Vous êtes sur la page 1sur 14

Journal of Petroleum Science and Engineering 116 (2014) 36–49

Contents lists available at ScienceDirect

Journal of Petroleum Science and Engineering


journal homepage: www.elsevier.com/locate/petrol

Analysis of multi-mode nonlinear coupled axial-transverse drillstring


vibration in vibration assisted rotary drilling
Ahmad Ghasemloonia n, D. Geoff Rideout 1, Stephen D. Butt 2
Advanced Drilling Group, Faculty of Engineering, Memorial University of Newfoundland, St. John's, NL, Canada, A1B 3X5

art ic l e i nf o a b s t r a c t

Article history: Unwanted vibration modes of a drillstring can result in inefficient drilling, and damage to the
Received 5 August 2012 drillstring, bit, BHA components, MWD tools and mud motors. Bottom-hole assembly (BHA) config-
Accepted 18 February 2014 uration design, shock sub parameter tuning, and establishing drilling parameters such as rotary speed
Available online 26 February 2014
and weigh-on-bit can be improved using computer simulation of a drillstring and its vibration modes.
Keywords: Drilling tools are under development to apply axial vibrations for the purpose of overcoming
drillstring model drillstring-wellbore friction, facilitating cutting removal and improving the rate of penetration (ROP)
vibration-assisted drilling of the bit. Predicting the effects (both desired and undesired) of such axial vibration generator tools is
coupled nonlinear axial-transverse vibration becoming increasingly important to industry. In this paper, the coupled nonlinear axial-transverse
multi-span BHA
dynamics of the entire drillstring are modeled and lateral instabilities are qualitatively studied. The
Bypassing PDE's
drillstring includes the pipes, a multi-span BHA with shock sub, and a force generator tool near the bit.
wellbore contact
multi-mode analysis The multi-span BHA model enables more accurate natural frequency prediction and multi-mode
contact analysis of the drillstring and wellbore. The governing equations are obtained using the
“Bypassing PDE's” method with the expanded Galerkin's method, which enables finding the symbolic
solution of the governing equations. The effects of mud damping, driving torque, and spatially varying
axial load are also included, along with nonlinearities due to geometry, axial stiffening, strain energy
and Hertzian contact forces. Simulations reveal resonant frequencies and show the relative severity of
the contact in each span of the BHA. The model features fast running time compared to a high-order
finite element model against which it is validated.
& 2014 Elsevier B.V. All rights reserved.

1. Introduction Several recent classes of drilling tools apply axial vibration to


the drillstring for the purpose of reducing drillstring-wellbore
Drillstring vibration is a primary cause of premature failure of friction (Barton et al., 2011), enhancing penetration mechanism
drillstring components, deterioration of the well trajectory, bit and (Manko et al., 2003; Babatunde et al., 2011) and facilitating cutting
stabilizer wear, lower penetration rate, deteriorating accuracy of removal. Vibration drilling may transmit power to the bit more
measurement while drilling (MWD) systems and decreased effi- efficiently than rotary drilling. Vibration tools improve drilling
ciency. Bottom-hole assembly (BHA) design and control strategies performance by various means, and collectively for the purpose of
to reduce unwanted drillstring vibration require enhanced this investigation their use is called “vibration assisted rotary
dynamic models that can reveal the modal characteristics and drilling” (VARD).
dynamic time response of the entire drillstring. This study is Introduction of vibration in rotary drilling must be done care-
motivated by two factors: (1) the need to extend existing models fully. The implementation of a VARD force above the bit excites the
to include more realistic effects such as multiple-span BHA's and drillstring axially, and as a result of coupling effects, the lateral
nonlinearities and (2) the increasing trend towards downhole mode will also be excited. Therefore, the vibration behavior of the
tools that generate intentional axial vibration. drillstring under the effect of axial force generators such as jars,
agitators, and higher-frequency tools for ROP enhancement in
rotary drilling must be predicted with accurate models. The model
n
Corresponding author. Tel.: þ 1 709 8642573; fax: þ 1 709 8648975. should incorporate drilling parameters such as weight-on-bit
E-mail addresses: a.ghasemloonia@mun.ca (A. Ghasemloonia), (WOB), rotary speed, and torque; should be able to predict natural
g.rideout@mun.ca (D. Geoff Rideout), sdbutt@mun.ca (S.D. Butt).
1 frequencies for realistic BHA configurations, capture coupled axial
Tel.: þ1 709 8643746; fax: þ 1 709 8648975.
2
Tel.: þ1 709 8648955; fax: þ1 709 8648975. and lateral vibrations, incorporate VARD tools, and manage the

http://dx.doi.org/10.1016/j.petrol.2014.02.014
0920-4105 & 2014 Elsevier B.V. All rights reserved.
A. Ghasemloonia et al. / Journal of Petroleum Science and Engineering 116 (2014) 36–49 37

