IADC/SPE-199556-MS

Directional Drilling Automation: Human Factors and Automated

Decision-Making

Bill Chmela, Helmerich & Payne Technologies; Sarah Kern, Helmerich & Payne IDC; Tyler Quarles, Sashmit
Bhaduri, Randy Goll, and Collin Van, Helmerich & Payne Technologies

Copyright 2020, IADC/SPE International Drilling Conference and Exhibition

This paper was prepared for presentation at the IADC/SPE International Drilling Conference and Exhibition held in Galveston, Texas, 3–5 March 2020.

This paper was selected for presentation by an IADC/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 International Association of Drilling Contractors or the Society of Petroleum Engineers and are subject to correction
by the author(s). The material does not necessarily reflect any position of the International Association of Drilling Contractors or 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 International Association of Drilling
Contractors or 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 IADC/SPE copyright.

Abstract
Inconsistent directional drilling performance has cost the oil and gas industry billions of dollars in drilling
costs, missed production potential, and increased lifting costs. While some directional drillers perform at a
high-level, others often fail to properly compensate for multiple variables seen while drilling. Automation of
the directional drilling service including automated decision-making is proving to be a viable solution to this
problem and has been implemented in the drilling of thousands of wells in North America in recent years.
A joint industry project (JIP) affiliated with the IADC (International Association of Drilling Contractors),
SPE DSATS (Society of Petroleum Engineers Drilling Systems Automation Technical Section), AUVSI
(Association for Unmanned Vehicle Systems International), Southwest Research Institute, Energistics, and
the OPC (Open Platform Communications) Foundation has created a Drilling Systems Automation (DSA)
Roadmap to help the industry understand the direction of drilling systems automation. In this roadmap, the
authors suggest that the transition from humans to automation in the general drilling space can occur across
four cognitive functions: acquiring information, analyzing and displaying information, deciding action, and
implementing action. They also suggest that there is value in partial automation.
This assessment is accurate when applied to directional drilling; over the past few years many individual
directional drilling tasks have become either fully automated or partially automated, each bringing
significant value to the operation. These individual automated tasks systematically link together, moving
toward the fully automated rig.
Continuous real-time updating of the bit position is one of the many critical analysis steps that can be
automated, driving steering decisions. The practice of accurately predicting bit position and trajectory is
a skill that is artfully developed by expert directional drillers over many years. To demonstrate the value
of partial automation, a specific example is presented where the continuous calculation of the real-time bit
position and its trajectory is fully automated, showing that the analysis and decision-making performed by
the automated system is faster and more accurate than performed by human directional drillers. Although
the details of only a single automated task is the focus of the case study, it is important to realize that it is
just one of many automated tasks currently implemented in the field on the path to full automation.
2 IADC/SPE-199556-MS

In the transition from human to automated processes, roles and responsibilities must change both on the
rig and in the office in order to fully benefit from its value potential. With most of the heavy cognitive lifting
performed by a machine, a single directional driller can now work remotely and manage the directional
control of multiple wells simultaneously. This remote directional driler can perform data analysis in a
structured scientific manner. Automation incorporates the science, integrates previously siloed best practices
and individual knowledge, and allows for continual consideration of the economic consequences to the
asset from each decision.

