I N NOVATI N G WH I LE DR I LLI N G
us reduce that setup time even further. We’re putting a lot of our
focus into making that process even faster.”
Insights from in-bit sensors
Baker Hughes’ bit drilling simulation software can create
digital twins of the bit and the target formation. It then uses
proprietary cutter force models to evaluate cutter and bit
body interactions of the bit and the rock . Operators’ specific
performance-limiting challenges can be modeled to find
customizable solutions.

already built and modify them to meet current needs. Because a
lot of the upfront work has already been done, the whole process
is streamlined,” he said.

Baker Hughes said it still runs field tests, but it finds that some
of its long-time clients now favor using digital twins in place of
field tests. This approach requires a higher level of trust between
customer and supplier, however. “We spend a lot of time calibrating
our digital capabilities with past field data and lab data, and when
you add that to having strong relationships with your customers,
that means we can really work together to streamline the process.”
In some cases, digital capabilities have been effective in help-
ing operators set performance records without the need for field
testing. In 2021, for example, a company in the Permian Basin had
reached a ceiling with its ROP while drilling the 12 ¼-in. inter-
mediate hole section. Using computer simulations, Baker Hughes
pinpointed cutters as the primary ROP limiter and then lab-
tested different cutters and their placement across the bit face.

Ultimately, the company recommended the Prism shaped-cutter
technology with its Dynamus drill bit because the simulations
showed that the point-loading on the Prism cutter would be ideal
for maximizing ROP in the ductile pressurized shale formations
of the hole section.

In the field, the Dynamus bit with Prism cutters drilled the
entire intermediate section from 1,405 ft to 5,200 ft in 16.3 hours,
resulting in an average ROP of 232.8 ft/hr. This was 37% above the
operator’s target and even set a record for the hole section, accord-
ing to Baker Hughes.

Mr Nelms said the next step with its digital twin modeling is
to use machine learning to automate the building of the digital
twin once the drill bit design and field data have been input into
the system.

“A lot of digitalization efforts are centered around manually
creating the models and building out the digital twins, but over
time we’ve built out a library of simulations,” he explained. “This
means we can have a faster process because a lot of the upfront
work has already been done and we can adjust and modify previ-
ous models to help build out new models. Automation can help
30 Halliburton had previously relied on data captured at the sur-
face to feed into its drill bit designs, but with the introduction of
its Cerebro Force in-bit sensors in 2020, the company says it can
now better understand and document the motions of the drill bit
from well to well.

Cerebro Force captures weight, torque and bending measure-
ments directly at the bit, allowing the company to map the down-
hole motion of the bit and any associated drilling dysfunctions.

This can prove valuable when it comes to improving future bit
designs. “Cerebro is all about understanding the subsurface conditions,”
said David Sostarich, Strategic Business Manager at Halliburton.

“Surface data is gathered thousands of feet away from the drill
bit, but in-bit sensing allows us to see what’s happening from a
downhole perspective. It gives us a clearer picture of what’s going
on at the bit-to-rock interface and provides us with better insights
into design. We’re looking at where lateral vibration, axial vibra-
tion, stick-slip and whirl are happening, and then marrying that
information with bit design, selection and implementation.”
In one Williston Basin project, data from the in-bit sensors
helped the company to design a Cerebro drill bit . Field test data
showed the bit achieved the fastest ROP, while minimizing torque,
at points when the rolling element engaged with the formation.

As the rollers engaged, the torque was minimized, thus increasing
ROP. However, the sensor data also showed that rollers were only
engaged for small percentages of each test run.

Halliburton then decreased the depth of cut at which the roll-
ers were engaging, a move that enabled the rollers to engage for
a greater percentage of the run and increase ROP. Whereas the
highest ROP prior to the change was 113.0 ft/hr, ROP post-change
maxed out at 160.2 ft/hr.

More recently, Halliburton added another tool to its bit-design
toolbox – Oculus, which is an automated software program that
serves as bit dull grading system. Launched in 2021, the soft-
ware uses machine learning algorithms to capture precise dull
information for individual cutters on a bit. The program analyzes
images taken of the bit after it has been pulled out of hole and
catalogues the data in a cloud database. It then automatically
grades the wear on each cutter.

By automating the dull grading process, the subjectivity of
visual examinations by human inspectors is eliminated, and
Halliburton gets more precise measurements and classifications
of bit dulling. It also becomes easier to utilize the dull analysis
over a large number of bits, including during the testing phase of
the design process, Mr Sostarich added.

“When we would evaluate drill bits previously, we would see
a lot of subjectivity, even utilizing the IADC code. One individual
might grade a bit a 1-2, another might grade it a 1-3 and another
might grade it a 2-3. What we wanted to do with Oculus is take a
digital image of each cutter, input it into the system and give it a
more specific and objective grade, so we can take these very sub-
jective things and turn them into objective data,” he said.

M A R C H/A P R I L 2023 • D R I L L I N G C O N T R AC T O R