trade-off between accuracy and computation time. To complement in the equations, although it adds complexity. Since the rotation
high-order finite element models that may require extremely long speed of the drillstring is small (50–150 rpm in practice), the
simulation times, the authors have developed an analytical model gyroscopic effect may be negligible (Chen and Geradin, 1995; Yigit
from which a symbolic set of partial differential equations is and Christoforou, 1996; Heisig and Neubert, 2000; Hakimi and
generated, and then solved symbolically. The model presented in Moradi, 2009; Ghasemloonia et al., 2012, 2013). Current literature
this paper runs very quickly on a desktop personal computer, and reports no studies of the effect of VARD force generators on the
allows rapid sensitivity studies of things such as the effect of shock nonlinear coupled axial-transverse vibration of the drillstring,
sub stiffness, and VARD tool force amplitude and frequency, on either in the frequency or the time domain. The model described
vibration along the entire drillstring. The effects of mud damping, herein has the important benefits of providing accurate natural
driving torque, and spatially varying axial load are also included, frequencies, generating time domain response, and capturing
along with nonlinearities due to geometry, axial stiffening, strain wellbore contact with greater fidelity than simpler single-span
energy and Hertzian contact forces. BHA models.
The models are developed for vertical wellbores. Although the Determining natural frequencies is important because, from an
application of horizontal and directional drilling is increasing in operational standpoint, vibration severity can be reduced if rota-
low permeability formations, each deviated wellbore begins with a tion speed is kept away from these frequencies (Dareing, 1984a;
vertical section (vertical pilot well) which might be through a Gulyaev et al., 2007). In the last decades, several studies have been
difficult formation. Deviation from vertical to horizontal can take conducted to investigate the lateral natural frequencies of the
up to 2000 ft. The model which is developed in this study is useful drillstring or BHA. Approaches and simplifications include simple
for analysis and design of either vertical well drillstrings, or for the non-rotating beam models (Dareing, 1984a, 1984b), and single-
drillstrings of vertical sections of horizontal and deviated wells. span BHA's (differential quadrature method of Hakimi and Moradi,
The following section reviews relevant literature about importance 2009; transfer matrix method of Chen and Geradin, 1995;
of various vibration modes and physical phenomena associated undamped finite element model of Khulief and Al-Naser, 2005
with the drillstring; modeling of wellbore contact, and methods with no wall contact). While Gulyaev et al. (2007) included two
for equation derivation and solution. Section 3 derives governing orthogonal lateral modes and mud internal flow, they neglected
equations, including eigenfrequencies and eigenfunctions of a the contact, axial mode and damping effect. Lateral natural
multi-span BHA which are implemented using a Lagrangian frequencies based on buckling analysis for different BHA lengths
approach. Section 4 provides numerical simulation results show- from Gulyaev et al. (2009) were much lower than values measured
ing the ability of the model to predict axial and lateral response of in the field (Khulief et al., 2008), suggesting the need for extended
a drillstring with a VARD force generator. The new model models. The effect of fluid damping, added fluid mass, stabilizer
described in this paper is validated with a high-order finite clearance and the friction coefficient on the critical rotary fre-
element model. Sections 5 and 6 give discussion, conclusions quencies was investigated by Jansen (1991), but the effects of
and future research directions. torque, gravity and axial-lateral coupling were neglected and the
implementation of these terms was suggested for future studies.
In the present paper, the axial and orthogonal lateral resonance
2. Literature review frequencies of a drillstring, assuming a multi-span BHA, will be
calculated. The first four modes will be retained for the two
Drillstring vibration is not simply independent axial, torsional orthogonal lateral directions and one axial direction (12 general-
and lateral vibration. A typical drillstring vibrates in 3 major ized coordinate systems in Lagrange's equations). After imple-
coupled modes: lateral-axial, lateral-torsional and axial-torsional. menting the solution method, the FFT analysis will be applied on
Bit bouncing, stick-slip and whirling are extreme examples of each generalized coordinate system to extract the first four natural
coupled vibration dominated by axial, torsional and lateral frequencies in each direction. Considering a multi-span BHA will
motions respectively. Among these coupled modes, the coupled give results that are more accurate than those from a single span
transverse mode is a major cause of drillstring failures (Chin, 1994; BHA model. Moreover, the resonance frequencies which are
Sotomayor et al., 1997; Spanos et al., 2002; Ghasemloonia et al., extracted from the developed nonlinear model are more precise
2012, 2013) and wellbore washout which happens at low frequen- compared to the ones which are derived from idealized linear
cies. The deteriorating effect of the orthogonal lateral modes could model. In fact, instabilities may occur at rotary speeds which may
be explained through the wave speed phenomenon. Bending not be considered critical from an uncoupled linear analysis (Yigit
waves are not propagated to the surface via the drillstring as are et al., 1998).
torsional and longitudinal waves, due to the difference in the wave In addition to natural frequencies, the axial or lateral time
speed for different types of modes. The propagation speed for axial responses of the drillstring or BHA are desirable. FEM and modal
and torsional motions is quite high compared to the lateral motion analysis are two major methods used to predict the time response
(Chin, 1994). Therefore, there could be severe bending vibrations of the drillstring. Yigit and Christoforou (1996) investigated the
deep in the hole, which the surface measuring tools do not detect. axial-transverse behavior of the non-rotating BHA and verified
BHA–wellbore contact is the main excitation source for lateral nonlinear axial-lateral coupling due to nonlinear strain. Mud
vibration. In the case of VARD drilling, axial vibration plays an damping, rotation and the other orthogonal lateral modes were
important role since the high frequency VARD force directly not considered in their model. They implemented a force mode in
excites the axial modes and lateral instabilities may exist due to their one-mode approximation assumed mode method to accel-
axial excitation (Berlioz et al., 1996). Therefore, in order to erate the convergence rate. Since the axial load in the BHA was
precisely model the VARD drillstring, two orthogonal coupled assumed constant in their study, one static axial deformation
transverse modes along with the axial mode will be considered mode was added to the assumed mode approximation. Spanos
in this study. In non-VARD applications, bit-rock interaction will et al. (1997) addressed the effect of contact and added fluid mass
lead to axial vibrations near the bit, which could excite axial generating the deformed shape of the BHA in the lateral modes,
modes. Natural frequencies in these directions are also influenced based on natural mode analysis. Based on the transfer function of
by the driving torque (Yigit and Christoforou, 1996, 1998; Gulyaev the BHA lateral vibration in the single orthogonal plane, frequency
et al., 2009; Liao et al., 2011; Ghasemloonia et al., 2012, 2013). The and mud density-dependent damping was proposed. Torque, axial
torque couples the two orthogonal lateral modes and must be kept force and axial displacement were not included.
38 A. Ghasemloonia et al. / Journal of Petroleum Science and Engineering 116 (2014) 36–49

