Download as pdf or txt
Download as pdf or txt
You are on page 1of 29

SPE/IADC-184611-MS

Improving Torque and Drag Prediction Using the Advanced Spline Curves
Borehole Trajectory

Mahmoud F. Abughaban and Alfred W. Eustes, Colorado School of Mines; John P. de Wardt, DE WARDT AND
COMPANY; Marc Willerth, Scientific Drilling International

Copyright 2017, SPE/IADC Drilling Conference and Exhibition

This paper was prepared for presentation at the SPE/IADC Drilling Conference and Exhibition held in The Hague, The Netherlands, 14–16 March 2017.

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 paper have not been reviewed by the Society of Petroleum Engineers or the International Association of Drilling Contractors and are subject to correction
by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers or the International Association of Drilling Contractors,
its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Society of Petroleum Engineers or
the International Association of Drilling Contractors is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations
may not be copied. The abstract must contain conspicuous acknowledgment of SPE/IADC copyright.

Abstract
Torque and drag [T&D] are critical elements in drilling deep vertical wells with unnoted tortuosity along the
borehole and in extended-reach and deviated wells with high and repeated tortuosity. Current T&D models
generate unreliable predictions due to the assumption that borehole trajectories are composed of constant
curvature arcs between survey stations [the minimum curvature calculation method]. This assumption
causes the bending parameter in the T&D equilibrium balance equations to be nil. Today's T&D models are
either based on a continuous drillstring to wellbore contact [the soft-string model] or intermittent contact
due to drillstring stiffness [the stiff-string model]. In both cases, the wellbore trajectory is based on the
minimum curvature method.
The T&D model proposed is a phase II initiative of the non-constant curvature trajectory model:
the advanced spline curves [ASC] borehole trajectory model published at the 2016 IADC/SPE Drilling
Conference and Exhibition (Abughaban el al. 2016). This modified three-dimensional [3D] T&D model
[the ASC 3D T&D model] is a stiff-string model that includes geometric torsion, wellbore curvature, change
in the rate of wellbore curvature and drillstring bending stiffness in the T&D equilibrium balance equations.
The model has been validated using field cases with real-time forces that define T&D measured at the
surface. The calculated outputs from these wells provide more accurate view of the drilling conditions
downhole, including the downhole weight on bit and torque on bottom.
The novelty of the proposed model is the ability to estimate a more realistic bending effects, accurately
predict the contact forces between the drillstring and the wellbore, and solve T&D parameters from
surface to total depth in reasonable time using standard engineering computer. Accurately estimating
these parameters will allow drilling engineers to update the driller with surface weight on bit and torque
parameters to improve the drilling performance without taking undue risks with the drilling system such
as stuck pipe, casing and drillpipe wear and drillstring fatigue. Thus, the ASC 3D T&D model is not only
an alternative approach to accurately model downhole T&D parameters to be used in real-time operation
centers [RTOC]; it can also serve as a step toward drilling automation.
2 SPE/IADC-184611-MS

Introduction
In the scope of this study, a non-constant curvature trajectory model [the ASC model] is used to predict
T&D outputs. Unlike the current method of borehole trajectory calculation, the minimum curvature method
[MCM], this model does not make the potentially unrealistic assumption of a constant curvature arc
between survey measurements. This provides an advantage for trajectory calculation and accurate drillstring
modeling.
Today's T&D measurements while drilling are either from the soft-string model or the stiff-string model,
which is explained briefly in the background section. In both models, the wellbore trajectory is calculated
using the MCM, assuming a constant curvature arc between survey stations. With this assumption, the
geometric torsion [τ] and the change in the rate of wellbore curvature [dκ/ds] are nil (Mitchell and Miska
2011). This generates an unreliable estimation of T&D magnitudes, since the true wellbore curvature [κ]
does change. There has been a claim that the change in curvature could be reasonably small, hence, it
can be assumed constant. However, it has been proven that the effect of the bending moment in the T&D
equilibrium balance equation cannot be effectively analyzed based on this assumption.

Extended Reach and Deep Vertical Wells


Torque level and axial drag in extended-reach [ER] wells are generally more dependent on the build or drop
sections with high and repeated tortuosity. In these conditions, the drillstring is in more contact with the wall;
the longer the deviated section, the more accurate estimations of the downhole T&D conditions are required
to optimize the delivery of the specific energy and the power transmitted from the surface. In vertical holes,
estimated T&D are usually within tolerable levels, however, the deeper a vertical well is drilled, T&D values
will increase. This is caused by the unnoted tortuosity along the borehole. Such phenomenon will affect the
downhole weight on bit [DWOB] and torque on bottom [TOB] that are determined from surface sensors.
These downhole conditions are critical to determine if the well can be drilled within equipment capabilities.
The buckling of drillpipe results from excessive bending loads is another issue, particularly that may
impact ER wells. This phenomenon causes an increase in the contact force between the drillstring and the
wellbore. As the drillstring weight is concentrated at the buckled portion of the drillstring, a lock-up will
occur due to axial friction build up in the extended build section with high angle (Wu 1996). This can lead
to a helical buckling to develop, eventually preventing the pipe from moving in the hole. To avoid helical
buckling along with lost weight on bit and the potential for lock-up, the additional drag created in the post-
buckled portion must be accurately calculated in the T&D models and compensated for when choosing
drilling parameters.

