Vibration 05 00016 v3

vibration
Article
Observer-Based H∞ Controller Design for High Frequency
Stick-Slip Vibrations Mitigation in Drill-String of Rotary
Drilling Systems
Rami Riane 1 , Mohamed Zinelabidine Doghmane 1,2 , Madjid Kidouche 1 and Sofiane Djezzar 3, *
1 Laboratoire d’Automatique Appliquée, Faculté des Hydrocarbures et de la Chimie, Université M’Hamed

Bougara de Boumerdes, Boumerdes 35000, Algeria; r.riane@univ-boumerdes.dz (R.R.);
m.doghmene@univ-boumerdes.dz (M.Z.D.); mkidouche@univ-boumerdes.dz (M.K.)
2 Reservoirs Evaluation Department, OPE, Exploration Division, Sonatrach, Hassi Messaoud 30500, Algeria
3 Petroleum Engineering Department, University of North Dakota (UND), Grand Forks, ND 58202, USA
* Correspondence: sdjezzar@undeerc.org
Abstract: The drilling process is among the most crucial steps in exploration and production activities
in the petroleum industry. It consists of using mechanical mechanisms to crush rocks by the drill
bit to pass through the different geological layers. The drill-string continuously transforms the
rotational movement from the top drive motor to the drill bit through the drill pipes. Due to the
strong interactions with the rocks, aggressive vibrations can arise in the drill-string in its three
dimensions, and consequently, this may create three types of synchronous vibrations: axial, lateral,
and torsional. The severe status of the latter is known as the stick-slip phenomenon, and is the most
common in rotary drilling systems. Based on field observations, it has been inferred that the high
frequency stick-slip vibrations may lead to drill-string fatigues and even to premature rupture. In the
best case, it reduces the drilling efficiency by decreasing the rate of penetration, due to which the
Citation: Riane, R.; Doghmane, M.Z.;
drilling operations become proportionally expensive. The main novelties of this research work are the
Kidouche, M.; Djezzar, S. Observer-
design of an H∞ observer-based controller to mitigate the high frequency stick-slip vibrations, and
Based H∞ Controller Design for
High Frequency Stick-Slip Vibrations
the quantitative analysis of the vibrations’ severity for ten degrees of freedom model. The observer
Mitigation in Drill-String of Rotary is designed to estimate the non-measurable rotational velocity of the drill bit due to the severity of
Drilling Systems. Vibration 2022, 5, the vibrations, while the controller is dedicated to suppressing the vibrations by using the top drive
264–289. https://doi.org/10.3390/ inputs. Thus, many scenarios have been considered to test and analyze the observer performance and
vibration5020016 the controller robustness. Furthermore, a comparison with the LQG observer-based controller has
Academic Editor: Rajamohan
been conducted, where H∞ has demonstrated better efficiency in suppressing the stick-slip vibrations
Ganesan under unstructured perturbations.
Received: 5 March 2022 Keywords: petroleum industry; high frequency stick-slip vibrations; h∞ observer-based controller;
Accepted: 16 April 2022
rotary drilling systems; drill bit velocity
Published: 20 April 2022
Publisher’s Note: MDPI stays neutral

with regard to jurisdictional claims in
published maps and institutional affil- 1. Introduction
iations.
The drilling process is the ensemble of all operations that allow the digging of holes in
the earth either vertically, deviated, or even horizontally. The petroleum industry generally
uses rotary drilling systems to ensure the production of hydrocarbons (Figure 1a). Since
this process is one of the main essential parts of the well’s total cost, its efficiency, reliability,
Copyright: © 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
and performance are very important features for deep well drilling. The occurrence of
This article is an open access article
drill-string vibrations is one of the main restrictive factors of these drilling aspects because
distributed under the terms and
it reduces the borehole quality and generates premature wears of the drilling equipment.
conditions of the Creative Commons Moreover, in extreme cases, it encourages the drill-string breakage by holding the drilling
Attribution (CC BY) license (https:// operations. Therefore, mastering the dynamics of such vibrations and its suppression in a
creativecommons.org/licenses/by/ robust way have been attracting a lot of interest in the last few years by researchers due to its
4.0/). great economic outcomes in the petroleum industry [1]. To this end, the most common type
Vibration 2022, 5, 264–289. https://doi.org/10.3390/vibration5020016 https://www.mdpi.com/journal/vibration

Vibration 2022, 5 265
of vibration, known as stick-slip, has been considered in this research work. The stick-slip is
primarily a torsional vibration generated by the drill-string elasticity and the discontinuous
nature of the torque on bit (Tob) [2]. Even though the diameter of the borehole can be larger
than the drill pipes’ diameter, there will often be a direct contact between the drill-string
and the borehole of the well. This is caused by two main reasons: the first is due to the
bending of the long drill-string, which is caused by its weight and imbalance centrifuge
forces, while the second reason is due to the fact that the well deviation has become more
considerable due to deep target reservoirs. Hence, the frictional contact forces can be
considerable enough to cause the stick-slip vibrations along the drill-string.
Aiming to overcome the repercussions of the stick-slip vibrations appearance during
drilling, many contributions and approaches have been developed in the last few years.
From a practical point of view, the drillers have adopted a method based on manual drilling
parameters’ manipulation by reducing the Wob and increasing the top drive rotational
velocity (or torque). Even this procedure can free the drill bit from the stuck phase, it
takes considerable time in which the borehole can face too many vibrations [3]. Thence,
more reliable and robust automatic approaches have been developed and introduced as
plug-in software solutions. More accurately, the soft torque rotary system (STRS) was first
introduced in [4], and it is based on eliminating the stick-slip vibrations by the top drive
torque manipulation in a closed-loop form. This solution was subsequently adopted by
other companies under different names [5,6]. However, it necessitates a direct measurement
of the drill bit velocity [7], which is not possible if the measurement while drilling (MWD)
tool is not logged. From an academic point of view, several papers have been published
about different control strategies for stick-slip vibrations mitigation in the drill-string [8,9],
most of which have used a two degrees of freedom model of the rotary drilling system in
the controller design [10–14]. This model has proven to be simple, practical, and sufficient,
but only if the supposition given by [15], which indicates that stick-slip occurs at low
frequencies, is verified. A model with several degrees of freedom has been proposed
by [16], and a non-linear model by [6]. The latter works assumed that the drill bit angular
velocity is measurable, which necessitated the implementation of MWD tools for real-
time control purposes. More recent works in this topic proposed the use of observers
for drill bit velocity (and torque) estimation [17], then stick-slip mitigation based on PI
controller [18,19]. In these works, the top drive velocity and torque have been considered
as the control input signals to mitigate the stick-slip vibrations in the drill-string. Even
the observer estimations were acceptable, but the tuning process of the PI controller has
influenced the controller robustness. Another approach based on automatic variation of
the Wob has been adopted in [20] to eliminate these vibrations. However, its limitation
is mainly due to the slow dynamic caused by the Wob variation in comparison to the
dynamics of stick-slip vibrations [1,21].
This research work is focused on the estimation and control of drill bit velocity to
suppress the high frequency stick-slip vibrations in a more robust way by using the H∞
observer-based controller. The main novelty of this work is that an observer is designed on
the basis of a linear model to estimate the non-measurable states as well as the unknown
input of ten degrees of freedom model subject to nonlinear input. In addition to that, the
severity of the stick-slip vibrations has been quantitatively analyzed on the basis of the
outcomes of the designed observer. A comparison with LQG observer-based controller has
been conducted, for which H∞ has demonstrated better performance in suppressing the
high-frequency stick-slip vibrations under the unstructured perturbations. The manuscript
is organized as follows. In Section 2, the types of vibrations have been overviewed with
some focus on stick-slip vibrations whose severity has been quantitatively classified. The
open-loop model with ten degrees of freedom has been tested under different scenarios to
validate the model reliability and its estimated drill bit velocity. In Section 3, the mathemat-
ical design of H∞ observer-based controller has been detailed for the drill-string model,
where the estimation results allowed us to design and investigate the controller robustness.
In Section 4, different scenarios have been introduced to test the proposed observer-based
controller against the high frequency stick-slip vibrations severities. Moreover, the pro-
posed controller results have been compared with the LQG observer-based controller, for
which H∞ has demonstrated better performance than LQG, especially toward the un-
structured perturbations. Section 5 presents conclusions and recommendations for future
research works.
2. Drill-String Model under Stick-Slip Vibrations

Based on [18], the vibrations can be generally classified into three categories: free,
forced, and self-sustaining. In self-sustaining vibrations, a periodic force is regenerated,
which then excites the vibrations themselves. Thus, if the system is prevented from vibrat-
ing, then the excitement force disappears [22]. On the other hand, for forced vibrations, the
excitation force is independent of the vibrations, and it can persist even when the vibrations
are suppressed [23]. Fortunately, the drilling process vibrations regime is self-sustaining,
due to which a constant disturbance can lead to unstable oscillating regime, which in
its turn evolves towards a stable limit cycle [24,25]. Therefore, the stick-slip vibrations
themselves can generate their own excitement through the axial vibrations [26]. Hence, by
eliminating the high-frequency stick-slip vibrations as quick as possible, we can prevent
the generation of the other types of vibrations too [27].
2.1. Torsional Vibrations

The severe torsional vibrations manifest as the stick-slip phenomena, in which a
cyclic drill bit stop can be noticed [7]. During the stick phase, the drill-string shown
in Figure 1b, rotated by the top drive from the surface, is twisted due to its elasticity,
while the drill bit will remain stuck until the cumulated bottom torque becomes greater
than the static frictional torque [2]. The sudden drill bit sleep leads to abnormal drill-
string acceleration, and it may even exceed the top drive velocity by several times [28].
Even though the nonlinear dynamic of stick-slip vibrations is poorly understood, most
researchers have concurred that during this cyclic phenomenon, the drill bit velocity
variations are synchronized with resistant torque variations [1,17]. This is mainly due to the
difference between the static and dynamic frictions between the polycrystalline diamond
compact (PDC) cutting edges and the rocks of the geological layer being drilled [29–32].
The mathematical equations of motion for the drill-string under stick-slip vibrations are
generally established based on a torsional system with one or two degrees of freedom
as given in [4,12,29,30,33]. It has been concluded through these research works that the
torsional vibrations appear more when using the PDC drill bit because it consumes more
torque than the tri-cone bit, for example [27]. Since the stick-slip is a self-sustaining
phenomenon, it is necessary to develop a new approach for the reliable estimation, fast
detection, and robust suppression of such harmonics. The new mathematical formula
given by (1) has been proposed in this study to quantify the stick-slip vibrations’ severity
SS% [34].
∆ Speed
SS% = (1)
Mean (Speed)
The change in drill bit rotational velocity is normalized by its mean; the SS% is
classified into four levels as detailed in Table 1. So, the elimination of this type of vibration
will be based on its severity level.
Table 1. Stick-slip severity classification.
Stick-Slip Severity Level Classification SS%

