Page 311 - Intelligent Digital Oil And Gas Fields

Smart Wells and Techniques for Reservoir Monitoring 259

section. More importantly, the hydraulic simulator should be capable of
computing the effect of the ICV by using a valve flow coefficient (Cv) that
has been measured in lab tests for multi- and single-phase flow.
To simulate and reproduce physical models that interact with smart well
completions, an automated model-calibration workflow that optimizes
down-hole valve settings for maximized oil-recovery factor can be devel-
oped. In short, such an automated workflow enables reactive and proactive
decisions to control unwanted production (water/gas, solids, fines, etc.)
from the smart well. The objective of this workflow is to allow the data
exchange and iteration between subsurface and surface models—such as
the reservoir simulator, hydraulic completions, and nodal wellbore and net-
work analysis—and to periodically update the regional/stor reservoir model
associated with the drainage area of the smart well completion. Moreover,
the workflow is required to model a semi-analytical wellbore model to per-
form the optimization of vertical lift performance (VLP). Such a workflow
needs to facilitate the following engineering functionalities:
• Receive, update, and allocate the real-time production data from the
remote-controlled system.
• Perform multilevel matching and calibration of a history-matched
simulation model with observed pressures and rates at the smart well
completion interval, well tubing, and surface production.
• Execute reactive control by optimizing the down-hole valve setting in
response to local data and predicted short-term fluids behavior.
• Execute proactive control by optimizing the down-hole valve settings at
multiple intervals to maximize oil recovery using both local nodal
properties and simulated predictions.
An example of an automated workflow for smart well calibration is given in
Fig. 7.9.
Dynamic calibration of a simulation model with observed pressures and
rates at the smart well completion level (i.e., per individual segment of the
ICV) is enabled through design, integration, and reconciliation of a virtual
PLT profile, which represents a probability distribution of a parameterized
reservoir property (such as permeability, water saturation, etc.). Such a dis-
tribution is derived as discrete conditional probability of reservoir property
(m) given the fluid flow rate (f ) per perforation interval (i) of the ICV
segment (s) of a smart well (w) as expressed in Eq. (7.1):

w
Prob F ¼ f \M ¼ ^m
w

e m w s,i|f ¼ Prob M ¼ ^m j F ¼ f ¼ s (7.1)
s
ð
Prob F ¼ f Þ