Novelty of the ASC 3D Torque and Drag Model


The underestimation of T&D during the drilling process has been investigated through theoretical modeling.
In such cases, it was suggested that improved T&D models would require a robust, more advanced
3D trajectory model (Stockhausen et al. 2003). Accurately determining T&D parameters enables better
assessment of the drilling feasibility of ER and deep vertical wells. It will also serve as an evaluation
process for designing and running downhole completion and production equipment, as well as minimizing
the occurrence of catastrophic drill string failure such as stuck pipe, casing and drillpipe wear and drillstring
fatigue.
This paper describes a novel methodology to predict T&D for a non-constant curvature trajectory in
which the first and second derivatives of the curve exist [the ASC borehole trajectory model]. The physics
behind the modified 3D T&D model is obtained by equilibrium equations of forces and moments using a
stiff-string model. The model includes τ, κ, dκ/ds and the drillstring bending stiffness [EI] to predict more
accurate drillstring contact forces and a more realistic bending stiffness effect.
To validate the T&D model for a non-constant curvature trajectory model [the ASC 3D T&D model],
various field cases with real-time forces that define T&D measured at the surface are used. The calculated
SPE/IADC-184611-MS 3

outputs from these wells shows improved accuracy compared to current models that are based on MCM.
Using the ASC 3D T&D model while drilling will provide engineers and drillers with more realistic
estimations of the downhole conditions along the wellbore and better correlate the information collected
from surface measurement of hookload, surface torque and weight on bit [WOB] to what is happening
downhole. Hence, an improved computational model of T&D while drilling will serve as a step toward
drilling automation development.

Background
T&D software has existed since the 1990s; however, some confusion still exists over the validity of these
models that are used to characterize drilling operations, particularly in the applications for ER wells. The
mathematical models used to evaluate the drillstring behavior inside the wellbore, are either [1] soft-string
model, i.e. the bending stiffness of the drillstring is neglected and a continuous drillstring to wellbore contact
is assumed, or [2] stiff-string model, i.e. the bending moment is assumed to be constant along the wellbore
curved section with an intermittent drillstring to wellbore contact.
Virtually, all T&D models assume that the drillstring is made up of short elements joined by connections
that transmit tension, compression, and torsion, but not bending moment. The calculations start at the bottom
of the string and proceed upward to the surface. Each short element contributes increments of torque,
axial drag, and weight. In the computational process, forces and torque values are summed to produce the
cumulative loads on the drillstring. Initial conditions usually are specified starting from the bottom, and for
drilling operations the inputs would be WOB, hookload and torque measured at the surface for different
drilling operations such as rotating off bottom [ROB], rotary drilling [RD], pickup [PU] and slackoff [SO].
The rectangular coordinate system [x, y, z] with the conventional unit vectors i, j, k and the Frenet-Serret
local system of coordinates with its unit tangent, normal and binormal vectors [ ] are useful tools to
derive the force and moment equilibrium equations (Mitchell and Miska 2011). Figure 1 illustrates these
coordinate systems for a given point on a curve T.

Figure 1—Local [ ] and global [x, y, z] coordinate system in 3D

Soft-String Model
The soft-string model, often called a cable or a chain, neglects the drillstring-stiffness effects. This means
that the drillstring is treated as a heavy cable, chain, or rope lying along the wellbore. The axial tension
and torque forces are supported by the drillstring and the contact forces are supported by the wellbore. It
also assumes that the drillstring has continuous surface contact area with the low side of the hole and it
4 SPE/IADC-184611-MS

is deformed to the shape of the wellbore, which in some cases will lead to errors in T&D predictions. In
practice, the drillstring position relative to the wellbore may be on the high side, low side, right side or left
side depending on the wellbore section and drillstring operation.
Throughout the history of T&D modeling, one model has consistently been assumed to be the typical
drillstring model; this is the theory developed by Johancsik et al. (1984) and then standardized by Sheppard
et al. (1987). In 1984, Johancsik began with a soft-string T&D theory assuming a normal force was caused
from tension against the curvature and the weight of the pipe section based on Coulomb's friction model.
Later, Sheppard et al. (1984) changed the model to a differential form that included the effects of mud
pressure. This subsequent model has been used in many different facets of the oil and gas industry.
In 1986, H-S Ho (Aston et al. 1998) went a step further by attempting to create a more comprehensive
model that satisfied all the force-equilibrium equations and moment equilibrium equations by assuming that
the string has shear forces. These models were standardized by Mitchell and Samuel (2009).
This was accomplished by mathematically deconstructing previous models, where it was determined
that the impact of bending moment was substantial and not effectively accounted for. This was further
shown through two case studies using field survey data. The purpose of the experiment was to determine
the accuracy of the soft-string T&D model compared to the revised Mitchell equations. Although the results
were not definitive, significant deviation of the soft-string model from the field data was observed. One
major shortcoming of this "soft-string" model, as stated early, is that bending stiffness is neglected.

Stiff-String Model
The current alternative to the soft-string model is the stiff-string model, which considers the drillstring
bending stiffness and annulus clearance. Even though there have been many stiff-string models developed,
none of them has a standard formulation in the industry (Wu 1996). In 1962, Lubinski (Aston et al. 1998)
was the first to consider the calculation of pipe deflection while in tension using the Euler beam theory.
Later, Paslay and Cernocky extended the study for the pipe being in compression (Paslay and Cernocky
1991). Both models included the bending moment and the drillstring-stiffness. Furthermore, the deflection
of the pipe in the wellbore and the severity of the drillstring buckling has been analyzed by Paslay and G.
Handelman in 1964, Christman in 1976, Walker and Friedman in 1977, Chandra in 1986, Dareing in 1991,
Rocheleau in 1992, J. Wu and H. Juvkam-Wold in 1993 and Li Zifeng in 1994 using the same methodology
adopted by Lubinski.
In 2007, Mason and Chen (2007) analyzed numerous companies' stiff-string T&D software and
concluded that all yielded similar results given comparable inputs. Additionally, some of the confusion with
the validity of each model's conclusions was cleared up. For instance, hole size does not have a significant
effect on T&D results. Also, the friction factors for running casing have been estimated at twice that of
drilling operations. Finally, one of the more important findings is that the T&D models use a kinetic friction
and not a static friction. Mitchell and Samuel found that standard models perform poorly in short radius
and complex 3D wells.
Later in 2014, a new dynamic three-dimensional analytical approach was published (Tikhonov at el.
2014), which incorporates a new criterion that calculates accurate and more robust contact forces per
joint using a stiff-string model. This model follows the same assumption of the Lubinski-Paslay-Cernocky
bending-stress-magnification factor, where the deflection is not constant and is changing along the BHA.
The only limitation of this model is the use of a constant curvature trajectory calculation [the MCM], which
assumes constant bending and, hence, constant bending moment.

