International Journal of Mining Science and Technology xxx (xxxx) xxx

Contents lists available at ScienceDirect

International Journal of Mining Science and Technology

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

Drilling signals analysis for tricone bit condition monitoring

Hamed Rafezi ⇑, Ferri Hassani
Department of Mining and Materials Engineering, McGill University, Montreal H3A 0G4, Canada

a r t i c l e i n f o a b s t r a c t

Article history: This paper presents a novel approach to investigate the relations between drilling signals and bit wear
Received 22 May 2020 condition in real world full-scale mining operations. This research addresses the increasing demand for
Received in revised form 30 October 2020 automation in mining to increase the efficiency, safety, and ability to work in harsh environments. A cru-
Accepted 27 December 2020
cial issue in fully autonomous unmanned drilling is to have a system to detect the bit wear condition
Available online xxxx
through the drilling signals analysis in real time. In this work, based on extensive field studies, a novel
qualitative method for tricone bit wear state classification is developed and introduced. The relations
Keywords:
between drilling vibration as well as electric motor current signals and bit wear are investigated and
Drilling
Tricone bit
bit failure vibration frequencies, regardless of the geological conditions, are introduced. Bit failure fre-
Vibration quencies are experimentally investigated and analytically calculated. Finally, the effect of bit design
Wear parameters on the failure frequencies is presented for the application of bit wear condition monitoring
Condition monitoring and bit failure prediction.
Failure prediction Ó 2020 Published by Elsevier B.V. on behalf of China University of Mining & Technology. This is an open
access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction
Drilling and blasting are two primary tasks in mining and these
two operations together, constitute over 15% of the total cost in
surface mining [1]. Understanding the drill bit wear status has
been a challenging issue in drilling, especially when tricone bits
are used. Currently, this task is assigned to the operators, therefore,
it is affected by the operator experience, judgment, and accuracy.
Therefore, there is no practical approach for a consistent bit
replacement action (Fig. 1). In order to achieve fully autonomous
drilling, it is essential to have a condition monitoring system to
determine the bit wear status in real time. Furthermore, changing
the bit before its working life finished would increase the total dril-
ling cost. Excessive bit wear will lead to bit failure and catastrophic
bit failure results in losing bit parts down the hole. When this hap-
pens, the operator will try to recover the detached bit parts from
the hole. This procedure is costly, time consuming, and delays pro-
duction and if the bit parts cannot be recovered, they will remain
in the hole. Eventually, they may mechanically damage the rock
crusher, and this will result in significant costs and delay in
production.
Drilling is a type of mechanical excavation and blasting is a tra-
ditional type of rock fragmentation used to break down formations
into practical particle sizes [2]. The applications of drilling in the
mining industry include blastholes, raise boring, coring, explo-
⇑ Corresponding author.
E-mail address: hamed.rafezi@mail.mcgill.ca (H. Rafezi).
ration, and support installations (e.g. rock bolts). Mechanical dril-
ling is divided into two primary methods: rotary drilling and
percussion drilling. In rotary drilling, the rotational motion is sup-
plied by an electric or hydraulic driven gearbox known as the
rotary head. Another feed motor moves the rotary head and drill
string up and down. This motor generates the pulldown force to
provide adequate weight on bit (WOB). Compressed air is also con-
ducted to the cutting area through nozzles placed on the bit. The
compressed air is used primarily to flush out drilled rock cuttings
from the hole and clean the cutting area so that the bit can con-
tinue penetrating the rock. The air also cools down the cutting area
and may contain particles to lubricate the bit bearings (i.e. when
using tricone bits with open bearings). Percussion drilling uses
hammer energy in addition to rotation to penetrate the rock. In
the top hammer percussion method, the hammer energy is applied
to the drill string, whereas in the down the hole (DTH) method, the
hammer energy is applied directly on the bit. Therefore, the DTH
technique is more suitable for deeper holes. There is no solid
boundary to determine the best drilling method for a given opera-
tion. However, the top hammer is generally preferred for drilling
holes less than 15.24 cm (6 in) in diameter. For hard formations
(unconfined compressive strength >100 MPa), DTH drilling usually
provides a better rate of penetration (ROP). DTH drilling is limited
by the pressurized air supply. For example, a 20.32 cm (8 in) DTH
bit is designed to receive 2500 kPa air pressure to provide enough
impact energy. Therefore, rotary drilling with tricone bits is the
most cost-effective method for larger hole diameters and generally

