Professional Documents
Culture Documents
Samba BA, Maja Ignova, Kate Mantle, Adrien Chassard, Tao Yu, Sylvain Chambon, Ziad Akkaoui, Lu Jiang,
Richard Harmer, Olivia Barcelata, Jinsoo Kim, and Mustapha Rhazaf, Schlumberger
This paper was prepared for presentation at the SPE Middle East Oil & Gas Show and Conference, Manama, Bahrain, 28 November – 1 December, 2021. The event
was cancelled. The official proceedings were published online on 15 December, 2021.
This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents
of the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect
any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written
consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may
not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.
Abstract
Today, directional drilling is considered a mix between art and science only performed by experts in the
field. In this paper, we present an autonomous directional drilling framework using an industry 4.0 platform
that is built on intelligent planning and execution capabilities and is supported by surface and downhole
automation technologies to achieve consistently performing directional drilling operations accessible for
easy remote operations.
Intelligent planning builds on standard planning activities that are needed for directional drilling
applications and advances them with rich data pipelines that feed predictive and prescriptive machine-
learning (ML) models; this enables more accurate BHA tendencies, operating parameters, and trajectory
plans that ultimately reduce executional risk and uncertainty.
Intelligent execution provides technologies that facilitate decision-making activities, whether they be
from the wellsite or town, by leveraging the digital-drilling program that is generated from the intelligent
planning activities. The program connects planning expectations, real-time execution data from the surface
and downhole equipment, and generates insights from data analytics, physics-based simulations, and offset
analysis to achieve consistent directional drilling performance that is transparent to all stakeholders.
This new framework enables a self-steering BHA for directional drilling operations. The workflow
involves an automated evaluation of the current bit position with respect to the initial plan, automated
evaluation of the maximum dogleg capability of the BHA, and the capability to examine the health of the
BHA tools and, if needed, an automated re-planning of an optimized working plan. This is accomplished
on a system level with interdependencies on the different elements that make up the complete workflow.
This new autonomous directional drilling framework will minimize operational risk and cost-per-foot
drilled; maximize performance, procedural adherence, and establish consistent results across fields, rigs,
and trajectories while enabling modern remote operations.
Introduction
Today, the oil and gas industry, and more specifically well construction, is tackling unprecedented challenges
to come up with a more sustainable operational model while minimizing drilling costs, risks, and operational
2 SPE-204607-MS
uncertainties. Technology development in the past enabled us to address industry-shattering challenges and
will prove to be the key to tackle these challenges today. For well construction to survive, investing and
embracing industry 4.0 capabilities that have proven to be indispensable in other industries outside oil &
gas is a must.
Directional drilling is a major activity within well construction. Due to the complexity of the wells being
drilled today, it poses great risk and reward scenarios for proper planning and execution. Yet, directional
(Figure 1). The pieces of this ecosystem need to be instrumented, automated, connected, and integrated.
Data generated at each step must be recorded, curated, and analyzed in order to draw correlations fast and
to learn new process adjustments in a continuous way.
Like self-driving cars, autonomous directional drilling capabilities aim at achieving more consistent
results, in terms of well trajectory quality, reduction of tortuosity, optimization of time to reach the target,
and achieving a competent wellbore. By enabling remote monitoring and control workflows, personnel are
less needed at the wellsite, reducing their exposure to risk zones and simplifying logistical aspects. From
the office, experienced staff can supervise a wider range of operations and consult with experts, while less
experienced employees are assisted to make safe decisions. This reduces the overall carbon footprint while
promoting sustainability in well construction.
To connect the pillars, it is crucial to have an integrated data architecture that unifies all of them and
ensures complete harmony and cohesion across the various workflows that would exist underneath them.
Starting in the planning phase, all associated digital data is passed onto the intelligent execution pillar
to drive both surface and downhole automation. Once the execution phase is completed, we leverage the
same data pipelines to close a feedback loop that would drive further refinements of the ML models in
the intelligent planning phase to start the cycle all over again. All technology blocks within the pillars will
naturally evolve as part of our agile development process, and as such, will continuously improve with time
as we progress toward our ultimate goal of complete autonomy. The intelligent execution is at the center of
the system as it interacts with each of the components and manages the information sharing between them.