Li et al. (2007) investigated the important issue of the bit-rock approach. The fast convergence rate without the use of a number
boundary condition, which must be treated as a source of either a of force modes to accelerate the convergence, as compared to
displacement or a force when constructing a model. His simple other numerical models (Yigit and Christoforou 1996), enables
beam model results for axial vibration, when compared with field sensitivity analysis for each controllable parameter in the model
data, suggested that an input displacement at the bit is the with the selected solution scheme. Another advantage of the
appropriate excitation method. “Bypassing PDE's” is the use of conventional energy terms rather
Due to the higher stiffness, higher mass, and lower natural than the variational form of the energy equations, which further
frequencies of the BHA compared to the pipe section, the vibration simplifies numerical solution of the developed model at the final
behavior of the drillstring is strongly influenced by BHA vibration, step. Outcomes of the model and analysis method of this paper
especially vibration resulting from contact with the wellbore. include multi-mode analysis of the equations (up to the fourth
Modeling the impact is a crucial task to precisely evaluate the mode), coupled lateral equations in two orthogonal planes along
lateral dynamic response. Modeling the contact behavior of the with the axial motion, symbolic model solution, and inclusion of
drillstring has been approached in different ways by various all interacting forces on the drillstring. The developed model can
researchers. Hakimi and Moradi (2009) modeled drillstring- be numerically solved to predict modal characteristics and evalu-
wellbore contact as a series of springs between the drillstring ate dynamic response analysis of the entire drillstring, including
and the wellbore with a constant stiffness. Khulief et al. (2008) multiple contact points on the BHA.
implemented a continuous force-displacement law at the contact
point in their multi-body FEM model for axial-bending and
torsional-bending. Jansen (1991) modeled the contact point of a
3. Derivation of governing equations
rotating drillstring as a two DOF lumped element model in two
orthogonal transverse planes. Coulomb friction and nonlinear mud
The drillstring is a beamlike structure with a high slenderness
drag force was assumed at the contact point. Liao et al. (2011)
ratio. In this study, the structure is under the effect of gravity, mud
developed a reduced order FEM model at the contact point of the
hydrostatic force, contact with the wellbore, mud damping and the
drillstring and wellbore. The effect of mud damping on the lateral
VARD force. The Euler–Bernoulli beam theory is used to model
motion at the contact point was neglected and the model was only
coupled lateral-axial vibration of the drillstring, which is consid-
capable of predicting lateral motion at the contact point.
ered as a beam with a high aspect ratio (Beck and da Silva, 2010).
Christoforou and Yigit (1997) modeled the lateral behavior of the
The drillstring in this study includes a three span BHA (with a
drillstring at the contact point for parametric resonance studies.
different length for each span) with a long pipe section. The
Hamilton's principle was implemented to derive the equations
driving torque is applied on the rotary table and the VARD force
assuming mud damping and constant axial force along the drill-
generator is applied on the BHA. The drillstring is inside a wellbore
string for the axial-transverse coupling. A single-mode assumed
filled with mud. Two orthogonal lateral directions, u and v; and
mode approximation was used just for the BHA with a Hertzian
axial direction w are assumed. A schematic diagram of the drill-
contact force at the contact point. Impact was assumed at mid-
string is shown in Fig. 1.
span of a single-span BHA, and a multi-span BHA analysis was
The effects of the hook load, WOB, mud hydrostatic force and
recommended to achieve more accurate results.
self weight are presented as a spatially varying axial force along
In the present study, the Hertzian contact theory is implemen-
the drillstring. The buoyant force in the drillstring is derived based
ted at the contact points of multiple spans. The Dirac delta
on effective tension (Mitchell and Miska, 2011). At the last point of
function of the radial deflection is applied to ensure that at the
the collar section, there are two axial upward forces, namely the
contact time (when the radial deflection exceeds the borehole
clearance), the impact force will be applied to the multiple contact
points. Successive contact or single impacts can be accurately
predicted. The effect of friction (rolling friction) at the contact
point is not considered in this study. When the radial deflection
does not exceed the borehole clearance, the Dirac delta function
will be calculated automatically in the analytical code and no
impulsive force will be applied in no-contact instants.
The choice of equation implementation and solution method
determines the extent to which symbolic variables can be used,
and the ease with which nonlinearity can be included. Contact
with the wellbore, mud damping, coupled transverse modes via
torque, nonlinear strain energy and axial stiffening are the major
nonlinear contributions to drillstring vibrations assumed in this
study. Khulief and Al-Naser (2005) implemented the Lagrangian
approach with the finite element method to study lateral instabil-
ities. The gyroscopic effect, torsional-bending coupling and grav-
itational field were considered, but drillstring–wellbore contact
was not. The model developed in this study enables generation of
the axial and orthogonal lateral–lateral multi-mode time response
analysis of any desired point on the entire drillstring, including
multiple contact points on the multi-span BHA in presence of the
VARD tool. Lagrange's equations for the continuous drillstring with
contact are developed and solved using the “Bypassing PDE's”
method. This method, which has been proven accurate for non-
linear problems (Thomsen, 2003), is based on combining the
expanded Galerkin's technique with the Lagrange's equation for
continuous structures, instead of the conventional Hamiltonian Fig. 1. Schematic of the multi-span drillstring under the effect of the VARD tool.
A. Ghasemloonia et al. / Journal of Petroleum Science and Engineering 116 (2014) 36–49 39

WOB and the hydrostatic force at the lower cross section. At the compression fields on the collar section. The quadratic nonlinear
neutral point (intersection of the pipes and collars) the axial term retained in the equation is due to geometric nonlinearity. The
compression in the collars changes to tension in the pipe section. work done by the driving torque can be expressed as (Simitses and
The varying axial force in the collar section is Hodges, 2005)
F collar ¼ ρcollar Acollar gz  WOB  ρmud glAcollar ð1Þ Z l1   
1 ∂2 ∂
W torque ¼ T uðz; tÞ vðz; tÞ
The total length of the drillstring is l in the above equation and the 2 0 ∂z 2 ∂z
reference point is assumed at the bottom of the collar section. The  2   !
force in the pipe section can be expressed as ∂ ∂
þ vðz; tÞ uðz; tÞ dz ð5Þ
∂z2 ∂z
F pipe ¼ F hook  ρpipe Apipe gðl zÞ ð2Þ

This force is depicted in Fig. 2 for a drillstring with 800 m pipe The torque appears as the third order derivative of u in the v
section and a 60 m collar section. direction and vise versa. Therefore, torque causes coupling
The Bypassing PDE's method is implemented to derive the between two orthogonal transverse directions. The work done
governing equations. This method is based on using the Lagrange's by a VARD force, if applicable, is
equation with the expanded Galerkin's method, instead of the con- Z  2  2 !
1 l1 ∂ ∂
ventional Hamilton's approach for continuous systems. This method W VARD  f orce ¼  F VARD sin ðωtÞ uðz; tÞ þ vðz; tÞ dz
2 0 ∂z ∂z
has shown accurate results for nonlinear problems. The expanded
Galerkin's method will be used in the first step of the energy ð6Þ
equations. Therefore, the variational form of the energy equations is
where ω is the excitation frequency of the VARD force generator.
not required. In the next section, the energy terms for the drillstring
The VARD force is assumed sinusoidal. Any span on the collar
are derived, to be used in combination with this method.
section is under a spatially varying axial force as stated in Eq. (1).
The energy term due to the compressive axial force can be
3.1. Energy equations for the BHA and the pipe sections
expressed as
Z  2  2 !
The kinetic energy for the first span is 1 l1 ∂ ∂
W axial  f orce ¼  F collar uðz; tÞ þ vðz; tÞ dz ð7Þ
Z  2  2  2 ! 2 0 ∂z ∂z
1 l1 ∂ ∂ ∂
K energy ¼ ρAcollar uðz; tÞ þ vðz; tÞ þ wðz; tÞ dz
2 0 ∂t ∂t ∂t
The mud damping force as a result of hydrostatic drag is a
ð3Þ quadratic velocity-related force (Jansen, 1991). The dissipated
The strain energy due to axial and lateral deformations is energy of this force is
0 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Z  2 2  2 2 ! Z l1  2  2
1 l1 ∂ ∂ ∂ ∂
P energy ¼ EI collar uðz; tÞ þ vðz; tÞ W mud  damping ¼ ρmud C D Rcollar @ uðz; tÞ þ vðz; tÞ
2 0 ∂z2 ∂z2 0 ∂t ∂t
 2  2 !! 1
∂ 1 ∂ 1 ∂  
þ EAcollar wðz; tÞ þ uðz; tÞ þ vðz; tÞ dz ∂ ∂
∂z 2 ∂z 2 ∂z uðz; tÞ þ vðz; tÞ dzA ð8Þ
∂t ∂t
ð4Þ
The first term represents the elastic stiffening, while the second The contact energy in the first span is approximated based on
term captures axial stiffening due to the gravitational field the Hertzian contact theory using a piecewise function (Yigit
(EAð∂w=∂zÞ represents the gravitational force) and shows the and Christoforou, 1996). In the following equation, bcl is the
coupling between the axial and flexural deformations. The non- borehole clearance and K h is the contact stiffness which is
linear axial stiffening term accounts for the stiffening effect of the related to the material and geometry at the contact point. r is
tension field over the pipe section and softening effect of the the radial displacement which is related to two orthogonal lateral