https://doi.org/10.1016/j.ijmst.2020.12.025
2095-2686/Ó 2020 Published by Elsevier B.V. on behalf of China University of Mining & Technology.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Please cite this article as: H. Rafezi and F. Hassani, Drilling signals analysis for tricone bit condition monitoring, International Journal of Mining Science and
Technology, https://doi.org/10.1016/j.ijmst.2020.12.025
H. Rafezi and F. Hassani
Fig. 1. Two tricone bits (10⅝00 ) labeled as worn-out bit.
suits a wider range of applications and hole sizes in terms of diam-
eter and depth [3].
Although assessing tricone bit wear condition using measure-
ment while drilling (MWD) data can benefit the mining industry,
researches have been limited by the complexities of conducting a
comprehensive field study. MWD data analysis could cover a wide
range of applications from blast fragmentation optimization [4] to
process fault detection. One approach in the literature to study bit
wear relied on measuring energy consumption during drilling. Bit
wear will affect the amount of energy required: a dull tool tooth
requires more penetration force. The energy required using a dull
tool almost doubled to provide the same ROP in comparison with
a sharp tool [5] while the WOB, rotational speed, and bit diameter
were assumed to be constant. The challenge in the implementation
of this approach would be dynamic working conditions and geo-
logical variations.
A second approach used a correlation between axial bit vibra-
tion spectra data and bit tooth wear (Fig. 2) [6]. In Fig. 2, T0 repre-
sents no wear, whereas T8 corresponds to completely worn inserts.
As the inserts become worn, the circumference of the cone
decreases. Therefore, at a constant drill string rotational speed,
the cone will rotate faster to cover the same path. Hence, in the
bit vibration spectrum the frequency, peaks around 20 Hz move
to around 30 Hz as the new bit teeth become totally worn.
Drilling vibration has also been offered as an input for the con-
trol system of a drill rig to keep the vibration levels at an accept-
able level by adjusting drilling parameters [7]. Researchers have
also correlated the drill rod vibration with rock hardness in under-
ground roof bolt drilling [8]. Cooper proposed detecting bit tooth
wear by comparing measured drilling performance with theoreti-
cal bit performance. In his approach, which requires knowledge
Fig. 2. Axial vibration spectra for tricone at different tooth wear grades in a
constant rock type [6].
2
International Journal of Mining Science and Technology xxx (xxxx) xxx
of geological information, a model estimates a theoretical ROP
from drilling parameters and rock strength data. The theoretical
values are then compared with the ROP measured in the labora-
tory; any difference between the two indicates the bit wear state.
Cooper also proposed a similar approach for rock strength mea-
surement, assuming the driller has access to rock strength data
from the drill log. As shown in Fig. 3, the drill log estimation fol-
lows an incremental trend compared to the geological log estima-
tions. The rock strength estimated from drill logs erroneously
increases due to increasing wear on the bit teeth. Cooper’s earlier
research was based on laboratory experiments, whereas his more
recent work is based on real lithological data and synthetic drilling
data. The author emphasizes that precise estimation of ROP based
on rock type—or vice versa—is very challenging in practice because
of the large number of parameters involved [9,10].
Sheppard and Lesage [11] measured the load distribution
between the rows of new and half-worn bits (T0 and T3 states)
in controlled laboratory tests and concluded that the load distribu-
tion is sensitive to tooth wear. A total force of 8 kN was distributed
between the outer and inner rows as 6 and 2 kN, respectively, for
the T0 bit and 5 and 3 kN, respectively, for the T3 bit (i.e. for T0:
outer/inner = 3 and for T3: outer/inner = 1.66). Therefore, the
outer/inner row force distribution ratio decreased as the bit wear
increased to the T3 level. Applying these results to actual opera-
tions is limited by the fact that tooth wear is not uniform within
the same row or among different rows [11]. Other research has
suggested a method for tricone bit performance evaluation using
MWD data. The degradation of production and the initial and final
ROP (before the bit change) was studied in an operating mine in
Sweden by Ghosh et al. [12]. Production degradation was assumed
to be caused by wear progress on the bit. The rock strength varia-
tions in the mine were assumed to be negligible.
In the present work, in order to generate a deep understanding
of the relationship between bit wear and drilling signals, an exten-
sive MWD data analysis is done. Drill rigs were equipped in two
participating mines in Canada to collect real-world, full-scale dril-
ling data in a variety of geological conditions. The drill bit wear
condition was visually inspected during the field measurement to
label the collected data accordingly and a novel qualitative wear
grading method for tricone bits is proposed. The MWD data were
analyzed in time and frequency domains. The rotary motor current
and axial vibration signals were determined to be bit wear sensi-
tive. Bit vibration fault frequencies were mathematically and
experimentally investigated and are introduced for tricone bit
wear monitoring.
Fig. 3. Rock compressive strength from geological logs and drill logs versus depth
[10].
H. Rafezi and F. Hassani
2. Bit wear
During the drilling process, generating rock cuttings as a result
of the contact between the tool and rock causes changes to tool
external and internal components. These changes are known as
bit wear.
The industry standard to measure the insert/tooth wear of tri-
cone bits is introduced by the international association of drilling
contractors (IADC) [13]. Based on this grading, tooth wear is graded
into eighths or increments of 1/8 of the insert missing height (from
0 to 8) as shown in Fig. 4. Therefore, a new bit tooth would have a
grade of zero, a bit with half worn insert would have a grade of
four, and a worn-out bit with no teeth would have a grade of eight.
The possible wear mechanisms for tricone bits include [1]: wear
on the gauge row inserts that usually occurs when drilling in soft-
medium but very abrasive formations, uniform wear on all teeth,
heavy wear on the shirttail that can occur when drilling with a
bent pipe that results in very high side thrust on the bit, broken
inserts on the gauge row, broken lug, worn-out teeth on the surface
of cone that touched the bottom of the drilled hole due to locked
cone, and loss of one or more cones in the hole due to bearing
failure.
The IADC recognizes bearing failure as the dominant failure
mode for tricone bits. The bit must be pulled when there is a good
reason to suspect bearing failure. Premature bearing failure (before
the anticipated bit lifespan) can occur due to unsuitable operation
setpoints, incorrect cutting structure selection, and excessive axial
or torsional vibrations [14]. If the bearing failure is not detected in
time, the chance of cone detachment is high. Risks associated with
leaving detached bit cones in the hole are serious and will lead to
costly ‘‘fishing” efforts.
The lifetime and wear mechanism of tricone bits differ depend-