0 Verylow 0 to 50%
1 Low 50% to 100%
2 Mean 100% to 150%
3 High >150%
Stick-Slip Severity Level Classification SS%
0 Verylow 0 to 50%
1 Low 50% to 100%
2 Mean 100% to 150%
Vibration 2022, 5 3 High >150%267
(a) (b)
Figure
Figure 1.1.Rotary
Rotarydrilling
drilling system:
system: (a) (a) Operational
Operational rig from
rig from Algeria,
Algeria, (b) Its drill-string
(b) Its drill-string scheme. scheme.
2.2. Drill-String
2.2. Drill-StringModel
Model
The
Thedrill-string model
drill-string developed
model in thisinstudy
developed this isstudy
basedis
onbased
dividingonthe drill-string
dividing theinto
drill-string
ten parts connected in series, where each part is presented by a torsion pendulum subjected
into ten parts connected in series, where each part is presented by a torsion pendulum
at its ends to viscous and dry frictions. The upper part is attached to the top drive, while
subjected at its ends to viscous and dry frictions. The upper part is attached to the top
the lower part is related to the borehole assembly (BHA), which is under viscous friction
drive,
and thewhile
Tob. Thethemathematical
lower part is related
model to drill-string
of the the borehole assembly
can be (BHA),
studied in which is under
two equations:
viscous friction and the Tob. The mathematical model of the drill-string
the first equation is dedicated to the dynamic of the Top drive as given by Equation can be(2),
studied in
two
whileequations:
the secondthe first equation
is dedicated is dedicated
to the BHA as givento bythe dynamic
Equation of the Top drive as given by
(3) [21].
Equation (2), while the
( second
.. is dedicated to. the. BHA
as. given by Equation (3) [21].
Jt ϕt = nKI − k∅ − C ϕt − ϕb − Ct ϕt
𝐽 𝜑 = 𝑛𝐾𝐼
. − 𝑘∅ − 𝐶 . 𝜑 − 𝜑 −𝐶 𝜑 (2)
L I = U − RI − nK ϕt (2)
𝐿𝐼 = 𝑈 − 𝑅𝐼 − 𝑛𝐾𝜑
.. . . . .
Jb ϕb = k∅ + C ϕt − ϕb − Cb ϕb − Tob ϕb , Wob (3)
𝐽 𝜑 = 𝑘∅ + 𝐶 𝜑 − 𝜑 . ..− 𝐶 𝜑 − 𝑇𝑜𝑏 𝜑 , 𝑊𝑜𝑏 (3)
where Jt is the equivalent moment of inertia, ϕt , ϕt , ϕt are the position, the angular velocity,
and the 𝐽acceleration 𝜑 ,𝜑 ,𝜑
where is the equivalent moment ofthe
inertia, .
are ..the position, the angular ve-
of the upper part of drill-string, ϕb , ϕ b , ϕb are the position, the
locity,
angularand the acceleration
velocity, of the upper
and the acceleration of thepart
lower ofpart
the of
drill-string,
the drill-string 𝜑 ,𝜑(BHA),
,𝜑 areCt isthe
theposition,
the angular
viscous velocity,
friction and the
coefficient, acceleration
k is the of the lower
stiffness constant, part
C is the of thefriction
viscous drill-string (BHA), 𝐶 is
coefficient
under
the viscous n is the torque
torsion,friction coefficient, 𝑘 is the
transmission ratio of the gearbox,
stiffness constant, K is𝐶the istorque constant,
the viscous I is
friction coef-
ficient under torsion, 𝑛 is the torque transmission ratio of the gearbox, 𝐾 is the torque
the current consumed by the electrical motor of the top drive, U is the power supply voltage,
and R and L
constant, 𝐼 are the armature
is the current resistance
consumed andbyinductance, respectively,
the electrical motor with of the∅= ϕt −
top ϕb . Jb𝑈
drive, , is the
Cb are respectively the equivalent moment of inertia and the viscous friction
and 𝑅 and 𝐿 are the armature resistance and inductance, respec- coefficient at
power supply. voltage,
the BHA. Tob ϕb , Wob is the unknown torque on bit, which is a function of the drill bit
tively, with ∅ = 𝜑 −
velocity and the weight on 𝜑 .bit𝐽 [35].
, 𝐶 The
are rock-bit
respectively the equivalent
interaction block diagram, momentwhichofincludes
inertia and the
viscous
the Tob, isfriction
shown in coefficient at the BHA.
Figure 2. Equations (2) and 𝑇𝑜𝑏
(3) 𝜑 , 𝑊𝑜𝑏
have been is the unknown
rewritten torque
in the state space on bit,
which is a by
form given function
(4). of the drill bit velocity and the weight on bit [35]. The rock-bit inter-
( .
x (t) = Ax (t) + BΓ(t)
(4)
y(t) = Cx (t)
have been rewritten in the state space form given by (4).
𝑥 𝑡 = 𝐴𝑥 𝑡 + 𝐵Γ 𝑡
(4)
𝑦 𝑡 = 𝐶𝑥 𝑡
Vibration 2022, 5 Where 𝑥 𝑡 𝜖ℝ is the states’ vector, Γ 𝑡 𝜖ℝ is the inputs’ vector, and y 𝑡268 𝜖ℝ is
the outputs’ vector, with 𝑥 𝑡 = ∅ 𝜑 𝜑 𝐼 , Γ 𝑡 = 𝑇𝑜𝑏 𝑈 , and y 𝑡 = 𝜑 𝐼
[36].
where x (t) e R4 is the states’ vector, Γ(t) e R2 is the inputs’ vector, and y(t) e R2 is the outputs’
0 1 −1 . 0. T T . T
, Γ(t)0= Tob

vector, with x(t) = ∅ ϕt ϕb I 0 U , and y ( t ) = ϕ t I [36].
⎛− − ⎞ ⎡ 0 0⎤ 0 1 0 0
𝐵 ⎢− 0⎥, 𝐶0 = 0 0 0 0 1

𝐴=⎜ 0 1 −⎟
1, =0
⎜  − k − − (Ct +C) 0 ⎟C nK ⎢ 
⎥ 0
 
0 

Jt Jt Jt Jt  0 1 0 0
A= ⎣ 0, B=⎦ − 1 0 , C = 0 0
 
⎝0 − 0 Jb −
− (C⎠
k C t +C ) 0 1
 Jb Jb 0  J b
 1
0
0 − nK L 0 − RL L
The mathematical model given by the previous matrix is for two degrees of freedom.
However, in this study,
The mathematical we have
model givenused
by thea previous
ten-degree of freedom
matrix is for two model
degreesgiven by [37] as
of freedom.
illustrated
However, inin Figure 3. we have used a ten-degree of freedom model given by [37] as
this study,
illustrated in Figure 3.
2.3. Open Loop System Responses
2.3. Open Loop System Responses
The scenarios carried out in this subsection have the objective of testing the model
The scenarios carried out in this subsection have the objective of testing the model
reliability against different top drive input torque signals. In addition, it is a crucial step
reliability against different top drive input torque signals. In addition, it is a crucial step
totoinvestigate
investigate and validate the
and validate theeffectiveness
effectivenessof of
thethe
H∞H∞ observer
observer before
before designing
designing the the
controller in cascaded
controller in cascaded structure
structure [38,39].
[38,39]. The variations
The input input variations were
were mainly basedmainly based on
on drillers’
drillers’ recommendations extracted from practical drilling experiences
recommendations extracted from practical drilling experiences [40,41]. [40,41].
Figure
Figure2.2.Simscape/Matlab
Simscape/Matlab Block diagramofofRock-bit
Block diagram Rock-bit interaction
interaction simulation
simulation model.
model.
2.3.1. Scenario 1: Constant Weight on Bit

In this simulation scenario, the drilling system has been driven under constant Wob
(154 kN ≈ 16 t) while varying the supply voltage of the top drive motor [42]. The obtained
Tob and the supply voltage are shown in Figure 4a,b, respectively, the velocities of top drive
and the BHA are illustrated in Figure 5a, and the stick-slip severity is given in Figure 5b.
It can be noticed that the vibrations were very high at the beginning (class 3), then from
t = 13 s they become within the permissible vibration range (classes 0 and 1). Because of
the staircase descent of the top drive supply voltage, the severe stick-slip vibrations have
been minimized to remain at a secure severity level, which cannot be guaranteed if we start
with a supply voltage of 250 v. A similar scenario has been run with the staircase ascent of
the voltage as shown in Figure 6a. The drilling system is under severe state of stick-slip
vibrations (class 3) with severity greater than 200%, as shown in Figure 7a,b. Figure 6b
shows the resultant torque on bit for this scenario. To exit the high frequency stick-slip
Vibration 2022, 5, FOR PEER REVIEW vibrations mode, the top drive must be operated with a practically high supply voltage, and 6
then gradually brought back to its nominal voltage (250 Vdc), as highlighted in Figure 6a.
Figure 3. Rotary drilling system simulation model with ten degrees of freedom.

Tob and the supply voltage are shown in Figure 4a, b, respectively, the velocities of top
drive and the BHA are illustrated in Figure 5a, and the stick-slip severity is given in
Figure 5b. It can be noticed that the vibrations were very high at the beginning (class 3)
then from t = 13 s they become within the permissible vibration range (classes 0 and 1)
Because of the staircase descent of the top drive supply voltage, the severe stick-slip vi-
brations have been minimized to remain at a secure severity level, which cannot be
guaranteed if we start with a supply voltage of 250 v. A similar scenario has been run
with the staircase ascent of the voltage as shown in Figure 6a. The drilling system is un-
der severe state of stick-slip vibrations (class 3) with severity greater than 200%, as shown
in Figure 7a,b. Figure 6b shows the resultant torque on bit for this scenario. To exit the
high frequency stick-slip vibrations mode, the top drive must be operated with a practi-
cally high supply voltage, and then gradually brought back to its nominal voltage (250
Figure
Vdc), 3.3.Rotary
Rotary drilling
as highlighted
Figure systemsimulation
in Figure
drilling system simulation
6a. modelwith
model withtenten degrees
degrees of freedom.
of freedom.

