Journal of Sound and Vibration (1998) 215(1), 167–181

Article No. sv981617

COUPLED TORSIONAL AND BENDING

VIBRATIONS OF DRILLSTRINGS SUBJECT TO
IMPACT WITH FRICTION
A. S. Y  A. P. C
Department of Mechanical and Industrial Engineering, Kuwait University, P.O. Box 5969,
Safat, 13060 Kuwait

(Received 8 January 1998, and in final form 13 March 1998)

A dynamic model for coupled torsional and bending vibrations of drillstrings is

presented. The bit/formation interactions are assumed to be related to the bit motion,
leading to a set of highly non-linear equations with a potential for self-excited behaviour.
Impacts with the borehole wall are accounted for by a consistent contact model which is
capable of capturing rolling both with and without slip of the drill collars along the
borehole wall. Simulation results show that the non-linear coupling significantly affects the
response. Due to the combined parametric and forcing excitations and the effect of strong
non-linearities, the response is not simple or easily predictable. At certain frequencies, there
is significant energy transfer between the two modes of vibrations. The simulation results
agree well with laboratory and field observations when the so called stick–slip vibrations
occur. This study suggests a new mechanism for this behaviour which was shown to be due
to static friction or speed dependent torque on bit.
7 1998 Academic Press

1. INTRODUCTION
It is well known that drillstring vibrations may lead to fatigue failures and abrasive wear
of tubulars, damage to the drill bit and the borehole wall. As a consequence, oilwell drilling
becomes inefficient and costly. On the other hand, measurements of these vibrations may
provide valuable information about the drilling assembly and formation characteristics.
Therefore, vibrations must be fully understood and their effects should be minimized in
any approach to drilling optimization.
Drillstring vibrations are generally quite complex. Axial, lateral and torsional vibrations
are all present and coupled. Phenomena such as bit bounce, stick–slip, forward and
backward whirl and parametric instabilities have all been shown to occur. Because of the
complexity, a full simulation covering all relevant phenomena is not practical, and a
common approach has been to study vibration mechanisms individually [1–5].
Some studies on drillstring dynamics were mostly concerned with axial and torsional
vibrations. The coupling between these two motions was also noted [1]. Later, it was shown
that the coupling mechanism between axial and torsional vibrations can lead to large
torsional vibrations [6]. Kyllingstad and Halsey [7] studied the stick–slip vibrations due
to static friction at the bit/formation interface. Brett [8] showed that the dependence of
bit torque on the bit speed may also cause similar vibrations. Halsey et al. [9], have shown
that torque feedback can help eliminate stick–slip vibrations. Another instability
mechanism for coupled axial and torsional vibrations was proposed by Elsayed et al. [10]
which is due to variation in the phase angle between surface undulations of subsequent
cutters.

0022–460X/98/310167 + 15 $30.00/0 7 1998 Academic Press

168 . .   . . 
Lateral vibrations which occur in the forms of whirling and parametric resonance have
also been studied extensively [11–16]. In these studies, the torsional vibrations are not
considered and the drillstring is assumed to rotate at a constant speed. Consequently, the
excitation due to bit/formation interaction was assumed to be a prescribed function of time
(i.e., a mono-frequency harmonic excitation). In the current work, however, the full
coupling between torsional and lateral vibrations is considered and the excitation at the
bit is assumed to be related to the rotation of the bit. This makes the coupling between
the torsional and bending vibrations much stronger and non-linear. Furthermore, a
consistent impact model is used to account for impacts of drill collars with the borehole
wall which may involve rolling with and without slip. The results show that the torsional
vibrations can significantly affect the dynamic behaviour. The findings are in qualitative
agreement with laboratory and field observations.