ing on bit design and working conditions. For example, a tricone
with sealed bearings will last much longer than the open bearing
type in hard formations. The seal does not allow dust and rock par-
ticles to enter the rotational structure, which extends the bearing
life significantly. In soft ground conditions, however, bearing type
does not make the same significant difference. Sealed bearing tri-
cone bits usually experience cone wear when working in hard for-
mations. Continued drilling with a bit with worn cones exerts
excessive forces on the bearing elements and ultimately results
in bearing failure.
Rolling elements comprise the connection between cones and
lugs in tricones and bit catastrophic failure happens as a result of
progressive wear on the bearing elements. Tricone bits are usually
designed in a way that the middle ball bearing holds the cone con-
nected to the lug structure. Therefore, middle ball bearing failure
could result in separation of the cone. Based on the extensive field-
work, studying the operators experiences and inspection of worn
tricone bits in scrap yards at several mine sites, a novel grading
method is introduced as an alternative to the T0 to T8 method
for tricone bits. This classification approach focuses on the rota-
tional structure condition qualitatively to grade the bit wear. This
grading method addresses the dominant failure mode and divides
the bit lifespan into five stages as follows: (1) Class 1: new bit, (2)
Fig. 4. IADC bit tooth wear grading [13].
3
International Journal of Mining Science and Technology xxx (xxxx) xxx
Class 2: appearance of slight wear on teeth tips and cone edges, (3)
Class 3: bearing(s) beginning to loosen because of damaged outer
roller bearing, progressive tooth wear, and missing teeth, (4) Class
4: deterioration stage—loose bearings because of the damaged
outer roller and ball bearing, accelerated bearing, and tooth wear,
and (5) Class 5: failure stage—excessive bearing looseness because
of severe damage to the ball bearing; bit change is required to
avoid catastrophic failure.
The field data in this research were labeled according to this
novel grading method.
3. Drilling vibration
Vibration always exists in drilling operations. High vibration
levels can be detrimental to the operation, causing premature bit
failure, early failure of drill string components, additional wear
and tear to the rotary head and hoist motor, and reduction in the
ROP. There are various sources of vibration in drilling operations.
The overall vibration on the drill string depends on the frequency,
amplitude, distance of the excitation source, and the system damp-
ing [15]. If the excitation frequency is close to the drill string nat-
ural frequency, drill string resonance will occur, and fatigue
loading will result in damage to the system and catastrophic failure
[16]. Drill string vibration comprises torsional, lateral, and axial
vibration [17].
Torsional vibration is excited by the frictional torque applied on
the bit or drill string inside the hole. It will manifest as accelera-
tions and decelerations in string rotary motion, resulting in a
non-uniform rotation of the bit down the hole. The stick–slip phe-
nomenon is an extreme case of torsional vibration and is usually a
concern in long drill strings used in the oil and gas industry. Early
models to investigate the torsional behavior of drill strings used a
torsional pendulum and assumed the pipe inertia to be negligible
[18]. Later studies investigated the effect of the bit-rock interaction
on the stick–slip phenomenon—a jerking motion when two objects
slide against each other [19], but are limited by simplifications and
assumptions [20]. Stick-slip vibration frequency in the drill string
is usually between 0.05 and 0.50 Hz [21].
Lateral vibration will cause the bit and drill string to rotate
around an axis other than the string geometrical center, causing
the bit to repeatedly strike the hole wall. The extreme case of lat-
eral vibration is bit whirling or walk of the bit around the hole.
Whirling mostly occurs with polycrystalline diamond compact
(PDC) drill bits, which are broadly applied in the oil and gas drilling
industry. Lateral vibration is usually damped along the drill pipe
and would not be seen on the surface. However, due to interactions
with the hole wall, high amounts of whirling will cause surface
detectable torsional and axial vibration [22]. Depending on the
string rotational speed and the number of cutters on the PDC bit,
whirling ranges between 5 and 100 Hz [21].
Axial vibration occurs as a result of bit-rock interactions. Bit
bouncing occurs when the WOB lifts the bit off the hole bottom
and then the bit drops repeatedly. This phenomenon is usually
associated with tricone bit drilling. Bit bouncing can generate an
excitation three times the tricone bit rotational speed. The fre-
quency of axial vibration is usually 1–10 Hz. Bit bouncing can be
sensed at the surface when drilling at shallow depths [21]. This
interaction adds a dynamic part to the axial load and causes fluctu-
ations in the actual WOB [23].
Indentions formed in the rock by bit teeth are also a source of
axial vibration. Laboratory tests by Ma and Azar to determine the
contact condition on roller cone bit teeth down the hole and tooth
velocity and position showed that the relationship between the
depth of penetration (DOP) and bit rotation exhibits a repeated
step shape (Fig. 5). A series of craters are formed under the bit after
H. Rafezi and F. Hassani
Fig. 5. DOP versus revolutions of bit [24].
a given number of rotations, then the bit suddenly drops and
begins creating a new series of indentions.
The step height (Dp) is the mean depth of the craters down the
hole and is calculated as [24]
Dp ¼ ROP  nx =N ð1Þ
where ROP is the rate of penetration in m/min; nx the number of
rotations; and N the bit rotational speed in rpm.
In a simplified model of tricone bit-rock interactions, Sheppard
and Lesage assumed that at an instant, only one tooth from each
row of a cone is in contact with the rock and the bit load is equally
distributed among all rows of the cone [11]. Their laboratory tests
measuring the torque and forces on every row of an instrumented
tricone bit showed that the rotational speed of the cones around
the bearing axis is 1.25–1.31 times that of the bit [11]. Hardage
used these results to define a periodic behavior for tricone bits
[25].
The axial forces on the tooth rows of a tricone bit can be approx-
imated by Eqs. (2)–(4) [17].
F ðtÞ  F m þ F 0 sinxrow t ð2Þ
F m ¼ ðF max þ F min Þ=2 ð3Þ
F 0 ¼ ðF max  F min Þ=2 ð4Þ
where F(t) is the periodic axial forces; Fm and F0 the mean and
dynamic forces, respectively; xrow the cone angular velocity; and
Fmax and Fmin the maximum and minimum amounts of force
applied on a row of the bit, respectively.
In 2012, Naganawa derived an equation of motion for tricone bit
vertical displacement during the drilling from the equilibrium of
forces (WOB and tooth forces interacting with rock (Fr,ijk)) exerted
on the bit rigid body [6]. The suffix ijk represents the k tooth on the
row j of bit cone i. Assuming the bit as an elastic model with one
degree of freedom and a spring constant of Ks, the WOB consists
of a static weight (W0) and a dynamic component (Fig. 6). Having
Fig. 6. Dynamic model of a tricone bit [6].
4
International Journal of Mining Science and Technology xxx (xxxx) xxx