Torque and Drag General Vector Differential Form


Torque, which is the moment of a force, can be defined as the measure of the forces' tendency to produce
torsion and rotation about an axis. These forces can be generated around the wellbore from different
sources, namely: mechanical torque, frictional torque, and bit torque. Vibrations of the drillstring could also
SPE/IADC-184611-MS 5

contribute to additional torque. In the process of modeling torque, it is assumed that the pipe will climb up
the side of the wellbore as it rotates as shown in Figure 2 [a].

Figure 2—Force and moment balance FBD: [a] torque – looking downhole; [b] 3D T&D model balance

Drag, on the other hand, is an axial force that is generated only when the pipe is moved in an axial
direction. It is proportional to the hookload and dependent on the friction factor coefficient. It always acts
in the opposite direction to pipe movement. A free body diagram [FBD] is shown in Figure 2 [b].
For a drillstring, the internal forces applied consist of the tangential [axial] force [Ft] and two shear forces–
one in the normal direction [Fn] and one in the binomial direction [Fb]. And the external forces consist of
the pipe effective weight [Wb], the side force [Wc] that is normal to the pipe contact force and, lastly, the
drag force [Wd] that is force induced due to the friction. All six forces are added to obtain the general vector
differential form of the balance equilibrium of forces, given by equation [1].

[1]

For a drillstring with bending stiffness [EI], the general form of the moment [Mt] balance equilibrium
equation is given by equation [2]:

[2]

The general vector differential form of the balance equilibrium of forces and moments becomes useful
when the wellbore is assumed to be modeled as a spline using the ASC borehole trajectory model.

Torque and Drag Model for the Advanced Spline Curve Trajectory
In this model the friction forces will be modeled using the Coulomb friction concept. This system is subject
to several assumptions, boundary constraints and load conditions. The required assumptions and guidelines
are as follows:
1. Assumptions:
a. The drillstring is a simple beam with homogenous properties with a circular cross sectional
shape.
b. The pipe curvature and torsion are assumed to be the wellbore curvature and torsion.
6 SPE/IADC-184611-MS

c. The drillstring diameter and material properties can be grouped depending on the bottom
hole assembly [BHA] design.
d. The rotary speed [RPM], WOB and pump rate [SPM] at the surface are constant [steady
state].
e. The external fluid loads are assumed to be zero [Δwef = 0].
f. The drillstring internal loads are not considered [PiAi = 0].
2. Guidelines:
a. The model is for a non-constant curvature trajectory [The ASC borehole trajectory model]
in τ and dκ/ds exist.
i. The effect of tool joints, couplings and wellbore irregularities and tortuosity are
considered.
ii. The bending moment is interchanging resulting in an equivalent and interchanging
tangent, normal and binormal forces along the wellbore.
iii. The tangent, normal and binormal unit vectors are modeled to the approximating
circle [see appendix B].

Torque and Drag Representation


When the wellbore trajectory is modeled using the ASC model, the wellbore positions are computed. From
this trajectory, τ, κ and dκ/ds are determined. These parameters will be included in the force and moment
equilibrium equations as follows:

• Tangent Component

[3]

[4]

Note that from equation [3]: the plus [+] means hoisting and minus [-] means lowering of the
pipe, which dictates the direction of the friction, such that drag always acts in the opposite direction
that the drillstring is moving.
• Normal Component

[5]

[6]
• Binormal Component

[7]

[8]

From equations [4], [6] and [8] the magnitude of the ASC contact force [lbf/ft]:

[9]

And the direction [degrees]:


SPE/IADC-184611-MS 7

[10]

Introducing the stiffness and bending moment terms in this model will have the potential to greatly reduce
the under-prediction of T&D parameters. Contact forces will be amplified and more accurately estimate the
downhole conditions. This model becomes more important in ER wells with large side turns where contact
side forces on the drillstring are not properly estimated using the traditional stiff-string model.

Estimating Downhole Weight on Bit and Torque


In this study the ASC 3D T&D model is used to estimate the DWOB and TOB while considering the friction
factor induced from both the axial and rotational loadings. These loads are the primary contributor to energy
losses from the surface (Aarrestad et al. 1994). Better estimated DWOB and TOB parameters will allow
drilling engineers to update the driller with target surface WOB and torque parameters to improve the rate
of penetration [ROP].
The DWOB is reduced primarily by the axial friction corresponding to the drillstring position and the
static weight of the drillstring immersed in the fluid. Rotational friction will reduce the TOB from the contact
between two rotating bodies. These expressions can be illustrated in equations [11] and [12]:

[11]

And,
[12]
The axial and rotational friction between the drillstring and the wellbore in this study depend primarily on
the normal force induced by the drillstring weight and the tension along the drillstring. Applying increased
WOB will cause a reduction in the tensile force since some portion of the drillstring is in compression. This
phenomenon will significantly reduce the rotational forces (Aston et al. 1998).
In a perfect vertical well, the surface WOB and surface torque could be equal to DWOB and TOB as the
axial and rotation friction could be nil. In contrast, ER and deep vertical wells, especially those with some
degree of dogleg severity, will have a significant reduction in the true DWOB and TOB due to friction force
along the wellbore. There is a necessity to have a more accurate model to better predict downhole conditions.
An academic example has been selected to illustrate how DWOB and TOB can be estimated. The use of
downhole sensors that measures real-time DWOB is required to validate the accuracy of these estimations.

Quantification of the ASC 3D Torque and Drag Model


The field data analysis for the verification and validation of the modified T&D model and the estimation
of DWOB and TOB from surface hookload measurements are presented. Two field cases have been used
to demonstrate the challenges during different drilling operations [ROB, RD, PU and SO]. The testing and
validating procedure of the model accuracy consists of two approaches:
1. Semi analytical approach: Data generation from the ASC borehole trajectory model at a high
resolution.
2. Using the ASC 3D T&D model.
The two approaches applied in this study are shown to be consistent and give a close match for accuracy
against field data compared to all models.
8 SPE/IADC-184611-MS

Semi-Analytical Approach
The ASC borehole trajectory model utilized in this analysis is a cubic polynomial piece-wise function with
a non-constant curvature trajectory assumption in which dκ/ds and τ exist. By setting specific boundary
conditions (Abughaban, et al. 2016), a collection of equations is obtained. Once the system is solved, each
component of the cubic function [T] is known over the entire wellpath. The general spline equation is used
to calculate measured depth [MDi], inclination [ϕi] and azimuth [θi] between any given survey data using a
semi-analytical approach. The generated data output from the ASC borehole trajectory model is then used as
an input in an industry-approved T&D software [Program A]. The mathematical steps of the semi-analytical
approach are presented in Appendix A.