In the next chapter, we will provide detail information regarding major capabilities within the pillars and
how combine in the directional-drilling workflow (Figure 3).
SPE-204607-MS 5
Figure 4—Illustration of the hybrid model that combines both the physics-based model and machine learning.
The industry 4.0 platform revealed in the first chapter enables the implementation of this hybrid model
directly inside the new well construction planning portion of the platform. With the automated engineering
analysis system and data analytics pipeline, the engineering design will be automatically validated by
comparing the predicted tool steering performance versus the trajectory plan with no human intervention
required. And with the model, the BHA, trajectory, and operating parameters can be optimized for the
application and delivered digitally for wellsite execution.
Figure 5 depicts an example of a plan derived for a particular run, taking into consideration the approved
initial trajectory, the BHA capability, and the operating parameters using the hybrid model. As mentioned
above, an important part of planning for execution is to identify the relevant offset wells to consider for the
subject well to be drilled. During the well execution, the ML algorithm will learn from both these relevant
offset wells and from the current subject well that is being drilled.
6 SPE-204607-MS
If we define p as the number of the relevant offset wells, we need to define for the ML scheme how
much importance or weight do we assign to these offset wells in comparison to the weight of the subject
well. Noting Oi the weight for relevant offset well i, and if we assume for each offset well i, Oi=w (uniform
weight for all p offset wells), noting C as the weight assigned for the current well, we will have the following
equation:
(Eq 1)
We can then define , the ratio between the weight of the current well and the weight of an offset
well. We have the following property:
– If τ > 1, current well has more weight than 1 offset well.
– If τ < 1, current well has less weight than 1 offset well.
This gives an opportunity to balance out the importance given to the current well compared to offset
wells based on its data quality, the level of trust on the current well, and how much experience has been
gathered on historical datasets. The more wells drilled, the better the understanding of steering responses
will be and the more accurate the model will be.
The results of the planning workflow are then combined in a digital-drilling program (DDP) and used by
the intelligent execution applications to deliver the needed workflows.
Intelligent Execution
Execution focuses on the analytical activities for directionally steering a well. The intelligent execution
supplies insights regarding trajectory delivery by leveraging descriptive, diagnostic, predictive and
prescriptive analytics. In essence the advisor has complete awareness of the current state of the BHA,
understands how the BHA has performed, and can continuously project its future performance and how to
affect that performance.
This high-level workflow can be generically applied to mud motors, rotary steerable systems (RSS) and
at-bit steerable systems (ABSS)—regardless of tool or hole size. The complexity comes with integrating
the specificity inside each step to ensure optimal results at the end of the workflow. Specificities such as:
SPE-204607-MS 7
• information from planning with offset analysis and adapting the hybrid model for real-time
execution (as mentioned in the previous section)
• various downhole automation possibilities at a given time to optimize the recommended
working trajectory execution, taking maximum advantage of the tool capabilities and minimizing
unnecessary surface actions
• real-time data and derived tool health as well as tool state estimation at a given point in time
• allowable deviation from the original plan, both in terms of position and angular deviation
• recommended constraints by automatic plan analysis that are adjustable by the users, based on
their preferences
• allowable tortuosity, risk measures, hole quality, confidence level, etc.
At the end of the second step, one can have either a single best candidate or few best candidates that
could be exposed to either a user or a ML scheme, depending on the level of automation.
Assuming the trajectory generator would give all possible types of trajectories or paths that one could
think of, we would concentrate our description here on the ranking system portion. The system is based
on a list of classification items that define features selected in order to rank the candidates. These features
are called candidate properties. Example of these properties include: trajectory length, ROP, total steering
length, toolface (TF) orientation, maximum steering ratio, average deviation from the plan, risk level, target
constraints, angular deviation, tortuosity, directional drilling index (DDI) (Alistair W. Oag 2000), hole
quality, tool wear, number of downlinks, geomechanics, confidence level, and even production level index
(Thormod E. Johansen 2016). For each property, we associate a defined weight (Wi) based on the trajectory
context. This weight is associated with the cost function of the candidate to produce the total cost for each
candidate trajectory as:
(Eq 2)
SPE-204607-MS 9
Note that by choosing this ranking system, we have a methodology that is quite modular and extensible.
