼ 0; . . . . . . . . . . . (D-14) uous at survey points. Clearly, in the case of a build section adja-
ds
cent to a hold section, the curvature changes from a fixed build rate
Fn ¼ 0; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (D-15) to a zero build rate. But what about a smoother curve that is
reasonably well modeled by the minimum-curvature method?
Fb ¼ kMt ; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (D-16) Appendix F is a sample calculation that shows the magnitude
of the bending-stress discontinuity for a constant-pitch helix
and approximated by two minimum-curvature segments.
d The tangent vector for a constant-pitch helix is
Mt  mwc ro ¼ 0: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (D-17)
ds
e1 þ sinðbsÞeY2 þ aeY3 ; . . . . . . . . . . . . . . . . . (F-1)
tY ¼ rb½cosðbsÞY
Because bz is a constant for the minimum-curvature trajectory,
Eqs. D-14 and D-16 have the analytic solution where r is the radius of the helix in the eY1 –eY2 plane, b is the
  rotation rate in this plane, and a is the displacement in the eY2 axial
s
Fb ¼ Fb exp
0
þ ro mf wbp bz : . . . . . . . . . . . . . . . . . (D-18) direction. That tY is a unit vector requires that
ro m f

Because r0 mf is typically small relative to measured depth s, a2 ¼ 1  r 2 b2 : . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (F-2)


the exponential term in Eq. D-18 either grows very large
(running in) or very small (pulling out changes sign of mf ). The derivative of the tangent vector gives
The most reasonable choice for F0b is zero. This will result in a
discontinuity in Fb at survey points. We have already observed tY ¼ rb2 ½ sinðbsÞeY1 þ cosðbsÞeY2  ¼ kY
n
: . . . . . . . . . . . . . (F-3)
that the bending moment may also be discontinuous at survey k ¼ rb2
points, so this is another negative point for the minimum-
curvature trajectory. The binormal vector associated with the helix is

Appendix E—Equivalent Forms of the Contact b ¼ tY nY


: . . . . . . . . . . . . . (F-4)
Force ¼ a½cosðbsÞeY1 þ sinðbsÞeY2  þ rbY
e3
Sheppard et al. (1987) expresses the contact force as The definition of bending moment gives
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2
wbp sin f  Fe f0 þðFe sin f#0 Þ : . . . . . . . . . . . (E-1) MY b ¼ EIrb2b:
Y . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (F-5)
2
wc ¼

We wish to show that Let us define three tangent vectors from the helix definition,
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2  2 Eq. F-1, at increments of bDs ¼ 1=2p:
wc ¼ wbp nz þ Fe k þ wbp bz . . . . . . . . . . . . . . . . . . . (E-2)
e1 þ aeY3
tY1 ¼ rbY
is an equivalent expression. (Note that this is true for any curve. e2 þ aeY3 : . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (F-5)
tY2 ¼ rbY
Also note that this contact force is not sufficiently general for e1 þ aeY3
tY3 ¼ rbY
torque/drag with variable curvature). First, we need an expression
for k in terms of inclination f and azimuth #. The Serret-Frenet Recall and apply Eq. B-4:
equations give
d k1 ¼ cos1 ðY
t1  tY2 Þ=Ds
tY ¼ k~
n; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (E-3)
ds k2 ¼ cos1 ðY
t2  tY3 Þ=Ds
and because nY is the unit normal vector, tY1  tY ¼ tY2  tY3 ¼ 1  r2 b2 : . . . . . . . . . . . . . . . . . . . (F-6)
pffiffiffi
d d 2 2 2
~
t ~t ¼ k2 ~ n ¼ k2 ; . . . . . . . . . . . . . . . . . . . . . . . . . . . (E-4)
n~ k1 ¼ k2 ffi rb ffi 0:90rb2
ds ds p
we can evaluate k by differentiating Eq. B-1 and evaluating The curvature for minimum curvature is approximately 90% of
Eq. E-4. We get the actual helix curvature for this case. The binormal vectors
associated with the two segments can be determined to be perpen-
k2 ¼ f02 þ sin2 f #02 : . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (E-5) dicular to the tY1  tY2 plane :
pffiffi
If we expand Eq. E-2, substituting Eqs. E-3 and E-5 into it, we bY1 ¼ 2=2½aeY1  aeY2 þ rbY e3  . . . . . . . . . . . . . . . . . . . . . . . (F-7)
obtain
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi and perpendicular to the tY2  tY3 plane :
wc ¼ F2e k2 þ 2Fe knz þ w2bp n2z þ w2bp b2z pffiffi
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi : . . . . (E-6) bY2 ¼ 2=2½aY
e1  aeY2 þ rbY e3 : . . . . . . . . . . . . . . . . . . . . . . . . (F-8)
¼ F2e ðf02 þ sin2 f#02 þ 2Fe t0z þ w2bp ð1  t2z Þ
The change in bending moment from Segment 1 to Segment 2 is
From Eq. B-1, the z-component of the tangent vector ~ t is cos f, so D MY ¼ EIk½bY2  bY1 
1  t2z ¼ 1  cos2 f ¼ sin2 f pffiffiffi
: . . . . . . . . . . . . . . . . . . . . . (E-7) ¼ EIk 2aeY1 : . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (F-9)
t0z ¼ ðcos fÞ0 ¼ f0 sin f
ffi EI p4 rb2eY1
Substituting Eq. E-7 into Eq. E-6 yields
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi The discontinuity in bending moment created by modeling the
 2 helix with minimum curvature is 27% greater than the helix bend-
wbp sin f  Fe f0 þðFe sin f#0 Þ : . . . . . . . . . . . (E-8)
2
wc ¼ ing moment itself.

70 March 2009 SPE Drilling & Completion


Production Research Company from 1973 to 1980. Mitchell
SI Metric Conversion Factors holds BA, MME, and PhD degrees from Rice University and is a
ft  3.048* E-01 = m registered professional engineer in Texas. Robello Samuel has
been a principal technical advisor in the drilling division of
in.  2.54* E+00 = cm Halliburton. He has over 20 years of experience in domestic
in.2  6.451 6* E+00 = cm2 and international oil/gas drilling and completion operations,
lbf  4.448 222 E+00 = N research, and teaching. He is also an adjunct professor at
psi  6.894 757 E+00 = kPa the University of Houston teaching courses in advanced-
*Conversion factor is exact. drilling engineering, downhole-drilling tools and technologies,
and complex-well architecture. Samuel has published
more than 65 technical papers and is presently serving as
a review chairman on the SPE Drilling & Completion Journal
Robert F. Mitchell is a Halliburton Fellow in the Halliburton Dril- and Journal of Petroleum Science and Engineering editorial-
ling and Evaluation Division. He has published over 70 papers review committees. He holds BS and MS degrees in mechan-
on drilling and well-completion problems and was the drilling ical engineering from Madurai College of Engineering, Guindy
editor for the SPE Petroleum Engineers Handbook. Currently, and MS and PhD degrees in petroleum engineering from
he is the executive editor of the SPE Drilling & Completion the University of Tulsa. He is also the author of Downhole
Journal. He won the SPE Drilling and Completions Award for Drilling Tools—Theory and Practice and coauthor of Drilling
2005, and he was vice president of Enertech Engineering and Engineering and Advanced Drilling Engineering—Principles
Research Company from 1980 to 1996 and worked at Exxon and Designs books.

March 2009 SPE Drilling & Completion 71

Вам также может понравиться