zb as the axial motion of the bit and setting z b as the equilibrium
position when the dynamic force is not applied, the equation of
motion is stated as [6]


 Xn X X
nr;i nt;ij
W 0  K s zb  z b  F r;ijk ¼ 0 ð5Þ
i¼1 j¼1 k¼1
Since the tooth forces change with bit penetration, in a realistic
representation of bit dynamics, the force needs to be formulated as
a function of time.
4. Method and experiments
The main goal of this study was to determine if a relationship
existed between drilling signal behavior and bit wear by analyzing
the vast MWD dataset from several benches at two mines. Wear-
labeled data obtained from frequent inspection of bit wear grade
during the operation was used to correlate signal trends with bit
changes over time. To achieve a deeper understanding of the sig-
nals and obtain signal features, time domain statistical analysis,
and frequency spectrum analysis were applied to the signals. Sig-
nals were exported to MATLAB software for analysis.
MWD signals were preliminarily studied in the time domain. To
analyze MWD data as random signals statistically, moments of the
random variable were extracted, and their trends were assessed
during the bit lifespan. The probability distribution of X as a ran-
dom discrete signal is characterized by determining the probabili-
ties that random variable X = xi for every data point (xi). The
probability distribution of a random discrete variable X is
described by P[X = xi]. Four moments of random signal i.e. mean,
variance, skewness, and kurtosis were calculated as signal descrip-
tive features. In addition, to study the frequency spectrum of vibra-
tion signals and to provide insights regarding the signs of bit wear,
fast Fourier transform (FFT) was applied to transform the signals
from the time to frequency domain.
The sensors, data acquisition, signal processing, and decision-
making units are the components of a condition monitoring (CM)
system (Fig. 7). The type and number of sensors depend on the nat-
ure of the process and the target of the CM system. Sensors include
a wide range of instruments (e.g. thermometers, accelerometers,
and acoustic emission transducers).
In this research, Bucyrus 49 HR drill rigs (Fig. 8) were used (cur-
rently recognized under the name CAT MD6640). The machine can
provide up to 64000 kg of pulldown bit load and covers a range of
blasthole diameters from 244 to 406 mm (9.6 to 16.0 in). This rig is
capable of angle hole drilling up to 25° in increments of 5°. In blast-
hole drilling, however, vertical hole drilling is more common. The
rig unit can drill to a depth of 21.3 m in single-pass and up to
85.3 m in multiple passes [26].
The rigs were equipped with two DATAQ Model DI-718Bx-S
units installed inside the control cabinet. Each unit has 16 signal
conditioned analog inputs and a sampling frequency up to
14400 Hz to record the digitized signal on a memory card.
The MWD data recorded using the DATAQ units include: hoist
motor current, rotary motor current, head encoder, bailing air pres-
sure (set-point), hoist voltage (set-point), rotary voltage (set-
point), lower-mast vibration, and upper-mast vibration signals.
The hoist voltage request and rotary voltage request determine
the WOB and rotational speed, respectively. The vibration is sensed
using the two accelerometers installed at approximately 2/3 the
drill mast height as well as the mast base (Fig. 9).
At a preliminary phase, constant setpoints (i.e. WOB and rota-
tional speed) were applied on the bits in a bench with relatively
uniform geology. The four configurations are given in Table 1.
H. Rafezi and F. Hassani
Fig. 7. Bit CM diagram.
Fig. 8. Bucyrus 49 HR drill rig at Bloom Lake mine.
Fig. 9. Accelerometer on the mast base with the protective cover.
Table 1
Drilling configurations at the preliminary phase.
Configuration # WOB (kN) Rotational speed (rpm)
1 415 60
2 415 80
3 300 60
4 300 80
In the next phase, in comprehensive fieldwork, the normal
operation of the Bucyrus 49 HR drill on several patterns yielded
more than 16600 m of blastholes and drilling signals correspond-
ing to the entire life cycle of five tricone bits. Working at different
benches provided drilling data in a wide variety of working condi-
tions and geology formations. Bit wear condition corresponding to
the MWD signals was assessed and recorded for data labeling.
The first bit in operation was a 12¼00 Sandvik sealed-bearing tri-
cone bit (Fig. 10a). It drilled a total length of 7158 m and failed
after 157.9 h of operation. In the last few shifts of operation before
the failure, the heel and gauge inserts were missing and eventually
bearing failure occurred on all three cones.
The second Sandvik bit was a trimming bit with a diameter of
10⅝00 . It drilled a total of 3435 m in 76.6 h of operation. It began
losing inserts from the center row of the cones and during the sub-
5
International Journal of Mining Science and Technology xxx (xxxx) xxx
sequent two shifts, the bit bearings failed catastrophically, losing
the three cones (Fig. 10b).
Three 10⅝00 Rotacan bits were tested. The first one drilled
1841 m during a total working time of 56 h. It experienced failure
of the roller and ball bearings in one cone (Fig. 10c). The second
one drilled 1726 m during a total working time of 47.7 h. It ulti-
mately ended up with loose bearings on all three cones and was
pulled. The third bit drilled 2805 m during 68 h of operation. It
maintained a healthy condition until two shifts before failure when
