Modeling, Identification and Control, Vol. 41, No. 1, 2020, pp.

1–11, ISSN 1890–1328

Automated drilling algorithms implementation on

a laboratory drilling rig
Erik Andreas Løken, Jens Løkkevik, Dan Sui

Energy and Petroleum Engineering Department, University of Stavanger, Stavanger, Norway. E-mail: dan.sui@uis.no

Abstract

Considering the state of art technology that exists today and the significant resources that are being
invested into the technology of tomorrow, an idea of intelligent and fully automated machineries working
on a drilling floor that is capable of consistently selecting best decisions or predictions based on realtime
information available and providing drillers and operators with such recommendations, becomes closer to
a reality every day.
This work shows results of the research carried out on the topic of drilling automation. Its objectives
are to design and test proof of concept technologies conducted on a laboratory-scale autonomous drilling
rig developed at University of Stavanger, Norway. Main contribution of the study is on drilling speed
(ROP) optimization with considering operational safety to personnel and environment (HSE) and drilling
efficiency along with a digitized drilling program for directional drilling. The case studies are presented
to show the different scenarios for drilling vertical wells and inclined wells.

Keywords: Drilling rig, drilling automation, drilling algorithm, drilling optimization

1. Introduction leaving only major decision points to be determined

by drillers. The latter automation level described for
the drilling scene is likely still several years away from
In recent years, the concept of drilling automation has being deployed to fields. A timeline that highlights ar-
advanced from primarily being automation of rig floor tificial intelligence applications in drilling practices is
equipment to novel solutions that can be rapidly de- given in Bello (2015).
ployed to the rig environment and assist drillers in a In our opinion, short term advances in drilling automa-
variety of operations. Aside from providing an early tion lie in developing digital and robust tools for drillers
warning to drillers, and if necessary, for instance per- to strengthen the understanding of the operations dur-
forming a controlled shut-in procedure once a kick is ing critical phases, and if possible, automate some of
migrating towards surface that gets detected, such in- the routine tasks that can be easily controlled by ma-
telligent systems could improve efficiency and reduce chines. In the same way that a Tesla can self-operate in
financial costs through continuous monitoring and in- known terrain, so should a drilling rig, enabling drillers
teractions with drillers and supervisors. Smart and to remain one step ahead, maintaining an overview of
digital drilling systems could also be used to suggest all on-going events.
operational parameters to drillers through correlating The study covers some of the topics including software
real-time drilling data with vast amounts of historic framework design for drilling automation system, con-
data stored in a virtual environment, popularly re- trol framework implementation, autonomous search al-
ferred to as a digital twin, or even exert full control gorithms development that allows our control system
of all rig equipment if permissible (top drive, draw- to maximize the rate of penetration (ROP) while min-
works, mud pumps, elevator, rough neck and so on)

doi:10.4173/mic.2020.1.1
c 2020 Norwegian Society of Automatic Control
 Modeling, Identification and Control

imizing the risk of running into drilling incidents and

drilling with sub-optimal parameter setpoints, and a
digital drilling algorithm proposed that has been im-
plemented to autonomously drill a deviation well using
a closed-loop steering system. Finally the results and
discussions are given to show the performance of our
autonomous drilling rig.