Directional Drilling: Human to Automated

One of the most important tasks of the directional driller has historically been to decide when to rotate the
pipe from the surface in an attempt to drill relatively straight, and when to stop rotation, point a bent-sub in
the direction he would like to steer toward, and determine how long to drill using only the downhole mud-
motor (sliding); and then implementing that decision on the rig.
While some directional drillers perform this task at a high-level, others often fail to properly compensate
for multiple variables including variations in rotary walk and build, motor yield variations, tortuosity risks,
target uncertainty, and deflections. Even expert directional drillers rarely account for reduced hydrocarbon
production potential related to drilling accuracy on a survey by survey basis. In an attempt to stay on a
defined path defined by geoscientists and reservoir engineers, there is simply too much data for directional
drillers to analyze in real-time to consistently make good decisions.
Automation of the directional drilling process, including this automated decision-making, is proving to
be a viable solution and has been successfully implemented in the drilling of thousands of wells in North
America in recent years. The transition to automated directional drilling is complicated by the fact that
directional drilling has historically been understood as more of an art form than a science. Tasks that are
better done by machines, and that can be automated, need to be identified as the first steps toward full
automation.
A Drilling Systems Automation (DSA) Roadmap has been created to help the industry understand the
direction of drilling systems automation. In this roadmap, the authors suggest that the transition from humans
("art") to automation ("science") in the general drilling space can occur across each of the four cognitive
functions below and further suggest that there is value in partial automation.
This assessment is accurate; many individual directional drilling tasks have become either fully
automated or partially automated, each bringing significant value to the operation. Further, these individual
automated tasks systematically link together, moving toward the fully automated rig.
The following individual tasks have been automated in a bit guidance system that operates across the
four cognitive functions detailed in the DSA roadmap.
1. Acquire Information
a. Automated data gathering from EDR (Electronic Drilling Recorder) via WITS (Wellsite
Information Transfer Standard) Level 0 and downhole sensors
b. Automated Survey Corrections (IFR (In-Field Referencing) and multi-station analysis)
2. Analyze and Display Information
a. Automated determination of rig activity
b. Automated computation of BHA (Bottom Hole Assembly) tendencies, motor performance, and
motor potential
c. Automated GR (Gamma Ray) correlation (FTD (Formation Top Detection))
d. Automated integration with geosteering
e. Automated accurate and truthful reporting including performance indicators
IADC/SPE-199556-MS 3

f. Automated real-time visual data displays available to all well stakeholders

g. Automated determination of bit position and trajectory utilizing BHA tendencies, motor
performance, rig activity and other real-time data
3. Decide Action
a. Automated Slide/Rotate decision-making, with millions of potential options taken into account,
and based on costs associated with drilling time, tortuosity risk, and lost production potential
4. Implement Action
a. More recently automated sliding technology has been introduced to the field that executes return
to bottom, oscillation, and toolface orientation control automatically during slide sequences
b. Automated autodriller and stick slip mitigation applications automatically optimize rotate
sequences
Although the tasks across the full process (listed above and captured in the workflow in Figure 1) have
been automated, the next section of this paper focuses on automation of just one of the analysis steps as an
example of how automating individual tasks brings value.

Figure 1—Directional Drilling Process Flow

Automated Analysis and Decision-Making: A Case Study

A universal standard procedure for continuously estimating the position of the bit and its trajectory relative
to the well plan has never been defined in the industry. However, it is important that this estimated subsurface
bit position is determined as accurately as possible in real-time because it is a key input for making real-
time steering decisions (precisely when to initiate and complete slide sequences and at what orientation).
As input to performing the bit position estimate, the directional driller receives a series of measurements
that contain the inclination, the azimuth, and measured depth of the downhole sensors that reside in the BHA.
In the majority of US shale plays, standard directional sensors are located approximately 60 ft behind the bit.
Predicting the bit's position and trajectory via these measurements is complicated and is further compounded
by the uncertainties and limitations associated with the measurement tool. Furthermore, the directional
driller typically receives survey data from this sensor only while making a connection, approximately once
every 90 feet of drilling. From this sparse collection of data points, the directional driller estimates the
position of the path the bit has taken and the position of the bit ahead of the last survey using a mathematical
projection. The directional driller must then continue to estimate the bit position while drilling until the
next survey is taken. Figure 2 shows these typical survey locations relative the current estimated position
4 IADC/SPE-199556-MS

of the bit. The position of the bit needs to be estimated over large distances where no certified positioning
measurements are taken.

Figure 2—Typical Survey Locations

In an example from South Texas, the directional driller is building a curve to land at a precise point
defined by geologists and reservoir engineers to maximize the production of the well (Figure 3). Landing
long or a "soft landing" (further away from the well than planned) will reduce the ultimate production of
the well, while landing short will require a large dogleg to correct, reducing the ultimate production or
potentially require a plug-back and sidetrack.