Adaptation of the T&D Model for a Non-Constant Curvature Trajectory Model


The main task is to calculate the axial effective force in the drillstring. If the axial force [Ft ] and the normal
i+1

force [Fn ] are known, the axial force at the top segment can be calculated using equation [13] as follows:
i+1

[13]
Where,

[14]

And,

[15]

However, Mitchell and Miska (2011) have conducted previous analysis stating that the z-component of
the tangential and normal vector is a non-linear function between the survey stations at a given measured
depth. Consequently, a numerical approach is required to calculate the drag force, and a closed form solution
will not be applicable for the ASC 3D T&D model.
In this case, a first approximation of the unit contact force must be obtained using equation [9], from
which the first approximation of the axial force at the top can be calculated. An average axial force is
calculated to obtain a second approximation of the contact force; thus, a second approximation of the axial
force and normal force is found. This iterative process is repeated until the axial force approximation reaches
a tolerance level less than 1%. Final values are then reported.
To effectively validate the ASC 3D T&D model, real-time forces that define T&D measured at the surface
and at the bottom of the drillstring are required to evaluate the results' accuracy. The field data includes
hookload and surface torque measured at the surface for different rig operations such as ROB, RD, and PU
and SO while rotating. Given comparable inputs for an industry-approved T&D program [Program A] and
ASC 3D T&D model, the T&D outputs show good agreement of results until a high-angled, high-tortuous
section of the wellbore is encountered.
Field Case 1. Well 1 is an extended reach well with a double build trajectory as shown in Figure 3. The
first build section begins at 1,600-ft [490-m] and ends at 6,900-ft [2,100-m] at a maximum inclination of
8°. The second build to horizontal was drilled over a 1,400-ft [425-m] section starting at 6,900-ft [2,100-
m] until 8,200-ft [2,500-m] with a maximum dogleg of 11°/100-ft. The well remains horizontal until a total
depth [TD] of 23,000-ft [7,010-m] with a maximum dogleg of 5°/100-ft.
SPE/IADC-184611-MS 9

Figure 3—Field case 1: ER well vertical section

The well is assumed to be an 8.5-in hole with 10-ppg mud weight. A friction factor of 0.20 is used in
the cased-hole section from the surface until 8,200-ft [2500-m], and a friction factor of 0.25 is used in the
open-hole section. These friction factor values were chosen to standardize the model validation process.
The BHA consists of an 8.5-in polycrystalline diamond compact [PDC] bit, a 6.75-in mud motor, a 6.25-in
DS-110 91 lbm/ft drillcollar, a 5-in heavy weight drillpipe with NC-44 connections, and a 5-in S-135 19.5
lbm/ft drillpipe with NC50 connections. The BHA components for field case 1 are summarized in Table 1.

Table 1—Field case 1: Drillstring design

Field Case 1

Drill Component OD [in] ID [in] Length [ft] Count Weight per foot [lb/ft]

PDC Bit 8.5 N/A 0.7 1 N/A


Mud Motor 6.75 N/A 30 1 86.80
NC-44 DC 6.25 2.25 30 1 91.00
MWD/LWD 6.75 2.00 44 1 130.00
NC-44 DC 6.25 2.25 30 2 91.00
HT-50 HWDP 5.00 3.00 29.9 40 59.00
NC-50 DP 5.00 4.276 30.917 695 19.50

The comparison of hookload, axial loading and surface torque parameters calculated using different
models, including the ASC 3D T&D model indicates good agreement of results. However, major differences
are observed when the wellbore is experiencing high and repeated doglegs and tortuosity [red box shown in
the contact force figure]. The ASC 3D T&D model, shown by the blue line in Figure 4, shows a high contact
force distribution between the wellbore and the drillstring throughout both the highly tortuous zone from
5,000ft [1,525 m] to 8,000ft [2,440 m] and the moderately tortuous zone [the second build] from 10,000
ft [3,050 m] to 12,000 ft [3,660 m]. All other models predict no contact in the second build and fail to
10 SPE/IADC-184611-MS

capture the additional drillstring to wellbore contact force. With closer examination, for the basic drilling
operations, the following findings can be highlighted:

Figure 4—Field case 1: ER well contact force. DLS in units of degrees/100 ft

Rotating off Bottom. For hookload results, the difference between the T&D programs used are within
tolerable limits [2 to 3%] since the friction is in the circumferential direction with no effect from the axial
direction (Tikhovo et al. 2014). The rotating torque shown in Figure 5 has a somewhat higher difference
compared to previous models at a maximum variation of 8% [approximately 10 klbf-ft difference]. The axial
load illustrated in Figure 6 has experienced the highest variation, among all models, of 12% [approximately
20 klbf difference]. One of the main reasons for these differences is the discrepancy in the contact forces
distribution.
SPE/IADC-184611-MS 11

Figure 5—Field case 1: ER well torque results while ROB. DLS in units of degrees/100 ft

Figure 6—Field case 1: ER well axial load while ROB

Pickup. The hookload has experienced high discrepancy of 17% [approximately 120 klbf difference].
This is due to the additional loads from the tool joint, wellbore irregularity and tortuosity considerations in
the force equilibrium equation. The rotating torque did not experience a major effect with a difference of
less than 2%. The axial load has experienced a higher difference at 14% [approximately 40 klbf difference]
as shown in Figure 7.
12 SPE/IADC-184611-MS

Figure 7—Field case 1: ER well axial load while pickup. DLS in units of degrees/100 ft

Slackoff. The hookload did not experience a major effect, compared to other models, with difference less
than 3 to 4%. The rotating torque showed a slight difference with a maximum discrepancy of 6%. The same
trend was observed for the axial load illustrated in Figure 8 with good agreement of results from surface
to TD.