2. Drilling Rig
The drilling rig, see Figure 1, consists of in total four
hardware systems, in addition to the control system.
These are: rotation, hoisting, circulation and power
systems that are introduced below.
Rotation System: There exist two rotational systems:
a conventional top drive, used for vertical drilling and
a downhole motor used for directional drilling. This
set-up allows us to circulate drilling fluid (either wa- Figure 1: Drill rig.
ter, water-based mud or oil-based mud ) all the way
from the top of derrick to the drill bit nozzles. The Parameter Description and Unit
top drive provides rotational speeds up to 1500 RPM. Material Aluminium, 6061-T6 alloy
By varying two analog voltage signals one can control Pipe length 914.4 mm
the RPM and torque from the top drive. The signals Pipe OD 9.53 mm
are transmitted from programmable logic controllers Pipe ID 7.75 mm
(PLCs) to a driver for the top drive. Wall thickness 1.2446 mm
Hoisting System: In order to simulate drilling opera-
tions, the system is equipped with a hoisting system
that consists of three actuators, each equipped with a Table 1: Drill pipe description
stepper motor. The motors raise/lower the top plate
that resembles the drill floor where the top drive and
downhole motor and drill bit. While the knuckle joint
other components are mounted. The three actuators
can be used to control the dogleg by varying the WOB
are controlled by its own stepper motor with a step-
so that a spring gets compressed (bend) or left un-
angle of 1.8 degrees, where for each 1.8 degrees of step-
compressed (no bend), the mechanical angle of the top
ping 10 micro-steps get transmitted (resulting in a to-
drive can be varied through pulsing to change the az-
tal of 2000 steps/rev). High accuracy for the actua-
imuth, if an offset is registered. The bit RPM is con-
tors is a key in order to ensure the required weight on
trolled by throttling the air flow coming from a com-
bit (WOB) control. Such powerful stepper motors can
pressor to the downhole pneumatic motor. The alu-
provide an approximate maximum WOB up to 500 N.
minium drill pipe used for drilling has the following
To address the challenge of vibrations in the system,
dimension:
a rigid coupling is used between the stepper motors
Pipe limit calculations and destructive testing of the
and actuators to reduce the overall vibrations making
drill pipe are given in Hagen et al. (2018), which are
the system more sturdy. More information and details
summarized below:
are found in Løken and Trulsen (2017), Hagen et al.
(2018), Hjelm and Nilsen (2018). • critical slenderness ratio of the pipe is 70.2, and
Circulation System: The circulation system consists of slenderness ratio of the drill pipe is approximately
two pumps alternating for 30 seconds each. This pre- 280.2,
vents the pumps from over-heating and should one
pump fail, the other will kick in to ensure the required • buckling limit is 280.5 N,
velocity margin for adequate hole cleaning and cooling • maximum torque before the pipe yields is approxi-
of the bit. The pumps have a maximum flow rate of mately 19.0 Nm, while the maximum torque before
19 L/min at a working pressure of 3.1 bar, see Hagen the pipe shears is approximately 24.7 Nm,
et al. (2018).
The drill string assembly consists of: aluminium drill • maximum tensile load is approximately 10.331 kN
pipe, downhole sensor sub, knuckle joint, pneumatic and maximum compressive load is approximately