2. EQUATIONS OF MOTION
The equations of motion for coupled torsional and bending vibrations are obtained
through a simplified model with equivalent lumped parameters shown in Figure 1. The
stabilized portion of the Bottom Hole Assembly (BHA) is modelled as a simply supported
shaft for the transverse motion of the drill collars and the whole drillstring is assumed to
be fixed at the top and free at the bit for torsional motion. The BHA is assumed to be
rigid with respect to torsional vibrations since the drill collars are much stiffer than the
drill pipe. By applying Newton’s law in polar co-ordinates attached to the borehole centre,
the following equations are obtained:
(m + mf )(r̈ − ru 2) + k(f)r + ch =v=ṙ = (m + mf )e0 [f 2 cos (f − u) + f sin (f − u)] − Fr ,
(1)
(m + mf )(ru + 2ṙu ) + ch =v=ru = (m + mf )e0 [f sin (f − u) − f cos (f − u)] + Fu ,
2
(2)
Jf + kT (f − frot ) + cv f + ch =v=ṙe0 sin (f − u) − ch =v=ru e0 cos (f − u)
= −T(f) + Fu [R − e0 cos (f − u)] − Fr e0 sin (f − u). (3)
,
,,
,,,

,
,
,
,
,,

,
,
,
,
,,
Y
rot

Drillpipe
l3 G
e0
R
r
Stabilizers X
0
Drillcollars

l2
l1
A A Drillbit

T( )
Drill collar Borehole wall
F( )
Section A–A
Figure 1. The sketch of the system.
      169
T 1
Parameters used in the simulations
Drillstring Drilling mud Borehole
E = 210 GPa rf = 1500 kg/m3 E = 210 GPa
r = 7850 kg/m3 Cd = 1 r = 7850 kg/m3
do = 0·2286 m Ca = 1·7 dh = 0·4286 m
di = 0·0762 m mf = 0·2 Ns/m2 m = 0·3
e0 = 0·0127 m Sy = 600 MPa
l1 = 19·81 m, l2 = 200 m
l3 = 1000 m

do = 0·127 m, 
di = 0·095 m

Weight and torque on bit

P0 = 100 kN, Pf = 100 kN; T0 = 4 kNm, Tf = 4 kNm

Here frot is the angular displacement at the rotary table (top of the drillstring), f is the
angle of rotation of the drill collars, v is the velocity of the geometric centre of the drill
collar section, e0 is the eccentricity of the centre of mass with respect to the geometric centre
of the collar section, J, m, mf , k(f), kT , ch and cv are the equivalent mass moment of inertia,
mass, added fluid mass, transverse stiffness, torsional stiffness, and hydrodynamic and
viscous damping coefficients which are derived from the associated continuous model of
the drill string by using a Lagrangian approach (see Appendix A), and R is the radius of

3
Bit speed
Normalized bit speed and TOB

TOB

–1
0 10 20 30 40 50 60 70 80 90 100
Normalized time
Figure 2. Effect of torsional vibrations on the bit speed and the TOB; ——, with torsion; – – –, without torsion.
170 . .   . . 
1.0

0.8
Normalized displacement

0.6

0.4

0.2

0.0
0 10 20 30 40 50 60 70 80 90 100
Normalized time
Figure 3. Effect of torsional vibrations on the lateral vibration; key as Figure 2.

the drill collars. In the present study, the angular velocity of the rotary table is assumed
to be given as
v(t) = v0 + va sin vr t, (4)
where v0 is the prescribed angular velocity of the rotary table, and va and vr are the
amplitude and the frequency of angular velocity fluctuations which depend on the control
characteristics of the rotary table drive system. By integrating equation (4) with respect
to time, the angular displacement of the rotary table can be obtained as
frot = v0 t + (va /vr )(1 − cos vr t). (5)
It should be noted that with the model described above, the rotary speed at the bit will
not be the same as the rotary table speed since it is one of the degrees of freedom of the
system. Consequently, the excitation forces due to the bit/formation interactions are no
longer prescribed functions of time as commonly assumed [1–5, 11–16]. Therefore, the
torque on bit (TOB), T(f), and the weight on bit (WOB), F(f) are assumed to be
T(f) = T0 + Tf sin nf, F(f) = F0 + Ff sin nf, (6, 7)
where n is the bit factor which depends on the type of the bit used, and, the subscripts
0 and f denote the mean and the amplitude values, respectively. The only continuous
external dynamic excitation comes from the prescribed motion of the rotary table. As a
consequence, linear phenomena such as simple and parametric resonance [14, 15], and
whirling [11–13], which are based on the assumption of harmonic excitations, are not
straightforward. Other intermittent external excitations are Fr and Fu which are the radial
and transverse contact forces, respectively, resulting from impacts of the drill collars with
the borehole wall. When there is no contact between the collars and the borehole wall,
these forces are zero.
      171
3. ANALYSIS OF CONTACT WITH THE BOREHOLE WALL
Previously, the impact between the drill collars and the borehole wall was modelled
through a linear or Hertzian contact stiffness [13, 16]. In these studies the effect of friction
was accounted for with the assumption of continuous sliding between the two surfaces.
Clearly, there might be situations where the collars roll along the borehole wall without
slip. In this study, to account for this type of contact situations, a different impact model
is used. This impact model is based on the impulse–momentum principle and it is assumed
that the impact is instantaneous [17, 18].
The impulse–momentum equations are obtained by integrating equations (1)–(3) with
respect to time during the contact duration Dt = t2 − t1 :