Tob and the supply voltage are shown in Figure 4a, b, respectively, the velocities of top
drive and the BHA are illustrated in Figure 5a, and the stick-slip severity is given in
Figure 5b. It can be noticed that the vibrations were very high at the beginning (class 3),
then from t = 13 s they become within the permissible vibration range (classes 0 and 1).
Because of the staircase descent of the top drive supply voltage, the severe stick-slip vi-
brations have been minimized to remain at a secure severity level, which cannot be
guaranteed if we start with a supply voltage of 250 v. A similar scenario has been run
(a) the staircase ascent of the voltage as shown in Figure
with (b) 6a. The drilling system is un-
der
Figure severe
4. Open
Figure
state
4. Openloop
of Responses
stick-slipofvibrations
loop Responses of
TopTop
(class
drivedrive
3) with
motormotor for severity
for Scenario Scenario greater
1.1:on(a)
1.1: (a) Torque
than 200%,voltage.
bit,Torque
(b) Inputon
as shown
bit, (b) Input
in Figure 7a,b. Figure 6b shows the resultant torque on bit for this scenario. To exit the
voltage.
high
2.3.2.frequency
Scenario 2:stick-slip
Constantvibrations
Power Supply mode, the top drive must be operated with a practi-
Voltage
cally high supply voltage, and then gradually brought back to its nominal voltage (250
In this second scenario, the drilling system is driven under constant power supply
Vdc),
voltageas highlighted in Figure
(200 Vdc) while 6a. the weight on bit [35]. At the beginning, the Wob is
varying
considered constant, and perturbation is applied at t = 30 s (Figure 8b). This directly affects
the torque on bit, as shown in Figure 8b, which causes the occurrence of torsional vibrations,
as shown in Figure 8c. Consequently, the system momentarily enters the class (3) severity
level as illustrated in Figure 8d. These vibrations have been remarkably amplified by
increasing the Wob at t = 60 s when the system enters class (3) severity level permanently.
Vibration 2022, 5, FOR PEER REVIEW 7
(a) (b)
(a) (b)
Figure(a)
5. Scenario 1.1 open loop responses: (a) Rotational (b)
Velocities of Top drive and Drill bit, (b)
Figure 5. severity.
Stick-slip Scenario 1.1 open loop responses: (a) Rotational Velocities of Top drive and Drill bit, (b)
Figure 5. severity.
Figure
Stick-slip Scenario 1.11.1
5. Scenario open loop
open responses:
loop responses:(a)
(a)Rotational
Rotational Velocities
Velocities ofofTop
Topdrive
drive and
and Drill
Drill bit,bit, (b)
Stick-slip severity.
(b) Stick-slip severity.
(a) (b)
(a) (b)
Figure(a) (b)the load, (b) Torque on bit (Tob).
6. Responses of open loop scenario 1.2: (a) Voltage of
Figure 6. Responses
Figure ofofopen
6. Responses openloop
loop scenario 1.2:(a)
scenario 1.2: (a)Voltage
Voltage of the
of the load,
load, (b) Torque
(b) Torque on bit on bit (Tob).
(Tob).
Figure 6. Responses of open loop scenario 1.2: (a) Voltage of the load, (b) Torque on bit (Tob).
(a) (b)
(a) (b)
Figure(a)7. Scenario
Figure 1.2 1.2
7. Scenario open loop
open responses:
loop Rotational(b)
responses:(a)(a)Rotational Velocities ofTop
Velocities of Topdrive
driveand
and Drill
Drill bit,bit, (b)
Figure 7. severity.
Stick-slip Scenario 1.2 open loop responses: (a) Rotational Velocities of Top drive and Drill bit, (b)
(b) Stick-slip
Figure severity.
7. severity.
Scenario 1.2 open loop responses: (a) Rotational Velocities of Top drive and Drill bit, (b)
Stick-slip
In the scenario
2.3.2. Scenario 2: Constant 2.2, the Wob Supply
Power is ramped up to 154kN, which is the threshold where
Voltage
2.3.2. Scenario
stick-slip 2: Constant
vibrations Power
occurred Supply
in the Voltage
scenario 2.1, as shown in Figure 9a. At this limit, the
2.3.2.InScenario
Wob this second scenario,
2: Constant
is maintained constant. the
Power
Thus,drilling
Supply system
Voltage
the drilling is driven
system is drivenunder constant
without power
stick-slip supply
or even
voltage In this second scenario, the drilling system is driven under constant power supply
In (200
torsional Vdc) while
this vibration
second varying
as given
scenario, in the
theFigureweight ont bit
9c,d.system
drilling At is[35].
= 60 s, a At the beginning,
disturbance
driven under the
takes place
constant Wob
on theis
power con-
supply
voltage
sidered
Wob, (200
after Vdc)
constant,
which while
and
the varying
perturbation
drill-string the
is isweight
applied
subjected on
to bit
at t
severe =[35].
30
high sAt the
(Figurebeginning,
frequency8b). This
stick-slip the Wob
directly
vibrations is con-
affects
voltage (200 Vdc) while varying the weight on bit [35]. At the beginning, the Wob is con-
sidered
the torque constant,
on and and
bit, the
as perturbation
shown the is
in Figure applied atcauses
t =in30 s (Figure 8b). as
This directly affects
is 8b, which at t = 30 the occurrence of torsional
directly vibra-
(Figure 9c), system enters class (3) severity a permanent way, demonstrated
sidered constant, and perturbation applied s (Figure 8b). This affects
the in
tions,torque
Figure on
as shown9d.bit,
inas shown
Figure 8c.inConsequently,
Figure 8b, which the causes
system the occurrenceenters
momentarily of torsional
the classvibra-
(3)
the torque on bit, as shown in Figure 8b, which causes the occurrence of torsional vibra-
tions, aslevel
severity shown
as in Figure 8c.
illustrated in Consequently,
Figure 8d. Thesethevibrations
system momentarily enters the class
have been remarkably (3)
ampli-
tions, as shown in Figure 8c. Consequently, the system momentarily enters the class (3)
severity
fied by level as illustrated
increasing the Wob inatFigure
t = 60 8d.
s Thesethe
when vibrations
system have been
enters classremarkably
(3) severityampli-
level
severity level as illustrated in Figure 8d. These vibrations have been remarkably ampli-
fied by increasing the Wob at t = 60 s when the system enters class (3) severity level
permanently.
fied by increasing the Wob at t = 60 s when the system enters class (3) severity level
permanently.
permanently.
(a) (b)
(a) (b)
(c) (d)
Figure 8. Scenario 2.1 responses: (a) Weight on bit, (b) Torque on bit, (c) Rotational velocities, (d)
Stick-Slip severity.
In the scenario 2.2, the Wob is ramped up to 154kN, which is the threshold where
stick-slip vibrations occurred in the scenario 2.1, as shown in Figure 9a. At this limit, the
Wob is maintained constant. Thus, the drilling system is driven without stick-slip or even
torsional vibration as given in Figure 9c,d. At t = 60 s, a disturbance takes place on the
(c) (d)
Wob, after which the drill-string is subjected to severe high frequency stick-slip vibra-
Figure
tions 8. Scenario
(Figure
Figure 8. 2.1 responses:
9c), and
Scenario 2.1 responses:
the system (a)Weight
(a) Weighton
enters onbit,
the bit,
(b)(b)
class Torque
(3)
Torque on on
bit,bit,
severity (c)in(c)
a Rotational
permanent
Rotational velocities,
way, as(d)
velocities,
Stick-Slip
demonstrated severity.
in Figure 9d.
(d) Stick-Slip severity.
In the scenario 2.2, the Wob is ramped up to 154kN, which is the threshold where
stick-slip vibrations occurred in the scenario 2.1, as shown in Figure 9a. At this limit, the
Wob is maintained constant. Thus, the drilling system is driven without stick-slip or even
torsional vibration as given in Figure 9c,d. At t = 60 s, a disturbance takes place on the
Wob, after which the drill-string is subjected to severe high frequency stick-slip vibra-
tions (Figure 9c), and the system enters the class (3) severity in a permanent way, as
demonstrated in Figure 9d.
(a) (b)
(a) (b)
(c) (d)
Figure 9. Responses of Scenario 2.2: (a) Weight on bit, (b) Torque on bit, (c) Rotational Velocities,
(d) Stick-slip severity.
(c) (d)
Figure 9. Responses of Scenario 2.2: (a) Weight on bit, (b) Torque on bit, (c) Rotational Velocities,
(d) Stick-slip severity.

3. H∞ Observer-Based Controller Design
The H∞ observer-based controller designed in this study is a dynamic output feed-
back3.approach with two-stage
H∞ Observer-Based cascaded
Controller Design structure as shown in Figure 10a,b [39,43]. In the
first stage, the control system generates
The H∞ observer-based controller designed an estimate of drill
in this study bit velocity
is a dynamic outputtofeedback
be controlled
using the measured
approach speedcascaded
with two-stage and current of the
structure top-drive
as shown outputs
in Figure and theInknown
10a,b [39,43]. the first input,
which is the control
stage, top-drive
systempower supply
generates voltageof[42].
an estimate This
drill bit estimation’s
velocity generation
to be controlled using is ob-
theby
tained measured speed and
the designed current with
observer of theunknown
top-drive outputs and the
Tob input. known
Then, theinput, which isis itera-
estimation
thecorrected
tively top-drive till
powerthesupply
estimatedvoltage [42].are
states Thisclose
estimation’s
enoughgeneration is obtained
to the measured by the
states with an
designed observer with unknown Tob input. Then, the estimation is iteratively corrected till
acceptable threshold error. Since in this study the constructed model has ten degrees of
the estimated states are close enough to the measured states with an acceptable threshold
freedom,
error. the
Sinceresisting torque
in this study appears at model
the constructed the drillhasbit
tenin the form
degrees of cutting
of freedom, and frictional
the resisting
torques
torque[17]. Henceforth,
appears at the drillthe
bit control
in the formstrategy
of cuttingthat
andwill be considered
frictional torques [17].isHenceforth,
the drill bit ve-
locity
thestabilization
control strategyaround
that willa be
given nominal
considered value
is the drill under the stabilization
bit velocity presence ofaroundthe resisting
a
given
torque nominal value under the presence of the resisting torque [25].
[25].
(a) (b)
Figure 10. 10.
Figure Observer-Based
Observer-Based controller
controller forfor rotary
rotary drilling
drilling system:system: (a) Scheme,
(a) General General(b)Scheme, (b) H∞ ap-
H∞ approach.
proach.
3.1. H∞ Approach
3.1. H∞ Approach
The development of an observer-based controller using the H∞ approach is refor-
mulated to solving particular case of standard H∞ problem. Therefore, the mathematical
The development of an observer-based controller using the H∞ approach is refor-
Vibration 2022, 5, FOR PEER REVIEW synthesis of the state model (4) is mandatory for designing an observer with unknown
mulated to solving particular case of standard H∞ problem. Therefore, the mathematical
input cascaded with the controller [39]. Figure 11 shows the closed loop form of drill-string
synthesis
model of thethestate
with model
designed H∞(4) is mandatory
observer-based for designing
controller an observer
in the rotary with unknown
drilling system.
input cascaded with the controller [39]. Figure 11 shows the closed loop form of
drill-string model with the designed H∞ observer-based controller in the rotary drilling
system.
3.1.1. Observer Synthesis with Unknown Inputs by H∞

Consider the augment system (5) obtained from the original state model (4). The
system has been rewritten in this state space form in order to analytically verify its ob-
servability and controllability through solving the Riccati equations. In addition to that,
this form can be seen as a full H∞ control problem, from which the gain of the H∞ ob-
server-based controller will be calculated [21,36].
𝑥 𝑡 𝐴 𝐸 𝑥 𝑡 𝐵
⎧ = +𝐵 𝑊 𝑡 + 𝑢 𝑡
⎪ 𝑇 𝑡 0 0 𝑇 𝑡 0
Figure ⎪ Scheme
11.General
General Scheme of the rotary drilling
systemsystem with Observer-based H∞ design.
Figure 11. of the rotary drilling
𝑥 𝑡
with Observer-based H∞ design.
𝑧 𝑡 =𝐶 (5)
⎨ 𝑇 𝑡
In model (5), the matrices B1, C1g, D21 are chosen in an appropriate way to fulfill
necessary and ⎪
⎪ sufficient conditions of the 𝑥H∞ 𝑡 approach. Moreover, the matrices Bi r
𝑦 𝑡 = 𝑐 0 +𝐷 𝑊 𝑡
⎩
resent an unknown input observer designed 𝑇 by𝑡 the tuning procedure. Therefore, we ta
𝐵 = 𝑏 0 ,𝐷 = 0 𝐼 , and
1𝑒 0 0 0 0 1 0 0 0 0
⎛ 0 1𝑒 0 0 0 ⎞ ⎛0 1 0 0 0 ⎞
3.1.1. Observer Synthesis with Unknown Inputs by H∞

Consider the augment system (5) obtained from the original state model (4). The sys-
tem has been rewritten in this state space form in order to analytically verify its observability
and controllability through solving the Riccati equations. In addition to that, this form
can be seen as a full H∞ control problem, from which the gain of the H∞ observer-based
controller will be calculated [21,36].
 . !
x (t)

 A E x ( t ) B2

 . = + B1 W (t) + u(t)



 T ob (t) 0 0 Tob (t) 0



x (t) (5)
z ( t ) = C 1g


 Tob (t)


x (t)


y ( t ) = c2 0 + D21 W (t)


Tob (t)

In model (5), the matrices B1 , C1g , D21 are chosen in an appropriate way to fulfill
the necessary and sufficient conditions of the H∞ approach. Moreover, the matrices Bi
represent an unknown input observer designed by the tuning procedure. Therefore, we
take B1 = b1 05×2 , D21 = 02×5 I2×2 , and
1e−6
   
0 0 0 0 1 0 0 0 0
 0 1e−6 0 0 0   0 1 0 0 0 
1e−6
   
C1g =
 0 0 0 0  with b1 = 
  0 0 1 0 0 
 (6)
 0 0 0 5 0   0 0 0 1e2 0 
0 0 0 0 0 0 0 0 0 27e3
The vector W(t) constitutes the perturbations on the states of the model and the
measurements. It should be noted that there is no requirement on disturbances for the H∞
approach unlike the LQG approach for instance [24,44]. Therefore, the observer ensuring
the performance criterion is given by (7).
 .

 x̂ z (t) = A g − Lh Cz x̂z (t) + Bz u(t) + Lh y(t)

ẑ(t) = C1g x̂z (t) (7)



ŷ(t) = Cz x̂z (t)

A E
with A g = Az = and Lh is the gain observer such that
0 0
Lh = −Y∞ CzT (8)
Y∞ is the solution of the Riccati Equation (9), which guarantees the observability of
the expanded system [36].