2
 Løken et al., “Automated Drilling”

985.72 N. a window-size of 5 samples, which combined with 60

Hz frequency has proven to be reliable for the WOB
control.
3. Control System Architecture
The drilling system is a collection of several systems
including mechanical, hydraulic, electrical, hardware, 4. Software Architecture
software, and data systems. For our system, the hard-
ware architecture is comprised of three levels: PLCs on
The software architecture is configured as a multi-
level 1, computers on level 2 and cloud-based storage
threading system, meaning that several client/server
of time-based data (optionally also depth-based data)
modules, referred to as threads that run on multiple
on level 3.
cores on the central processing units (CPUs), run in
On level 1, the control aspect (PLCs) for the main sys-
parallel using a gRPC-API (see Section 4.1) to han-
tems on the rig are located to control rotational sys-
dle the communication between each module / thread.
tem, hoisting system and hydraulic mud system. Each
In addition, an API based on OPC UA (see Section
system gets controlled through its own microcontroller
4.2) has been developed to support remote connectivity
(PLC), that both executes commands to the control-
and remote event handling, allowing external partners
lable systems on the rig (actuators, motors, pumps and
to execute commands remotely to the control system,
so on) and gathers information from the equipped sen-
and receive data from the rig in real-time upon request.
sors. On level 2, PCs are located to handle the decision-
The system is described further in Sand (2018) and
making logic. This logic involves carrying out the dig-
Guggedal and Steinstø (2019).
ital detailed operating procedure (DDOP), see Section
6.3 for detailed information, ROP search algorithms
(see Section 6.1), steering algorithm (see Section 6.2),
detection of drilling incidents and carrying out reme- 4.1. gRPC
dial actions, as well as pushing the data to a configured
database for post-analysis and data extraction (level The gRPC is an open source Remote Procedure Call
3). A key factor has been to ensure that all compo- (RPC) developed by Google. RPC is a communication
nents that make up the drilling system are compatible platform where the program performs remote opera-
for real-time control. The implemented control system tions on a server. The framework allows for communi-
dataflow with PLCs is illustrated in Figure A.1. cation between a variety of platforms and programming
The PID controller is used controller in the closed-loop languages, and while traditional APIs (applicable pro-
systems as: gramming interface) share data through formats read-
Z t able by humans, the gRPC uses protocol buffers for
de(t) transparent communication between client and server
u(t) = u0 + Kp e(t) + Ki e(τ )dτ + Kd (1)
0 dt applications. Protocol buffers are communication plat-
where u0 is the bias, u(t) is the error between true mea- forms that are not readable by humans. The user cre-
surement and the setpoint, Kp is the controller gain for ates the server client implementation. This kind of
proportional controller, Ki is the controller gain for in- structure is beneficial for many reasons, but the main
tegral controller and Kd is controller gain for derivative one is that both size of buffer files are smaller and
controller. When drilling on the laboratory rig, the communication times are much higher than traditional
hoisting system controls the WOB to drill the forma- communication platforms, see Guggedal and Steinstø
tion. It is important that the selected WOB setpoint is (2019).
constant during drilling in order to keep a steady drill
rate, and the use of a PID controller ensures that the
error between the WOB setpoint and measured hook 4.2. API - OPC UA
load is minimized, seeing that the WOB is calculated
as the difference between free hanging weight and hook An Open Platform Communications Unified Architec-
load. The Kp , Ki and Kd gains can get defined and up- ture (OPC UA) is used for the API implementation on
per and lower thresholds allow the operator to adjust the rig in order to integrate a plug-and-play concept on
how prone the controller should be to axial vibrations. the drilling rig. This enables the option for the client
The rig operator can also select different distances and to remotely execute two different types of interactions
speeds for the actuators to move either to increase or with the rig: request information such as drilling data
decrease the WOB. The WOB data that is sampled in or exert commands to control the rig, see Guggedal and
real-time is filtered using a moving median filter with Steinstø (2019).

3
 Modeling, Identification and Control

4.3. Graphical User Interfaces (GUIs) our requirements where both leverages the features of
relational database systems, and also offers great query
Two GUIs have been developed for this laboratory
optimization for time series data. This is convenient in
scale drilling rig. These are a visualization GUI to
order to not have to implement two different databases;
track the progress of the autonomous control system
one to store sensor data and another to store rig state
and drilling performance and a downhole well trajec-
data.
tory environment, see Figure A.2, also showing the
In order to easily manage the data that gets stored in
real-time position, inclination and orientation of the
the database, an application that serves as the inter-
bit.
face between the users and the database is required. A
separate small application is also required: responsible
5. Data Management for pushing the data to the database. The sensors’ sig-
nals are sampled at a specific frequency and these data
5.1. Sensors are stored in text files on the computer.
To make the data in the database easily accessible, a
With the main task being developing an autonomous user-friendly platform to access the data was created
drilling rig, the sensor implementation is an area that in the team. This is solved by writing a web appli-
has received a lot of attention. The computer relies cation, where users can log in and manage the stored
on good data from sensors implemented to make deci- data. Users can access and download the data gathered
sions and provide users with valuable information. Our from either specific experiments, or a particular subset
drilling rig system has a total of 17 sensors, including of those data. They can then choose to visualize the
three load cells capable of providing measurements for data by generating graphs, query experiments using
x, y and z-axis; two torque sensors, one integrated in different parameters, manage which variables that the
the top drive and one external for torque measurement database shall contain, register the drilled rocks (la-
at the top of the drill string. The design, manufactur- bel the rock types, experiments, add comments etc.),
ing and implementation of a downhole measurement manage access to the database, and so on. In addition,
tool including accelerometer, gyroscope and magne- the data can easily be down-sampled to for instance
tometer have also been conducted in our work. Drill 10% of the raw data. When working with datasets of
string vibration is a complex phenomenon that often millions of rows (measurements) and between five and
results in Non-Productive Time. The complexity lies thirty columns (sensors or features), this immediately
in the coupled action of the three vibrational modes: allows the user to select which data-frequency to work
axial, lateral and torsional. The modes excite one an- with, rather than having to import a large dataset,
other, which makes it difficult to recognize a particular carry out either a linear down-sampling process and
mode and prevent destructive vibrations. A downhole then save the data as a new file.
measurement system allows us to capture the moment
when vibrations occur and adjust operational parame-
ters (the WOB and the RPM) in real-time to mitigate 6. Autonomous Drilling
vibrations. The other function of downhole measure-
ment is to do he bit position tracking and control. The concept is a closed-loop autonomous control sys-
tem that combines several models with a digital de-
tailed operational drilling procedure developed for di-
5.2. Database
rectional drilling. A simplified flow chart illustrating
During each drilling experiment, the sensors installed the system is shown in Figure 2. In addition, a novel
on the rig gather a large amount of time-stamped data. voting system has been designed to prioritize which
There have been many drilling experiments conducted commands that should receive priority if an event, or
over the years and a lot of data has been collected, incident gets detected or if conflicting commands from
either in real-time or for post analysis using a high- models get received at the same time. The voting sys-
frequency data acquisition (DAQ) module. tem has been configured so that critical incidents that
TimescaleDB is a faster and more reliable solution. can risk damage to either equipment or the well re-
It is created by heavily modifying the architecture of ceive the highest priority. Maximizing the ROP and
PostgreSQL, a well-known RDBMS. TimescaleDB still ensuring that downhole vibrations are low are located
preserves all the features and capabilities of a con- at the lowest priority since these tasks are not critical
ventional RDBMS. Data is organized in tables with to handle immediately to meet the objective.
columns, rows, indexes, primary keys, foreign keys, There are however several challenges to perform an op-
constraints and all the desirable features usually re- timal ROP search. In the drilling scene, there is lacking
served for relational databases. TimescaleDB satisfies information with regards to the destination that the