g
t2

{(m + mf )(r̈ − ru 2) + k(f)r + ch =v=ṙ} dt

t1

g
t2

= {(m + mf )e0 [f 2 cos (f − u) + f sin (f − u)] − Fr } dt, (8)

t1

g
t2

{(m + mf )(ru + 2ṙu ) + ch =v=ru }dt

t1

g
t2

= {(m + mf )e0 [f 2 sin (f − u) − f cos (f − u)] + Fu } dt, (9)

t1

1.0

0.8
Normalized displacement

0.6

0.4

0.2

0.0
0 50 100 150 200 250 300
Normalized time
Figure 4. Lateral vibrations with impacts with the borehole wall.
172 . .   . . 
4
Bit speed

Normalized speed and torque Table speed

2

TOB
–1 Top torque

–2
0 50 100 150 200 250 300
Normalized time
Figure 5. Torsional vibrations with impacts with the borehole wall.

g
t2

{Jf + kT (f − frot ) + cv f + ch =v=ṙe0 sin (f − u) − ch =v=ru e0 cos (f − u)} dt

t1

g
t2

= {−T(f) + Fu [R − e0 cos (f − u)] − Fr e0 sin (f − u)} dt. (10)

t1

As the contact duration approaches zero, which is the main assumption here, and upon
noting that the system configuration is continuous, equations (8)–(10) can be shown to
yield the momentum balance equations [17]
(m + mf )Dṙ = −Pr + (m + mf )e0 Df sin (f − u), (11)
(m + mf )rDu = Pu − (m + mf )e0 Df cos (f − u), (12)
JDf = [R − e0 cos (f − u)]Pu − e0 sin (f − u)Pr , (13)
where Dṙ and Du are the jump discontinuities in the velocities, and Pr and Pu are the
impulses in radial and transverse directions, respectively, defined as

g
t2

P = lim F(t) dt. (14)

Dt:0
t1

The radial and transverse impulse values for a particular impact are calculated by using
the two-dimensional rigid-body impact-friction model proposed by Wang and Mason [19].
For completeness, the details of this model as applied to the system studied here are given
in Appendix B.
      173
The response of the system before an impact is obtained by numerical integration of
the equations of motion (equations (1)–(3)). At each integration time step, the contact
conditions are checked by using a contact algorithm. If an impact is detected to occur,
the integration is stopped and the momentum balance equatons (equations (11)–(13)) are
formed and solved for the jump discontinuities in velocities. Then, the integration is
restarted again with the new initial conditions.

4. DIMENSIONLESS FORMULATION

For convenience, the governing equations are expressed in dimensionless form by

defining the variables

t = v0 t, r̄ = r/bc , (15)

where bc is the clearance between the borehole wall and drill collar. Substitution of
equations (15) into equations (1)–(3) lead to the dimensionless equations

r + v̄b2 (1 − o sin nf)r̄ + c̄h =v=r − r̄u 2 = ē0 f 2 cos (f − u) + ē0 f sin (f − u) − Fr , (16)

r̄u + 2r u + c̄h =v̄=r̄u = ē0 f 2 sin (f − u) − ē0 f cos (f − u) + Fu , (17)

f + v̄s2 (f + ff sin nf) + 2zv̄s f + c̄*

h =v̄=[r sin (f − u) − r̄u cos (f − u)]

= v̄s2 (frot − f0 ) + T
u [R − ē0 cos (f − u)] + T
r ē0 sin (f − u), (18)

Bit speed

3
Normalized speed and torque

2 Table speed

Top torque TOB

–1

–2
0 5 10 15 20 25 30 35 40 45 50
Normalized time
Figure 6. Torsional vibrations with impacts with the borehole wall (close-up view).
174 . .   . . 