Fig. 2. Spatially varying axial force along an 860 m drillstring.


40 A. Ghasemloonia et al. / Journal of Petroleum Science and Engineering 116 (2014) 36–49

deflections: Z l2  2  2 !
1 ∂ ∂
! þ F VARD sin ðωtÞ uðz; tÞ þ vðz; tÞ dz
2 ∂z ∂z
K h ðr  bcl Þ3=2 bcl rjrj l1
contact energy ¼  r ð9Þ  2
0 otherwise Z l2
1 ∂
þ ðρAcollar gz  WOB  ρmud glAcollar Þ uðz; tÞ
The other two spans of the BHA are under the same effects. Thus 2 l1 ∂z
the energy terms are the same, except for a change in the  2 !

integration limits, namely l1  l2 for the mid span and l2  l3 for þ vðz; tÞ dz ð13Þ
∂z
the top span. The contact locations are different in each span. The
equations for the other two spans are not shown here due to space
limitations. Lagrangian span 3
The first few lateral vibration modes in the lower frequencies Z  2  2  2 !
1 l3 ∂ ∂ ∂
will not be excited in the pipe section (Hakimi and Moradi, 2009). ¼ ρAcollar uðz; tÞ þ vðz; tÞ þ wðz; tÞ dz
2 l2 ∂t ∂t ∂t
Therefore, the pipe is assumed to undergo axial vibration only. The
kinetic and potential energy terms of the pipe section are Z l3  2  2 !
1 ∂2 ∂2
Z  2 Z  2  EI collar uðz; tÞ þ vðz; tÞ
1 l4
∂ 1 l4 ∂ 2 l2 ∂z2 ∂z2
energy of pipe ¼ ρApipe wðz; tÞ dz þ EApipe wðz; tÞ dz
2 l3 ∂t 2 l3 ∂z  2  2 !!
ð10Þ ∂ 1 ∂ 1 ∂
þEAcollar wðz; tÞ þ uðz; tÞ þ vðz; tÞ dz
∂z 2 ∂z 2 ∂z
The energy of the VARD force and of the tensile axial load in the
Z   
pipe section are 1 l3
∂2 ∂
Z    T uðz; tÞ vðz; tÞ
1 l4 ∂ 2 l2 ∂z2 ∂z
WpipeV ARD  f orce ¼ F VARD sin ðωtÞ wðz; tÞ dz
2 l3 ∂z    !
Z   ∂ 2

1 l4 ∂ þ vðz; tÞ uðz; tÞ dz
Wpipeaxial  f orce ¼ ðF pipe Þ wðz; tÞ dz ð11Þ ∂z2 ∂z
2 l3 ∂z
Z l3  2  2 !
1 ∂ ∂
The Lagrangian of the three-span BHA and the pipe section are as þ F VARD sin ðωtÞ uðz; tÞ þ vðz; tÞ dz
2 l2 ∂z ∂z
below:
Z l3  2
Lagrangian span 1 1 ∂
Z  2  2  2 ! þ
2
ðρAcollar gz  WOB  ρmud glAcollar Þ
∂z
uðz; tÞ
1 l1 ∂ ∂ ∂ l2
¼ ρAcollar uðz; tÞ þ vðz; tÞ þ wðz; tÞ dz
2 0 ∂t ∂t ∂t  2 !

Z  2 2  2 2 ! þ vðz; tÞ dz ð14Þ
1 l1 ∂ ∂ ∂z
 EI collar uðz; tÞ þ vðz; tÞ
2 0 ∂z2 ∂z2
 2  2 !! Lagrangian pipe
∂ 1 ∂ 1 ∂
þ EAcollar wðz; tÞ þ uðz; tÞ þ vðz; tÞ dz Z  2 Z  2
∂z 2 ∂z 2 ∂z 1 l4 ∂ 1 l4 ∂
¼ ρst Apipe wðz; tÞ dz  EApipe wðz; tÞ dz
Z l1  2   2 l3 ∂t 2 l3 ∂z
1 ∂ ∂
 T uðz; tÞ vðz; tÞ Z l4  
2 0 ∂z2 ∂z 1 ∂
 F VARD sin ðωtÞ wðz; tÞ dz
 2   ! 2 l3 ∂z
∂ ∂
þ vðz; tÞ uðz; tÞ dz Z  
∂z2 ∂z 1 l4 ∂
 2  2 !  ðF H  ρst Apipe gðl  zÞÞ wðz; tÞ dz ð15Þ
Z 2 l3 ∂z
1 l1 ∂ ∂
þ F V ARD sin ðωtÞ uðz; tÞ þ vðz; tÞ dz
2 0 ∂z ∂z
At this step of the Bypassing PDE's method, the expanded
Z  2 Galerkin's method is applied to the equations. So, u, v and w are
1 l1 ∂
þ ðρAcollar gz  WOB  ρmud glAcollar Þ uðz; tÞ assumed as comparison functions multiplied by mode participa-
2 0 ∂z
 2 ! tion factors. Since the comparison functions for each span are
∂ different, the Lagrangian should be assumed for each span sepa-
þ vðz; tÞ dz ð12Þ
∂z rately and after substituting the comparison functions, they will be
added together as the total Lagrangian of the system. Thus, u, v
Lagrangian span 2 and w could be expressed separately for each span. Therefore, for
Z  2  2  2 ! the first span:
1 l2 ∂ ∂ ∂
¼ ρAcollar uðz; tÞ þ vðz; tÞ þ wðz; tÞ dz
2 l1 ∂t ∂t ∂t 4
Z  2 2  2 2 ! wðz; tÞ ¼ ∑ χ r ðzÞ U pr ðtÞ
1 l2 ∂ ∂ r¼1
 EI collar uðz; tÞ þ vðz; tÞ
2 l1 ∂z2 ∂z2 4
uðz; tÞ ¼ ∑ φr ðzÞ Uηr ðtÞ ð16Þ
 2  2 !! r¼1
∂ 1 ∂ 1 ∂
þ EAcollar wðz; t Þ þ uðz; tÞ þ vðz; tÞ dz 4
∂z 2 ∂z 2 ∂z vðz; tÞ ¼ ∑ φr ðzÞ Uλr ðtÞ
Z l2  2   r¼1
1 ∂ ∂
 T uðz; tÞ vðz; tÞ
2 l1 ∂z2 ∂z The subscript r depends on the desired mode shapes according to
 2   ! the frequency range of interest. For this problem the first four
∂ ∂
þ vðz; tÞ uðz; tÞ dz modes will be retained to conduct the multi-mode analysis. The
∂z2 ∂z
expanded Galerkin's method for the second span is
A. Ghasemloonia et al. / Journal of Petroleum Science and Engineering 116 (2014) 36–49 41