4
 Løken et al., “Automated Drilling”

tion that depends on various parameters such as WOB,

torque on bit (TOB), RPM, flow rate (Q), rock for-
mation strength (UCS), drill bit properties and so on.
The machine exerts control over the bit side proper-
ties. The formation strength and drilling environment
on the other hand is heterogeneous, and changes in
the drilling environment must be responded to in real-
time. The two most important control variables are
WOB and RPM. By varying either of the control vari-
ables (that are coupled, meaning that if one is varied
so becomes the other) a difference in ROP, drilling ef-
ficiency and system response can be expected.
A gradient descent (multivariate) search algorithm to
optimize the ROP has been developed in Geekiyanage
et al. (2019). The search algorithm now calculates the
ROP gradient from evaluating the change in the depth
gradient (ROP) over the last five seconds, and a learn-
ing rate function can be used to determine the step size
to move for each iteration. The algorithm is terminated
when the following condition is no longer satisfied:

0 < X ≤ Xi,max . (2)

Denoting the drilling parameters WOB = x1 , TOB

= x2 , RPM = x3 , etc., the ROP X = (x1 , x2 , ..., xn )
Figure 2: Simplified illustration of the logic behind the where n ∈ <+ . ROP can then be expressed as:
autonomous drilling system.
ROP = f (X) = f (x1 , x2 , ..., xn ) (3)

The search algorithm will attempt to find a set of coor-

machine needs to reach for the maximum ROP and dinates (x∗1 , x∗2 , ..., x∗n ), so that the ROP becomes max-
the most optimal route to get there. With several lo- imum. The operational state space of the drilling ma-
cal maximums in the state space, the machine needs to chine constitutes the search space of the algorithm,
not only locate a peak, but the global peak to ensure satisfying the condition Xi,min < X < Xi,max , for
that the highest ROP is achieved. Another challenge instance 0 < RP M < 2000 and 0 < W OB < 50.
occurs when considering the route that the machine The depth gradient, referred to as instantaneous ROP,
follows in the search. Perhaps would the shortest path is calculated using linear regression as can be shown
in an operation cause the machine to drill with highly in Equation (4), where N represents the number of
suboptimal controller set points; for instance a very samples considered to obtain an instantaneous gradi-
high RPM combined with an unnaturally low WOB ent and n is always ≥ 2.
which can result in severe lateral vibrations and pos-
sibly also damage either the drill string, BHA or bit. ROPinstantaneous = ∇f (Xk )
A third challenge is the step size that the algorithm Pn Pn Pn
selects. By keeping the step size very small, the time N [ Xk,t × f (Xk,t )] − Xk,t × f (Xk,t )
t=1 t=1 t=1
to move from the start point to the goal state increases = Pn n
P (4)
as the machine would need to not only vary the con- N (Xk,t )2 − [ Xk,t ]2
troller setpoint but also drill at those setpoints in order t=1 t=1