A g Y + YA Tg + Y γ−2 C1g
T
C1g − CZT Cz Y + Bz BzT = 0 (9)
3.1.2. Controller Synthesis by H∞

The drill bit velocity controller has been developed based on a reference model [45],
i.e., the velocity of the drill bit is forced to follow the proposed reference velocity model.
Thus, the model given by (7) is augmented to the model described by (10).
  .  
ρ(t)
     
αm 0 0 ρ(t) bm 0 0
 x. (t)   r ( t )


= 0 A 0  x (t)  +  0 b1  +  B2 u(t)



. w(t)
  
−




 ζ ( t ) αc m αcr 0 ζ (t) 0 0 0


  

 ρ(t)
z(t) = Cm  x (t)  + D12 u(t) (10)

ζ (t)





  
ρ(t)




y(t) =  x (t) 





ζ (t)
where αm , bm , Cm are the state matrix, control, and observer of the reference model respec-
tively, with αm = −0.7, bm = 0.7, Cm = 1, r (t) is the input of the reference model, and
ρ(t) is the output of the reference model, which the drill bit velocity should follow. ζ (t) is
the integral of tracking error, and α is the controller tuning parameter, with α = 10. The
matrices b1 , D12 are chosen in an appropriate way. Moreover, the two matrices C1m and
D12 constitute the tuning matrices required to bring back the response of the designed
controller to the desired performance, where:
 
1 0 0 0
 0 1 0 0  0 0 0 10 0 5 0
b1 = 
 0
, C1m = and D12 = (11)
0 1 0  0 0 0 0 0 0 1
0 0 0 1
In a more compact form, the system (11) is rewritten as:

 .

 x m (t) = Am xm (t) + β 1 W (t) + β 2 u(t)

z(t) = C1m xm (t) + D12 u(t) (12)


y(t) = xm (t)

Then, the control ensuring internal stability [25] and guaranteeing the performance
criterion is written as:
u(t) = Fh xm (t) (13)
with
Fh = − β T2 X∞ (14)
where X∞ is the solution of the Riccati Equation (15), which guarantees the controllability
of the system [36].

Am T
X + XAm + Xm γ−2 β 1 β T1 − β 2 β T2 Xm + C1m
T
C1m = 0 (15)
4. Results and Discussion

Many scenarios have been considered in this part in order to study and test the
robustness of the designed H∞ observer-based controller for different situations in the
drilling field. First, the estimation precision of the designed observer without any controller
has been investigated, and then the controller has been activated in order to evaluate the
complete H∞ observer-based controller bloc. The parameters used in these scenarios are
namely the moments of inertia of the rotating parts, the stiffness of the drill pipe, and
the viscous friction coefficients. These parameters have been calculated from the basic
equations and the fundamental rig parameters, such as: the length of the drill pipes and
the drill collars, the diameter and thickness of the drill-string and drill presses, the mud
viscosity, etc.
4.1.ObserverPerformanceTests
4.1.ObserverPerformanceTests
In these tests, the controller has been disabled while applying the following
Vibration 2022, 5 275scenar-
In these tests, the controller has been disabled
ios to study the estimation precision of the designed observer.while applying the following scenar-
ios to study the estimation precision of the designed observer.
4.1.1.
4.1.Scenario 3: ConstantTests
Observer Performance Wob with Step Voltage
4.1.1. A
Scenario
constant 3: Constant
In these tests,
Wob the
equalWob
controllerwith
to 0.7has Step
ofbeen
the Voltage
disabled
overall while applying
drilling thewhich
weights, following scenarios to 29
is equivalent
toAstudy
tons, has the estimation
constant
been Wob equal
applied precision
over tothe of
of the
thedesigned
0.7 simulation overall observer.
drilling
time. A step weights,
voltagewhich is equivalent
that rises from 300toV29 to
tons,
450 Vhas at tbeen
= 10 applied
s has been overalsotheconsidered
simulationastime. shown A step voltage
in Figure thatFigure
12a,b. rises from 300 V the
13a shows to
4.1.1. Scenario 3: Constant Wob with Step Voltage
450 V at t =and
measured 10 sthehasestimated
been alsodrill considered as shown
bit velocities within theFigure 12a,b.errors.
estimation FigureHence,
13a showsit canthe
be
A constant Wob equal to 0.7 of the overall drilling weights, which is equivalent to
measured
noticed and
that the estimated drill bit velocities with the estimation errors. Hence, it can be
29 tons, hasthe
been system
appliedsuffers
over thefrom high time.
simulation frequency
A step stick-slip
voltage thatvibrations
rises from 300 of class
V to (3) as
noticed
450 V at t = 10 s has been also considered as shown in Figure 12a,b. Figure 13a shows the (3) the
demonstratedthat the in system
Figure suffers
13b. from
Then, by high frequency
increasing the stick-slip
top drive vibrations
power supply of class
voltage, as
demonstrated
vibrations
measured havein Figure
and become 13b. Then,
of class
the estimated drill(1)by
bitat increasing
t = 23 s,with
velocities the
which top
the is drive power
a tolerable
estimation supply
region.
errors. Hence, voltage,
This the
operation
it can
vibrations have become of class (1) at t = 23 s, which is a tolerable
takes a considerable time (13s) to eliminate the vibrations. Due to their high frequency
be noticed that the system suffers from high frequency stick-slip region.
vibrations of This
class operation
(3) as
takes
content,a considerable
demonstrated in
such vibrations timecan
Figure (13s)
13b. Then,to eliminate
create by too muchthe
increasing the vibrations.
top
damage drive Due
power
during to their
supply
such a timehigh
voltage, frequency
the
period vi-[45,46].
brations
content, suchhave become
vibrations of class
can (1)
createat t =
too23 s, which
much is a
damage tolerable
In addition to that, the observer has provided good estimation of the unknown Tob, duringregion.
suchThisa operation
time takes
period [45,46].
In a considerable
addition to that,timethe
(13s)observer
to eliminate has the vibrations.good
provided Due to their high frequency
estimation of the content, Tob,
unknown
which is treated as a disturbance on the drill-string dynamic model, as given by Figure
such vibrations can create too much damage during such a time period [45,46]. In addition
which
14. to is
It that, treated
is also as
noticeablea disturbance on the drill-string dynamic model, as given by Figure
the observer has that the measurable
provided good estimation states
of theofunknown
the system, namely
Tob, which the Top
is treated as drive
14. It
velocityis also
and noticeable
a disturbance current,
on the have thatbeen
thedynamic
drill-string measurable
estimated states
with
model, asgood ofaccuracy,
given theFigure
by system,
as It
14. namely
demonstrated the Top
is also noticeable drive
in Figure
velocity
15athat
andthe and current,
Figure have
statesbeen
of theestimated
15b, respectively.
measurable system, namely with the
good Topaccuracy, as demonstrated
drive velocity and current, have in Figure
15a been
and Figure
estimated 15b, respectively.
with good accuracy, as demonstrated in Figure 15a,b, respectively.
(a) (b)
(a) (b)
Figure 12. Test inputs of the observer for Scenario-3: (a) Top drive motor voltage, (b) Applied
Figure
Weight 12. Test
on12.
Figure Testinputs
bit. of the
inputs of theobserver
observer for Scenario-3:
for Scenario-3: (a) Top(a) Top
drive drive
motor motor
voltage, (b) voltage, (b) Applied
Applied Weight
Weight on bit.
on bit.
(a) (b)
(a) (b)
Figure 13.13.
Figure Measured
Measured and estimated
and estimatedvariables by the
variables by thedesigned
designedobserver:
observer:(a)(a) Rotational
Rotational velocities, (b)
velocities,
Figure
(b) 13.
Stick-slip Measured
severity.
Stick-slip and estimated variables by the designed observer: (a) Rotational velocities, (b)
severity.
Vibration 2022, 5, FOR PEER REVIEW

Figure 14. Measured and estimated Toque on bit by H∞ observer.
Figure 14. Measured and estimated Toque on bit by H∞ observer.