4 There are 12 unknowns in the equations. Six boundary conditions


wðz; tÞ ¼ ∑ χ r ðzÞ Upr ðtÞ
r¼1
are zero deflections at the supports, while 2 boundary conditions
4 are zero bending moments at both ends. The remaining four
uðz; tÞ ¼ ∑ ψ r ðzÞ U ηr ðtÞ ð17Þ boundary conditions are slope and bending compatibility equa-
r¼1
tions at the two middle supports. For the nontrivial solution of the
4
vðz; tÞ ¼ ∑ ψ r ðzÞ U λr ðtÞ system of equations, the determinant of the coefficient matrix is
r¼1 set to zero. The result is the frequency equation:
  
For the last span of the BHA it is 1 sinhð10βÞ
 2 sin ð10βÞ cos ð10βÞsinhð10βÞ
sin ð10βÞsinhð10βÞ sin ð10βÞ
4
wðz; tÞ ¼ ∑ χ r ðzÞ Upr ðtÞ 3 cos ð10βÞsinh ð10βÞ2
r¼1  
sinhð10βÞ
4 þ sin ð10βÞ2 coshð10βÞ
uðz; tÞ ¼ ∑ θr ðzÞ Uηr ðtÞ ð18Þ sin ð10βÞ
r¼1
þ 2 sin ð10βÞcoshð10βÞsinhð10βÞ
4  
vðz; tÞ ¼ ∑ θr ðzÞ U λr ðtÞ sinhð10βÞ
r¼1 þ sin ð10βÞsinhð10βÞ coshð10βÞ
sin ð10βÞ
In the above expressions χ; φ; ψ and θ are comparison functions for þ sin ð10βÞ cos ð10βÞsinhð10βÞ
 
axial and orthogonal lateral motions of the first, second and last cos ð10βÞ  coshð10βÞ
þ sin ð10βÞsinh ð10βÞ2
span of the BHA, respectively. pr ðtÞ; ηr ðtÞ and λr ðtÞ are mode parti- sin ð10βÞ
 
cipation factors for axial motion (w), and lateral motions (u) and cos ð10βÞ  coshð10βÞ
 sin ð10βÞ2 sinhð10βÞ ¼0 ð21Þ
(v), respectively. The boundary condition for the axial motion of sin ð10βÞ
the drillstring is assumed as fixed at the top and free at the
bottom, and the spans are assumed as pinned-pinned boundary The equation was solved numerically using the Newton–Raphson
conditions (location of the stabilizers) in the lateral direction. algorithm. The first four values for β are 7.171, 12.57, 13.77 and
Jogi et al. (2002) verified the natural frequency analysis of several 16.64. The values of β will be substituted in 12 equations (these 12
modeling packages with field results. They proved that the equations are the results of applying 12 boundary conditions to
simplification of boundary conditions in mathematical models Eq. (20)) to find the first four mode shapes of each span, with
agrees well with the field results for almost all modes, especially the z-coordinate of the last spans measured from the very
for pinned-pinned boundary conditions in the lateral mode. The right point of the span. The first four mode shapes of the first
comparison function for axial motion, considering the above span are
mentioned boundary condition, is expressed as ϕ1 ¼ sin ð0:119516 zÞ  0:3340965884 sinhð0:119516 zÞ
  ϕ2 ¼ sin ð0:2095 zÞ þ 0:00007849517415 sinhð0:2095 zÞ
ð2r 1Þπz
χ r ðzÞ ¼ sin ð19Þ ð22Þ
2l ϕ3 ¼ sin ð0:22616 zÞ þ 0:01671568731 sinhð0:22616 zÞ
ϕ4 ¼ sin ð0:274 zÞ þ 0:02704677997 sinhð0:274 zÞ
where l is the length of the drillstring. Since the BHA is assumed as
a three span beam, the comparison function of each span is The first four mode shapes of the second span can be expressed as
required. The exact mode shapes of a three span beam will be
ψ 1 ¼  2:461446386 sin ð0:119516z 0:19516l1 Þ
derived in the following section and they will be implemented
þ 1:927004330 sinhð0:119516z  0:19516l1 Þ
in the Lagrangian equation as the corresponding comparison
functions. þ 0:9754692378 cos ð0:119516z  0:19516l1 Þ
 0:9754692378 coshð0:119516z  0:19516l1 Þ
3.2. Eigenfunctions and eigenfrequencies of a three span beam ψ 2 ¼ 1:093320970 sin ð0:2095z  0:2095l1 Þ
with different lengths  0:001075050855 sinhð0:2095z  0:2095l1 Þ
 0:0009073462857 cos ð0:2095z  0:2095l1 Þ
As it was discussed in the previous section, the mode shapes þ 0:0009073462857 coshð0:2095z  0:2095l1 Þ
of a simply supported three span beam are required to proceed ψ 3 ¼ 0:7786943701 sin ð0:22616z  0:22616l1 Þ
with the Bypassing PDE's method. A schematic of a 3 span beam is  0:2780537428 sinhð0:22616z  0:22616l1 Þ
shown in Fig. 3.
 0:2482829965 cos ð0:22616z  0:22616l1 Þ
Separate coordinate systems are assumed for each span and the
þ 0:2482829965 coshð0:22616z  0:22616l1 Þ
normal mode for each span can be written as
ψ 4 ¼ 0:2184545015 sin ð0:274z  0:274l1 Þ
φðxÞ ¼ a cos ðβxÞ þb sin ðβxÞ þ coshðβxÞ þ d sinhðβxÞ
 0:8338446736 sinhð0:274z  0:274l1 Þ
ψðyÞ ¼ e cos ðβyÞ þ f sin ðβyÞ þg coshðβyÞ þ h sinhðβyÞ ð20Þ  0:8239843412 cos ð0:274z  0:274l 1 Þ
θðzÞ ¼ i cos ðβzÞ þ j sin ðβzÞ þ k coshðβzÞ þl sinhðβzÞ
þ 0:8239843412 coshð0:274z  0:274l1 Þ ð23Þ
The eigenfunctions of the last span are
θ1 ¼  6:090745568 sin ð 0:119516z þ 0:119516lÞ
 0:1451462187 sinh ð  0:119516z þ 0:119516lÞ
θ2 ¼ 1:046660710 sin ð  0:2095z þ 0:2095lÞ
þ 0:000007081079825 sinh ð 0:2095z þ0:2095lÞ
θ3 ¼ 0:9019326856 sin ð  0:22616z þ 0:22616lÞ
þ 0:0009810187433 sinh ð  0:22616z þ 0:22616lÞ
θ4 ¼ 0:3072276254 sin ð  0:274z þ 0:274lÞ
Fig. 3. Schematic of a 3 span BHA. þ 0:0001544631224 sinhð 0:274z þ 0:274lÞ ð24Þ
42 A. Ghasemloonia et al. / Journal of Petroleum Science and Engineering 116 (2014) 36–49