1.4 Bit speed

Table speed

1.2

Normalized speed and torque

1.0

0.8

0.6

0.4
Top torque TOB

0.2

0.0
0 20 40 60 80 100 120 140
Normalized time
Figure 7. Torsional vibrations with a fluctuating table speed.

with the following dimensionless parameters:

0 1
vb2 pT F pTf + l1 Ff
v̄b2 = 1− 0 − 0 , o= , (19)
v02 l1 Fcr Fcr l1 Fcr − pT0 − l1 F0

Fr Fu
Fr = , Fu = , (20)
(m + mf )bc v02 (m + mf )bc v02
u = Fu bc /Jv02 ,
T r = Fr bc /Jv02 ,
T (21)
ē0 = e0 /bc ,  = R/bc ,
R (22)
2
c̄h = ch bc /(m + mf ), c̄*
h = ch bc e0 /J, (23)
v̄s = vs /v0 , (24)
z = cv /2Jvs , f0 = T0 /kT , ff = Tf /kT . (25)
Here vb and vs are the natural frequencies for bending and torsion of the unloaded
drillcollar for the associated linear uncoupled problem, respectively, and Fcr is the Euler
buckling load given as
vb2 = EIp 4/2(m + mf )l13 , vs2 = kt /J, Fcr = p 2EI/l12 , (26)
It is clear that equations (16)–(18) constitute a highly non-linear coupled set of ordinary
differential equations as opposed to differential equations with time varying coefficients
used in previous studies. Since the excitation due to the bit/formation interaction is
assumed to be related to the rotation of the bit the coupling between the torsional and
bending vibrations is much stronger. Therefore, it is expected that when the torsional
      175
vibrations are excited, they will play a dominant role in the dynamic behaviour of
drillstrings. In the following section a number of simulations are presented to demonstrate
the use of the proposed model in predicting the dynamic behaviour.

5. SIMULATION RESULTS AND DISCUSSION

The dynamic response of a drillstring is obtained by a numerical solution of equations
(16)–(18) with appropriate initial conditions until an impact occurs with the borehole wall.
The effect of impact on the response is accounted for by using the momentum balance
method described in section 3. Because of impacts and other non-linearities, a numerical
solution procedure based on a variable-step variable-order predictor–corrector differential
equation solver, which is suitable for stiff problems, is used to integrate the equations of
motion [20]. As is well known, the accuracy of the numerical solution depends on the
choice of the time step. As a rule of thumb, the time step used was 1/10 of the smallest
natural period of the associated linear problem. In addition, the time step is reduced
internally in the integrator to avoid discontinuities in the response. The time step is reduced
by ten times if the relative error bound specified (10−8) is exceeded. This is especially
important in order to handle chatter-like behaviour.
The parameters used in the following simulations are shown in Table 1, and represent
a typical case in oil well drilling operations. The rotary table speed is varied between 6
and 15 rad/s which is a common operating range. From a simple linear and uncoupled
analysis for this set-up, the critical frequency for torsional resonance is found to be
1·85 rad/s, while the critical frequency for whirling resonance (due to bending) is found
to be 6·14 rad/s. Since whirling has been investigated earlier [11–13, 16], only the
simulations showing the effects of torsional vibrations are presented in this paper.

1.0

0.8
Normalized displacement

0.6

0.4

0.2

0.0
0 100 200 300 400 500
Normalized time
Figure 8. Lateral vibrations with a fluctuating table speed.
176 . .   . . 
3.0

Bit speed
2.5

Normalized speed and torque 2.0 Table speed

1.5

1.0

0.5

0.0

TOB
Top torque
–0.5

–1.0
0 100 200 300 400 500
Normalized time
Figure 9. Torsional resonance due to fluctuating table speed.