Figure 14.Measured
Figure 14. Measuredand
and estimated
estimated Toque
Toque onbybitH∞
on bit byobserver.
H∞ observer.
(a) (b)
Figure 15. Measured and estimated variables by the designed observer: (a) Top-drive c
(a) Top-drive velocity. (b)
Figure
Figure(a)15.
15. Measured
Measured and estimated
and estimatedvariables
variablesby
bythe (b)observer:
thedesigned
designed (a) Top-drive
observer: (a) Top-drivecurrent,
current, (b)
Top-drive 4.1.2. Scenario 4: Random Wob with Constant Voltage
velocity.
(b) Top-drive
Figure velocity. and estimated variables by the designed observer: (a) Top-drive current, (b)
15. Measured
Through this scenario, the observer behavior against random Wob varia
Top-drive velocity.
4.1.2. Scenario 4: Random
RandomWob
4.1.2. Scenario 4:examined.
been The
Wob with ConstantVoltage
randomness
with Constant Voltage
of this input is stimulated by the dynamic a
Through
4.1.2.Throughdrill-string
Scenario this
this [47].the
4:scenario,
scenario,
Random For
the
Wob this,
withthe
observer
observer inputs
behavior
behavior
Constant shown
against in Figure
against
Voltage random random 16a,b
Wobhave
Wob variations hasbeen
been applied
variations has
examined.
been examined.observer has
The randomness provided good
of this input estimation
is stimulated results for the
by the dynamic drill bit velocity
axis of theaxis as
drill- shown
Through The this randomness
scenario, theofobserver this input is stimulated
behavior againstby the dynamic
random Wob variations of thehas
drill-string 17a,
string [47].[47].
For Forandthis,
this, forinputs
the the inputs
the unknown
shown shown Tobin16a,b
in Figure input,
Figure as16a,b
have shown
been in been
applied.
have Figure 17b.
The applied.
H∞ The
observer
The measur
been
has
examined.
provided good
The randomness
estimation results
offor
thisthe
input
drill
is
bit
stimulated
velocity as
by
shown
the indynamic
Figure
axis H∞
17a,
of the
observer shown
has provided in Figure
good 18a,b
estimation of the top
results drive
for have a fluctuating
the drill16a,b
bit velocity behavior
as shown around
in Figure 500
drill-string [47].
and for thecurrent), For
unknownand this,
Tobaroundthe inputs
input, as15 shown shown
in in
Figure Figure
17b. The have
measured been
inputs applied.
shown in The H∞
17a, and for the unknown Tob input,rd/sas (for
shownthe velocity).
in Figure The
17b. large
The peak of theinputs
measured current
observer
Figure 18a,bhasofprovided
the is
top goodhave
drive estimation results
a fluctuating for the around
behavior drill bit500A
velocity as shown
(for the current),in Figure
shown in ginning
Figure 18a,b mainly
of the top duedrive
to the direct
have start of the motor without activation ofthe
the c
17a, and
and around for15therd/sunknown Tob input,
(for the velocity). asa shown
The large
fluctuating
peak of inthe behavior
Figure
current17b. around 500A (for
Thebeginning
at the measured is inputs
current), Further, the velocity andvelocity).
the current fluctuations have been well estimated
mainly and
shown due around
to
in Figure 15 start
the direct
18a,b rd/s (for
of
of the topthe
the motor have aThe
drivewithout large ofpeak
activation
fluctuating the of thearound
controller.
behavior current
Further,atthe
500A the be-the
(for
ginning ceptable
is mainly due accuracy,
to 15
the even
direct with the presence of unknown Wob characterized by
velocity
current), and
andthearound
current (forstart
fluctuations
rd/s theof
have the motor
been well The
velocity). without
estimated activation
large with
peak of theofcurrent
acceptable the controller.
accuracy,at the be-
Further, the dynamics,
velocity as
and shown
the in Wob
Figure 18a,b.
even with
ginning isthe presence
mainly due ofto the current
unknowndirect fluctuations
characterized
start of the motor have
by been dynamics,
arbitrary
without well estimated
activation asof thewith
shown in ac-
controller.
Figure accuracy,
ceptable 18a,b. even with the presence of unknown Wob characterized by arbitrary
Further, the velocity and the current fluctuations have been well estimated with ac-
dynamics,
ceptableasaccuracy,
shown ineven Figure with18a,b.
the presence of unknown Wob characterized by arbitrary
dynamics, as shown in Figure 18a,b.
(a) (b)
Figure
Figure 16. Test 16.ofTest
inputs inputs for
the observer of Scenario-4:
the observer for drive
(a) Top Scenario-4: (a) Top
motor voltage, (b)drive motor
Applied voltage, (b
Weight
(a)
on bit. Weight on bit. (b)
(a) Test inputs of the observer for Scenario-4: (a) (b)
Figure 16. Top drive motor voltage, (b) Applied
Weight on bit.
Weight on bit.
ibration 2022, 5, FOR PEER REVIEW 14
bration 2022, 5, FOR PEER REVIEW 14
(a) (b)
(a) 17. Measured and estimated variables by the designed
Figure (b) observer for Scenario-4: (a) Rota-
tional velocities, (b) Stick-slip severity.
Figure 17. Measured and estimated variables by the designed observer for Scenario-4: (a) Rota-
Figure 17. Measured and estimated variables by the designed observer for Scenario-4: (a) Rotational
tional velocities,
velocities, (b)(b) Stick-slip
severity.
(a) (b)
(a) 18. Measured and estimated variables by the(b)
Figure
Figure 18. Measured and estimated variables by the designed designed
Observer forObserver
Scenario-4: for Scenario-4: (a)
(a) Top-drive
Top-drive current,
Figurecurrent, (b) Top-drive
(b) Top-drive
18. Measured velocity.
and velocity.
estimated variables by the designed Observer for Scenario-4: (a)
Top-drive current, (b) Top-drive velocity.
4.2. Scenario5: Disturbed
4.2. Scenario5: Measurements
Disturbed Measurements
In this
4.2. Scenario5:
In this scenario,
Disturbed
scenario, thesystem
system has
Measurements
the has been
beendriven
driven with
witha constant voltage
a constant equal to
voltage 650 V
equal to 650V
and In and constant
constant Wob
Wob equal equal to 103
to 103 kN,kN, as demonstrated in Figure 19a,b. Then, disturbances
this scenario, the system has asbeendemonstrated
driven withina Figure constant 19a,b.
have been introduced to the top drive current and velocity with non-zero mean signal with
voltageThen, disturbances
equal to 650V
have been
and constant introduced to the top drive current and velocity with non-zero mean signal
frequencyWob equal
of 1kHz to 103
as given kN, as20a,b
in Figure demonstrated in Figure
[48]. As expected, the H∞ 19a,b.
observerThen,
has disturbances
arrived
with to
have frequency
been of 1kHz
filterintroduced to as
the unstructured given in Figure
theperturbations
top drive 20a,b
current
unlike and
other[48]. As expected,
velocity
observer, theKalman
withasnon-zero
such the H∞ mean
observer has
filter.signal
arrived to
with frequency filter
This is mainly the unstructured
of caused
1kHz by as the
given perturbations
in Figure
supposition put20a,b unlike
on the[48].
choiceother observer,
Asofexpected,
covariancethe such as the
H∞ observer
matrices, Kalman
whilst has
filter.H∞This is
design mainly
did not caused
require by
any the supposition
restrictions in termsput
of on
their the
arrived to filter the unstructured perturbations unlike other observer, such as thethe choice
choices. of
Figure covariance
21a,b show matrices,
Kalman
estimated
whilstThisH∞isdesign velocity and torque of the drill bit, and Figure 22a,b demonstrates the estimated
filter. mainlydid not require
caused by the any restrictions
supposition put in onterms of their
the choice choices. Figure
of covariance 21a,b
matrices,
show current
the and velocity
estimated of the top
velocity and drive
torquerespectively.
of the It can
drill beand
bit, concluded
Figure that
22a,b H∞demonstrates
estimation the
whilstperformance
H∞ designhas did notnot
beenrequire
affectedany restrictions
by the unstructured in terms of theirthat
perturbations choices. Figure
can appear on 21a,b
estimated
show thetop
the
current
estimated and velocity
velocity
drive inputs during and ofdrilling
the top
thetorque of drive
the drill
process.
respectively.
bit, and
Hence,
It can22a,b
Figure
the controller
be concluded
demonstrates
robustness can be
that the
H∞
estimation
estimated
maintainedperformance
current evenandin suchhas
velocity notofbeen
situations, affected
thesince
top drive
the by the unstructured
respectively.
observer estimations Itresults
can beperturbations
canconcluded
have a direct thatH∞
that can
appear on
influence
estimation the top drive inputs
on the controller
performance has not during
robustness. the drilling process. Hence, the
been affected by the unstructured perturbations that can controller robust-
appear on the top drive inputs duringsituations,
ness can be maintained even in such the drilling since the observer
process. Hence,estimations
the controller results can
robust-
havecan
ness a direct influence even
be maintained on theincontroller robustness.
such situations, since the observer estimations results can
have a direct influence on the controller robustness.
show the estimated velocity and torque of the drill bit, and Figure 22a,b demon
estimated current and velocity of the top drive respectively. It can be conclude
estimation performance has not been affected by the unstructured perturbation
appear on the top drive inputs during the drilling process. Hence, the controll
Vibration 2022, 5 ness can be maintained even in such situations, since the observer estimations
278 r
have a direct influence on the controller robustness.
bration 2022, 5, FOR PEER REVIEW 15
Vibration 2022, 5, FOR PEER REVIEW
Weight on bit.
(a) 19. Test inputs of the observer for Scenario-5:
Figure (b)(a) Top drive motor voltage
Weight on bit.
Figure 19. Test inputs of the observer for Scenario-5: (a) Top drive motor voltage, (b) Applied Weight
on bit.
(a) (b)
(a)drive measurements, (a) Rotational velocity, (b)(b)
Figure 20. Noises on Top The current.
Figure
Figure 20. Noises 20.drive
on Top Noises on Top drive
measurements, (a)measurements, (a)(b)
Rotational velocity, Rotational velocity, (b) The curren
The current.
(a) (b)
(a) (b)
FigureFigure
21. Measured
21. Measured and
andestimated variables
estimated variables by by
the the designed
designed Observer
Observer for Scenario-5:
for Scenario-5: (a) Rota-
(a) Rotational
tionalvelocities,
velocities, (b)Figure
Tob.
(b) Tob. 21. Measured and estimated variables by the designed Observer for Scenar
tional velocities, (b) Tob.
4.3. Scenario6: Parametric Variation
This time the system’s inputs are as follows: a constant supply voltage of 650 V with
a constant Wob of about 103 kN as shown in Figure 23a,b [49]. However, a variation
of the parameters, namely resistance (r), inductance (l), length, and mass of string, has
been applied. These are related to the calculation of the moments of inertia and the
viscous friction coefficients [50]. As shown in Figures 24 and 25, the designed observer has
demonstrated good performance against the parametric variations. However, it showed
sensitivity to the torque constant variation (K), as given in Figure 24b. The torque constant
has a deviation of 3% from its nominal value [49].
(a) (a) (b) (b)

Figure 22. Measured and22.estimated
Figure Measuredvariables by thevariables
and estimated designedbyObserver for Scenario-5:
the designed (a) S
Observer for
Top-drive current, Top-drive
(b) Velocity.
current, (b) Velocity.
(a) (b)
Figure 21. Measured and estimated variables by the designed Observer for Scenario-5: (a) Rota-
Vibration 2022, 5 tional velocities, (b) Tob. 279
ibration 2022, 5, FOR PEER REVIEW 16
(a)REVIEW
Vibration 2022, 5, FOR PEER (b)
Figure 22. 22.
Figure Measured
Measured andandestimated
estimated variables
variables by the Observer
by the designed designed
for Observer forTop-drive
Scenario-5: (a) Scenario-5: (a)
Top-drive current, (b)
current, (b) Velocity. Velocity.
4.3. Scenario6: Parametric Variation

This time the system’s inputs are as follows: a constant supply voltage of 650 V with
a constant Wob of about 103 kN as shown in Figure 23a,b [49]. However, a variation of
the parameters, namely resistance (r), inductance (l), length, and mass of string, has been
applied. These are related to the calculation of the moments of inertia and the viscous
friction coefficients [50]. As shown in Figures 24 and 25, the designed observer has
demonstrated good performance against the parametric variations. However, it showed
sensitivity
(a) to the torque constant variation (K), as(b) given in Figure 24b. The torque con-
stant has a deviation of 3% from its nominal value [49].
(a)of the observer for Scenario-6: (a) Top drive motor
Figure 23. Test inputs (b) voltage, (b) Applied
Weight on bit.
Figure
Figure 23. Test inputs23. Test
of the inputsforofScenario-6:
observer the observer
(a) Topfor Scenario-6:
drive (a) (b)
motor voltage, Top drive Weight
Applied motor voltag
on bit. Weight on bit.
(a) (b)
Figure 24. Measured
Figure 24. Measured(a)
andandestimated variables
estimated variables by designed
by the the designed Observer
Observer (b)
for Scenario-6:
for Scenario-6: (a) Rota-
(a) Rotational
tionalvelocities,
velocities,(b)(b)
Tob.Tob.
Figure 24. Measured and estimated variables by the designed Observer for Scenar
tional velocities, (b) Tob.
4.4. Performance Tests of Observer-Based Controllers
In this part, the performances of the full H∞ observer-based controller have been
discussed. Keeping in mind that the controller receives the estimated state in order to
provide output feedback to the top drive [43], the main objective of the controller is to
impose a predefined dynamic on the top drive that ensures the tracking of the desired
reference by the drill bit [51].
(a) (b)
(a) (b)
Figure 24. Measured and estimated variables by the designed Observer for Scenario-6: (a) Rota
Vibration 2022, 5 tional velocities, (b) Tob. 280
(a) (b)
ibration 2022, 5, FOR PEER REVIEW Figure 25.

Figure 25. Measured
Measured and and estimated
estimated variables
variables by theObserver
by the designed designed Observer(a)for
for Scenario-6: Scenario-6: 17(a
Top-drive
Top-drive
current, current, (b) Velocity.
(b) Velocity.
4.4.4.4.1. Scenario 7:
Performance Voltage
Tests Step Reference Tracking
of Observer-Based Controllers
Figure 26a,bIn thisshows thethe
scenario, velocity
control and
systemvoltage of top
is subjected to drive responses
a step reference thattorises
thisfrom
first input
In this part, the performances of the full H∞ observer-based controller have bee
0 to 7 It
scenario. rd/s, then from
is clear that the7 rd/s to 15 rd/shas
controller without applyingthe
guaranteed anytracking
additional ofweight on the with
the reference
discussed.
bit. Figure Keeping
26a,b showsin mind
the that the
velocity and controller
voltage of receives
top drive the estimated
responses to this state
first in order t
input
acceptable error and within a short time. Figure 27a,b shows the controller output for the
provide output feedback to the top drive [43], the main objective
scenario. It is clear that the controller has guaranteed the tracking of the reference withof the controller is t
top drive current and velocity respectively. This demonstrates that the controller has
impose a predefined
acceptable error and dynamic on the
within a short topFigure
time. drive27a,b
thatshows
ensuresthe the tracking
controller outputof the
for desire
provided
the smooth
top drive values
current andwith fewer
velocity fluctuations.
respectively. This Thus, this signal
demonstrates that cancontroller
the be handled has by the
reference by the drill bit [51].
top drive without
provided smooth any abnormalities.
values Moreover,Thus,
with fewer fluctuations. the increase ofcan
this signal current consumption
be handled by the (at t
= 15 top
s) is caused
drive withoutby any
theabnormalities.
increase in viscous
Moreover,friction due of
the increase tocurrent
the increased
consumption speed(at of the
4.4.1. Scenario 7: Voltage Step Reference Tracking
t = 15 s) which
drill-string, is caused by thethat
means increase in viscous friction
H∞ controller can be due to the increased speed
straightforwardly of the
implementable in
In this
drill-string,scenario,
which
the top drive machines. the
means control
that H∞ system is
controller subjected
can be to a step reference
straightforwardly that
implementable rises infrom 0 t
the top
7 rd/s, thendrive machines.
from 7 rd/s to 15 rd/s without applying any additional weight on the bi
(a) (b)
Figure 26. Observer-based
Figure 26. Observer-basedcontroller responses:
controller responses: (a) (a) Controlled
Controlled drill
drill bit bit velocity,
velocity, (b)voltage.
(b) Control Control volt-
age.
4.4.2. Scenario 8: Wob step Tracking Reference
This time, the control system is subjected to a Wob step reference in order to test
its robustness. To do this, the Wob has been increased from 25 kN to 180 kN at t = 10 s
(Figure 28a) with a reference velocity of 7 rd/s. It can be noticed from Figure 29a that this
increase caused a complete stop of the drill bit for 2 s, followed by high frequency stick-slip
vibrations (Figure 28b). Then, the designed controller forced the drill-string to leave this
stuck phase and mitigate the vibrations by tracking the desired outputs in 7 s (Figure 29).
Moreover, the control signals sent to the top drive are shown in Figure 30a,b.
(a) (b)
Figure 27. Observer-based controller responses for Top Drive: (a) Current, (b) Velocity.
(a) (b)
Figure 26. Observer-based controller responses: (a) Controlled drill bit velocity, (b) Control volt-
age.