Fig. 4. The first four mode shapes of a three span beam.

Table 1
Parameters used in the simulations.

T ¼ 4000 Driving torque (N m) l4 ¼ 800 Length of the pipe section (m)


ρmud ¼1500 Mud density (kg/m3) FVARD ¼ 20 000 VARD force amplitude (N)
CD ¼ 1 Hydrodynamic drag coefficient ωVARD ¼ 600 VARD tool frequency (rad/s)
Kh ¼6.78  1011 Hertzian stiffness (N m  1.5) Acollar ¼0.02639 Collar cross sectional area (m2)
WOB ¼100 000 Weight on bit (N) E¼ 210  109 Young's modulus (Pa)
Apipe ¼0.00471 Pipe cross sectional area (m2) bcl ¼0.1 Borehole clearance (m)
l1 ¼15 Length of the BHA first span (m) FH ¼ 320 000 Hook load (N)
l2 ¼15 Length of the BHA second span (m) ρst ¼ 7860 Pipe and collar density (kg/m3)
l3 ¼30 Length of the BHA last span (m)

The above mode shapes are shown in Fig. 4 for a three span BHA Table 2
with 15, 15 and 30 m lengths for the first to the third spans, The first four natural frequencies for coupled axial-transverse-transverse modes.
respectively.
Direction First Second Third Fourth
mode (Hz) mode (Hz) mode (Hz) mode (Hz)
3.3. Lagrangian and equations of motions
Lateral "u" 1.34 1.95 2.35 3.15
Lateral "v" 1.37 1.92 2.30 3.10
Substituting the corresponding mode shapes in the Lagrangian Axial "w" 7.65 22.25 38.59 60.30
of each span and adding the corresponding terms, the total
Lagrangian of the system is derived. Lagrange's equations are
implemented for each mode participation factor. Substituting the
comparison functions and integrating the resulting equations over
the drillstring length domain, and using the mode orthogonality
relations, gives twelve second order coupled nonlinear time time response, the initial time step was set to 10  9 swhich is
differential equations. One of the major advantages of the current well below the smallest natural period in the system. The para-
model is that the equations are kept symbolical, up to this step. meters and numerical values used this study are shown in Table 1.
This symbolic approach allows a high speed sensitivity analysis for The FFT of each specific generalized coordinate system was
each controllable parameter, especially the VARD force amplitude derived, which reveals the first four natural frequencies for the
and frequency, which are of great interest for VARD drilling two orthogonal lateral planes and the axial direction. The results
studies. The mathematical model and the above procedure have are shown in Table 2.
been implemented in Maples. Since lateral constraints (stabilizers) are assumed for the BHA,
the flexural frequencies are higher than in other studies that do
not assume multi-mode contact at the BHA (Khulief et al., 2008).
4. Numerical results There is a small variation between resonance frequencies in
the u and v directions as a result of the numerical solution. The
This system was numerically solved using a Fehlberg fourth- maximum difference is 0.05 Hz, which is a negligible difference
fifth order Runge–Kutta method with degree four interpolant, in rotary drilling (around 3 rpm). From a practical drilling
which is an adaptive numeric procedure for solving the initial standpoint, the rotational speed should be adjusted so that it
value problems combining fourth-order and fifth-order Runge– does not correspond to one of the eigenfrequencies, since
Kutta techniques. The main advantage of this method is the resonance causes large amplitude oscillations. There might be
dynamic step reduction strategy compared to the fixed-step fourth- other frequencies which cause large-amplitude response due
order Runge–Kutta method. In order to avoid discontinuities in the to chaotic motion of the drillstring, bit-formation interaction
A. Ghasemloonia et al. / Journal of Petroleum Science and Engineering 116 (2014) 36–49 43

Fig. 5. Axial deflection near the hook point.

Fig. 6. Axial deflection, midpoint on span 1 of the BHA.

Fig. 7. Axial deflection, midpoint on span 2 of the BHA.


44 A. Ghasemloonia et al. / Journal of Petroleum Science and Engineering 116 (2014) 36–49

Fig. 8. Axial deflection, a point close to the bit on span 3 of the BHA.

Fig. 9. Axial velocity, a point close to the bit on span 3 of the BHA.

Fig. 10. Phase plane, a point close to the bit on span 3 of the BHA.
A. Ghasemloonia et al. / Journal of Petroleum Science and Engineering 116 (2014) 36–49 45

Fig. 11. Lateral deflection and velocities, a no-contact point on the top span of the BHA.