In the first set of simulations, which are shown in Figures 2–5, the rotary table speed
is assumed to be constant. This assumption has been commonly used in most drilling
dynamics studies [1–5, 11–16]. Figures 2 and 3 show the effect of torsional vibrations on
the bit speed, Torque On Bit (TOB), and lateral vibrations when the rotary table speed
is equal to 9 rad/s. As clearly seen in Figure 2, when torsional vibrations are not
considered, the bit speed is the same as the rotary table speed, and the TOB is a pure
sinusoid. In other words, the bit speed and the TOB are completely prescribed, which is
generally the approach taken in most earlier studies dealing with lateral vibrations of
drillstrings [15, 16]. When there are considerable torsional vibrations, however, the bit
speed is significantly different from the rotary table speed. In fact, field measurements have
shown that, when there are significant torsional vibrations, the bit speed differs from the
rotary table speed by as much as three times [8, 9], which is also in agreement with the
present results. As a consequence, the TOB time history contains finite time intervals
during which the bit is non-rotating (constant f) followed by high frequency oscillations.
This type of behaviour is an indication of stick-slip motion which will be examined later
with further simulations. As expected, the torsional vibrations also affect the lateral
vibrations, as is clearly seen in Figure 3. With the bit speed reaching almost three times
the rotary table speed, because of the rotating unbalance, the amplitudes of transverse
vibrations are much higher. Note that the rotary table speed of 9 rad/s is away from any
linear critical speed for either torsional or lateral vibrations. Apparently, the non-linear
coupling between torsional and lateral vibrations complicates the dynamics, and requires
further investigation.
Figures 4 and 5 show the dynamic response when the rotary table speed is equal to
8 rad/s. The lateral vibrations shown in Figure 4 indicate an unstable response and
consequently, impacts with the borehole wall occur. As can be seen, in addition to impacts
      177
there may be periods of continuous rubbing and rolling along the borehole wall. It should
be noted that the impact model used is instrumental in capturing this type motion
accurately. The instability of the response can also be seen in the torsional vibrations
shown in Figure 5. Since the growth of amplitudes are far from being linear it is probable
that parametric resonance occurs due to the non-linear stiffness term in equation (18). It
seems that the growth of torsional vibrations lead to large lateral vibrations and impacts.
Impacts with friction, however, reduce the torsional vibrations, and the torsional
instability manifests itself in the lateral vibrations, and energy is transferred from torsional
to lateral motion.
The effect of large torsional vibrations on the TOB is also evident in Figure 5. As
mentioned earlier, when the torsional vibrations are significant, the stick–slip behaviour
is observed in the TOB time history. When the torsional vibrations are reduced due to
impacts, the TOB becomes a pure sinusoid. In order to investigate the motion of the bit
further, a close-up view of Figure 5 is shown in Figure 6. It is seen that when the bit speed
is approaching zero, the TOB trace is becoming flat, indicating a sticking behaviour. Each
flat region is followed by a sinusoidal pattern whose frequency varies according to the bit
speed. During this period, the drill pipe is twisted by the rotary table causing the surface
torque to increase as shown in Figure 6. This is followed by a release of the torsional energy
stored during the sticking period causing the bit to speed up to more than twice the rotary
table speed before it slows down and again comes to a complete stop. The period of these
alternating modes of relaxation and torsion is approximately equal to the period of
torsional vibrations. The field data observed by Halsey et al. [9] closely resembles the
results presented in Figure 5. This is interesting, because in this study there are no
externally prescribed stick–slip conditions at the bit/formation interface. Instead, this
behaviour is the result of the model used for TOB and is entirely due to the coupled