Figure 3—Landing Position Comparison

The directional driller performs his calculations based on the most recent survey station and estimates
that the bit is currently drilling at an inclination of 29 degrees. He determines that to land the well in the
precise location defined in the wellplan he should continue to slide for 44 feet and then rotate the rest of the
stand. The directional driller's estimation of the bit position was based on mathematical projections and the
assumption that that the driller's performance while sliding has been and will continue to be consistent.
Often, however, the slide performance of the driller is often not consistent. Maintaining a high toolface
precision while drilling through varied geological formations require a directional driller to have specific
local knowledge and experience. The lack of this can result in sections of both good and poor toolface
control while sliding.
Using the automated bit guidance tool, the data is analyzed in real-time and indicates that the driller
had relatively precise toolface control while drilling prior the last survey (72%) and poor toolface control
IADC/SPE-199556-MS 5

while sliding after the last survey (54%) (Figure 4). Toolface precision often varies because of numerous
variables, but it is known that a toolface precision greater than 70% delivers motor yields close to the full
motor potential, while toolface control at less than 50% delivers much lower motor yields.

Figure 4—Toolface Precision and Motor Yields

The aforementioned poor toolface control is similar to the response shown in Figure 5, with the green
spikes indicating toolface orientation. Note how the orientation is relatively consistent and aligned in the
shallower portion of the well and gets much more erratic at the depth identified by the arrow.

Figure 5—Erratic Toolface Orientation

In this case, the traditional calculations which estimated the bit inclination of 29 degrees were incorrect
because the assumption of consistent performance was violated; and the motor yield ahead of the last survey
was much lower.
Knowing this, is it possible to more precicely estimate the inclination and current position of the bit?
Since it is possible to determine and record the rig state (sliding or rotating) along with the real-time toolface
6 IADC/SPE-199556-MS

orientation data versus time and depth, there is enough information to replace a mathematical projection
based on assumptions with automated forward modeling using real data. The system utilizes the following:

• Automated rig state detection

• Toolface orientation

• Recent Dynamic Motor Yield & Motor Potential while sliding

• Recent BHA (Bottom Hole Assembly) tendencies while rotating

This fully automated forward modeling shows that the 29 degrees inclination projection that was
calculated via legacy methods by the human directional driller in this case was incorrect, and its actual value
was 25 degrees inclination. The higher accuracy of the bit position and trajectory by the bit guidance system
determined a longer slide interval to precisely land the curve (the automated bit guidance system determined
a 67 ft slide would be required. This is 25 ft longer than determined using a non-automated process).

Figure 6—Bit Inclination Inconsistencies

Introducing new technologies in the field, such as an automated bit guidance system, is complicated from
a change management perspective. In this case, should one base the slide interval decision on the completely
automated advice from a machine or should it be based on manual methods that have been in use for decades
and have served us relatively well?
The driller decided to split the difference between the two methods (manual and automated) and slide 57
feet. After drilling the next stand, a survey was acquired showing an inclination of 32.6 at the sensor and a
"build rate to land" was calculated at this point as 7.8 degrees per hundred feet. This required nearly 100%
sliding to land the curve. If the driller had performed the shorter slide as determined using legacy projection
methods, he would have landed short, which would have led to a sidetrack and additional costs.
In this case, the automated system estimated the position of the bit with more accuracy because it used
more of the available data along with automated forward modeling algorithms to utilize this data in real-
time. In the case study, the directional driller did not fully follow the algorithm, however he did increase
the footage slid. In this instance, the algorithm actually saved a bit trip.
Determination of bit position and trajectory was highlighted here as an example of just one of many tasks
that have been improved through automation.

Documenting Inconsistency and Improving Performance

In this example, the legacy method to predict bit position and its trajectory would have been acceptable
if the performance of the driller was consistent throughout the interval. In order to understand the relative
frequency of similar situations, the ability of drillers to hold precise toolface orientation was studied on a
small sample set of 10 wells in South Texas with no automation being used. A toolface precision of 77% was
measured in these wells, with a standard deviation of 20%. This variation in the ability to control toolface
IADC/SPE-199556-MS 7

suggests that postioning errors such as those we saw in the case study (and based on assumptions) occur
more often than one might expect.
An additional 230 wells in the area were then analyzed with automated feedback to the driller. By
providing real-time feedback to the driller, the precision improved from 77% to 83%, shown in Figure 7,
and the standard deviation decreased from 20% to 17%,. Reduction of the standard deviation improves the
ability of directional drillers to make better decsisions; and improving the precision ultimately allows for
better decsions, for lower bend motors to be used, and leading to lower tortuousity with reduced drilling
costs.