it started losing inserts on the gauge row. After more than 80% of
gauge row inserts were gone, bearing failures occurred on two
cones.
Overall, the dominant failure mode in the field was bearing fail-
ure; even when other wear mechanisms were happening, the bit
usually experienced bearing failure if it was not pulled by the oper-
ator. Over-usage of a bit will result in direct production losses
including lower rates of penetration, decreased hole finishing pre-
cision, and long-term costs for the operation (i.e. maintenance and
downtime).
5. Results and discussions
5.1. MWD analysis
The MWD data were labeled by frequent inspection and accord-
ing to the novel tricone wear grading method introduced in Sec-
tion 2 in order to relate signal trends with bit changes over time.
The preliminary signal to discuss was the ROP, followed by the
rotary electric motor current and the drilling vibration.
5.2. ROP analysis
The effect of WOB and rotational speed on the ROP was evalu-
ated based on the test configurations used (Table 1) and data col-
lected in the preliminary fieldwork. The ROP trends presented in
Fig. 11 correspond to a new Rotacan tricone bit drilling on a rela-
tively uniform bench. Therefore, the ROP variation is assumed to
solely result from WOB and rotation speed values. By having
healthy bit nozzles and enough air bailing pressure and volume,
the perfect cleaning condition is assumed to be valid.
The ROP was increased by increasing the WOB and rotation
speed as illustrated in Fig. 11. Increasing the WOB from 310 to
420 kN at 60 rpm increased the average ROP by 32%. By compar-
ison, increasing the rotational speed from 60 to 80 rpm at 310
kN increased the average ROP by <15%.
Field data under controlled working conditions showed that the
average ROP differed little (<4%) between drilling with Class 1 ver-
sus Class 3 bits. However, the average ROP was 22% lower for the
Class 5 bit than the Class 1 at a uniform geological condition
(Fig. 12).
In a routine mining operation, however, at a dynamic subsur-
face medium, the ROP is dominated by the geological condition
and a decremental trend in ROP as a result of bit wear is not iden-
tified (Fig. 13).
H. Rafezi and F. Hassani
Fig. 10. Tricone bits in various conditions.
Fig. 11. ROP at different setpoints at a uniform geological condition.
Fig. 12. Measured ROP at maximum setpoint values (WOB = 420 kN, N = 80 rpm) for
Class 1, Class 3, and Class 5 bits.
Fig. 13. ROP trend during the lifetime of a tricone bit.
5.3. Motor electric current analysis
Analysis of the preliminary field data on electric current signals
of rotary and hoist motors found that the rotary motor current was
sensitive to bit wear in uniform geological conditions. As bit wear
6
International Journal of Mining Science and Technology xxx (xxxx) xxx
increased, the current signal scattered and fluctuated compared to
the signal generated when drilling with a healthy bit (Fig. 14).
A similar pattern with a higher intensity was observed for the
progressively wearing bit. To quantify this behavior, the trend in
signal statistical features was analyzed. Among these features,
the root mean square (RMS) and the variance of the rotary motor
current signal demonstrated meaningful trends during the bit lifes-
pan, as shown in Fig. 15a and b, respectively. The RMS showed an
incremental trend in the initial wear stages (15% increase from
Class 1 to 3) and in the bearing wear failure zone, a significant
jump in RMS was observed (76% increase from Class 1 to 5) as plot-
ted in Fig. 15a. The defective bearings affected by the progressive
wear, require more torque to maintain a stationary rotational
speed. Progressive wear in the rolling elements resulted in a sub-
stantial increase in the signal variance (over 1300% increase from
Class 3 to 5) as plotted in Fig. 15b. These trends in rotary motor
current signal features are potential tools for implementation in
a tricone bit wear monitoring system.
Eq. (6) explains the relationship between electric current and
torque in the electric rotary motor where more current is required
to maintain a constant rotational speed at a higher torque.
T ¼ 30IVEf =pN ð6Þ
where T is the torque in Nm; I the current in A; V the applied
voltage in V; Ef the motor efficiency; and N the rotational speed
in rpm.
5.4. Vibration analysis
Vibration data collected from the accelerometers were analyzed
in the frequency spectrum to assess the effects of bit condition on
frequency components in different working conditions. Fast Four-
ier transform (FFT) was applied to transform the signals and ana-
lyze the signs of bit wear and failure in the frequency domain.
5.4.1. Drill pipe natural frequency
The drill string—including the pipe(s) and the tricone bit
model—built in SolidWorks software was imported to ANSYS for
modal analysis. In the study, the drill strings consisted of two
pipes. However, shorter and longer strings were also analyzed to
determine their natural frequency. The first and second frequency
modes of axial vibration in three types of boundary conditions for a
string consisting of one, two, and three pipes are presented in
Tables 2 and 3, respectively. For a drilling depth of approximately
15 m, two drill pipes are required. The rotational speed range is 60
to 120 rpm which is equal to 1–2 Hz. Therefore, the axial vibration
H. Rafezi and F. Hassani International Journal of Mining Science and Technology xxx (xxxx) xxx