1.0

0.8
Normalized displacement

0.6

0.4

0.2

0.0
0 100 200 300 400 500
Normalized time
Figure 10. Unstable lateral vibrations with a fluctuating table speed.
178 . .   . . 
dynamics rather than Coulomb friction alone. It is worth noting that the results presented
here may explain some of the observed stick–slip vibrations with tricone bits.
Since no speed control system is perfect some fluctuation in the rotary table speed is
inevitable. In fact, there are field observations showing a sinusoidally varying table speed
[8]. It is therefore, important to investigate the effect of these speed fluctuations on the
coupled dynamics. The results of this investigation are shown in Figures 7–10. Figures 7
and 8 show a typical response when the nominal table speed is 12 rad/s. The amplitude
and the frequency of the speed fluctuation are 1 rad/s and 4 rad/s, respectively. In this case
the table and the bit speeds are very similar to the ones observed by Brett [8] with some
small harmonics present in the bit speed which is due to the coupling with the lateral
vibrations. Figures 9 and 10 show the response when the frequency vr is decreased to
2 rad/s, which is close to the linear torsional natural frequency (1·846 rad/s). As can be
seen in Figure 9, the growing torsional vibration amplitudes indicate a resonance
behaviour. The instability in the torsional vibrations leads to growing lateral amplitudes
and eventual impacts with the borehole wall, as seen in Figure 10. As in the case of
parametric instability presented in Figures 4 and 5, the energy responsible for the large
lateral motion is derived from the torsional vibrations which are reduced due to friction
with the borehole wall.
As demonstrated, the model presented here can be used to predict the coupled response
of drillstrings for a particular set of operating conditions. It has been found that torsional
vibrations have a strong influence on the lateral instabilities. Since the torsional vibrations
can be measured easily at the surface, the coupled model can be used to monitor and
control the downhole vibrations. The results of the current study explain clearly why the
stick–slip vibrations are eliminated with a torque feedback [8, 9]. Surface torque is a
measure of torsional vibrations, and controlling this torque is equivalent to controlling the
torsional vibrations. When the torsional vibrations are small, there are no stick–slip
vibrations. Furthermore, reducing the torsional vibrations will eliminate a number of
instabilities, thus leading to smoother drilling.

6. CONCLUSIONS
A dynamic model for studying coupled torsional and bending vibrations of drillstrings
has been presented. In the proposed model the WOB and the TOB are assumed to be
functions of the bit rotation. Consequently, the torsional vibrations affect these excitations
leading to self-excited behaviour. The impacts with the borehole wall are accounted for
by using the momentum balance method with appropriate coefficients of restitution and
friction. The impact model is capable of capturing both rolling with and without slip of
the drill collars along the borehole wall.
It has been shown that the non-linear coupling significantly affects the response.
Instabilities may occur at rotary speeds which may not be considered critical from an
uncoupled linear analysis. At certain frequencies, there is significant energy transfer
between the two modes of vibrations. The simulation results are in qualitative agreement
with laboratory and field observations when the so called stick–slip vibrations occur. This
study suggests a new mechanism for this behaviour which has been shown to be due to
static friction or speed dependent TOB. It also explains why controlling the surface torque
instead of rotary table speed eliminates stick–slip vibrations. The model proposed can be
used for predicting the non-linear dynamics as well as controlling the coupled bending and
torsional vibrations in drillstrings.
      179
ACKNOWLEDGMENT

This work was supported by Kuwait University, Research Administration, Grant

Number EM-122.

