SPE/IADC 106935

Integrated BHA Modeling Delivers Optimal BHA Design

David C.-K. Chen, Halliburton Sperry Drilling Services

Copyright 2007, SPE/IADC Middle East Drilling Technology Conference & Exhibition
This paper was prepared for presentation at the SPE/IADC Middle East Drilling Technology
Conference & Exhibition held in Cairo, Egypt, 2224 October 2007.
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, as presented, have not been reviewed by the Society of Petroleum Engineers or
International Association of Drilling Contractors and are subject to correction by the author(s).
The material, as presented, does not necessarily reflect any position of the SPE, IADC, their
officers, or members. Electronic reproduction, distribution, or storage of any part of this paper
for commercial purposes without the written consent of the Society of Petroleum Engineers
and 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 where and by whom the paper
was presented. Write Librarian, SPE, P.O. Box 833836, Richardson, TX 75083-3836 U.S.A.,
fax 1.972.952.9435.

Abstract
This paper presents the development of a new bottomhole
assembly (BHA) program and how it is used to deliver
optimal BHA design. The new program allows users to run
static and dynamic models in the same application which will
save time and minimize errors. For the static model, the
program uses the newly developed generic algorithm for
solving the Lubinskis BHA equations directly. The strengths
of the new algorithm are accuracy and computation efficiency,
as compared to the conventional finite-element based BHA
programs. In addition, the program is designed to run fast on a
personal computer (PC) with an intuitive user interface so that
it can be used to optimize BHA design in the office as well as
at the rig site.
The static model is primarily designed for directional
drilling applications, such as optimal BHA design for
maximum steerability, bending moment calculations to
minimize fatigue failure, and BHA sag corrections to improve
survey quality. The dynamic model is based on a hybrid of
analytical and finite-element methods to calculate the critical
rotary speeds of the BHA. This paper describes the
significance of applying these features in a user-friendly
application to improve the optimization process and maximize
drilling performance.
Introduction
Bottomhole assembly (BHA) modeling is always an essential
component of directional drilling. A state-of-the-art BHA
program enables many critical features, such as designing a
BHA to optimize directional performance, optimizing
stabilizer locations to minimize vibration and increase
downhole tool reliability, and improving survey data by
correcting the BHA sag. Since the 1950s, several methods
have been developed and applied in the drilling industry to
build the BHA models. 1-7

In general, the challenges encountered in the development

of a computationally efficient, flexible, and accurate BHA
model can be summarized as follows:
x Nonlinear differential equations
x Unknown upper boundary conditions at the tangential
point (location and orientation)
x Unknown boundary conditions at stabilizers
x Collars and wellbore wall contacts
x Large deformation caused by bent housing motor or the
bend in rotary steerable tools
The most commonly used method in BHA modeling is
probably the finite-element method because it is easy to
develop and use. However, to the knowledge of the author,
many commercial finite-element based BHA programs are still
based on the small deformation theory. As a result, they have
been shown to lack the accuracy required to model steerable
assemblies, such as motor or rotary steerable systems. Finiteelement modeling is also cumbersome in handling the collars
and wellbore contact. To accurately model steerable systems,
the semi-analytical methods are usually required, but semianalytical methods are inflexible and difficult to program.
They are often designed to analyze some specific BHA models
and are limited to BHAs with rather simple configurations.
Objectives
In addition to the challenges in finding an accurate and
flexible static BHA model, most of the programs are
cumbersome to run because of the complicated user interface
or the lengthy calculation time. Thus, the objectives of
developing a state-of-the-art BHA program include the
following:
x Accuracy. It should provide the most accurate modeling
results possible, especially for the motors and rotary
steerable systems in which conventional finite-element
based programs fail.
x Flexibility. The new BHA program should be capable
of modeling any BHA configuration without restriction,
as well as of handling the collar and wellbore contacts.
The program should also allow users to model special
drilling tools, if needed.
x Computational efficiency. The program will primarily
be used on laptop computers by engineers. The
computational time should be less than a few seconds
for the most complex BHA model.