Fig. 14. Time domain signal of the rotary motor current signal for a Class 1 and Class 3 bit.

Fig. 15. RMS and variance of rotary motor current for the three bit classes.

Table 2
First fundamental frequency (Hz) for drill string consisting of one, two, and three gresses to Class 2 and increases this frequency component. This
pipes. conclusion, however, is based on a uniform contact force distribu-
Boundary condition One pipe Two pipes Three pipes tion, which cannot be generalized to geological conditions in blast-
Fixed top – fixed bit 104.09 29.57 13.74
hole drilling. Therefore, application of this frequency component to
Fixed top – supported bit 71.56 20.33 9.44 wear detection is not practical.
Fixed top – free bit 16.36 4.65 2.16 Based on the comprehensive field data collected in a variety of
geological conditions, the axial vibration 3 rpm frequency peak or
tricone bit bouncing frequency was determined to be the forma-
Table 3 tion drillability indicator. At a constant bit wear level and the same
Second fundamental frequency (Hz) for drill string consisting of one, two, and three drilling setpoints (WOB and rpm), a decrease in the ROP resulted
pipes.
from hitting harder rock formations. It was observed that drilling
Boundary condition One pipe Two pipes Three pipes in harder formations would excite the 3 rpm component in the
Fixed top – fixed bit 286.72 81.44 37.84 axial vibration spectrum.
Fixed top – supported bit 232.35 66.00 30.66 A series of harmonics of the cone rotational speed (CRS) was
Fixed top – free bit 104.09 29.57 13.74 found in the axial vibration frequency as the bit reached Class 3
wear. These peaks start from 2x CRS and were detectable up to
approximately 70 Hz. In addition, the frequency band ranging from
fundamental frequencies of the drill string in all the three bound- 40 to 60 Hz was strongly excited by a worn bit starting at Class 3
ary conditions are well above the pipe rotation frequency. (Fig. 16a). This frequency range followed an incremental trend as
the bit reached Class 4 wear and increased up to 300% when the
5.4.2. Drilling vibration frequency spectrum analysis bit reached Class 5 wear (Fig. 16b).
Analysis of vibration signals collected during the lifespans of
bits in the frequency domain using FFT showed that the amplitudes 5.4.3. Tricone bit vibration frequencies
of some frequency bands were affected by bit wear condition Bearings comprise the connection between cones and lugs in
changes. These vibration excitations occurred as bit wear tricone bits. The names of these bearings as used in this paper
approached the worn condition (or potential catastrophic failure) are introduced in Fig. 17. Each bearing has unique fundamental fre-
regardless of changes in geology and working conditions. In addi- quencies based on its design, geometry, and speed of operation.
tion, low-frequency vibration was sensitive to geology and work- These frequencies are calculated using Eqs. (7)–(10) [27].
ing condition. BPFI ¼ ðNBR=2Þ  U  ð1 þ ðB=Pd Þ  coshÞ ð7Þ
Vibration signals in the lateral and axial directions from the
accelerometers were analyzed in the frequency domain to locate BPFO ¼ ðNBR=2Þ  U  ð1  ðB=Pd Þ  coshÞ ð8Þ
the signal frequency bands sensitive to bit wear. Results from the
preliminary fieldwork showed that tooth wear, which is geometri- FTF ¼ ðU=2Þ  ð1  ðB=Pd Þ  coshÞ ð9Þ
cal changes on the teeth and tooth breakage, results in a non-

uniform distribution of cutting forces exerted on each cone. This BSF ¼ ðPd =2BÞ  U  1  ððB=Pd Þ  coshÞ2 ð10Þ
phenomenon unbalances the rotation and excites the 1 rpm in
the axial vibration frequency spectrum. Therefore, the wear pro-
7
H. Rafezi and F. Hassani International Journal of Mining Science and Technology xxx (xxxx) xxx

Fig. 16. Bit axial frequency spectrum for a Class 3 and Class 5 bit.

Fig.18. Class 5 bit vibration frequency spectrum at 60 rpm and 260 kN of force.