REFERENCES

1. I. F and J. J. B 1960 American Society of Mechanical Engineers Journal of Engineering
for Industry, May, 129–135. An experimental study of drillstring vibration.
2. P. R. P and D. B. B 1963 American Society of Mechanical Engineers Journal of
Engineering for Industry, May, 187–194. Drill string vibrations due to intermittent contact of
bit teeth.
3. D. W. D and B. L 1968 American Society of Mechanical Engineers Journal of
Engineering for Industry, October, 1–9. Longitudinal and angular drillstring vibrations with
damping.
4. R. F. M and M. B. A 1985 World Oil, March, 101–104. Lateral vibration: the key
to BHA failure analysis.
5. J. K 1993 Proceedings of the 14th Biennial ASME Design Technical Conference on Vibration
and Noise, Albuquerque, NM, DE-60, 355–362. On the generation of bending vibration in a drill
rod.
6. T. V. A and A. K 1988 Society of Petroleum Engineers Drilling Engineering
3, 12–18. An experimental and theoretical study of a coupling mechanism between longitudinal
and torsional drillstring vibrations at the bit.
7. A. K and G. W. H 1988 Society of Petroleum Engineers Drilling Engineering
3, 369–373. A study of slip stick motion of the bit.
8. G. F. B 1992 Society of Petroleum Engineers Drilling Engineering 7, 168–174. The genesis
of torsional drillstring vibrations.
9. G. W. H, A. K and A. K 1988 Proceedings of the 63rd Society of
Petroleum Engineers Drilling Engineering Annual Technical Conference and Exhibition, Houston
TX, 277–282. Torque feedback used to cure slip–stick motion.
10. M. A. E, D. W. D and M. A. V 1997 American Society of Mechanical
Engineers Journal of Energy Resources Technology 119, 11–19. Effect of torsion on stability,
dynamic forces, and vibration characteristics in drillstrings.
11. R. J. S 1989 Ph.D. Dissertation, Massachusetts Institute of Technology, Department of Ocean
Engineering, Bending vibrations of rotating drillstrings.
12. J. K. V, J. W. N and R. J. S 1990 Society of Petroleum Engineers Drilling
Engineering 5, 282–190. Case studies of the bending vibration and whirling motion of drill
collars.
13. J. D. J 1991 Journal of Sound and Vibration 147, 115–135. Non-linear rotor dynamics as
applied to oilwell drillstring vibrations.
14. V. A. D, F. A and A. J 1993 Society of Petroleum Engineers Drilling
and Completion 8, 84–92. Dynamic stability of drillstrings under fluctuating weight on bit.
15. A. B, J. D H, R. D and E. D 1996 American Society of Mechanical
Engineers Journal of Vibration and Acoustics 118, 292–298. Dynamic behavior of a drill-string:
experimental investigation of lateral instabilities.
16. A. P. C and A. S. Y 1997 Journal of Sound and Vibration 206, 243–260. Dynamic
modelling of rotating drillstrings with borehole interactions.
17. Y. A. K and A. A. S 1986 Journal of Sound and Vibration 104, 187–207. Impact
responses of multi-body systems with consistent and lumped masses.
18. A. S. Y and A. P. C 1998 American Society of Mechanical Engineers Journal
of Vibration and Acoustics 120, 47–53. The Efficacy of momentum balance method in transverse
impact problems.
19. Y. W and M. T. M 1992 American Society of Mechanical Engineers Journal of Applied
Mechanics 59, 635–642. Two-dimensional rigid-body collisions with friction.
20. G. D. B and A. C. H 1975 AMC Transactions on Mathematical Software 1, 71–76.
A polyalgorithm for the numerical solution of ordinary differential equations.
21. A. S. Y and A. P. C 1994 Composite Engineering 4, 1143–1152. On the impact
of a spherical indenter and an elastic-plastic transversely isotropic half-space.
180 . .   . . 
APPENDIX A: A LUMPED PARAMETER MODEL OF THE DRILLSTRING
The dynamics of a continuous structure can be described with a finite number of degrees
of freedom if it is assumed that the displacements can be described as a superposition of
finite number of modes. Equivalent system stiffness, damping and mass parameters for
each mode of the simplified model can be obtained from the properties of the original
structure by means of a virtual work approach. In this work, the stabilized section of the
BHA is modelled as a simply supported beam for transverse motion of the collars, and
the whole drillstring is assumed to be fixed at the top, and free at the bit for torsional
motion. The BHA is assumed to be rigid for torsional vibrations. This assumption is
justified since the drillcollars are much stiffer than the drill pipe in torsion. Assuming one
mode approximation for both the transverse and torsional vibrations, the equivalent
system parameters can be obtained as
J = 2rIl2 + (1/3)rIp l3 , m = rp(d o2 − d i2 )l1 /8, (A1, A2)
EIp 4 T(f)p 3 F(f)p 2
mf = prf (d i2 + Ca d o2 )l1 /8, k(f) = − − , (A3, A4)
2l13 2l12 2l1
kT = GIp /l3 , ch = (2/3p)rf Cd do l1 , cv = pmf d o3 l2 /2(dh − do ), (A5–A7)
where r, E and G are the density, Young’s modulus, and shear modulus of the drillstring
material, respectively, I = p(d o4 − d i4 )/64 is the area moment of inertia for the cross-section,
do and di are the outside and inside diameters of the drill collars, respectively, rf and mf
are the density and the viscosity of the drilling mud, Ca is the added mass coefficient due
to the displaced mass of the mud inside the drillstring, Cd is the drag coefficient for the
hydrodynamic damping due to the mud, dh is the borehole diameter, Ip = p(d  4o − 
d 4i )/32
is the polar area moment of inertia of the drill pipe section, and  do and  di are the outside
and inside diameters of the drill pipe, respectively.

