American Oil and Gas Reporter - September 2024 - 41

SpecialReport: Hydraulic Frac Tech
Huge Uplift
The gains can be substantial. When
AIF's real-time optimization is applied
to a well that already would have strong
proppant distribution using today's techniques,
it increases first-year production
by at least 6%. For more problematic
wells that otherwise would have uneven
proppant distribution, the increase in firstyear
production tends to be around 20%.
To reach these estimates, Hess and
ResFrac conducted two modeling studies
comparing today's stages to near-perfect
ones. The first study involved a fully
coupled hydraulic fracture reservoir simulation
model, while the second looked
at a much simpler fracture-reservoir simulation
model to ensure the learnings
were not model-dependent.
The fully coupled model was calibrated
using a comprehensive dataset
from Hess' Observation Lateral project
in the Bakken. Conducted in 2019 and
2020, the project studied how various
completion designs influence fracture
geometry and drainage. It took place
on a six-well pad containing a 10,000foot
observation lateral through the
FIGURE 1
Fracture Geometries (top) and Drainage Patterns (bottom) from Fully Coupled Model
Three Forks formation that was equipped
with cemented pressure gauges and
fiber optics. An offset lateral in the
middle Bakken used fiber optics to
gather cluster-level completion measurements
and detect frac hits. A second
offset housed geophones to enable microseismic
mapping.
In any case, the fully coupled model
accurately reproduced the fracture geometry,
morphology and unique fracture
pressure behavior from the observation
lateral's gauges. The reservoir model was
calibrated to match the production history,
bottom-hole pressure behavior and
drainage pressures. The calibration process
resembles the one illustrated in SPE
209164, " Observation Lateral Project:
Direct Measurement of Far-Field Drainage
in the Bakken, " though that paper discusses
a loosely coupled model rather than a
fully coupled one.
To check whether the conclusions
drawn from the fully coupled model depend
on the details of the Observation
Lateral site's fracture geometry, Hess and
ResFrac conducted a second study that
paired the same reservoir simulation
model with a much simpler fracture
model. The simple model specified fracture
geometries based on rules of thumb rather
than trying to deduce geometries from
detailed data.
Specifically, the simple model calculated
fracture lengths using cluster-level
fluid volumes, then estimated the propped
length to be a percentage of the total
length. The model assumed that fracture
conductivity rose in proportion to cluster-level
fluid volumes, though it did account
for fracture conductivity falling
with effective normal stress.
The results from the simpler model
should be less accurate than the ones
from the fully coupled model. However,
it is realistic enough to compare with
the more robust model's predictions
and ensure that the learnings are not
model-dependent.
Scenarios Considered
All the models were scaled down
" sectors " to reduce runtime. The models
included five Middle Bakken wells and
1,200-foot lateral sectors with 33-foot
cluster spacing (36 clusters total). The
Perfect UI Case
Base UI Case
µ = 0.97
σ = 0.04
µ = 0.79
σ = 0.08
Low UI Case
µ = 0.25
σ = 0.22
SEPTEMBER 2024 41
American Oil and Gas Reporter - September 2024

Table of Contents for the Digital Edition of American Oil and Gas Reporter - September 2024
