Page 263 - Intelligent Digital Oil And Gas Fields

Integrated Asset Management and Optimization Workflows 213

Monte Carlo (McMC) (Maucec et al., 2007, 2011, 2013a,b). These tech-
niques are outlined in more detail in later sections.
In 2001, Begg et al. (2001) outlined the need for a holistic, integrated
approach to assess and manage the impact of uncertainties for the optimiza-
tion of oil and gas assets and investment decision-making. They proposed
the concept of a stochastic integrated asset model (SIAM) embedded in a
decision-support system. While the SIAM ties to a classical domain models
via a simulation engine (reservoir simulator) as a key component to quantify
uncertainties and their nonlinear behavior, the new components of the
workflow-like scenario analysis, real options thinking/valuation, value of
flexibility, and decision optimization are introduced. In following years,
the E&P industry has adopted and applied a variety of approaches to asset
production optimization with uncertainty:
• Bailey et al. (2004) introduced the workflow for NPV optimization
under uncertainty through the utility function F λ ¼μ λσ,where μ
and σ are the mean and standard deviation of the collection of N
individual appropriately sampled NPV realizations, and λ represents
the risk-aversion factor. The utility function F λ implies a maximiza-
tion of NPV under the constraint of the variable-importance mini-
mization of its standard deviation, depending on the user’s own risk
preference.
• Cullick et al. (2004) designed an optimization system that consists of
three workflows: an outer optimization workflow (outer loop), an inner
scenario and uncertainty-management-simulation workflow (inner
loop), and a dispatcher for distributed computing. The optimizer uses
(meta)heuristic search methods and is validated with three problems
of asset optimization: intuitive solution, nonintuitive solution without
uncertainty, and solution with URM.
• Sarma et al. (2005) introduced a closed-loop production optimization
under uncertainty with polynomial chaos expansion for an efficient
uncertainty propagation and validated the workflow in real-time optimi-
zation of NPV for the reservoir-under-waterflood regime, production
constraints, and uncertain subsurface characterization.
• Liu and Reynolds (2015) use multiobjective optimization [weighted
sum (WS) and normal boundary intersection (NBI)] for jointly maximiz-
ing the expectation and minimizing the risk by solving a max-min prob-
lem and applying it to a well-placement optimization and optimal well
control. Here, the concept of minimizing the risk simply refers to
“maximizing the minimum value life-cycle NPV”.