SPE/IADC 106935

x Integrated static and dynamic analyses. The program

should be capable of running static and dynamic models
in the same application.
x User-friendly interface. Because the program will
likely be used in the office and at rig sites, the program
must be easy to run with an intuitive user interface.
General Features
To meet the objectives, a new BHA program, MaxBHA
(Fig. 1), has been developed. The following general features
are included in the program:
x Equilibrium Dogleg Severity (DLS) Rate Prediction
to predict the build/drop/turn rates (inclination/azimuth
DLS) for specified operating conditions
x Force Analysisto calculate the side force acting on the
bit and stabilizers/pads, as well as the deflection, slope
(angle), bending moment, and shear force along the
BHA
x Formation Index Calculationto calibrate the
formation index (characteristic) for a particular interval
x Predict Ahead Analysisto produce the trajectory from
the current location to the future location based on the
bit side cutting capability
x Sensitivity Analysisto allow users to vary a selected
parameter and determine the results
x Survey Sag Correction Calculationto calculate and
correct the misalignment of the BHA relative to the
wellbore at the survey sensor.
x WHIRL Softwareto calculate the lateral natural
frequencies (critical rotary speeds) and mode shapes of
the BHA
The first six features are all static BHA models. The last
feature is a dynamic model using the WHIRL software to
predict the critical rotary speeds of the BHA. Integrated static
and dynamic BHA models are possible because they share the
same boundary conditions determined by the static model.
Static Model Theory
The conventional finite-element method has been shown to
lack the accuracy required for static BHA modeling, especially
for the steerable systems, such as motors and rotary steerable
systems. Consequently, a decision was made to adopt the
semi-analytical method that will solve the Lubinskis BHA
equations directly.
The general solution, developed by Lubinski in the 1950s
to solve for the continuous beam-column model (fourth-order
nonlinear differential equations), can be expressed in the form
of:
x( z )

P1  P2 u cos(

W
W
H'
q sin D
z )  P3 u sin(
z) 
z
u z 2 (1)
EI
EI
W
2W

in which P1, P2, and P3 are three variables,

H ' is the

normalized side force acting on the left side of the segment, W

is the weight on bit, q is the unit weight of the drillstring, and

D is the inclination angle. Thus, for each segment in the BHA

model, there are four unknowns: P1, P2, P3 and H ' . Note that
P1, P2, and P3 are called the segment profiles in this paper.
A BHA model usually consists of components such as the
drill bit, stabilizers, crossovers, bends, and offset pads. For a
BHA with N components, there will be (N+1) sections in the
model. Note that the (N+1)th section has an unknown
boundary condition, bounded by the top component and the
tangential point where the collars are in contact with the
wellbore wall. In each section, there will be four unknowns as
solved by using Lubinskis BHA equations. Thus, for an Ncomponent BHA, there are approximately 4(N+1) unknowns
associated with 4(N+1) nonlinear equations in the model.
Deriving the 4(N+1) nonlinear equations or even iterating
4(N+1) unknowns is an unthinkable task, especially when N
becomes large. For example, a typical BHA with two or three
stabilizers, three or four crossovers, and one bend will have an
N value of at least 6. Thus, there are approximately 28
nonlinear equations to be derived and 28 unknowns to be
solved. As a result, two key challenges involved when using
the closed-form solutions are:
x Finding a process to construct the model dynamically
x Reducing the unknowns to a manageable number for
iteration
A new generic algorithm has been developed to selfstructure the BHA model. The first step is to divide a
complicated BHA configuration into independent modules and
segments. Modules are bounded by two stabilizers. Segments
are bounded by BHA components, such as stabilizers, bends,
and crossovers. A recursive process is used to derive solutions
for each segment and module, one at a time, starting from the
top of the BHA (tangential point) down to the bit. Only two
unknowns (and two nonlinear equations) are left when the
recursive process reaches the bit. The two unknowns are then
solved by iteration. As a result, the entire process is robust and
computationally efficient. A detailed discussion of the
algorithm is given in Ref. 8.
Dynamic Model Theory
The dynamic model is a hybrid of analytical and finiteelement methods based on the steady-state theory. The static
analysis is first run to obtain the boundary conditions, such as
the tangential point location and the contacts at stabilizers.
The boundary conditions are then used to construct the mass
and stiffness matrices of the BHA, as shown in Eq. 2 and Fig.
2. The finite element method is employed to calculate the
critical rotary speeds (eigen-values) and mode shapes (eigenvectors) of the BHA:
[M] {} + [K] {U} = {0}

(2)

in which {U} is the displacement vector, [M] is the mass

matrix with buoyancy effect, and [K] is the stiffness matrix
determined by BHA geometry (ID, OD, material), stabilizer
contact points, upper contact point, weight on bit (WOB), and