(a) (b)
4.4.2. Scenario 8: Wob step Tracking Reference

This time, the control system is subjected to a Wob step reference in order to test its
robustness. To do this, the Wob has been increased from 25 kN to 180 kN at t = 10 s
(Figure 28a) with a reference velocity of 7 rd/s. It can be noticed from Figure 29a that this
increase caused a complete stop of the drill bit for 2 s, followed by high frequency
stick-slip vibrations (Figure 28b). Then, the designed controller forced the drill-string to
leave(a)
this stuck phase and mitigate the vibrations by (b)tracking the desired outputs in 7s
(Figure 29). Moreover, the control signals sent to the top drive are shown in Figure 30a,b.
(a) 28. Observer-based controller responses for Scenario-8:
Figure (b) (a) Wob, (b) Torque on bit.
(a)28. Observer-based controller responses for Scenario-8:
Figure (b) (a) Wob, (b) Torque on bit.
Figure 28.28.
Figure Observer-based
Observer-basedcontroller responsesfor
controller responses forScenario-8:
Scenario-8:(a) (a) Wob,
Wob, (b) Torque
(b) Torque on bit.on bit.
(a) (b)
(a) 29. Observer-based controller responses for Scenario-8:
Figure (b) (a) Velocity, (b) Voltage.
(a)29. 29.
Figure
Observer-basedcontroller
controller responses
responses for
(b)(a)(a)
forScenario-8:
Scenario-8: Velocity,
Velocity, (b) Voltage.
(b) Voltage.
Figure 29. Observer-based controller responses for Scenario-8: (a) Velocity, (b) Voltage.
(a) (b)
(a) 30.30.Observer-based
Figure
Figure controllerresponses
Observer-based controller responses of Top
of Top drive (b)
drive for Scenario-8:
for Scenario-8: (a) Current,
(a) Current, (b) Veloc-
(b) Velocity.
ity.
Figure (a)30. Observer-based controller responses of Top drive (b)for Scenario-8: (a) Current, (b) Veloc-
ity.
4.4.3. Scenario
Figure 9: Stick-Slip
30. Observer-based Mitigation
controller responses of Top drive for Scenario-8: (a) Current, (b) Veloc-
ity.
4.4.3. In this scenario,
Scenario the controller
9: Stick-Slip performance’s improvement in suppressing the high
Mitigation
frequency stick-slip vibrations has been investigated. The drilling system has been driven
4.4.3. Scenario 9: Stick-Slip Mitigation

In this scenario, the controller performance’s improvement in suppressing the high
frequency stick-slip vibrations has been investigated. The drilling system has been driven
with a deactivated controller at the beginning and with Wob = 180 kN and a voltage of
300 V. These inputs cause the generation of the stick-slip phenomenon in the drill-string [9].
At t = 10 s, the controller has been activated, as shown in Figures 31 and 32a,b. These figures
Vibration 2022, 5, FOR PEER REVIEW demonstrated that after 5s from controller activation, the vibrations have been reduced 19
from class (3) to class (1). Then, they have been completely mitigated in a very short time
Vibration 2022, 5, FOR PEER REVIEW in comparison to manual manipulation procedures used nowadays in the drilling fields. 19
Moreover, the control law shown in Figure 33a,b is very smooth and can be injected into an
operational top drive without too many technical restrictions.
Figure 31. Controlled Torque on bit by H∞ approach.
Figure 31. Controlled Torque on bit by H∞ approach.

Figure 31.31.
Figure Controlled
ControlledTorque onbit
Torque on bitby
byH∞
H∞approach.
approach.
(a) (b)
Figure
(b) (a) Velocity, (b) Voltage.
Figure(a)
32. Observer-based
Figure 32. Observer-basedcontroller
controller responses
responses for (b)
forScenario-9:
Scenario-9:(a)(a) Velocity,
Velocity, (b) Voltage.
(b) Voltage.
Figure 32. Observer-based controller responses for Scenario-9: (a) Velocity, (b) Voltage.
(a) (b)
Figure
(a)33.33.
Observer-basedcontroller responsesforfor
controller responses Top
Top drive,
(b)
drive, Scenario-9:
Scenario-9: (a) Current,
(a) Current, (b) Velocity.
(b) Velocity.
Figure(a) (b)Scenario-9: (a) Current, (b) Velocity.

33. Observer-based controller responses for Top drive,
4.4.4. Scenario 10: Wob Disturbances Filtering
Figure 33. Observer-based controller
ability ofresponses for Top drive, Scenario-9:
to filter(a)
theCurrent, (b) Velocity.
4.4.4. In this
Scenario scenario,
10: WobtheDisturbances the designed
Filtering approach unstructured dis-
turbances on the Wob has been studied [24]. The perturbations are generally initiated by
InScenario
4.4.4. this scenario,
10: Wobthe ability of theFiltering
designed approach to filter the unstructured dis-
the axial dynamic of theDisturbances
drill-string, which has been previously neglected in the mathe-
4.4.4. Scenario 10: Wob Disturbances Filtering