Fig. 12. Lateral phase planes, trajectory and radial deflection, a no-contact point on the top span of the BHA.

uncertainties or whirling motion, but the resonance frequencies The time history of any desired point is achieved using
extracted in this study will provide the driller with rotary speeds the numerical solutions of the generalized coordinate systems
that must be avoided. (12 mode participation factors in this study) and the expanded
46 A. Ghasemloonia et al. / Journal of Petroleum Science and Engineering 116 (2014) 36–49

Galerkin's equations. The lateral behavior at the contact points is The lateral vibration behavior of any point on the BHA can be
of significant practical interest. Phase plane qualitative analysis is concisely depicted by radial deflection plots. The borehole–drill-
performed to determine the severity of contact at the contact string clearance is set in the Hertzian contact equation and radial
points, which should be minimized to avoid wellbore washouts displacement exceeding the borehole clearance results in a restor-
and joint failures. ing force. Fig. 11a and c presents the lateral deflections of the point
Axial deflection of the drillstring is an especially important in two orthogonal planes. Fig. 11b and d shows the corresponding
response due to the possible presence of a VARD force generator. lateral velocities. Fig. 12a and c demonstrates the phase planes of
Axial deflection at a point on the pipe, very close to the hook point this point in the u and v directions. These figures verify a stable
(where the draw-works cable is attached to the pipe), is shown behavior at the center of the wellbore. Fig. 12b shows the
in Fig. 5. The initial fluctuations are due to the imposed initial trajectory of that point. As it is clear from this figure, the point
conditions to the BHA. The deformation settles to a stable region spends a considerable amount of time near the wellbore center.
with a peak-to-peak value of 5 mm. The axial deflections of the Fig. 12d shows the corresponding radial deflection of that point.
midpoint of the first and second spans are shown in Figs. 6 and 7. The maximum radial deflection is around 0.05 m, which is in
The axial deflection of a point on the last span, close to the bit, is between the wellbore wall and the center of the wellbore.
shown in Fig. 8. The mean deflection magnitude is below zero, as a Fig. 13 shows the radial deflection at the contact point on the
result of the interaction of all axial forces and the assumed top span. At this point, the drillstring hits the wellbore irregularly
reference coordinate system, and converges to a 0.01 m peak-to- with a bouncing contact behavior. A phase portrait of the contact
peak region after entering the stability region. The corresponding point is shown in Fig. 14, which verifies highly irregular behavior
axial velocity of that point is depicted in Fig. 9. The phase portrait of at this location. The radial deflection of the contact point at the
this point is shown in Fig. 10, which demonstrates stable behavior middle of the second span is shown in Fig. 15. The contact is not as
in the axial mode. The amplitude and frequency of the VARD tool, as severe as for the top span. After hitting the wellbore for a period of
well as spatially varying axial force affect this behavior. time, the drillstring spends a period near the center, and then hits

Fig. 13. Radial deflection of the contact point, first span of the BHA.

Fig. 14. Phase portrait of the contact point on the first span of the BHA.
A. Ghasemloonia et al. / Journal of Petroleum Science and Engineering 116 (2014) 36–49 47

Fig. 15. Radial deflection of the contact point, second span of the BHA.

Fig. 16. Phase portrait of the contact point on the second span of the BHA.

Fig. 17. Radial deflection of the contact point, last span of the BHA.
48 A. Ghasemloonia et al. / Journal of Petroleum Science and Engineering 116 (2014) 36–49

Fig. 18. Phase portrait of the contact point on the last span of the BHA.

the wellbore again. The phase plane in Fig. 16 demonstrates that at 6. Conclusions
this location, the drillstring is not traveling strictly near the
wellbore. The contact behavior at the span closest to the bit is Predicting the dynamic behavior of a drillstring through com-
depicted in Figs. 17 and 18. The radial deflection shows less contact puter simulation is an essential step in designing BHA configura-
compared to previous spans. It seems that the damping behavior tions that will be less susceptible to vibration, for predicting safe
of the mud is affecting the contact at this point; however, the rotary speed ranges, and for ensuring that down-hole tools
length of the span is half that of the top span. The radial phase achieve their goals without creating excessive vibration in other
plane also shows a less irregular behavior compared to the other parts of the string. The coupled nonlinear axial-transverse beha-
two spans. vior and lateral instabilities of the drillstring under an applied
This model is capable of predicting dynamic behavior at any axial force were studied in this paper. A model was constructed
point on the drillstring, either on the pipe section or BHA. that extends existing models in the literature by including multi-
Additionally, the symbolical code enables sensitivity analysis of ple BHA spans, wellbore contact, mud damping, spatially varying
the controllable parameters. In practice, a VARD tool should be axial force, torque, and nonlinear coupling terms, torque, axial
tuned for different geotechnical formations, and simulation is a vibration, and transverse vibration in two orthogonal directions,
potentially powerful tool to investigate the behavior of the axial stiffening and torque were retained. The Bypassing PDE's
drillstring in the wellbore before drilling new formations. The method, along with the Lagrange's equation, was implemented to
symbolical Maple code for the new model allows running the derive the nonlinear equations. Using the expanded Galerkin's
equations with a new set of VARD tool parameters, as well as method, with the first four modes retained for each span, a multi-
different values for: span length, pipe length, mud properties, point contact analysis was performed in the time domain. The
pipe and collar section and material properties. The developed equations were solved symbolically. The multi-span analysis of the
model is not restricted by inclusion of the VARD elements. BHA provided more realistic predictions of the resonant rotary
The model can be rearranged by setting the VARD force to zero speeds. The rotary speed of the drillstring should be kept far
and can be implemented for drilling simulations without the enough from the natural frequencies to avoid excessive deflections
VARD tool. and contact with the wellbore, both of which can cause premature
failure of bottom-hole assembly components. Phase plane analysis
of the contact points demonstrated a highly irregular contact at
the top span of the BHA, while the contact at the span closest to
5. Discussion the bit was not as severe. The symbolical model of this paper can
be used to conduct a sensitivity analysis of controllable parameters
Drillstring vibration models are continuing to grow in complex- for tuning a VARD force generator, determining rotary speed
ity and predictive ability. This paper presents a model that guidelines, and designing suppression methods, either for vibra-
incorporates important geometric features and nonlinearities, at tion in the axial or transverse directions.
least some of which are missing from other models in the
literature. No model can include all phenomena that might affect
dynamic response, especially when fast computation time and
symbolic equations are desired. This model does not account for Acknowledgment
formation properties and heterogeneity, BHA imbalance, misalign-
ment and friction factors. However, the model allows the engineer This research was conducted at the Advanced Drilling Technol-
to compare the sensitivity of different BHAs to vibrations. The ogy Laboratory at Memorial University of Newfoundland and was
accuracy of the model will be improved when calibrated using funded by the Atlantic Canada Opportunities Agency (AIF Contract
actual well logs, but for now has been validated against a finite No. 781-2636-1920044), the Research and Development Corpora-
element model (FEM) developed by the authors (Ghasemloonia tion of Newfoundland and Labrador, Husky Energy, and Suncor
et al., 2013a, 2013b). Energy. Also, the authors would like to acknowledge the valuable
A. Ghasemloonia et al. / Journal of Petroleum Science and Engineering 116 (2014) 36–49 49

