goal to manage water flooding of a reservoir so as to optimize oil production
DESCRIPTION
Closed-loop Reservoir Management Using a Reduced-order Model-based Control & Uncertainty Propagation Framework Grant Title: Closed-loop Reservoir Management Using a Reduced-order Model-based Control & Uncertainty Propagation Framework - PowerPoint PPT PresentationTRANSCRIPT
GOALTo manage water flooding of a reservoir so as to optimize oil production. To employ an optimal model-based control framework that uses uncertain parameter updating and a particular reduced-order model based on the update.
INTRODUCTIONThe accuracy of the parameters impacts the prediction capability of the reservoir model that is used for closed-loop reservoir management. It is justified to update the uncertain model parameters for any model-based application. Latin hypercube Hammersley sequence sampling (LHHS) technique is used for efficient propagation of the uncertain parameters through the reservoir model to quantify the uncertain parameters and their effects on the model results. Markov chain Monte Carlo (MCMC) is one approach to address parameter updating; however MCMC requires excessive executions of the first-principles reservoir model to generate the updates. To avoid this computational burden, a method such as Partial least squares (PLS) will be used to determine the relationships between the uncertain parameters and the model’s predictions. MCMC combined with PLS provides real-time efficient updates of the uncertain parameters. Another updating approach is to employ an ensemble Kalman filter (EnKF). The EnKF updates the uncertain parameters when new measured data are available. In this manner, estimates of the current state of the model account for the uncertainties that are in the state variables and the measurements. After updating, a reduced-order model can be identified to enable efficient real-time optimization.
Framework Example: 5-spot pattern reservoir
Closed-loop Reservoir Management Using a Reduced-order Model-based Control & Uncertainty Propagation Framework
Grant Title: Closed-loop Reservoir Management Using a Reduced-order Model-based Control & Uncertainty Propagation FrameworkGrant Number: 0927796 NSF Program: Dynamical Systems PI Name(s): Karlene A. Hoo
Updated Results•MCMC
•EnKF
•Production
REFERENCES•Chen, Y and KA Hoo (2010). Compt. & Chem. Enging., 34(10), 1597-1605•Chen, Y and KA Hoo (2010). ADCONIP, under review•Chen, Y and KA Hoo (2010). Int. J. Sys. Sci. , under review
True porosity
MCMC approach: Ym(n): measurements, Θ: uncertain parameter. EnKF approach: Yf: forecast, Ya: assimilated estimates of the states.
Schematic of a two-dimensional reservoir and wells
↓: water injection well↑: oil production well
Posterior distribution
Updated porosity
Prior distribution
True permeability
Updated permeability
MCMC: (◇) oil; (△) water
EnKF: (+) oil; (▽) water
Reference: (o) oil;
(*) water
MCMC: blue
EnKF: green
Reference: red
PlantModel-based
Controller
Reduced-order Model
YYmm
Model
YYaa(n(n))
YYff(n)(n)
ΘΘkk
MCMC
LHHS
ParameterUpdates
PLS Model
ΘΘk+1k+1
Parameter
UpdatesEnKF
YYmm YYmm
Y(n)Y(n)