SPE/IADC 106935

inclination. Unlike the static analytical model, the bend in the

BHA has little effect on the mass and stiffness matrices.
Consequently, the finite element method has been proven to
be an effective and accurate method for the dynamic model.
Field Implementations
Case #1: Comparing new BHA Model with Field Data and
FE-based Model. In addition to the analytical verifications,
results of the MaxBHA program have been verified by
comparing with field data. In general, the directional
tendencies observed from the field have closely matched the
modeling predictions. However, field data have been
consistently different from those using the finite-element (FE)
method, especially for the steerable systems. Generally
speaking, the MaxBHA program and the FE-based BHA
program produce similar results for the conventional rotary
assemblies. When a bend is introduced, either through the
steerable motors or rotary steerable system, the discrepancy
between the MaxBHA program and the FE program becomes
large, particularly when the bend is greater than one degree.
The discrepancy may result from the use of the small
deformation theory in the FE-based programs. The following
paragraphs describe two of the field verifications.
The first example is a steerable motor assembly, including
an 8 -in. PDC bit, a 6 -in. motor with a 1.5 deg. bend, 6
in. drill collars, and three stabilizers. The gauge and location
of the three stabilizers are: 8.45 in. (1.25 ft), 8.25 in. (7.25 ft),
and 8.25 in. (33.8 ft). The steerable motor was used to kick off
from vertical. The maximum build rate observed from the
survey data was consistent at approximately 11.5 deg/100 ft.
The new MaxBHA program predicted a maximum build rate
of approximately 10.8 deg/100 ft, whereas the commercial
finite-element based BHA program predicted a maximum
build rate of less than 6 deg/100-ft, as shown in Fig. 3.
The second example is a point-the-bit rotary steerable
system (RSS) assembly9. The RSS was used to kick off from
vertical in another well. The build rate of the RSS is observed
to increase with the inclination from both field data and the
MaxBHA program. Although the finite-element program
successfully predicted the build rate at low inclinations, it
failed to predict the trend of increased build rate with
inclination, as shown in Fig. 4.
Case #2: Optimizing Rotary Steerable Systems. A pointthe-bit rotary steerable system (RSS) with full-service
LWD/MWD tools was used to drill a directional well in the
Gulf of Mexico (GOM). The BHA schematic is shown in Fig.
1. The MaxBHA program is used to optimize the following
drilling performances:
Increase DLS capability during kick-off. The BHA design
was studied to increase the DLS capability during kick-off.
The easiest way to accomplish this goal is through optimizing
the gauges of two string stabilizers. Using the sensitivity
analysis, the relationship between DLS capability and
stabilizer gauge can be obtained. Fig. 5 shows the DLS vs.
stabilizer gauge for the two string stabilizers, indicating a
maximum DLS at 8.1 in. stabilizer #3 and at 8.5 in. stabilizer

#4. The results allow engineers to optimize the BHA design to

increase the DLS capability during kick-off.
Minimize bending stresses on RSS. To minimize the
bending stress of the RSS tool and to minimize fatigue failure,
the MaxBHA program was used to study the effect of a flex
sub placed above the RSS tool. Fig. 6 illustrates the sensitivity
analyses on bending moment and DSL vs. the length of flex
sub at 5 degree and 90 degree inclinations. The data allows
engineers to optimize the flex sub design to minimize bending
moment at the RSS tool without sacrificing too much on the
DLS capability.
Manage BHA misalignment. The misalignment between
the hole axis and the MWD directional sensor will result in a
survey error. The magnitude of the error, depending on the
BHA type and geometry, sensor location, and hole size, has
been shown to be the major contributor of the TVD error. Fig.
7 shows the misalignment (slope) along the BHA. The results
indicate that BHA misalignment varies from -0.2 degree to 0.4
degree. If it is not possible to change the survey sensor to
provide a zero misalignment, the survey correction module
can be performed to obtain the correct inclination. When
performing the survey correction module, the original survey
data and tool setting parameters (such as tool face) are
imported into the MaxBHA program. The correct inclinations
(shown in blue in Fig. 8) are obtained by performing the force
analysis at each survey station. The correct survey data then is
used to re-calculate the TVD. 20 feet of TVD error from the
BHA misalignment in a typical designed well is not
uncommon and sometimes the error could be even greater.
Thus, it is imperative that the implications of the BHA
misalignment be fully understood by the industry.
Avoid critical rotary speeds. Running the critical rotary
speeds of the BHA creates severe vibrations, resulting in
BHA/bit and MWD tool failures. The critical speed
calculation module (the WHIRL software) is integrated in
the MaxBHA program and does not require users to enter any
new data. Fig. 9 shows the critical rotary speeds for this BHA
to be at 14, 89, 163, and 199 RPM. Avoiding the critical
speeds was a major contributor for a very low vibration
environment.
Case #3: Optimizing Steerable Motor System in a ShortRadius Well. A 6-1/8-in. slim hole was drilled with a 4 -in.
short-radius motor system with curves up to 60 degrees. After
the hole was drilled, a conventional 4 -in. motor system with
LWD (see Fig. 10) was run in the curves to drill the lateral
section. The objective of the BHA modeling was to study and
optimize the 4 -in. BHA so that it can be pushed through the
short radius curves without bending failure. Force analysis
was used to optimize the BHA through minimizing the side
forces and bending moment (see Fig. 11).
Because a substantial length of collars are in contact with
the wellbore wall in short-radius wells, the challenges in BHA
modeling are to find the correct boundary conditions and to
reach a solution for the severely bent BHA. The new BHA
program shows excellent performances in both tasks.