In this scenario, the ability of the designed approach to filter the unstructured dis-
the axial dynamic of the drill-string, which has been previously neglected in the math-
ematical model of the drill-string [26]. First, the drilling system is driven with constant
and disturbed Wob, and with deactivated control. Then, at t = 60 s, the controller has
been activated. Figure 34 shows the attenuation of the velocity fluctuations of drill bit
by the activated controller. The stick-slip severity has also been reduced from class (3) to
Vibration 2022, 5, FOR PEER REVIEW class (1) by activating the controller, as shown in Figure 35a. Moreover, Figure 35b shows 20
the required controller voltage that should be applied to the top drive. It can be noticed
that this is a smooth and tolerable input.
Figure 34. Noised drill bit velocity without and with H∞ control activation.
Figure 34.34.Noised
Figure Noiseddrill
drillbit
bit velocity withoutand
velocity without andwith
with H∞
H∞ control
control activation.
activation.
(a) (b)
Figure (b) (a) Stick-Slip severity, (b) Voltage.
Figure 35. Observer-based controller responses for Scenario-10: (a) Stick-Slip severity, (b) Voltage.
Figure 35. Observer-based controller responses for Scenario-10: (a) Stick-Slip severity, (b) Voltage.
4.5. Scenario
4.5. Scenario11:
11:Performance Limitations
Performance Limitations
4.5. Scenario
InInthis 11: Performance
thisscenario,
scenario, the Limitations
the controllers’
controllers’ performances
performances havehave
beenbeentestedtested
underunder
extreme extreme
condi- con-
ditions
tions
In bybyimposing
this imposing
scenario, disturbances
disturbances on performances
on the
the controllers’ the Wob
Wob (Figure
(Figure and36)
36)have on
beenand
both ontheboth
tested top the top
drive
under drivecon-
velocity
extreme ve-
and
locity current
and (Figure
current 37a,b),
(Figure along
37a,b), with
along parametric
with variations
parametric on the
variations
ditions by imposing disturbances on the Wob (Figure 36) and on both the top drive ve- length
on of
the drill
lengthpipesof drill
and and
pipes drill collars. TheseThese
drill collars. variations are of the
variations areorder
of of 130%
the of
ofthe
ordervariations nominal
130% of value, plus avalue,
locity and current (Figure 37a,b), along with parametric onthe
the nominal
length of drill
variation in the torque constant of the order of 103%. Figure 38 shows that the designed
pipes and drill collars. These variations are of the order of 130% of the nominalthe
plus a variation in the torque constant of the order of 103%. Figure 38 shows that de-
value,
H∞ observer-based controller has provided a better estimation than LQG even for this
signed
plus H∞ observer-based
a variation in the torque controller
constant has provided
oflaw order ain
theshown ofbetter
103%.estimation
Figure 38 than
shows LQGthat even for
the de-
scenario [44]. Moreover, the LQG control Figure 39a is very disturbed and
this
signed scenario
cannot H∞ [44]. Moreover,
beobserver-based
implemented incontroller
the LQG
operational hascontrol
top
law
provided shown
a better
drive, while
in
the H∞
Figure
estimation 39a is very
than LQG
has provided
disturbed
even for
a better
andcontrol law as shown in Figure 39b. Figures 40–42 show the obtained results of H∞ ainbetter
this cannot
scenario be implemented
[44]. Moreover, in
theoperational
LQG top
control drive,
law shownwhile in the H∞
Figure has
39a provided
is very disturbed
control
and cannot
comparisonlaw be as shown
LQG forin
toimplemented Figure
torque 39b.top Figures
inonoperational
bit, top
drive 40–42 show
drive,and
current, topthe
while the obtained
driveH∞ results
has provided
velocity, ofa H∞
respectively. in
better
comparison
It is clear
control to shown
lawthrough
as LQG forinfigures
these torque
Figure on
that bit,Figures
the
39b. top drive
designed current,
controller
40–42 show hasand top
provided
the drive
obtainedbettervelocity,ofrespec-
estimation
results H∞ in
tively.
and It is
control clear
than through
the LQG these
even figures
for this that the
scenario designed
where controller
extreme
comparison to LQG for torque on bit, top drive current, and top drive velocity, respec- has
conditions provided
have been better
considered.
estimation
tively. It is and Therefore,
clearcontrol
through we can
than conclude
thefigures
these LQG eventhat the proposed
for designed
that the H∞
this scenario observer-based
where has
controller controller
extreme is
conditions
provided better
better
have beenthan LQG and can
considered. guaranteewe
Therefore, better
can robustness
conclude in operating
that the rotary drilling
proposed H∞ systems.
observer-based
estimation and control than the LQG even for this scenario where extreme conditions
controller
have been is better than
considered. LQG andwecan
Therefore, canguarantee
conclude better
that therobustness
proposed in H∞operating rotary
observer-based
drilling systems.
controller is better than LQG and can guarantee better robustness in operating rotary
drillingHowever,
systems.it should be taken into consideration that since the H∞ observer-based
controller
However, proposed in this
it should bestudy
takenwas intobased on solving
consideration thattwo cascaded
since the H∞ Riccati equations,
observer-based
Vibration 2022,
Vibration 5, 5FOR PEER REVIEW
2022, 284
Figure 36. Perturbed Weight on bit input for Scenario-11.
Figure 36. Perturbed Weight on bit input for Scenario-11.
Figure36.
Figure 36.Perturbed
Perturbed Weight
Weight oninput
on bit bit input for Scenario-11.
for Scenario-11.
(a) (b)
Figure
(a)37. Noised inputs of Top drive for Scenario-11: (a)(b)
Velocity, (b) Current.
Figure 37.37.
Figure Noised
(a) inputs
Noised inputsof
ofTop drivefor
Top drive forScenario-11:
Scenario-11:(a)(a) Velocity,
(b)
Velocity, (b) Current.
(b) Current.
Figure 37. Noised inputs of Top drive for Scenario-11: (a) Velocity, (b) Current.
(a) (b)
Figure
(a)38. Real and estimated Drill bit velocity with: (a) LQG
(b) controller, (b) H∞ controller.
Figure 38.38.
Figure Real
(a) and
Real andestimated
estimated Drill bitvelocity
Drill bit velocitywith:
with:(a)(a)
LQGLQG controller,
(b) (b) H∞
controller, (b)controller.
H∞ controller.
Figure 38. Real and estimated Drill bit velocity with: (a) LQG controller, (b) H∞ controller.
(a) (b)
Figure
(a) 39.
Figure 39.Top
Top drive voltageforfor
drive voltage Scenario-11
Scenario-11 with:
with: (a) LQG
(a) LQG (b) controller,
controller, (b) H∞(b) H∞ controller.
controller.
Figure 39.(a)
Top drive voltage for Scenario-11 with: (a) LQG controller,
(b) (b) H∞ controller.
Figure 39. Top drive voltage for Scenario-11 with: (a) LQG controller, (b) H∞ controller.
(a) (b)
Figure(a) (b)controller, (b) H∞ controller.
40. Real and estimated Torque on bit under: (a) LQG
(a)40. Real and estimated Torque on bit under: (a) LQG
Figure (b) controller, (b) H∞ controller.
Figure 40.40.
Figure Real and
Real andestimated Torqueonon
estimated Torque bitbit under:
under: (a) LQG
(a) LQG controller,
controller.
(a) (b)
(a) Real and estimated Top drive current under: (a) (b)
Figure 41. LQG controller, (b) H∞ controller.
Figure (a)
41.41.
Figure Real and
Real andestimated Topdrive
estimated Top drivecurrent
current under:
under: (a)(b)
(a) LQG LQG controller,
controller.
Figure 41. Real and estimated Top drive current under: (a) LQG controller, (b) H∞ controller.
(a) (b)
Figure (a) 42.
42.
Figure Real and
Real estimated
and estimated Top drivevelocity
Top drive velocity under:
under: (a)(b)
(a) LQG LQG controller,
controller.
Figure (a)
42. Real and estimated Top drive velocity under: (a)(b)
LQG controller, (b) H∞ controller.
However, it should be taken into consideration that since the H∞ observer-based
5. Conclusions
Figure 42. Real
controller and estimated
proposed in this Top drive
study wasvelocity
based onunder: (a)two
solving LQGcascaded
controller, (b) H∞
Riccati controller.
equations,
5. Conclusions
This work wascovariance
devotedmatrices
to the B design of an observer-based controller in rotary
the choice of the 1 , B2 , and D12 is very crucial for the approach
drilling systems.
5. Conclusions
This work Hence,
convergence. The motivation
was devoted
it is highly behind
torecommendablethis choice
the design oftoan runof control strategy
observer-based
many tests includes
controller
and scenarios therotary
in
with tech-
nical and practical
different
drilling covariance difficulties that arise
matrices values in thewhen measuring
drilling field in the down
order hole
to guarantee drilling
that no torque
Thissystems.
work was The motivation
devoted to behind the design this choice
of an of control
observer-basedstrategy includes
controller inthe tech-
rotary
and matter
speed. what
For the
thistopreason,
drive inputan signals are,with
observer the control
unknown systeminput
will converge
was to its required
inserted into the con-
nical
drilling and practical
systems. difficulties
The motivation that arise
behind when
this measuring
choice of controlthe down
strategy hole drilling
includes torque
the tech-
trol dynamic
loop. in
This the most
observer robust way.
withanunknown Based on the
input obtained
was capable results in this
of was research
producing work,
estimations of
and
nical speed.
and and Forto
practical
in order
this reason,
difficulties
preserve
observer
that
the obtained arise with
whenunknown
robustness ofmeasuring
input
the down
the H∞ observer-based
inserted
hole into
drilling
controller,
the con-
we torque
the
trol real
loop.states ofobserver
the drilling system, in addition to an estimate of a primordial quantity
speed.This
andrecommend Forthethis reason, with
implementation anunknown
observer input
with
of signal filters was
unknown
that capable
can normalizeinputof producing
was
and inserted
adapt estimations
the top into
drivethe con- of
of
thethe drilling
real states process,
of the which
drilling is the
system, torque
in on
additionbit. The
to an approach
estimate adopted
of a in
primordialthis work
quantity to
trolinput
loop.signals,
This namely
observer thewith
torqueunknown
and the angularinputvelocity, in an appropriate
was capable of producingway similar to
estimations of
mitigate
of the
the the high
drilling
simulation frequency
process,
scenarios which stick-slip
is
provided the
in vibrations
torque
this on
research bit.
work. in a approach
The robust and efficient
adopted in waywork
this in the to
the real states of the drilling system, in addition to an estimate of a primordial quantity
drill-string
mitigate of high
the rotaryfrequency
drilling systems was the H∞ observer-based controller. A model of
of the drilling process, which stick-slip
is the torque vibrations
on bit. Thein aapproach
robust and efficient
adopted way
in this in the
work to
ten degrees
drill-string of
of freedom
rotary of
drilling the drill
systems string
was was
the H∞developed based
observer-based on the Simscape
controller. A model envi-of
mitigate the high frequency stick-slip vibrations in a robust and efficient way in the
ronment
ten degrees of Simulink/Matlab. Moreover, a model of the rock–bit
based interaction was estab-
drill-string ofofrotary
freedom of the
drilling drill string
systems was the wasH∞ developed
observer-based on the Simscape
controller. A model envi-of
lished
ronment underof the same environment
Simulink/Matlab. Moreover, along a with the
model of top rock–bit
the drive model. To estimate
interaction was the
estab-
ten degrees of freedom of the drill string was developed based on the Simscape envi-
5. Conclusions
This work was devoted to the design of an observer-based controller in rotary drilling
systems. The motivation behind this choice of control strategy includes the technical and
practical difficulties that arise when measuring the down hole drilling torque and speed.
For this reason, an observer with unknown input was inserted into the control loop. This
observer with unknown input was capable of producing estimations of the real states
of the drilling system, in addition to an estimate of a primordial quantity of the drilling
process, which is the torque on bit. The approach adopted in this work to mitigate the
high frequency stick-slip vibrations in a robust and efficient way in the drill-string of rotary
drilling systems was the H∞ observer-based controller. A model of ten degrees of freedom
of the drill string was developed based on the Simscape environment of Simulink/Matlab.
Moreover, a model of the rock–bit interaction was established under the same environment
along with the top drive model. To estimate the downhole states, especially the bit speed
and the torque on bit, an observer was first designed and tested under different scenarios
by comparing the measured and estimated states to ensure the reliability of the observer
before designing and discussing the controller robustness and performance. The obtained
results of the observer were very promising. Thus, the controller was implemented in
cascade to ensure the tracking task. This controller’s purpose was to drive the drill bits
peed, relying upon the unknown in put observer outcomes in such a way to keep the bit
speed steady even under extreme conditions of the weight on bit, model uncertainties,
and measurement noises. That is why the obtained observer-based controller has been
tested under different possible and probable scenarios from the drilling field. The obtained
results have demonstrated that the controller robustness is very high, for which the high
frequency stick-slip vibrations were suppressed in less than 5 s. This is a very short
time in comparison to the manual suppression adopted nowadays in the drilling fields.
Moreover, the severity of stick-slip vibrations has been quantitatively evaluated through the
proposed empirical equation. Furthermore, the H∞ observer-based controller performance
has been qualitatively compared to the LQG observer-based controller. The proposed
approach demonstrated good performance even for an extreme case, in which it provided
a rejection to unstructured perturbations unlike the LQG approach. Therefore, based on
the obtained results, it is highly recommendable to implement the designed approach on
an operating rotary drilling system towards the development of the so-called smart rotary
drilling systems.
Author Contributions: Conceptualization, M.Z.D. and M.K.; methodology, M.Z.D. and S.D.; software,
R.R.; validation, M.K., S.D. and M.Z.D.; formal analysis, R.R.; investigation, M.Z.D.; resources, M.K.;
data curation, R.R. and S.D.; writing—original draft preparation, R.R. and M.Z.D.; writing—review
and editing, S.D. and M.Z.D.; visualization, R.R.; supervision, M.Z.D. and S.D.; project administration,
M.K. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: There are no acknowledgments for this paper.
Conflicts of Interest: The authors declare no conflict of interest.
References
1. Moharrami, M.J.; Martins, C.d.; Shiri, H. Nonlinear integrated dynamic analysis of drill strings under stick-slip vibration.
Appl. Ocean Res. 2021, 108, 102521. [CrossRef]
2. Liu, X.; Long, X.; Zheng, X.; Meng, G.; Balachandran, B. Spatio-temporal dynamics of a drill string with complex time-delay
effects: Bit bounce and stick-slip oscillations. Int. J. Mech. Sci. 2019, 170, 105338. [CrossRef]
3. Zarabimanesh, Y.; Saffari, P.R.; Saffari, P.R.; Refahati, N. Hydro-thermo-mechanical vibration of two vertically aligned single-
walled boron nitride nanotubes conveying fluid. J. Vib. Control 2021. [CrossRef]
4. Jansen, J.D. Nonlinear Dynamics of Oil Well Drill Strings; Delft University Press: Delft, The Netherlands, 1993; p. 223,
ISBN 90-6275-880-0.
5. Runia, D.J.; Dwars, S.; Stulemeijer, I.P.J.M. A Brief History of the Shell “Soft Torque Rotary System” and Some Recent Case Studies.
In Proceedings of the SPE/IADC Drilling Conference, Amsterdam, The Netherlands, 5–7 March 2013; p. 10. [CrossRef]
6. Johannessen, M.K. Stick-Slip Prevention of Drill Strings Using Nonlinear Model Reduction and Nonlinear Model Predictive
Control. Ph.D. Thesis, Norwegian University of Science and Technology, Trondheim, Norway, 2010; p. 99.
7. Liu, X.; Vlajic, N.; Long, X.; Meng, G.; Balachandran, B. Nonlinear Oscillations of a flexible rotor with a drill bit: Stick-slip and
delay effects. Nonlinear Dyn. 2013, 72, 61–77. [CrossRef]
8. Mendil, C.; Kidouche, M.; Doghmane, M.Z. Hybrid backstepping sliding mode controller for stick–slip vibrations mitigation in
rotary drilling systems. IETE J. Res. 2021. [CrossRef]
9. Mendil, C.; Kidouche, M.; Doghmane, M.Z. Hybrid sliding PID controller for torsional vibrations mitigation in rotary drilling
systems. Indones. J. Electr. Eng. Comput. Sci. 2021, 22, 146–158. [CrossRef]
10. Jansen, J.D.; van den Steen, L. Active damping of self-excited torsional vibrations in oil well drillstrings. J. Sound Vib. 1995, 179,
647–668. [CrossRef]
11. Abdulgalil, F.; Siguerdidjane, H. Nonlinear Friction compensation design for suppressing stick slip oscillations in oil well
drillstrings. IFAC Proc. 2004, 37, 769–773. [CrossRef]
12. Navarro-Lopez, E.M.; Suarez, R. Practical Approach to Modelling and Controlling Stick-Slip Oscillations in Oilwell Drill Strings.
In Proceedings of the 2004 IEEE International Conference on Control Applications, Taipei, Taiwan, 2–4 September 2004; Volume 2,
pp. 1454–1460. [CrossRef]
13. Li, L. Time-varying sliding mode adaptive control for rotary drilling system. J. Comput. 2011, 6, 564–570. [CrossRef]
14. Karkoub, M.; Zribi, M.K.M.; Lamont, L.E.L. Robust µ-synthesis controllers for suppressing stick-slip induced vibrations in oil
well drill strings. Multib. Syst. Dyn. 2010, 23, 191–207. [CrossRef]
15. Vromen, T.G.M. Control of Stick-Slip Vibrations in Drilling Systems. Ph.D. Thesis, Eindhoven University of Technology,
Eindhoven, The Netherlands, 2015; p. 256.
16. Doghmane, M.Z.; Bacetti, A.; Kidouche, M. Stick-Slip Vibrations Control Strategy Design for Smart Rotary Drilling Systems.
In Artificial Intelligence and Renewables Towards an Energy Transition, Proceedings of the ICAIRES 2020, Tipaza, Algeria, 22–24 November
2020; Hatti, M., Ed.; Lecture Notes in Networks and Systems; Springer: Cham, Switzerland, 2020; Volume 174. [CrossRef]
17. Riane, R.; Kidouche, M.; Illoul, R. Unknown resistive torque estimation of a rotary drilling system based on Kalman filter. IETE J.
Res. 2020. [CrossRef]
18. Hong, L.; Girsang, I.P.; Dhupia, J.S. Identification and control of stick–slip vibrations using Kalman estimator in oil-well drill
strings. J. Pet. Sci. Eng. 2016, 140, 119–127. [CrossRef]
19. Fu, M.; Zhang, P.; Li, J.; Wu, Y. Observer and reference governor based control strategy to suppress stick-slip vibrations in oil well
drill-string. J. Sound Vib. 2019, 457, 37–50. [CrossRef]
20. Canudas-de-wit, C.; Corchero, M.A.; Rubio, F.R.; Navarro-l, E. DOSKIL: A New Mechanism for Suppressing Stick-Slip in Oil Well
Drillstrings. In Proceedings of the 44th IEEE Conference on Decision and Control, and the European Control Conference, Seville,
Spain, 12–15 December 2005; pp. 8260–8265.
21. Riane, R.; Kidouche, M.; Doghmane, M.Z.; Illoul, R. Modeling of Torsional Vibrations Dynamic in Drill-String by Using PI-
Observer. In Proceedings of the 4th International Conference on Electrical Engineering and Control Applications, ICEECA 2019,
Constantine, Algeria, 17–19 December 2019; Bououden, S., Chadli, M., Ziani, S., Zelinka, I., Eds.; Springer: Singapore, 2021;
Volume 682. [CrossRef]
22. Saffari, P.R.; Fakhraie, M.; Roudbari, M.A. Size-dependent vibration problem of two vertically aligned single-walled boron nitride
nanotubes conveying fluid in thermal environment via nonlocal strain gradient shell model. J. Solid Mech. 2021, 13, 164–185.
[CrossRef]
23. Mamaghani, A.E.; Khadem, S.E.; Bab, S. Vibration control of a pipe conveying fluid under external periodic excitation using a
nonlinear energy sink. Nonlinear Dyn. 2016, 86, 1761–1795. [CrossRef]
24. Yongkun, W.; Mingwei, S.; Zhenghui, W.; Zhongxin, L.; Zengqiang, C. A novel disturbance-observer based friction compensation
scheme for ball and plate system. ISA Trans. 2014, 53, 671–678. [CrossRef]
25. Zheng, X.; Agarwal, V.; Liu, X.; Balachandran, B. Nonlinear Instabilities and control of drill-string stick-slip vibrations with
consideration of state-dependent delay. J. Sound Vib. 2020, 473, 115235. [CrossRef]
26. Gupta, S.K.; Wahi, P. Bifurcations in the axial torsional state-dependent model of rotary drilling. Int. J. Non-Linear Mech. 2018, 99,
13–30. [CrossRef]
27. Fleury, P.; Mathieu, J. Vibrations Mecaniques, Acoustique; Sér. Physique Generale et Experimentale [Series]; Eyrolles: Paris, France,
1962.
28. Kessai, I.; Benammar, S.; Doghmane, M.Z.; Khelifa, S. Preventive Maintenance Optimization of Top Drives in Smart Rotary
Drilling Systems. In Artificial Intelligence and Renewables Towards an Energy Transition, Proceedings of the ICAIRES 2020, Tipaza,
Algeria, 22–24 November 2020; Hatti, M., Ed.; Lecture Notes in Networks and Systems; Springer: Cham, Switzerland, 2020;
Volume 174. [CrossRef]
29. Pelfrene, G. Rôle du Processus de Forabilité des Roches dans les Vibrations de Torsion des Systèmes de Forage Pétrolier. Ph.D.
Thesis, École Nationale Supérieure des Mines de Paris, Paris, France, 2010.
30. Richard, T.; Germay, C.; Detournay, E. A Simplified model to explore the root cause of stick-slip vibrations in drilling systems
with drag bits. J. Sound Vib. 2007, 305, 432–456. [CrossRef]
31. Detournay, E.; Richard, T.; Shepherd, M. Drilling response of drag bits: Theory and experiment. Int. J. Rock Mech. Min. Sci. 2007,
305, 432–456. [CrossRef]
32. Detournay, E.; Defourny, P.A. Phenomenological model for the drilling action of drag bits. Int. J. Rock Mech. Min. Sci. Geomech.
Abstr. 1992, 29, 13–23. [CrossRef]
33. Mendil, C.; Kidouche, M.; Doghmane, M.Z.; Benammar, S.; Kong, F.T. Rock–bit interaction effects on high-frequency stick-slip
vibration severity in rotary drilling systems. Multidiscip. Model. Mater. Struct. 2021, 17, 1007–1023. [CrossRef]
34. Young, P.C.; Willems, J.C. An approach to the linear multivariable servomechanism problem. Int. J. Control 1972, 15, 961–979.
[CrossRef]
35. Zhen, C.; Xuzhi, L.; Min, W.; Chengda, L.; Luefeng, C. Equivalent-input-disturbance-based robust control of drilling trajectory
with weight-on-bit uncertainty in directional drilling. ISA Trans. 2021; in press. [CrossRef]
36. Doyle, J.; Glover, K.; Khargonekar, P.; Francis, B. State space solutions to standard h2 and h∞, control problems. IEEE Trans.
Autom. Control 1989, 34, 831–847. [CrossRef]
37. Alif, A.; Boutayeb, M.; Darouach, M. On the Design of Robust H8 Memory and Memoryless Tracking Controllers for a Class
of Linear Time Delay Systems with Time-Varying Uncertainties. In Proceedings of the American Control Conference 2006,
Minneapolis, MN, USA, 14–16 June 2006; pp. 3867–3872. [CrossRef]
38. Doghmane, M.Z.; Kidouche, M.; Eladj, S.; Belahcene, B. Design of Optimal Decentralized Controller Using Overlapping
Decomposition for Smart Building System. In Artificial Intelligence and Renewables Towards an Energy Transition, Proceedings of the
ICAIRES 2020, Tipaza, Algeria, 22–24 November 2020; Hatti, M., Ed.; Lecture Notes in Networks and Systems; Springer: Cham,
Switzerland, 2020; Volume 174. [CrossRef]
39. Amiri, S.; Keyanpour, M.; Masoudi, M. Observer-based output feedback control design for a fractional ODE and a fractional PDE
cascaded system. ISA Trans. 2021; in press. [CrossRef]
40. Doghmane, M.Z.; Kidocuhe, M.; Riache, S.; Aibeche, A. Hybrid Adaptive Backstepping-Sliding Mode Control Design for
Non-linear Under-Actuated Systems. In Artificial Intelligence and Heuristics for Smart Energy Efficiency in Smart Cities, Proceedings of
the IC-AIRES 2021, Tipasa, Algeria, 24–26 November 2021; Hatti, M., Ed.; Lecture Notes in Networks and Systems; Springer: Cham,
Switzerland, 2022; Volume 361. [CrossRef]
41. Mendil, C.; Kidouche, M.; Doghmane, M.Z. Modeling of Hydrocarbons Rotary Drilling Systems Under Torsional Vibrations:
A Survey. In Artificial Intelligence and Renewables Towards an Energy Transition, Proceedings of the ICAIRES 2020, Tipaza, Algeria,
22–24 November 2020; Hatti, M., Ed.; Lecture Notes in Networks and Systems; Springer: Cham, Switzerland, 2020; Volume 174.
[CrossRef]
42. Besselink, B.; Vromen, T.; Kremers, N.; van de Wouw, N. Analysis and control of stick-slip oscillations in drilling systems. IEEE
Trans. Control Syst. Technol. 2016, 24, 1582–1593. [CrossRef]
43. En-Zhi, C.; Bao-Lin, Z.; Zhihui, C.; Qing-Long, H.; Binrui, W. Observer-based state feedback control for offshore steel jacket
structures under denial-of-service attacks. ISA Trans. 2021, 115, 46–60. [CrossRef]
44. Sarker, M.; Rideout, D.G.; Butt, S.D. Butt, Advantages of an LQR Controller for Stick-Slip and Bit-Bounce Mitigation in
an Oil Well Drill String. In Proceedings of the 2012 ASME IMECE, Houston, TX, USA, 9–15 November 2012; pp. 1305–1313.
[CrossRef]
45. Doghmane, M.Z.; Kidouche, M.; Ahriche, A. Decentralized overlapping control design with application to rotary drilling system.
IETE J. Res. 2021. [CrossRef]
46. Kessai, I.; Benammar, S.; Doghmane, M.Z. Dynamic failure analysis and lifetime estimation of tool-string in rotary drilling system
under torsional-axial coupled vibrations. Eng. Fail. Anal. 2022, 134, 106037. [CrossRef]
47. Kessai, I.; Benammar, S.; Doghmane, M.Z.; Tee, K.F. Drill bit deformations in rotary drilling systems under large-amplitude
stick-slip vibrations’. Appl. Sci. 2020, 10, 6523. [CrossRef]
48. AlZibdeh, A.; AlQaradawi, M.; Balachandran, B. Effects of high frequency drive speed modulation on rotor with continuous
stator contact. Int. J. Mech. Sci. 2017, 131–132, 559–571. [CrossRef]
49. Mendil, C.; Kidouche, M.; Doghmane, M.Z. A Study of the Parametric Variations Influences on Stick-Slip Vibrations in Smart
Rotary Drilling Systems. In Artificial Intelligence and Renewables Towards an Energy Transition; M. Hatti, Ed.; Springer International
Publishing: Cham, Switzerland, 2021; pp. 707–715. [CrossRef]
50. Doghmane, M.Z.; Kidouche, M. Optimal Decentralized State Control of Multi-Machine Power System Based on Loop Multi-
overlapping Decomposition Strategy. In Proceedings of the 4th International Conference on Electrical Engineering and Control
Applications, ICEECA 2019, Constantine, Algeria, 17–19 December 2019; Bououden, S., Chadli, M., Ziani, S., Zelinka, I., Eds.;
Springer: Singapore, 2021; Volume 682. [CrossRef]
51. Kessai, I.; Benammar, S.; Doghmane, M.Z.; Tee, K.F. Estimation of circular arc crack depths and locations in rotary drilling pipes
subjected to free vibrations. Vibrations 2022, 5, 165–182. [CrossRef]