to measure the response. If the step size is too large An objective function J can be defined using Euclidean
however, this reduces the time required to reach the norm, so that:
destination but increases the likelihood of possibly by-
passing the destination (overshooting). min J(X) = kf (X) − ROPsetpoint k2 . (5)
X

The ROP gradient vector can now get calculated from:

6.1. ROP Optimization
−
→ ∂f (X) ∂f (X) ∂f (X) T
g(X0 ) = ∇f (X0 ) = [ , , ..., ]
To optimize the ROP, one first needs to consider what ∂x1 ∂x2 ∂xn X=X0
ROP really is. ROP is a multi-variable vector func- (6)

5
 Modeling, Identification and Control

At any given combination of the control parameters, For every 1 mm measured depth (MD), the relation-
the ROP gradient always points to the maximal in- ship between horizontal build and true vertical depth
crease of the ROP function, and the gradient is al- (TVD) gets calculated so that:
ways perpendicular to the ROP hyper-surface contour
f (X) = c, where the constant c is an arbitrary real H horizontalbuild[mm]
= (10)
number. The algorithm is now capable of determining TV D T V D[mm]
A look up table has been created which contains the
H/T V Dplanned . The look up table values are then
called every mm, and the relationship between the ac-
tual build and planned build is checked by the equa-
tion:
H/T V Dactual
Statusbuild = . (11)
H/T V Dplanned
If the build status is ≥ 1, the system will be in a drop
angle mode, and if the build status is ≤ 1, the machine
will be in a build angle mode and needs to increase the
build-rate by increasing the WOB.
While the downhole RPM only depends on the gra-
dient search algorithm and the different phases of the
operation, the voting system in the control system will
continuously evaluate whether or not the system is in
a build or drop mode. If for example the ROP gradient
search proposes 14 kg WOB, but the system should be
in a drop mode, the WOBsetpoint will be overwritten
Figure 3: Constraints and ROP contours in the state to 12 kg, which is the highest permissible WOB in the
space of the drilling machine. drop mode. While a higher WOB from experiments
have yielded a higher ROP, another control algorithm
the search direction from a previous operating point. exists that evaluates whether the recommended con-
The optimization now is reduced to a univariate line troller setpoint (within the range identified for build or
search, moving along the local gradients. The gradient drop mode) gets forwarded to the PID controller, or
descent method can be shown as: whether an even lower setpoint should be used in the
event of severe downhole vibrations or if the system is
→
− →
− −→
X k+1 = X k − ηgk Xk , (7) in a critical phase.

where η represents the learning rate; i.e. the step size 6.3. Digital Detailed Operating Procedure
to move in each iteration, and k denotes the iteration
number. The step size can be selected by the equation The following digital detailed operating procedure
below, considering that we define φ(ηk ) = f (Xk + ηgk ) (DDOP) has been developed to autonomously execute
and 0 <  < 12 , which satisfies the Wolfe condition: the drilling of a directional well. The DDOP is ar-
ranged in eight phases, as visualized in Figure 4.
0
φ(ηk ) ≥ (1 − )φ (0). (8)

The termination criterion to abort the search if the 7. Rig Performance

ROP increases at k + 1 is less than k can be defined In this section, three well logs of a short inclined sec-
by: δ the tolerance specified for when gradient search tion, a vertical pilot hole and a complete inclined well-
shall terminate bore section are presented. The plotted figures have
been downsampled to 1 Hz (from a 60 Hz sampling
kf (Xk+1 ) − f (Xk )k ≤ δ. (9)
frequency). To smoothen the plots for inclination, az-
imuth and weight on bit, a Kalman filter has been used.
6.2. Downhole Closed-loop Steering
7.1. Experiment 1: Inclined Well Section
The concept of closed-loop steering in the control sys-
tem is currently limited to inclination steering through The first well section with logging tools in the BHA was
active management of knuckle joint bend in the BHA. drilled from 355 mm MD to 440 mm MD in homoge-

6
 Løken et al., “Automated Drilling”

difficult to monitor and confirm bit whirl with the cur-

rent configuration.