It is easy to add additional candidate properties as long as one can give a mathematical description of it. For
example, using a ML algorithm, we can use historical data to map how easy or difficult it was to drill a given
path and define a drilling difficulty index from the surface automation, adding it to the above equation (Eq
2). Once the optimal path has been selected and the steering commands to best achieve this path are known,
there is an additional optimization problem defining the best link with surface automation to efficiently
Surface Automation
The power of the surface automation has been discussed extensively in the paper (Brennan Goodkey 2020).
In particular, it was demonstrated how a drilling automation system is used to deliver efficient, consistent,
and safe drilling operations. In the context of ADD, the surface automation connects the intelligent execution
with the rig control system, and as such, represents the executing arm of the autonomous directional drilling.
It consists of a digital acquisition system (DAS), an automated downlinking and an automated energy
management system that connect to the rig control system. In our current journey, with the DAS largely
in place, the focus is put into the last two components and the connection between those components and
the intelligent execution.
Depending on the level of automation presented in Figure 2, different surface automation schemes could
be used. In cases where we have access to the rig interface and can independently operate the rig control
system, this will be used to automatically execute the directional drilling commands, be it in terms of
steering commands, downlink to the downhole tools, or the execution of the selected operating parameters
for optimized operations. In cases where the interface to the rig control system is not available, an automated
downlinker is currently being developed. This will consist of a hardware system installed at the edge, which
will provide direct access to the flow line for automated downlinking with very little disturbance of the
rest of the drilling process. Thus, intelligent execution will have direct access to send commands to the
downhole automation system.
Downhole Automation
Intelligent execution will control steering commands, whether it is for motor steering, RSS steering, or
ABSS steering. There is, however, an important additional control layer when running an RSS or ABSS
tools. It is the availability of direct downhole trajectory automation. Downhole trajectory automation refers
to a series of closed-loop control modes for drilling the different trajectory types in a well, see Figure 8.
Closed-loop control modes are managed by intelligent execution, which defines when to enter or exit certain
modes, or simply runs the downhole tool in manual mode. When a link is available, surface automation
helps intelligent execution control settings of the manual mode.
10 SPE-204607-MS
• borehole quality
• operations efficiency
• compensation (demanning)
• operations integrity
In terms of the currently available downhole automation control modes, we note four main modes:
• HIA is a closed-loop hold control of inclination and azimuth for tangents and laterals
• curve control mode is the latest addition to these close-loop modes and controls curves, be they
2D or 3D.
Figure 9 depicts a typical downhole closed-loop controller. Vertical drilling RSS, IH, and HIA modes
have evolved as mature projects to automatically maintain a given attitude. Until 2020, the curve portion of
the well was the only substantial part of the well that did not have a closed-loop automated control. It was
deemed too difficult given the unavailability of reliable downhole depth measurements and other challenges.
This has now been addressed with the addition of a curve control mode, which has been successfully field
tested and will be ready for commercialization in 2021. The curve control mode enables the RSS or ABSS
tool to autoadjust its steering with a given DLS (°/100 ft) and TF (°) to maintain a smooth curve with steady
DLS.
SPE-204607-MS 11
• Recommendations evaluation.
The results for the different parameters of state estimation and slide detection show a good accuracy of
the models evaluating the different parameters in between surveys (Figure 10). As revealed before, this
accuracy is crucial to generating a valid recommendation
All these results have been computed without continuous direction and inclination (D&I) information
available during the run. When adding continuous D&I information in the computation of state parameters,
bit position, and steering capacity, the accuracy is increased by about 20% on average, and the impact is
more important on the wells with the highest estimation errors.
Diving into the predictive and prescriptive functionality, we look at the 551 recommendations generated
by the directional drilling advisor for these 10 wells (Figure 11). The results showed that more than 95%
Figure 12—Example of a Motor run with slide/rotate sequence recommendation in the curve
To facilitate transparency within the directional drilling workflow, we derive a metric called steering
efficacy (Figure 13) that can be used in numerous ways to support all stakeholders. It provides information
on how effective each steering sequence was relative to the recommendation and reveals a clear and concise
steering history.
SPE-204607-MS 13
The field test showed that ROP was faster than in the offset wells during all curve-control test runs.
And in some cases, ROP was 60% faster and the trajectory was controlled as required with significantly
fewer surface interventions (number of downlinks) than conventional manual methods for drilling the curve.
