amid the incessant rhetoric over the energy transition, operators continue to wrangle
tricky wells and exploit the latest in digital wizardry to beef up drilling performance
in ostensibly straightforward wellbores.
whether it's effectively drilling extended-reach wells with all the inherent
instability and lost circulation issues or applying digital technologies to
deliver consistency across all rigs within an operator's asset, the outside
noise takes a back seat to maximizing the value of wells throughout their
productive life. At least, that was the message delivered last November at
IADC's Drilling Engineering Committee (DEC) technology forum, which pointed out
how "engineers are increasingly focusing on designing and
delivering wells to maximize the return on investment, while dealing with a
more stringent regulatory environment."
comparison of two widely diverse drilling projects highlighted that
“Walking a fine
successful drilling of a shallow, extended-reach gas well offshore Malaysia
exemplified the balancing act that operators must often accept when trying to
optimize drilling parameters, while corralling the cornucopia of high-angle
"These wells are notoriously
challenging, requiring ECD management, hole cleaning and the prevention and
management of lost circulation. We had to walk a fine line on balancing the
risks between wellbore instability and lost circulation," Michael Yao, a
rock mechanics advisor for operator Hess, told the hybrid forum. "All the
challenges were lumped together, particularly the pore pressure versus the
collapse pressure. Highly deviated horizontal wells also tend to have more
breakouts and require higher mud weights."
Pre-drilling planning began with an
evaluation of two offset wells with similar trajectories, both drilled with an
11.4 lb/gal mud weight. The first well had been drilled trouble-free at an
average ROP of 100 ft/hr, while the second was drilled at 150 ft/hr, resulting
in lost circulation, with a 13.7-to-13.8 lb/gal ECD overwhelming the equivalent
static density (ESD) of 11.8 lb/gal.
The targeted well was programmed for
an 8½-in. hole, with TD at 12,642 ft, MD (3,925 ft, TVD), and 3,300 ft of open
hole. The 9⅝-in. casing was to be landed at 9,282 ft, with a 78o
inclination. Before reaching reservoir sand in the lateral, the drilling team
would have to maintain the stability of roughly 1,000 ft of pressured and
heated shale that would be penetrated at high angle.
Priority was placed on optimizing
both mud weight and the drilling parameters, to manage ECD and hole cleaning. To
prevent losses in the weaker zone, well designers programmed Hess's own formulation
of the widely used stress cage wellbore-strengthening methodology that provides
pressure containment by treating fragile areas with drilling fluids with
engineered, particulate lost circulation materials (LCM). It became clear that
to successfully land the well as programmed, however, some trade-offs would
have to be considered.
"How much breakout can be tolerated,
and how can we minimize ECD with optimal ROP and flowrate, without compromising
hole cleaning? How much wellbore strengthening can we achieve to push the
fracture gradient higher, and which side of the risk do we want to take? If you
don't have a choice, you have to decide if you'd rather deal with wellbore
collapse or lost circulation," Yao said.
To reconcile the ECD
requirement on lost circulation, Hess selected a 10.8 lb/gal surface mud weight
for the 8½-in. hole. The final drilling parameters called for the well to be
drilled at 100 ft/hr, but this would drop to 50 ft/hr upon implementing the stress
"We would use
Hess's stress cage formulation to prevent losses, in case the low-side pore
pressure fracture gradient is encountered. Whether to strip out the stress cage
would be determined, based on results from real-time pressure measurements,
once the competent sand is exposed," he said.
The well was drilled to
TD uneventfully, under controlled drilling parameters, with no losses or
wellbore stability issues. While the stress cage was implemented as planned, lower-than-expected
ECD values later rendered it unnecessary.
uniformity. In a far less extreme drilling environment,
ConocoPhillips looked for a way to develop uniform rig performance in the
vertical sections within its South Texas Eagle Ford asset, which takes in the
fragile Wilcox sands. "The vertical section, historically, is where we've
had some really high-performers, but we just didn't have uniformity in ROP
across all our rigs," said Eric Muller, drilling engineer turned data
analyst. "We wanted to push the threshold as far as we can, without
breaking down the Wilcox."
To that end, the company turned to
machine learning and advanced modeling techniques, with the aim of optimizing
ROP without incurring BHA damage. "We wanted our data scientists to look
at all our ERD and formation data to see if there was something our high-
performing rigs were doing that could be used throughout," he said.
Using big data to first look at
historical drilling performance, the effort began with a dataset incorporating
260 wells with more than 8.7 million data points, later filtered down to 116
wells and 138,000 data points. The process evolved into identifying the factors
that could be controlled—primarily differential pressure—followed by flowrate
and weight on bit (WOB).
The eventual model concluded that
maintaining a minimum differential pressure of 850 psi would yield an
incremental 31 ft/hr, which would require the motors and BHA to be modified
accordingly. A minimal flowrate of 600 gpm and a minimal WOB of 25,000 lb (30,000
lb maximum) were also predicted to increase ROP by 10 ft/hr and 4 ft/hr,
The guidance developed through the data-driven
machine learning model accurately predicted ROP in vertical sections, with a
mean absolute percentage error of around 12.5%. After rollout across the fleet,
the operator has seen a consistent 26% increase in ROP in the vertical sections,
without incurring any BHA damage. "Almost overnight, we saw the
performance of our rigs skyrocket. We set like nine vertical records in a row
in the first three weeks," he said. "It was more of a cultural shift
and (created) competition between crews."
ConcoPhillips is now developing
similar models for the curve and lateral sections, with a vision to use
artificial intelligence (AI) and machine learning to eventually create closed
loop drilling and completions. WO
JIMREDDEN@SBCGLOBAL.NET / JIM REDDEN, a Houston-based consultant and a journalism graduate of Marshall University, has more than 40 years of experience as a writer, editor and corporate communicator, primarily on the upstream oil and gas industry.