Figure 8—Field case 1: ER well axial load while slackoff


SPE/IADC-184611-MS 13

Rotary Drilling. Several noticeable differences are observed between the models for all three parameters.
The hookload has experienced a high discrepancy of 21% [approximately 25 klbf difference]. The rotating
torque also experienced a high discrepancy of 18% [approximately 9 kft-lbf maximum difference] as shown
in Figure 9. A similar trend is observed for the axial load [Figure 10] with a 23% discrepancy [approximately
18 klbf difference].

Figure 9—Field case 1: ER well torque while drilling @ 120 RPM and 60 klbf WOB. DLS in units of degrees/100 ft

Figure 10—Field case 1: ER well axial load while drilling @ 120 RPM and 50 klbf WOB
14 SPE/IADC-184611-MS

The observed discrepancies for all basic drilling operations are primarily due to the low contact force
predicted. Thus, accurately predicting the contact forces is important to prevent drillstring failures and
catastrophic events such as stuck pipe, casing and drillpipe wear and drillstring fatigue. The results of torque
and hookload for the different models are summarized in Table 2.

Table 2—Field case 1: T&D results comparison

ROB PU SO RD
Rig Operation
ST [kft-lbf] HL [klbf] HL [klbf] HL [klbf] ST [kft-lbf] HL [klbf]

New T&D Model 54 166 249 112 67 192

Program A – ASC 47 162 242 102 60 189

Program A - MCM 43 156 214 99 56 157

All parameters are rounded to the nearest whole number.


Key: HL = hookload; ST = surface torque; kft-lbf = 1000 ft-lbf.

Field Case 2. Well 2 is a backward-nudge horizontal well [Figure 11]. The nudge begins at 1,600-ft [490-
m] and ends at 3,800-ft [1,160-m]. The first build was drilled over a 1,400-ft [425-m] section starting at
2,840-ft [865-m] to 6,600-ft [2,010-m] with a maximum dogleg of 2°/100-ft. The second build was drilled
over a 4,000-ft section from 3° inclination until the well became horizontal [90° inclination] at a 10,670-ft
[3,250-m] and remained horizontal until 17,122-ft [5,220-m], with a maximum dogleg of 18°/100 ft.

Figure 11—Field case 2: Backward-nudge horizontal well vertical section

The 9.625-in casing set at 7,704-ft [2,350-m] has a friction factor of 0.2. The 8.75-in open hole below
the shoe has a friction factor of 0.25. The mud weight is 10.20 ppg. The BHA consists of an 8.75-in PDC
bit, a 6.75-in mud motor, a 6.25-in DS-110 91 lbm/ft drillcollar and a 5-in s-135 19.5-lbf/ft drillpipe with
NC50 connections.
Similar trend of results was observed for field case 2 for all three parameters [hookload, axial loading and
torque] indicating good agreement among programs. However, major differences of the contact force are
observed across the high and repeated tortuosity zones [red box shown above the contact force figure] from
SPE/IADC-184611-MS 15

9,000-ft [2,745-m] to 17,000 ft [5,182-m]. None of the models used had the ability to capture the additional
drillstring to wellbore contact force across that zone except the ASC 3D T&D model [shown in Figure 12].

Figure 12—Field case 2: Backward-nudge horizontal well contact force. DLS in units of degrees/100 ft

The following findings from the basic drilling operations can be highlighted:
Rotating off Bottom. The results among the T&D programs used are within acceptable variation. The
hookload has experienced a difference of less than 2 % [approximately 5 klbf difference], since there is no
effect from the axial direction. The rotating torque predicted using the ASC 3D T&D model indicates the
lowest value in the vertical section due to the contact force distribution effect. A higher torque is estimated at
the build section with high tortuosity, as expected, with a 20% maximum discrepancy [approximately 3 kft-
lbf difference]. The fluctuation of torque values is illustrated in Figure 13. Axial loading showed a higher
difference with a maximum discrepancy of 8% [approximately 20 klbf difference] as shown in Figure 14.
16 SPE/IADC-184611-MS

Figure 13—Field case 2: Backward-nudge horizontal well torque while ROB

Figure 14—Field case 2: Backward-nudge horizontal well axial loading while ROB. DLS in units of degrees/100 ft

Pickup. The hookload estimations showed a large difference between models with a maximum
discrepancy of 12% [approximately 9 klbf difference]. The rotating torque showed a smaller difference
SPE/IADC-184611-MS 17

with a maximum discrepancy of 4%. A similar trend is observed for the axial load in Figure 15 with good
agreement [2% discrepancy] across different models from surface to TD.

Figure 15—Field case 2: Backward-nudge horizontal well axial loading while PU

Slackoff. The surface hookload did not see a major change with a difference of 3%. Similar observations
are noted for the rotating torque and the axial load from surface to TD with a difference of 6% and 2%
respectively.
Rotary Drilling. The surface hookload has experienced a major effect with a maximum discrepancy of
22% [approximately 14 lbf difference]. The rotating torque, from the ASC 3D T&D model, predicted the
lowest estimation at the build section, unlike all the others. The difference ranges between 20 and 32%
[approximately 4 kft-lbf difference] as shown in Figure 16. Observations show that in high-angled wells,
the overall torque tends to drop while drilling because most of the drillstring is in compression and the
tension profile in the build section is reduced (Paune and Abbassian 1996). This prediction of reduced
contact force in the initial vertical portion of the well [Figure 12] leads to a lower torque estimation. The
axial load has a noticeable difference across the build section of 18% [approximately 30 klbf difference],
shown in Figure 17.
18 SPE/IADC-184611-MS

Figure 16—Field case 2: Backward-nudge horizontal well torque while


drilling @ 75 RPM and 25 klbf WOB. DLS in units of degrees/100 ft

Figure 17—Field case 2: Backward-nudge horizontal well axial loading while drilling @ 75 RPM and 25 klbf WOB

Comparing the hookload, the surface torque and the axial load from different models and inputs helps
validate the ASC 3D T&D model. These basic operations generally agree with the ASC 3D T&D model
with acceptable variation to one another [shown in Table 3].
SPE/IADC-184611-MS 19

Table 3—Field case 2: T&D results comparison

ROB PU SO RD
Rig Operation
ST [kft-lbf] HL [klbf] HL [klbf] HL [klbf] ST[kft-lbf] HL[klbf]

New T&D Model 8 25 38 17 24 66

Program A - ASC 9 30 46 19 23 72

All parameters are rounded to the nearest whole number.