contributions to the software implementation made by Dr. Hakimi, H., Moradi, S., 2009. Drillstring vibration analysis using differential
Muzychka at Memorial University. quadrature method. J. Petrol. Sci. Eng. 70, 235–242.
Heisig, G., Neubert, M., 2000. Lateral drillstring vibrations in extended reach wells.
In: IADC/SPE Drilling Conference, New Orleans, USA.
References Jansen, J.D., 1991. Nonlinear rotor dynamics as applied to oil well drillstring
vibrations. J. Sound Vib. 147 (1), 115–135.
Jogi, P.N., Macpherson, J.D., Neubert, M., 2002. Field verification of a model-derived
Babatunde, Y., Butt, S.D., Molgard, J., Arvani, F., 2011. Investigation of the effects of
natural frequencies of a drillstring. J. Energy Resour. Technol. 124 (3), 154–163.
vibration frequency on rotary drilling penetration rate using diamond drag bit.
Khulief, Y.A., Al-Naser, H., 2005. Finite element dynamic analysis of drillstrings. J.
In: 45th US Rock Mechanics/Geomechanics symposium (ARMA), San Francisco,
Finite Element Anal. Des. 41, 1270–1288.
USA.
Khulief, Y.A., Al-Sulaiman, F.A., Bashmal, S., 2008. Vibration analysis of drillstrings
Barton, S., Baez, F., Alali, A., 2011. Drilling performance improvement in gas shale
plays using a novel drilling agitator device. In: SPE North American Unconven- with string-borehole interaction. J. Mech. Eng. Sci. (IMechE, Part C) 222,
tional Gas Conference and Exhibition, Texas, USA. 2099–2110.
Beck, A.T., da Silva , C.R.A., 2010. Timoshenko versus Euler beam theory: pitfalls of a Li, Z., Yanshan, U., Guo, B., 2007. Analysis of longitudinal vibration of drillstring in
deterministic approach. Struct. Saf. 33, 19–25. air and gas drilling. In: SPE Rocky Mountain Oil and Gas Technology Sympo-
Berlioz, A., Der Hagopian, J., Draoui, E., 1996. Dynamic behavior of a drillstring: sium, Colorado, USA.
experimental investigation of lateral instabilities. J. Vib. Acoust. 118, 292–298. Liao, C.M., Balachandran, B., Karkoub, M., Abdel-Magid, Y.L., 2011. Drill-string
Chen, S.L., Geradin, M., 1995. An improved transfer matrix technique as applied to dynamics: reduced -order models and experimental studies. J. Vib. Acoust.
BHA lateral vibration analysis. J. Sound Vib. 185 (1), 93–106. 133, 041008 (1-041008-8).
Chin, W.C., 1994. Wave Propagation in Petroleum Engineering. Gulf publishing Co., Manko, I.K., Pilipenko, V.V., Zapolsky, G.Z., 2003. Use hydrodynamic cavitation for
Houston increase of efficiency of process of well drilling. In: Fifth International
Christoforou, A.P., Yigit, A.S., 1997. Dynamic modeling of rotating drillstrings with Symposium on Cavitation (CAV 2003), Osaka, Japan.
borehole interactions. J. Sound Vib. 206 (2), 243–260. Mitchell, R.F., Miska, S.Z., 2011. Fundamentals of Drilling Engineering. SPE, TX
Dareing, D.W., 1984a. Drill collar length is a major factor in vibration control. J. Simitses, G., Hodges, D., 2005. Fundamentals of Structural Stability. Elsevier, Oxford
Petrol. Technol. 36 (4), 637–644. Sotomayor, G.P.G., Placido, J.C., Cunha, J.C., 1997. Drillstring vibration: how to
Dareing, D.W., 1984b. Guidelines for controlling drillstring vibrations. J. Energy identify and suppress. In: 5th Latin American and Caribbean Petroleum
Resour. Technol. 106 (2), 272–277. Engineering Conference, Rio de Janerio, Brazil.
Ghasemloonia, A., Rideout, D.G., Butt, S.D., 2012. Coupled transverse vibration Spanos, P.D., Payne, M.L., Secora, C.K., 1997. Bottomhole assembly modeling and
modeling of drillstrings subjected to torque and spatially varying axial load. J. dynamic response determination. J. Energy Resour. Technol. 119, 153–159.
Mech. Eng. Sci. (IMechE, part C) 227 (5), 946–960. Spanos, P.D., Chevallier, A.M., Politis, N.P., 2002. Nonlinear stochastic drillstring
Ghasemloonia, A., Rideout, D.G., Butt, S.D., 2013. Vibration analysis of a drillstring in vibrations. J. Vib. Acoust. 124 (4), 512–519.
vibration-assisted rotary drilling: finite element modeling with analytical Thomsen, J.J., 2003. Vibrations and Stability, second ed. Springer, Berlin
validation. J. Energy Resour. Technol. 135 (3), 032902 (18 pp.). Yigit, A.S., Christoforou, A.P., 1996. Coupled axial and transverse vibrations of oil
Gulyaev, V.I., Lugovoi, P.Z., Belova, M.A., Solov'ev, I.L., 2007. Effect of the length of a well drillstrings. J. Sound Vib. 195 (4), 617–627.
rotating drillstring on the stability of its quasistatic equilibrium. Int. J. Appl. Yigit, A.S., Christoforou, A.P., 1998. Coupled torsional and bending vibrations of
Mech. 43 (9), 1017–1023.
drillstrings subject to impact with friction. J. Sound Vib. 215 (1), 167–181.
Gulyaev, V.I., Gaidaichuk, V.V., Solovjov, I.L., Gorbunovich, I.V., 2009. The buckling
of elongated rotating drillstrings. J. Petrol. Sci. Eng. 67, 140–148.