APPENDIX B: IMPACT MODEL AND COEFFICIENT RESTITUTION

In impact of rigid bodies, the deformation history is assumed to consist of two phases:
compression phase and restitution phase. The compression phase is the period from the
instant of contact to the point of maximum compression when the relative approach
velocity becomes zero. The restitution phase begins at the end of compression phase and
ends at the instant of separation. In general, for an oblique impact, Mason and Wang [19]
have identified five contact modes: (1) sliding, (2) sticking in compression phase
(C-sticking), (3) sticking in restitution phase (R-sticking), (4) reversed sliding in
compression phase (C-reversed sliding), and (5) reversed sliding in restitution phase
(R-reversed sliding). By adopting the Poisson method for the coefficient of restitution,
these contact modes and the associated radial and transverse impulses can be calculated.
The necessary equations to calculate the impulses are as follows:
1. Sliding: if Pd q (1 + e)Pq ,
C0
Pr = −(1 + e) , Pu = −smPr ; (B1, B2)
B2 + smB3
2. C-sticking: if Pd Q Pq and m q =ms =,
B1 C0 + B3 S0 B3 Pr − S0
Pr = −(1 + e) , Pu = ; (B3, B4)
B1 B2 − B32 B1
      181
3. R-sticking: if Pd Q Pq Q (1 + e)Pq and m q =ms =,
C0 B3 Pr − S0
Pr = −(1 + e) , Pu = ; (B5, B6)
B2 + smB3 B1
4. C-reversed sliding: if Pd Q Pq and m Q =ms =,

$ % $ %
1+e 2smB3 S0 2S0
Pr = − C0 + , Pu = sm Pr − ; (B7, B8)
B2 − smB3 B3 + smB1 B3 + smB1

5. R-reversed sliding: if Pd Q Pq Q (1 + e)Pq and m Q =ms =,

$ %
C0 2S0
Pr = −(1 + e) , Pu = sm Pr − , (B9, B10)
B2 + smB3 B3 + smB1

where

6 7
S0 /=S0 = if S0 $ 0
s= . (B11)
1 if S0 = 0

Here S0 and C0 are the initial values of sliding and compression velocities calculated as
S0 = ru + Rf , C0 = −ṙ, (B12, B13)
B1 , B2 and B3 are constants which depend on the geometry and mass properties as
B1 = 1/(m + mf ) + [R − e0 cos (f − u)]2/J, (B14)
B2 = 1/(m + mf ) + [e0 sin (f − u)] /J, 2
(B15)
B3 = e0 sin (f − u)[R − e0 cos (f − u)]/J. (B16)
Pd , Pq and ms are given as
Pd = (B2 + smB3 )sS0 , Pq = −(mB1 + sB3 )C0 , ms = −B3 /B1 . (B17–B19)
Here m and e are the friction and restitution coefficients, respectively. The friction
coefficient is assumed to be a constant which depends only on the properties of contacting
materials, whereas the coefficient of restitution depends on the impact conditions as well
as the material and geometric properties and is obtained from the contact law given in
reference [21].
It was shown that the coefficient of restitution varies as

6 7
1 b E 0·893
e= , (B20)
0·972b−1/4 b q 0·893

where b is the normalized impact velocity given as

0 1
1/2
15·735(m + mf )(E*)4
b= ṙ, (B21)
p 5d h3 Sy5

where E* is the effective Young’s modulus for the contacting bodies given by
1 1 − n12 1 − n22
= + , (B22)
E* E1 E2
where ni and Ei are the Poisson’s ratio and Young’s modulus for the drill collars and the
casing/formation, respectively, and Sy is the yield strength of the softer material.