Key: HL = hookload; ST = surface torque; kft-lbf = 1000 ft-lbf.

Next, the ASC 3D T&D model is validated by comparison against field data.
Depth-Based Field Data from Field Case 2. Depth-based data for field case 2 were obtained from daily
morning reports. These data provide an entire day's operations [24 hours of drilling] by a single data point
that limits its value in representing the day's operation. Nevertheless, these data are the only consistent
measurements of drilling T&D acquired during the drilling operation, and thus will be used to validate the
ASC 3D T&D model. The available data represent 9-days of drilling operation over a 6,500-ft [1,980-m]
drilled section.
Rotating off Bottom. The surface hookload should match the buoyed weight of the drillstring measured
at the top-drive assembly. Any difference observed would likely be from the mud weight error, as the model
assumes a single mud weight value for the entire well. Error from the entered components of the drillstring
and the BHA in the model compared to the ones used in the field could contribute to the differences. The
results from ROB hookload [Figure 18] show an accurate agreement at the build section with high tortuosity
from 10,000-ft [3,048-m] to 11,000-ft [3,350-m] MD with predictions within 3% for both the ASC 3D T&D
model and the MCM input, which is within the industry-acceptable variation.

Figure 18—Field case 2 Backward-nudge horizontal well hookload while ROB


20 SPE/IADC-184611-MS

Pickup. The axial loading and friction factors will directly impact the hookload. This is shown by the data
in Figure 19. The ASC 3D T&D model estimates showed an overestimated difference of 8% [approximately
+30 klbf difference] compared to drilling data following the high tortuosity zone. In contrast the MCM
model significantly underestimates the hookload by 17% [approximately -80 klbf difference]. This large
difference can be explained by the MCM model's failure to account for the contact force across the high
tortuosity interval.

Figure 19—Field case 2: Backward-nudge horizontal well hookload while pickup

Slackoff. The hookload shows an accurate agreement at the build section with high tortuosity [less than
1% difference]. However, an underestimated parameter, using the ASC 3D T&D model, was observed
around 12,500-ft [-9.5% average difference]. Then overestimated onward over the horizontal section
[+19.2% average difference] as illustrated in Figure 20. Similar observation is noted from the MCM
input with an underestimated parameter of 13% difference at the high tortuosity zone, followed by an
overestimation of 14% difference at the horizontal section. Since ASC model is assuming a greater side force
due to high tortuosity, a lower slack-off weight should be expected because more of the drillstring is being
supported by the wellbore friction rather than the hookload. Hence, the qualitative differences between the
ASC 3D T&D model and Program A [MCM] vs. field data are not well understood, although the qualitative
character of models prediction appears to be correct in the author's view.
SPE/IADC-184611-MS 21

Figure 20—Field case 2: Backward-nudge horizontal well hookload while slackoff

Rotary Drilling. The comparison of surface torque against the field data is shown in Figure 21. Generally,
the results from the ASC 3D T&D model shows close agreement at a maximum overestimation of 11%
compared to the MCM input at a maximum underestimation of 22%. This variation could be attributed to
the model's predicted contact forces, particularly at the high tortuous zone. It should be noted that again,
the MCM model significantly underestimates the torque due to failure to capture the contact forces in the
high tortuosity interval.

Figure 21—Field case 2: Backward-nudge horizontal well torque while drilling @ 75 RPM and 25 k-lbf WOB

For the basic drilling operations, the percentage difference between the field data and model's predictions
are summarized in Table 4.
22 SPE/IADC-184611-MS

Table 4—Field case 2: Percentage difference between field data and model predictions

MD [ft] HL while ROB, ASC HL while PU, ASC HL while SO, ASC ST while RD, ASC
From To 3D T&D Model 3D T&D Model 3D T&D Model 3D T&D Model

8256 8729 -8.2 -27.9 +13.5 -1.6


8729 9269 -3.8 -26.4 +19.18 -1.5
10639 11111 -0.4 +7.5 -7.2 +15.5
11111 11678 -0.9 +6.3 -9.0 +21.0
11678 12434 -0.9 +9.0 0.0 +16.6
12434 13939 -0.9 +9.2 0.0 +12.4
13939 14974 +1.8 +8.9 +9.8 +11.8
14974 15445 +5.3 +9.0 +19.2 +17.4
15445 17122 +6.7 +6.0 +19.2 +14.7

Key: HL = hookload; ST = surface torque; [+] = overestimated parameter; [-] = underestimated parameter

Validating the Estimated Downhole Weight on Bit and Torque on Bottom


Once the axial and rotational effective forces against the drillstring movement and rotation are calculated,
a new developed model will estimate DWOB and TOB. These parameters are estimated using surface
hookload and torque measurement during the basic drilling operations [ROB, PU, SO and RD]:
Step 1: Calculation of the forces acting on the bottom of a drillstring can be computed for a given
drillstring specification, survey data and friction coefficient.
Step 2: The estimations of DWOB and TOB are an iterative process, such that different downhole
values will be estimated until the measured and calculated hookload reaches a tolerance level less
than 1%.

Step 3: The final estimated DWOB and TOB values are chosen.
The academic example [10,000-ft MD s-shape synthetic well] shows losses of 8-klbf in WOB at the
bottom [an estimated 32% loss], as illustrated in Table 5. This claim will require the use of downhole
sensors that measure real-time DWOB to validate the accuracy of this estimation. However, this analysis
was conducted using the modified T&D model that offers a better estimation of the stiffness and bending
moment and allows a more accurate prediction of the actual downhole parameters. This will enable drilling
engineers to update the driller with optimized WOB and surface torque parameters to improve the ROP
without taking undue risks to the drillstring, particularly in high angle horizontal wells.