Figure 7—Toolface Orientation Precision

In this case study, it has been shown that automating a more accurate prediction of the bit position and its
trajectory, along with providing easily digestable feedback to the driller, improves performance, delivering
wells at lower cost and with higher production potential.

Reconfiguration of Roles
The bit guidance system automates across the four cognitive functions: acquiring data, analyzing data,
deciding actions and implementing actions; and can provide additional previously untapped value by
reconfiguring roles, responsibilities and workflows.
With most of the heavy cognitive lifting performed by a machine, a single directional driller can now work
remotely and manage the directional control of multiple wells simultaneously. Automation also provides
the directional driller time for futher analysis of data in a structured scientific manner.
Automated data-gathering and reporting allows further study of:

• Important bit wear, formation, and rheological effects

• Local effects caused by the bit/rock formation

• Individual differences in directional drilling/drilling skillsets

The proficiency accrued via in-field experience can automatically inform decisions made by the system
such that there is limited loss in competence due to shift or personnel changes. Thus, the knowledge and
data gained by drilling previous wells, or with a specific tool configuration or a specific basin can be used
to automatically improve future wells. Mistakes can be reduced by near instant transference of expeience
8 IADC/SPE-199556-MS

across wells or versions. Automated bit guidance can therefore function as risk-minimizing insurance
against reaching the full production potential of the well or in non-productive drilling time.
In addition, a remote operations team can perform further optimization by live monitoring and acting
as the voice of the customer by incorporating operational biases via bit guidance system configuration,
which is especially important when there is sparse information for a specific well or region. These teams
can hedge their skillsets and experience greatly augmented by software-based advanced drilling analytics,
combined with on the fly forward modeling and the recovery of unknown drilling parameters and latent
unobservable state via inverse modeling. A combination of these tools can be used to both extrapolate and
interpolate between gaps in knowledge and/or data, or to automatically correct common heuristics that may
fit some global tendencies but may not correctly explain local variation. All of this allows the automated
bit guidance to act as a force multiplier and change the economics of directional drilling, by eliminating the
need to keep a directional driller on site all the time.
A more scalable model can be built where the automated decision command and control performed by the
bit-guidance system can result in more consistent results with fewer in-field personnel. A remote operations
analyst has considerably fewer decisions per hour to perform during the course of each well due to reduced
cognitive load, allowing for a scaling of the number of rigs they can simultaneously supervise effectively.
Instead of focusing on the minutiae of second by second decisions that occur during sliding and rotating,
a remote operations analyst focuses on high leverage/high risk situations more apropos to their expertise.
The remote directional driller has more time to validate automated decisions based on his experience while
being supported by other experts working remotely in the same center.
With volatility in oil prices and rig counts, the traditional model's limiting effects in reducing the cost
of drilling are compounded as more experienced directional drillers leave the industry or retire upon every
business cycle. Instead of directional drilling becoming a lost art, all the inductive biases and knowledge
can instead be programmed into the system and optimization procedures can be used to adjust the decision
and control so over time, so more optimal decisions are made.

Conclusion
Over the past few years many individual directional drilling tasks have become either fully automated
or partially automated, each bringing significant value to the operation. These individual automated tasks
systematically link together, moving toward the fully automated rig.
This paper has demonstrated that partial automation of the directional drilling process and automated
feedback to the driller are both examples of where automation leads to better decisions, reduced drilling
and lifting costs, and improved production potential. In the case study, this is a result of higher accuracy
of both bit position projection and bit trajectory.
Automated decision-making is proving to be a viable solution to inconsistent directional drilling
performance and has been implemented in the drilling of thousands of wells in North America in recent
years. Automated bit guidance has been implemented not just in bit positioning, but across all four cognitive
functions: acquiring data, analyzing data, deciding actions and implementing actions.
Roles, responsibilities and processes must change both on the rig and within remote support centers to
capture the full value of automated decision-making. By integrating previously siloed best practices and
individual knowledge, automation also allows for continual consideration of the economic consequences to
the asset from each decision, while harnessing the tribal knowledge so common in the industry: a blended
mix of art and science.