Fig. 17. Bit rotational structure: the outer and inner roller bearings as well as the According to Eqs. (11) and (12), bit design parameters have a
ball bearing. minor effect on the fault frequencies and scaling the size of the
bit components (e.g. balls or rollers) and pitch diameter does not
shift the failure frequencies. Depending on the bit geometrical
where BPFI is the ball/roller pass frequency of inner race in Hz;
design, the CRSR usually ranges from 1.25 to 1.31, which is equiv-
NBR the number of balls/rollers; U the rotational speed difference
alent to a maximum 5% potential growth in the frequency value.
between outer and inner race in Hz; B the ball diameter in mm;
The most influential parameter on bit fault frequencies is bit rota-
Pd the pitch diameter in mm; h the bearing contact angle; BPFO
tional speed, which can range from 50 to 150 rpm in soft forma-
the ball/roller pass frequency of outer race in Hz; FTF the funda-
tions like siltstone and 40 to 80 rpm for extremely hard
mental train frequency in Hz; and BSF the ball/roller spin frequency
formations like hematite and quartzite. In practice, the most com-
in Hz.
monly used rotational speed range is 60–90 rpm [3]. Taking both
During the drilling operation, as a bit reaches Class 3 condition,
rpm and CRSR ranges into consideration, tricone failure frequen-
damage occurs to the cone and lug edges. Therefore, the outer race-
cies range from 45 to 77 Hz. Fig. 19 shows the growing trend of
way of the outer roller bearings on each cone is initially prone to
the bit failure frequencies based on the bit rotational speed at
damage. Field data analysis shows that in a Class 3 bit, due to dam-
two extreme values of CRSR based on the bit design at a fixed con-
age to the outer race on the outer roller bearing, the harmonics of
tact angle equal to 30°.
BPFO of the outer roller bearing are excited and the 5x harmonic
Because of complex loading conditions in bearings, accurate
can be clearly sensed on the drill mast. This frequency component
measurement of the contact angle is not feasible. However, some
is the named outer roller bearing failure frequency (ORBF).
theoretical approaches have been developed to study the contact
As the operation continues, the loose bearings with higher
mechanism. The relationship between contact angle and parame-
clearance allow dust and tiny rock chips to penetrate the bearings
ters including axial force, rotational speed, and friction coefficient
mechanism. Even tricone bits with sealed bearings experience
in the bearing for various applications has been a topic of interest
sealing breakage and are not safe from bearing deterioration. As
for researchers [28].
the fault reaches the middle ball bearing in a Class 4 bit, the har-
To address the uncertainty related to contact angle, the entire
monics of BPFO of the ball bearing are excited and the 5 harmonic
range of 5° to 45° was considered in the tricone bit bearing failure
is detectable on the mast and this frequency is named the middle
analysis. Fig. 20 shows the effects of contact angle variation at its
ball bearing failure frequency (MBBF). Excessive wear on a ball
bearing leads to Class 5 wear. Failure of the middle ball bearing will
result in bit catastrophic failure and possibly detachment of the
cone.
Eqs. (11) and (12) are proposed to calculate tricone bit failure
frequencies.

ORBF ¼ ðNR=24Þ  N  CRSR  ð1  ðR=PRBÞ  coshÞ ð11Þ

MBBF ¼ ðNB=24Þ  N  CRSR  ð1  ðB=PBBÞ  coshÞ ð12Þ

where ORBF is the outer roller bearing failure frequency in Hz;
NR the number of rollers; CRSR the cone to bit rotational speed
ratio; R the roller diameter in mm; PRB the roller bearing pitch
diameter in mm; h the bearing contact angle; MBBF the middle ball
bearing failure frequency in Hz; NB the number of balls; and PBB
the ball bearing pitch diameter in mm.
Failure frequencies of a Class 5 bit at a rotational speed of
60 rpm and 260 kN of operating force are illustrated in Fig. 18. Fig. 19. Failure frequencies trend based on bit rotational speed and CRSR.