Vibration 05 00016 v3

Uploaded by

Copyright:

Available Formats

Vibration 05 00016 v3

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Vibration 05 00016 v3

Uploaded by

Copyright:

Available Formats

vibration

1 Laboratoire d’Automatique Appliquée, Faculté des Hydrocarbures et de la Chimie, Université M’Hamed

Publisher’s Note: MDPI stays neutral

Vibration 2022, 5, 264–289. https://doi.org/10.3390/vibration5020016 https://www.mdpi.com/journal/vibration

2. Drill-String Model under Stick-Slip Vibrations

2.1. Torsional Vibrations

Table 1. Stick-slip severity classification.

Stick-Slip Severity Level Classification SS%

2.3.1. Scenario 1: Constant Weight on Bit

2.3.1. Scenario 1: Constant Weight on Bit

2.3.1. Scenario 1: Constant Weight on Bit

Vibration 2022, 5 272

3.1.1. Observer Synthesis with Unknown Inputs by H∞

3.1.1. Observer Synthesis with Unknown Inputs by H∞

Lh = −Y∞ CzT (8)

3.1.2. Controller Synthesis by H∞

In a more compact form, the system (11) is rewritten as:

4. Results and Discussion

Vibration 2022, 5, FOR PEER REVIEW 13

Vibration 2022, 5, FOR PEER REVIEW 13

Figure 14. Measured and estimated Toque on bit by H∞ observer.

Figure 14. Measured and estimated Toque on bit by H∞ observer.

Vibration 2022, 5 277

bration 2022, 5, FOR PEER REVIEW 15

Vibration 2022, 5, FOR PEER REVIEW

(a) (a) (b) (b)

ibration 2022, 5, FOR PEER REVIEW 16

4.3. Scenario6: Parametric Variation

ibration 2022, 5, FOR PEER REVIEW Figure 25.

Vibration 2022, 5, FOR PEER REVIEW 18

Vibration 2022, 5, FOR PEER REVIEW 18

Vibration 2022, 5, FOR PEER REVIEW 18

4.4.2. Scenario 8: Wob step Tracking Reference

4.4.3. Scenario 9: Stick-Slip Mitigation

Figure 31. Controlled Torque on bit by H∞ approach.

Figure 31. Controlled Torque on bit by H∞ approach.

Figure(a) (b)Scenario-9: (a) Current, (b) Velocity.

4.4.4. Scenario 10: Wob Disturbances Filtering

Vibration 2022, 5, FOR PEER REVIEW 21

Figure 36. Perturbed Weight on bit input for Scenario-11.

Figure 36. Perturbed Weight on bit input for Scenario-11.

Vibration 2022, 5 285

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.