SPE/IADC 106935

3. Direct comparisons between the new code and the finite-

Case #4 Verifying the Dynamic BHA Model (WHIRL)

with MWD Vibration Sensor. A rotary assembly with a
1417.5 in. bi-centered bit with two 14 in. stabilizers was
used to drill a vertical section in the GOM. Rotary speed was
operated close to and away from critical RPMs (CRPMs) to
assess validity of the model using a commercial MWD
vibration sensor10. Good correlation between CRPM and the
increased lateral vibrations were seen as shown in Fig. 12.
This time-based plot shows drilling parameters, including
rotary speed, in the 4th column along with calculated CRPMs
#3 to #5. In the 5th column, accelerations (in gs) show the peak
measurements of each of the three accelerometers of the
downhole vibration sensor. These curves are shaded according
to the severity of vibration present: green for low, yellow for
medium, and red for high severity. According to
recommended practice, a drilling assembly should never be
operated under high severity vibration because this commonly
results in component failure. During the period between 01:00
and 01:10, the rotary RPM had been reduced until at
approximately 105 RPM, it was close to the 4th CRPM. This
triggered high severity vibrations, as indicated by the vibration
data.
At 01:10, the rotary speed was increased to 130 RPM,
placing it between the 4th and 5th CRPMs, (a safe zone) and
the downhole vibration reduced immediately. Note that the
normal recommendation in the event of high severity vibration
is to decrease the RPM. If they had done so, not only could
vibration have increased by triggering the 3rd critical RPM, but
also the ROP would have been reduced. In this case, according
to the WHIRL software prediction, the RPM was increased to
remove the resonant excitation and the vibration stopped.
To further verify the predictions of the WHIRL software
model, the rotary speed was increased at 01:17 to
approximately 155 RPM, close to the 5th CRPM. As before,
there was an increase in vibration severity as a result of
resonance in the BHA, which continued until the RPM was
reduced at 01:32. This time the reduction in RPM did not
completely cure the vibration, only reduce its severity. The
downhole shocks only reduced to a benign level after the RPM
was further reduced at 01:48 when the string was picked up to
back-ream before making a connection. This observation
proves that it is sometimes necessary to stop drilling to fully
eliminate resonant vibration.
Summary and Conclusions
1. An integrated static and dynamic model has been

developed. The objectives of the new program include

providing the greatest accuracy possible, flexibility to
handle any BHA configurations, and computational
efficiency to run on laptop computers with a user-friendly
interface.
2. A new algorithm, based on solving the Lubinskis
equations directly, has been developed for the static BHA
model. Because of its use of closed-form equations, the
algorithm is believed to be the most accurate program in
the industry. The algorithm is also capable of handling any
complex BHA configuration.