8
H. Rafezi and F. Hassani
Fig. 20. Effect of bearing contact angle on ORBF and MBBF at 70 rpm rotational
speed and a CRSR of 1.3.
extreme limits on bit failure frequencies. The ORBF and MBBF
are affected by less than 7% and 2%, respectively. Consequently,
the failure frequency range does not exceed 54 and 60 Hz,
respectively.
6. Conclusions
The goal of the present paper is to study the drilling signal for
the purpose of tricone bit wear detection in mining industry appli-
cations. The analysis is based on real-world drilling data collected
from two Canadian mine sites. This paper reviews previous works
towards an understanding of bit wear condition from drilling sig-
nals and drilling vibration analysis. The accomplishments of this
work are summarized as:
(1) From the two fieldworks, an extensive bit wear-labeled dril-
ling dataset was generated, including multi-spot high-
frequency vibration.
(2) In the preliminary fieldwork, a drill rig in an iron ore mine
was equipped with a data acquisition unit and accelerome-
ters on several spots of the machine. Drilling signals were
measured during drilling with tricone bits at three levels
of wear condition.
(3) A comprehensive fieldwork was conducted in a copper mine.
Two heavy-duty high-frequency accelerometers were
installed on the drill rig in over 16 km of blasthole drilling.
Bit wear condition over the complete lifecycle of tricone bits
was inspected and documented during the entire MWD.
(4) A novel qualitative wear grading method for tricone bits is
introduced. The method focuses on bit catastrophic failure
caused by bit internal bearing failure. This method classifies
a new bit as Class 1 and a worn-out bit as Class 5.
(5) Rotary motor current signal statistical features are found to
be sensitive to bit wear. Rotary motor current RMS follows
an incremental trend with the progress of the wear and
the variance is sensitive to bearing deterioration. 52% of
increase in RMS and about 1300% growth in variance are
measured as the bit reaches Class 5 from Class 3 condition
in uniform geological condition.
(6) In axial vibration, frequency bands sensitive to bit wear and
drillability are identified. Bit failure frequencies are intro-
duced independent from the geological conditions. Mathe-
matical equations are proposed to calculate the failure
frequencies based on bit design parameters.
The results of this study support the development of bit condi-
tion monitoring systems in drilling that could be implemented in
blasthole drill rigs by mining operations and rig manufacturers.
9
International Journal of Mining Science and Technology xxx (xxxx) xxx
Declaration of Competing Interest
The authors declare that they have no known competing finan-
cial interests or personal relationships that could have appeared
to influence the work reported in this paper.
Acknowledgements
The authors appreciate generous supports from Canada Natural
Sciences and Engineering Research Council (NSERC - CRD grant#
461514, NSERC - I2I grant# 516232), McGill University Engine Cen-
tre as well as Teck Resources, ArcelorMittal and Rotacan
companies.
References
[1] Gokhale BV. Rotary drilling and blasting in large surface mines. CRC Press;
2011.
[2] Hartman HL, Mutmansky JM. Introductory mining engineering. John Wiley &
Sons; 2002.
[3] AtlasCopco. Blasthole drilling in open pit mining, 3rd ed.; 2012.
[4] Park J, Kim K. Use of drilling performance to improve rock-breakage
efficiencies: A part of mine-to-mill optimization studies in a hard-rock mine.
Int J Min Sci Technol 2020;30(2):179–88.
[5] Falconer IG, Burgess TM, Sheppard MC. Separating bit and lithology effects
from drilling mechanics data. In: Proceedings of the 1988 SPE/IADC Drilling
Conference. Dallas: Society of Petroleum Engineers; 1988.
[6] Naganawa S. Feasibility study on roller-cone bit wear detection from axial bit
vibration. J Petrol Sci Eng 2012;82:140–50.
[7] Aboujaoude CE. Feedback control of vibrations in surface rotary blasthole
drilling. Doctoral dissertation. McGill University 1997.
[8] Liu S, Fu M, Jia H, Li W, Luo Y. Numerical simulation and analysis of drill rods
vibration during roof bolt hole drilling in underground mines. Int J Min Sci
Technol 2018;28(6):877–84.
[9] Cooper GA, Lesage M, Sheppard M, Wand P. Interpretation of tricone drill bit
vibrations for bit wear and rock type. In: Proceedings of the 1987 Rapid
Excavation and Tunneling Conference. New Orleans, LA: Soc of Mining
Engineers of AIME; 1987.
[10] Cooper GA. A proposal for the real-time measurement of drill bit tooth wear.
Trans - Geothermal Resources Council 2002:189–92.
[11] Sheppard MC, Lesage M. Forces at the teeth of a drilling rollercone bit: Theory
and experiment. In: Proceedings of the 1988 SPE Annual Technical Conference
and Exhibition. Houston, TX: Society of Petroleum Engineers; 1988.
[12] Ghosh R, Schunnesson H, Kumar U. Evaluation of operating life length of rotary
tricone bits using Measurement While Drilling data. Int J Rock Mech Min Sci
2016;83:41–8.
[13] Halliburton. Roller Cone Drill Bits. IADC Dull Grading; 2009.
[14] Cobb T, Scott D, Dirksen R, Dourfaye A, Fleming C, Rivera R, et al. Drilling
Manual. 12th edition, 2015.
[15] Macpherson JD, Mason JS, Kingman JEE. Surface measurement and analysis of
drillstring vibrations while drilling. In: Proceedings of the 1993 SPE/IADC
Drilling Conference. Amsterdam, Neth: Society of Petroleum Engineers; 1993.
[16] Reid D, Rabia H. Analysis of drillstring failures. In: Proceedings of the 1995
SPE/IADC Drilling Conference. Amsterdam, Neth: Society of Petroleum
Engineers; 1995.
[17] Poletto F, Miranda F. Seismic while drilling: Fundamentals of drill-bit seismic
for exploration. Handbook of Geophys Exploration: Seismic Exploration
(Volume 2004;35):Elsevier.
[18] Lin YQ, Wang YH. Stick-slip vibration of drill strings. J Eng Industry 1991;113
(1):38–43.
[19] Besselink B, Van De Wouw N, Nijmeijer H. A semi-analytical study of stick-slip
oscillations in drilling systems. J Comput Nonlinear Dyn 2011;6(2):021006.
[20] Ghasemloonia A, Geoff Rideout D, Butt SD. A review of drillstring vibration
modeling and suppression methods. J Petrol Sci Eng 2015;131:150–64.
[21] Aadnoy BS, Cooper L, Miska S, Mitchell RF, Payne ML. Advanced drilling and
well technology. SPE 2009.
[22] Christoforou AP, Yigit AS. Active control of stick-slip vibrations: The role of
fully coupled dynamics. In: Proceedings of the SPE Middle East Oil Show.
Bahrain: Society of Petroleum Engineers; 2001.
[23] Dunayevsky V, Abbassian F, Judzis A. Dynamic stability of drillstrings under
fluctuating weight on bit. SPE Drill Complet 1993;8(2):84–92.
[24] Ma D, Azar J. Dynamics of roller cone bits. J Energy Res Technol 1985;107
(4):543–8.
[25] Hardage BA. Crosswell seismology and reverse VSP. Seismic applications
series, Vol. 1. Geophysical Press; 1992.
[26] Caterpillar. Cat Ò MD6640 rotary blasthole drill; 2016.
[27] Graney BP, Starry K. Rolling element bearing analysis. Mater Eval 2012;70
(1):78–85.
[28] Wang F, Jing M, Fan H, Wei Y, Zhao Y, Liu H. Investigation on contact angle of
ball bearings. In: Proceedings of the Institution of Mechanical Engineers, Part
K: Journal of Multi-body Dynamics. SAGE Publications Ltd; 2017.p.230–51.