Capacitance resistance models (CRMs) comprise a family of material balance reservoir that have been applied to primary, secondary and tertiary recovery processes. CRMs predict well flow rates based solely on previously observed production injection rates, producers’ bottomhole pressures (BHPs); i.e., geological model rock/fluid properties are not required. can accelerate the learning curve analysis by providing interwell connectivity maps corroborate features such as sealing faults channels,...

Summary Automating model calibration and production optimization is computationally demanding because of the intensive multiphase-flow-simulation runs that are needed to predict response real reservoirs under proposed changes in inputs. Fast surrogate models have been speed up reservoir-response predictions without compromising accuracy. Surrogate either derived by preserving physics involved processes (e.g., mass balance) provide reliable long-range or developed solely on basis statistical...

Abstract Reduced order modeling techniques have been investigated in the context of reservoir simulation and optimization past decade to mitigate computational cost associated with large-scale nature models. Although great progress has made basically two fronts, namely, upscaling model reduction, there not a consensus which method (or methods) is preferable terms trade-offs between accuracy robustness, if they indeed, result large savings. In particular case such as proper orthogonal...

Eduardo Gildin and Yalchin Efendiev acknowledge the partial support of US Department Defence Army ARO Project under grant number W911NF-12-1-0206. This publication also was made possible by National Priorities Research Program 7-1482-1278 from Qatar Fund (a member The Foundation). would like to thank Energy Office Science, Advanced Scientific Computing Research, Applied Mathematics program award DE-FG02-13ER26165.

Summary We present a global/local model reduction for fast multiscale reservoir simulations in highly heterogeneous porous media. Our approach identifies low-dimensional structure the solution space. introduce an auxiliary variable (the velocity field) our that achieves high compression of model. This is achieved because field conservative any low-order reduced framework, whereas typical global based on proper-orthogonal-decomposition (POD) Galerkin projection cannot guarantee local mass...

Physics-informed neural networks (PINNs) integrate physical principles into machine learning, finding wide applications in various scientific and engineering fields. However, solving nonlinear hyperbolic partial differential equations (PDEs) with PINNs presents challenges due to inherent discontinuities the solutions. This is particularly true for Buckley–Leverett (B-L) equation, a key model multiphase fluid flow porous media. In this paper, we demonstrate that PINNs, conjunction Welge's...

Linear Parameter Varying (LPV) Systems are a well-established class of nonlinear systems with rich theory for stability analysis, control, and analytical response finding, among other aspects. Although there works on data-driven identification such systems, the literature is quite scarce in terms that tackle LPV models large-scale systems. Since ubiquitous practice, this work develops methodology local global based nonintrusive reduced-order modeling. The developed method coined as DMD-LPV...

Summary Reservoir simulations for CO2 sequestration are computationally expensive because they rely on being run large timescales. Shortened, cheaper simulation timelines prevent the observation of gas leakages that might occur over a longer duration. In statistics community, exclusion these is called censorship. We propose fast simulator surrogate captures potentially unobservable long-term risks. The crux our approach survival analysis, branch tailored to handle censored data. Our proposed...

Abstract Objective/Scope Fast-objective function estimators (FOFE) are often used to speed up reservoir management. This work presents a FOFE constructed with the parametric Dynamic Mode Decomposition (DMDp) method for carbonate WAG-CO2 injection. The results then compared simulation analyze FOFE's efficiency. Method/Procedure/Process We present an example of how changes in production strategy can affect behavior. utilizes snapshots gas and water saturation numerical runs different sizes...

Summary This work introduces a new model for the production-decline analysis (PDA) of hydraulically fractured wells on basis concept induced permeability field. We consider case when hydraulic-fracturing operation—in addition to establishing fundamental linear-flow geometry in drainage volume—alters ability formation conduct fluids throughout, but with varying degrees depending distance from main fracture plane. show that, under these circumstances, reservoir response departs...

Abstract The Capacitance Resistance Model (CRM) is a fast way for modeling and simulating gas waterflooding recovery processes, making it useful tool improving flood management in real-time. CRM an input-output material balance-based model, requires only injection production history, which are the most readily available data gathered throughout life of reservoir. In this work, relationship explored by representing with state-space (SS) equations. linear system SS equations define between...

We propose an online adaptive local-global POD-DEIM model reduction method for flows in heterogeneous porous media. The main idea of the proposed is to use local indicators decide on global update, which performed via reduced cost multiscale basis functions. This unique combination allows (1) developing that are used both and updates (2) computing modes functions consist offline some approach constructing a system based Proper Orthogonal Decomposition (POD) Galerkin projection....

Abstract Large data volumes and complex physical processes coupling multiphase flow, numerics production assessment, as in the case of reservoir simulation, are difficult to analyze rapidly but needed guide operators for designing strategies subsequent management. The traditional simulation process is time consuming alternatives such data-driven proxy modeling can overcome computation complexity drawbacks. A machine learning technique called recurrent neural network (RNN) has been proved...

Abstract Despite great advances in reservoir simulation capabilities with the introduction of high-performance computing (HPC) platforms and enhanced solvers, high fidelity grid-based still remains a challenging task. This task is especially demanding for fine-resolved geological reservoirs multiphase multicomponents production optimization uncertainty quantification frameworks where several calls large scale model need to be performed. Model order reduction techniques have been applied...

Summary In this paper, we apply mode decomposition and interpolatory projection methods to speed up simulations of two-phase flows in heterogeneous porous media. We propose intrusive nonintrusive model-reduction approaches that enable a significant reduction the size subsurface flow problem while capturing behavior fully resolved solutions. one approach, use dynamic decomposition. This approach does not require any modification reservoir simulation code but rather post-processes set global...