Between 50-75% fewer downlinks were needed compared to previous offset wells. The new mode proved
easy to use and received positive feedback from directional drillers. Phase 2 of field testing is currently being
performed and the range of jobs has expanded with the whole curve routinely being drilled to landing point
(LP) where HIA mode is engaged. We have tested in larger hole sizes up to in with DLS requirement
as low as 2° /100ft.
14 SPE-204607-MS
Figure 15 depicts an example of a constant curvature drilled with the downhole curve control mode with
a target DLS = 7°/100 ft and a TF =36°. These target values of DLS and TF are converted to matching
target attitudes (inclination and azimuth); these values are shown in both time (Figure 15) and depth (Figure
16) as blue triangles. The continuous inclination and azimuth responses are superimposed on both figures
as green circles.
Figure 16—3D downhole curvature control plotted in depth domain. These were real-time data points sent to
the surface. Target attitudes are shown as blue triangles, while the actual attitudes are shown as green circles
Figure 15 presents the data dump as recorded in the tool, and Figure 16 presents the real-time surface data
as transmitted to the surface and converted to depth-domain. We can see that the downhole curve control
SPE-204607-MS 15
kept both expected inclination and azimuth within a very close range of the curve requirement. This resulted
in a very smooth curve being drilled with minimum surface intervention.
• missing intelligent planning would result in a poor learning with untapped design improvements,
lack of anticipation, and slow learning process
• missing intelligent execution would lead to developing ignorant system, which would omit
potentially important part of the system, resulting in suboptimal decisions
• missing surface automation would impact consistency with low procedural adherence and high
dependency on experience
• missing downhole automation would result in a very slow loop control with suboptimal trajectories
and too many interventions.
Conclusion
Directional drilling, well construction, and the broader oil & gas industry are being challenged from
multiple fronts. To respond to challenges we face today, we must develop technology that leverages industry
4.0 elements. Autonomous directional drilling is our answer to a more sustainable operational model,
minimizing drilling costs, risks, and operational uncertainty while providing consistent and optimized
performance, regardless of the field, rig, or trajectory. Intelligent planning provides valuable dynamic
insights related to BHA tendency and trajectory optimization that increase our confidence in the execution
phase. Intelligent execution, and more specifically the directional drilling advisor, provides descriptive,
diagnostic, predictive, and prescriptive analytics related to the steering process. This enhances our crews’
abilities to execute the jobs today, either from town or the wellsite, and enables us to advance to a future
where steering autonomy can be realized. Downhole closed-loop automation advancements, such as the
curve control mode, further enables improved steering response out of the BHA and minimizes the need
for surface intervention.
We have been active in deploying these building blocks in real-life applications across the world
and are encouraged by the value these technologies are bringing to directional drilling operations. The
directional drilling advisor has shown to be 95% accurate with the steering recommendations on real-life
16 SPE-204607-MS
wells. Downhole closed-loop steering for curves has been successful on various applications complexities,
independent of hole size and steering technology, and produced performance gains up to 60% while
minimizing surface intervention.
Acknowledgement
The authors would like to thank Geoff Downton, Magnus A Hedlund, Alexandre Jourde, Ling Li, and Jason
Bibliography
Alistair W. Oag, Mike Williams. 2000. "The Directional Difficulty Index - A New Approach to Performance
Benchmarking." SPE SPE59196.
Brennan Goodkey, Gerardo Hernandez, Andres Nunez, Mauricio Corona, Kamal Atriby, Mohammed Rayes, Rafael
Carvalho, Carlos Herrera. 2020. "Drilling in the Digital Age: Harnessing Intelligent Automation to Deliver Superior
Well Construction Performance in Major Middle Eastern Gas Field." SPE SPE203251.
Downton, G. C. 2005. "Directional Drilling System Response and Stability." IEEE.
Hendrik Suryadi, Tao Yu, Yuelin Shen, Jack Yang, Robin Chen, Wei Chen, Timur Khusaenov. 2019. "Development of
Automatic BHA Directional Tendency Prediction Based on Drill-Ahead Modeling." SPE SPE198350.
Thormod E. Johansen, Donald G. Hender, Lesley A. James. 2016. "Productivity Index for Arbitrary Well Trajectories in
Laterally Isotropic, Spatially Anisotropic Porous Media." SPE SPE184408.