element based BHA programs against the field data

confirm the improved accuracy, particularly of the new
code when modeling steerable motors and rotary steerable
systems.
4. A steady-state dynamic BHA model using a hybrid of
analytical and finite-element method has been developed
to predict the critical speeds of the BHA. The dynamic
BHA model has been incorporated into the BHA program
to allow users performing the static and dynamic BHA
analyses in the same application.
5. Running at one of the critical speeds increases the chance
of high vibrations resulting in downhole tool and bit
failures. Good correlation between the critical rotary speed
predictions and the onset of BHA and bit whirl confirm its
importance as well as the accuracy of the model.
6. A wide range of modules that are based on the new
algorithm have been developed in the new BHA program.
The new program has enabled engineers to optimize the
BHA design more efficiently for a maximizing drilling
performance.
Acknowledgements
The author would like to thank Mr. Min Wu for the
contribution to this project and Halliburton Energy Services
for their support and for their permission to publish this paper.
References
1. Lubinski, A., and Woods, H.B., Factors Affecting the Angle of
Inclination and Doglegging in Rotary Bore Holes, API Drilling
and Production Practice, 1953, 222-250.
2. Lubinski, A., and Woods, H.B., Use of Stabilizers in Controlling
Hole Deviation, API Drilling and Production Practice, 1953,
165-182.
3. Millheim, K.,Behavior of Multiple Stabilizer Bottom Hole
Assemblies, Oil and Gas Journal, January, 1979.
4. Walker, B.H. and Friedman, M.B., Three-Dimensional Force
and Deflection Analysis of a Variable Cross-Section Drillstring,
Journal of Pressure Vessel Technology, May, 1977.
5. Callas, N.P., and Callas, R.L., Boundary Value Problem is
solved, Oil and Gas Journal, December, 1980.
6. Ho, H-S., General Formulation of Drillstring under Large
Deformation and Its use in BHA Analysis, SPE paper #15562,
presented at the 61st SPE Annual Technical Conference and
Exhibition in New Orleans, 1986.
7. Jogi, P.N., Burgess, T.M., and Bowling, J.P., ThreeDimensional Bottomhole Assembly Improves Directional
Drilling, 1986 IADC/SPE #14768, presented at the 1986
IADC/SPE Drilling Conference in Dallas.
8. Chen, D. C-K, and Wu, M., Maximizing Drilling Performance
with State-of-the-Art BHA Program SPE paper #104502,
presented at the 2007 SPE/IADC Drilling Conference in
Amsterdam, February, 2007.
9. Yonezawa, T., et al., Robotic Controlled Drilling: A New
Rotary Steerable Drilling System for the Oil and Gas Industry,
SPE/IADC paper #74458, presented at the 2002 SPE/IADC
Drilling Conference in Dallas, February, 2002.
10. Zannoni, S.A., Cheatham, C.A., Chen, D. C-K., and Golla, C.A.,
Development and Field Testing of a New Downhole MWD
Drillstring Dynamics Sensor, SPE paper #26341 presented at the
1993 SPE Annual Technical Conference and Exhibition in
Houston.

SPE/IADC 106935

Fig. 1 BHA configuration and force analysis results for field study case #2.

Fig. 2 The WHIRL software (dynamic model) theory is based on modeling the BHA as a continuous beam. The static
model is used to predict the upper contact point (the tangential point), as well as the contacts at stabilizers. The finite
element method then is used to calculate the critical speeds and mode shapes of the continuous beam. Thus accurate
static BHA modeling is essential for the dynamic BHA model.

SPE/IADC 106935

Comparisons between Model Results and Field Data

14

12

Max Build Rate (deg/100')

10

2
FE based program

20

MaxBHA

40

Field Data

60

80

0
100

Inclination (deg)

Fig. 3 Max build rate predictions (deg/100-ft) vs. field data for a 8 -in. hole size steerable motor system.

Comparisons between Model Results and Field Data

7

Max Build Rate (deg/100')

1
FE based program

20

MaxBHA

40

60

Field Data

80

0
100

Inclination (deg)

Fig. 4 Max build rate prediction (deg/100-ft) vs. field data for an 8 -in. hole size point-the-bit RSS.

SPE/IADC 106935

Fig. 5 Sensitivity analyses on DSL vs. stabilizer gauges allows engineer to maximize DLS for kick-off. The results
indicate that maximum DLS is created with 8.1 gauge at the stabilizer #3 and full gauge at the stabilizer #4.

SPE/IADC 106935

Fig. 6 Sensitivity analyses on bending moment and DSL vs. the length of flex sub for two different inclinations. The
results allow engineer to optimize the flex sub design to minimize bending moment at the RSS tool without sacrificing too
much on the DLS capability.

SPE/IADC 106935

Fig. 7 Force analysis provides the slope (BHA misalignment) data allows engineer to see how it changes with different
BHA design and varies along the BHA. The data also is used to correct the survey error.

Fig. 8 In survey calculation mode, the correct inclinations (shown in blue) are calculated by performing the force
analysis at each survey station. The correct survey data then is used to re-calculate the TVD.

10

SPE/IADC 106935

Fig. 9 WHIRL analysis (dynamic model) provides critical speeds of the BHA (14, 89, 163, and 199 RPM) to be avoided.

Fig. 10 BHA model for running a steerable motor through 60 deg/100 DLS. The challenges in BHA modeling are to find
the correct boundary conditions and to find the solution for the severely bent BHA as a substantial length of collars are in
contact with the wellbore wall.

Fig. 11 Force analysis indicates the bending moment along the BHA.

SPE/IADC 106935

11

Fig. 12 Time-based drilling parameters and MWD vibration data were compared to critical RPM predictions (CRPM) from the
WHIRL software. The results show that critical rotary speeds correlated well with the downhole vibrations, which, in turn, validates
the accuracy of the software.