Summary
In this work, we present an industrial automation framework for control and
optimization of hydrocarbon-producing fields while satisfying business and
physical constraints. The all-encompassing reservoir-management problem is
decomposed into a hierarchy of decision-making problems at different time
scales.
We exemplify the proposed approach through a case study on a multiple-layer
reservoir with a classical waterflood problem, in which a numerical reservoir
model is used as a virtual field. A model-predictive control (MPC) strategy is
used to regulate well and field instrumentation at economically optimal set
points determined by an overlying supervisory control level. The study
demonstrates significant reduction in water-handling costs and increased oil
recovery.
This work is a starting point for further development in automatic
intelligent reservoir technologies, which capitalize on the abilities of
permanent instrumented wells and remotely activated downhole completions.
Introduction
Reservoir management today is facing remarkable challenges in optimizing
profitability while satisfying a number of constraints (physical, financial,
geopolitical, and human). To optimize profitability, engineers traditionally
have used mathematical models, field data, and domain expertise in an effort to
make decisions about the best operating scenario. To increase the opportunities
for profitability by greatly increasing the volume of available field data and
the number of potential operating scenarios, the industry has recently started
deploying sophisticated hardware for remote sensing and actuation of wells and
facilities.
However, the acquisition of domain expertise about an oil field is a lengthy
and often unstructured activity that cannot be undertaken easily on a
continuous basis. In addition, because of the complexity and magnitude of an
all-encompassing optimization problem for an entire oil field, decisions are
made in a fragmented way for various pieces of that oil field. The lack of
intelligent software applications exacerbates the situation. As a result, the
capabilities of new sensing and actuation hardware have not been fully
realized, making it difficult to justify the significant cost that such
hardware imparts. In fact, it is fair to say that not much can be expected from
a feedback-based decision-making loop unless all elements in the loop are
properly configured, connected, and functioning. The industry state of the art
is clearly far from this ideal end.
To address the above issues, we propose a fieldwide optimization and control
framework1 with the following key features:
• It uses a hierarchy of time scales to separate the levels over which
decision making is performed, thus rendering a complex problem solvable.
• It integrates field data for continuous learning of key reservoir
features, based on simplified empirical models suitable for real-time
operations.
• It continuously optimizes reservoir performance while satisfying all
business and physical (surface and subsurface) constraints.
• It uses an advanced feedback-control strategy, which can be implemented
easily on field controllers at the wellhead or downhole.
• Its multiscale structure can naturally host optimization levels such as
multilateral selection, well location, and portfolio optimization.
To exemplify the above framework, we develop in this work an MPC (receding
horizon) scheme that underlies a supervisory optimization level, which predicts
the best operating points of a hydrocarbon-producing field. The resulting
structure is a self-learning and self-adaptive scheme that optimizes multiphase
fluid migration in compartmentalized reservoirs while integrating downhole
completions, wellhead restrictions, and business objectives and
constraints.
To demonstrate the capabilities of the proposed approach, we develop a
software prototype and test it on a case study using a commercial reservoir-
and well-modeling environment as a virtual reservoir. Dynamic simulations show
that the proposed strategy results in significant reduction of water injected
and produced, with a simultaneous increase in overall oil recovery. For the
case study presented, the self-learning reservoir-management strategy is able
to reduce cumulative water production by almost 80% and reduce water injection
by 55%, increasing project profitability from 13 to 55%.
In the sections that follow, we first give a background of current
reservoir-management challenges as applied to continuous oilfield modeling and
decision-making processes, multivariable optimization, and automatic control.
We then present our proposed approach based on petroleum-system identification,
an MPC strategy, and a closed-loop linear programming-optimization level that
searches continuously for the best operating point of the field. Finally, we
offer suggestions for further work.
© 2005. Society of Petroleum Engineers
View full textPDF
(
2,253 KB
)
History
- Original manuscript received:
11 February 2004
- Revised manuscript received:
17 March 2005
- Manuscript approved:
8 August 2005
- Version of record:
15 December 2005
View Cart