7.2. Experiment 2: Vertical Well Section

The second well log, is shown in Figure 7. The RPM

is kept nearly constant at 1040, and the WOB is in-
creased from first 5, to 10 and later 15 kg. While the
increased WOB leads to an improvement in the ROP,
a slight build can be observed (according to measure-
ments and visual inspection), which is possibly caused
by the knuckle joint starting to bend at high WOB.
As can be recalled from Section 6.2, the range for
WOBBuildInclination starts at approximately 12 kg (up
to maximum 18 kg). While the MSE also for this well
log is calculated with 1 Nm torque constantly, it can
be observed that the MSE is reduced as the WOB in-
creases, suggesting less energy is wasted, and a higher
percentage of the combined energy usage goes into bit-
rock interaction.

Figure 4: Eight phases constitute the digital direc- 7.3. Experiment 3: Deviation Well with
tional drilling operation. WOB 5 to 20 kg

Continuing to vary the WOB, the same approach as

neous cement, using a 7 degree whipstock, see Figure 5. before is used in an inclined section in Figure 8. For
The WOB setpoint is kept constant at 5 kg, while the the first 235 mm, the bit is RIH, with a 5 kg WOB set-
RPM setpoint is varied between 925 and 1075 revolu- point and an RPM of between 730 and 1040. The pre-
tions per minute. The resulting ROP is approximately viously drilled well section that the whipstock had been
0.4 to 0.5 m/hr, which is considered a good result given landed in got drilled as shown in Experiment 2 above.
the increased wellbore friction and a low WOB set- Contrary to the last experiment, it is observable that
point. WOB setpoint has from experiments conducted while the ROP continues to increase with an increase
in the past been assessed to be the most important in the WOB setpoint, the MSE does not decrease. A
drilling parameter affecting the ROP. The MSE varies possible explanation to this is that increased friction
between 10 and 40 MPa, and was calculated by hard- between the non-rotating BHA and the wellbore in the
coding a constant torque value equal to 1 Nm. From inclined section leads to a lower ROP (approximately
the figure, it can be observed that even with high oscil- 1 m/hr) than expected. Horizontal build is calculated
lations in the downhole sensor measurements, approx- from measurements to approximately 100mm, with an
imately 7.5 mm horizontal build is achieved, with an offset on the azimuth of approximately 3 to 4 mm (oc-
offset on the azimuth of approximately 1 mm. In 6 curs at the end of the inclined section). Upon inspec-
below, a cross-section of the first deviation well that tion of the well, approximately 45 mm horizontal build
was drilled successfully is shown after a waterjet cut- was achieved. Possible explanations of both the MSE,
ting machine was used to split the rock sample. over-estimated horizontal build and only a moderately
From inspection of the rock, approximately 55 mm high ROP can be that; either the increased wellbore
horizontal build was achieved over 490 mm TVD. In friction reduces the overall efficiency of the drilling op-
the last 150 mm of drilling, the knuckle joint turned eration and that constant vibrations against the well-
slightly due to a twist in the connection between the bore influence the measurements for inclination calcu-
knuckle joint and sensor sub, causing an azimuth off- lation. Another possible explanation can be that only
set of approximately 10 mm. From evaluating the well half of the BHA is submerged inside of the well, leaving
integrity, it is probable that bit whirl has occurred, the BHA even more unconfined, which in turn affects
possibly caused by the undergauge BHA. It is however the vibration cycle of the entire assembly.