Table 5—DWOB academic example with 175 klbf measured HL

DWOB [klbf] Calculated Hookload [klbf] Hookload Tolerance < 1%

SWOB = 25 162 8
23 167 5
21 169 4
19 170 3
DWOB = 17 173 1
SPE/IADC-184611-MS 23

Summary of Observations
Based on the study analysis and field applications, the use of the current industry-standard T&D model has
validity of T&D predictions in low tortuosity wellbores. The higher the build angles and more extended the
reach, the impact of bending stiffness and the change in curvature becomes more of greater importance.
Two field cases have been tested to compare the variation of the ASC 3D T&D model against an industry
approved T&D software [Program A]. The input used in Program A are (1) actual field surveys with typical
drilling surveys at 90-ft [30-m] intervals [MCM input] and (2) semi-analytical surveys generated from the
ASC borehole trajectory model at a one foot [0.3-m] resolution [ASC input]. The absolute value relative
variation between the two methods in program A and the ASC 3D T&D model are summarized in Table 6.

Table 6—Average of absolute value percentage of variation between ASC 3D T&D model and Program A [MCM input and ASC input]

ROB Pickup Slackoff Rotary Drilling

Program A Program A Program A Program A

ASC MCM ASC MCM ASC MCM ASC MCM

HL 2 3 12 17 <4 <4 12 21

ST 8 20 3 6 <6 <6 17 20

AL 8 12 10 14 <3 <3 18 23

All percentages are rounded up to the nearest whole number.


Key: HL = hookload; ST = surface torque; AL = axial load

The ASC 3D T&D model and the Program A [MCM] were tested for accuracy against field data using
actual depth-referenced data taken from daily drilling reports. A single data point of this data represented
one day of drilling. The results are summarized in Table 7.

Table 7—Average value of variation percentage of ASC 3D T&D model and Program A [MCM input] vs. Field data

ROB Pickup Slackoff Rotary Drilling

Program Program Program Program


ASC 3D T&D ASC 3D T&D ASC 3D T&D ASC 3D T&D
A [MCM] A [MCM] A [MCM] A [MCM]

HL -1 -2 +8 -17 0 0 X

ST X X X +11 -22

Key: HL = hookload; ST = surface torque; [+] = overestimated parameter; [-] = underestimated parameter; × = not analyzed

From Table 6 and Table 7, input data from MCM shows the least level of accuracy in predicting T&D
parameters specifically in pick up and rotary drilling. The highest level of accuracy is achieved by using
the ASC 3D T&D model as it accounts for the change of the rate of curvature and more accurately predicts
the drillstring contact force.

Recommendations
The modified T&D model [the ASC 3D T&D model] provides improved capabilities to predict the contact
force along the drillstring in extended-reach and deviated / vertical wells with high tortuosity enabling more
accurate prediction of drilling parameters including hookload, torque and axial load. Table 8 summarizes
the accuracy of different models for specific well categories.
24 SPE/IADC-184611-MS

Table 8—Accuracy of models in predicting T&D parameters

Well Categories
Models
Low tortuosity Extended reach High tortuosity

Soft String ✓ X X
Stiff String (MCM) ✓ ✓ X
Stiff String (ASC) ✓ ✓ ✓
ASC 3D T&D Model ✓ ✓ ✓

Key: ✓ = acceptable estimation of T&D parameters; X = underestimation due to model assumptions

Since T&D are critical elements in drilling these wells, it is recommended to use the ASC 3D T&D model
to better model the downhole drilling forces and reduce the potential for catastrophic events while drilling,
such as stuck pipe, and drillstring fatigue and wear phenomena.

Conclusion
The following conclusions are made based on the study and field data:
1. The ASC 3D T&D model is a stiff-string model that includes the geometric torsion [τ], the wellbore
curvature [κ], the change in the rate of wellbore curvature [dκ/ds] and the drillstring bending stiffness
[EI] in the general force and moment equilibrium equations allows:
a. More accurate prediction of the drillstring contact force, and
b. Better understanding of the true effect of bending moment.
2. The ASC 3D T&D model was compared to current industry standard models and showed good
agreement until a region of high tortuosity is encountered. At that point, the ASC 3D T&D model
predicted additional axial and torsional loads. These forces are not predicted in current models that
assume constant curvature arcs [MCM].
3. The calculated T&D output from using a non-constant curvature trajectory showed significant
deviation from currently used models in high tortuosity sections. This was tested by comparing two
approaches:
a. A semi-analytical approach – survey data generation from the ASC borehole trajectory model
at one foot [0.3-m] resolution as an input in current industry-approved T&D software.
b. The ASC 3D T&D model compared to actual drilling data.
4. Two field cases are presented where the soft-string model [current industry standard] fails to predict
accurate and reliable T&D parameters under pickup and rotary drilling conditions.
a. Field case 1 is an extended reach well [23,000-ft MD]
b. Field case 2 is a backward-nudge horizontal well with high dogleg severity [18°/100-ft]
5. The ASC 3D T&D model predictions in the same field cases agree with the observed field data over
the basic drilling operations compared to current MCM T&D model.
6. The estimated DWOB and TOB can be more accurately predicted with the ASC 3D T&D model
enabling higher surface read out weight on bit applications without overloading the bit itself. This
will enable improved drilling performance when significant loss of real weight on bit occurs in highly
deviated or tortuous wells.

Acknowledgements
The authors like to express their appreciation to their respective companies for giving their permission to
publish the results. Additionally, we thank Kirwinlvinodaq S. Lawrence and Matthew Countryman from
SPE/IADC-184611-MS 25

the CSM 2016 Externship Program for simulating the T&D parameters using K&M's ERA software for
field case 1 ERD well.

Nomenclature

Symbol Unit Meaning

ASC - Advanced spline curves


BHA - Bottom hole assembly
DLS °/100ft Dogleg severity
DWOB klbf Downhole weight on bit
ER - Extended reach
HL - Hookload
MCM - Minimum curvature method
PDC - Polycrystalline diamond compact
PU - Pickup (drillstring moving upward)
ROB - Rotating off bottom
ROP - Rate of penetration
RPM - Rotary speed in revolution per minute
RTOC - Real-time operation center
SO - Slackoff (drillstring moving downward)
SPM - Pump rate in stroke per minute
TD ft Total depth
TOB kft-lbf Torque on bottom
T&D - Torque and drag
WOB klbf Weight on bit
26 SPE/IADC-184611-MS

Mathematical Symbols

Conversion Factors

SI Metric Conversion Factors

ft × 3.048* E-01 = m
in × 2.54* E-03 = cm
lbf × 9.869 233 E+00 = N
ft-lbf × 1.355 818 E+00 = N-m
psi × 6.894 757 E+00 = kPa

*Conversion factor is exact.

References
Aarrestad, T.V., Blikra, H. 1994. Torque and Drag-Two Factors in Extended-Reach Drilling, Journal of Petroleum
Technology, 271 (3): 800-803. SPE-27491-PA. http://dx.doi.org/10.2118/27491-PA
Abughaban, M.F., Bialecki, B., Eustes, A.W., de Wardt, J.P., & Mullin, S. 2016. Advanced Trajectory Computational
Model Improves Calculated Borehole Positioning, Tortuosity and Rugosity. Presented at the SPE/IADC Drilling
Conference, Fort Worth, Texas, 1-3 March. SPE-178796-MS. http://dx.doi.org/10.2118/178796-MS
Aston, M.S., Hearn, P.J., McGhee, G. 1998. Techniques for Solving Torque and Drag Problems in Today's Drilling
Environment. Presented at the SPE Annual Technical Conference and Exhibition, New Orleans, Louisiana, 27-30
September. SPE-48939-MS. http://dx.doi.org/10.2118/48939-MS
Johancsik, C.A., Friesen, D.B., Dawson, R. 1984. Torque and Drag in Directional Wells-Prediction and Measurement.
Journal of Petroleum Technology, 36 (6): 987-992. SPE-11380-PA.http://dx.doi.org/10.2118/11380-PA
Mason, C., and Chen, D.C.-K. 2007. Step Changes needed to Modernise T&D Software. Presented at the SPE/IADC
Drilling Conference, Amsterdam, Netherlands, 20-22 February. SPE-104609-MS. http://dx.doi.org/10.2118/104609-
MS
Mitchell, R.F. and Miska, S.Z. 2011. Directional Drilling. In Fundamentals of Drilling Engineering. Vol. 12, Richardson,
Texas: Textbook Series, Society of Petroleum Engineering.
Mitchell, R.F., & Samuel, G.R. 2009. How Good Is the Torque/Drag Model? Presented at the SPE/IADC Drilling
Conference, Amsterdam, The Netherlands, 20-22 February. SPE-105068-MS. http://dx.doi.org/10.2118/105068-MS
SPE/IADC-184611-MS 27

Paslay, P. R. and Cernocky, E. P. 1991. Bending Stress Magnification in Constant Curvature Doglegs With Impact on
Drillstring and Casing. Presented at the SPE Annual Technical Conference and Exhibition, Dallas, Texas, 6-9 October.
SPE-22547. http://dx.doi.org/10.2118/22547-MS
Payne, M.L., Abbassian, F. 1996. Advanced Torque-and-Drag Considerations in Extended-Reach Wells. Presented
at the SPE/IADC Drilling Conference, New Orleans, Louisiana, 12-15 March. 12-15 March. http://
dx.doi.org/10.2118/35102-MS
Sheppard, M. C., Wick, C., & Burgess, T. 1987. Designing Well Paths To Reduce Drag and Torque. SPE Drilling
Engineering, 2 (4): 344-350. SPE-15463-PA. http://dx.doi.org/10.2118/15463-PA
Tikhonov, V., Valiullin, K., Nurgaliev, A., Ring, L., Gandikota, R., Chaguine, P., Cheatham, C. 2014. Dynamic
Model for Stiff-String Torque and Drag. SPE Drilling & Completion, 29 (3): 279-992. SPE-163566-PA. http://
dx.doi.org/10.2118/163566-PA
Wu, J. 1996. Drill-Pipe Bending and Fatigue in Rotary Drilling of Horizontal Wells. Presented at the SPE Eastern Regional
Meeting, Columbus. Ohio, 23-25 October. SPE-37353-MS. http://dx.doi.org/10.2118/37353-MS
28 SPE/IADC-184611-MS

Appendix A
Semi-Analytical Approach for the Advanced
Spline Curve Borehole Trajectory Model
The purpose of this analysis is to increase the number of surveys [resolution] using a semi-analytical
approach from the ASC borehole trajectory model as follow:
Step 1: Compute the general spline curves equation:
[A-1]
Using the specifies boundary conditions (Abughaban et al. 2016) For i = 0, …, n – 1 where:

Step 2: Solve the function T(s) for the approximation of the wellbore positions–true vertical depth
[TVD], Easting [E] and northing [N]–at any step change in measure depth [si]:

[A-2]

Step 3: Compute the inclination [ϕi] and azimuth [θi] at each measured depth [MDi] selected as follow:

[A-3]

The high-resolution survey results output from the ASC borehole trajectory model are used as an input
in an industry-approved T&D software [Program A].
SPE/IADC-184611-MS 29

Appendix B
Radius of Curvature of a Spline Curve
Drawing a circle that closely fits nearby points on a local section of a spline curve, forcing the curve and
the circle to osculate will satisfy the conditions where both curves will have the same tangent and curvature
at the point where they meet as shown in Figure B-1.

Figure B1—Spline curve and circular arc osculation

Thus, the radius of curvature R can be expressed as follows:

[B-1]

Where Y″(s) Y‴(s) are computed as follow (Abughaban, et al. 2016):

And, the tangent, normal and binormal unit vectors can be expressed as follow for i = 1, …, n – 1 (Mitchell
and Miska 2011):

[B-2]

[B-3]

[B-4]

You might also like