7
 Modeling, Identification and Control

Figure 5: Experiment 1: Well log representing rig performance when drilling with downhole motor in inclined
section using a 7 degrees whipstock.

that contains best practice drilling parameters for the

different formation types, so that when a new forma-
tion type gets confirmed, the optimal drilling param-
eters can immediately get selected. The rig can then
make use of various search-algorithms in the reduced
state-space to identify the point at which the Founder
point of the system is found. This point describes the
optimal performance point when drilling with optimal
conditions related to WOB and RPM, and the point is
a linear function between these two parameters so that
when one of the two increases or decreases, equipment
might get damaged. Once the Founder point has been
located, for instance by gradient descent method , a se-
Figure 6: Well profile using 7 degrees whipstock in ho- ries of drilling incident models can be used to not only
mogeneous cement. classify the occurrence of a drilling incident, but also if
possible, predict that an incident is imminent (unless
measures get taken).
8. Conclusions While all drilling operations are unlikely to get au-
tomated in the near future, in our opinion the first
The resulting control system is capable of executing a step on the path is to provide the driller with powerful
series of control algorithms and models in parallel, such tools to increase safety, operational efficiency and re-
as the directional drilling program, the ROP search (by duce costs. It is therefore our recommendation to build
gradient descent method), inclination steering, down- a control system that promotes human-machine inte-
hole vibrations, incident and stuck pipe detection (and gration, where the machine for instance can suggest
handling), and so on. This is possible through the use optimal strategies going forward, and the driller can
of voting system that prioritizes pending commands select one of these or simply override the suggestion.
from the control algorithms depending on which ac- Furthermore, ensuring high quality of data in all steps
tion is most critical to execute. While tuning off the of the decision making process will increase reliability
PID controller and gradient descent method is diffi- and ensure consistently high performance.
cult with a bendable BHA developed to be used in
inclined well sections, the proof of concept has been
obtained by the algorithms selecting optimal drilling Acknowledgments
setpoints based on measured rig performance and re-
sponse to earlier changes executed. In terms of con- Thanks go to NORCE for their support in helping to
trol algorithms, the suggested practice for a complete develop the rig and the agent. We also thank all the
closed-loop system is that a database gets developed past and present members of the Drillbotics team at

8
 Løken et al., “Automated Drilling”

Figure 7: Experiment 2: Well log representing rig performance when drilling with downhole motor in vertical
pilot hole section for 170 mmTVD to later RIH with whipstock. Riser is used to ensure verticality.

Figure 8: Experiment 3: Well log representing rig performance when RIH to 240 mmTVD and then drilling
with downhole motor to 600 mmTVD. A 7 degrees whipstock is used to kick off.

the University of Stavanger. We acknowledge Equinor tecture and API Integration. Bachelor thesis, Uni-
academia program and Aker BP for funding the ongo- versity of Stavanger, 2019.
ing project.
Hagen, H., Jakobsen, A., and Khadisov, M. Laboratory
Drilling Rig Construction, Testing and Modeling for
References Optimization and Problem Management. Bachelor
thesis, University of Stavanger, 2018.
Bello, e. a. Application Of Artificial Intelligence Meth-
ods In Drilling System Design And Operations: A Hjelm, O. A. and Nilsen, S. J. Further Develop-
Review Of The State Of The Art. Journal of Artifi- ment and Testing of an Autonomous Drilling Rig
cial Intelligence and Soft Computing Research, 5(2), and Control Algorithms for Improved Drilling Per-
2015. pages 121–139. doi:10.1515/jaiscr-2015-0024. formance. Master’s thesis, University of Stavanger,
Geekiyanage, S. C. H., Loeken, E. A., and Sui, D. 2018.
Draft: Architectures and Algorithms of an Au-
tonomous Small-Scale Drilling Agent. Submitted for Løken, E. A. and Trulsen, A. Construction, Design
review, Elsevier, 2019. and Optimization of an Autonomous Laboratory-
Scale Drilling Rig. Bachelor thesis, University of
Guggedal, C. and Steinstø, M. Control System Archi- Stavanger, 2017.

9
 Modeling, Identification and Control

Figure A.1: Implemented control system illustrating layers and dataflow internally and externally with PLCs
and sensor equipment.

Sand, E. L. Design and implementation of a control

system for a fully automated drilling rig. Bachelor
thesis, University of Stavanger, 2018.

A. Appendix

10
 Løken et al., “Automated Drilling”

Figure A.2: GUI is equipped with gauges and parameters showing the control system setpoints and sensor
feedback as well as building a well log in real-time for interpretation of all stages of the drilling
operation.

11