Simulation Runtime Research Articles

DPM: A deep learning PDE augmentation method with application to large-eddy simulation

A framework is introduced that leverages known physics to reduce overfitting in machine learning for scientific applications. The partial differential equation (PDE) that expresses the physics is augmented with a neural network that uses available data to learn a description of the corresponding unknown or unrepresented physics. Training within this combined system corrects for missing, unknown, or erroneously represented physics, including discretization errors associated with the PDE's numerical solution. For optimization of the network within the PDE, an adjoint PDE is solved to provide high-dimensional gradients, and a stochastic adjoint method (SAM) further accelerates training. The approach is demonstrated for large-eddy simulation (LES) of turbulence. High-fidelity direct numerical simulations (DNS) of decaying isotropic turbulence provide the training data used to learn sub-filter-scale closures for the filtered Navier–Stokes equations. Out-of-sample comparisons show that the deep learning PDE method outperforms widely-used models, even for filter sizes so large that they become qualitatively incorrect. It also significantly outperforms the same neural network when a priori trained based on simple data mismatch, not accounting for the full PDE. Measures of discretization errors, which are well-known to be consequential in LES, point to the importance of the unified training formulation's design, which without modification corrects for them. For comparable accuracy, simulation runtime is significantly reduced. A relaxation of the typical discrete enforcement of the divergence-free constraint in the solver is also successful, instead allowing the DPM to approximately enforce incompressibility physics. Since the training loss function is not restricted to correspond directly to the closure to be learned, training can incorporate diverse data, including experimental data.

Modelling wind turbine tower-rotor interaction through an aerodynamic damping matrix

Current wind turbine modelling packages mainly adopt a complex methodology in which aerodynamic forces are coupled with the motion of the wind turbine components at every time step. This can result in long simulation run times, detrimental for the large number of simulations required for fatigue or reliability analyses. This contribution presents an efficient wind turbine modelling methodology based on blade element momentum theory and a linearization of the aerodynamic forces. This allows the wind-rotor interaction to be reduced to static forces applied at the tower top, with additional terms proportional to the tower velocities expressed as an aerodynamic damping matrix. This aerodynamic model was implemented as part of a finite element model of the tower and was successfully verified against the fully-coupled modelling package FAST. The damping matrix components explain key features of the coupling between fore-aft and side-side vibrations of the wind turbine. This coupling causes energy transfers between the two directions, complicating aerodynamic damping identification. The aerodynamic damping matrix offers novel insights and an efficient method to describe the aerodynamic damping of wind turbines.

Multiphysics for nuclear energy applications using a cohesive computational framework

With the recent development of advanced numerical algorithms, software design, and low-cost high-performance computer hardware, reliance on coupled multiphysics to predict the behavior of complex physical systems is beginning to become standard practice. This is especially true in nuclear energy applications where strong nonlinear interdependencies exist between reactor physics, radiation transport, multi-scale nuclear fuels performance, thermal fluids, etc. Resolving these nonlinear dependencies requires choices in multiphysics software approaches. Two main multiphysics modeling and simulation approaches have emerged. The first is based upon “code coupling” where disparate physics codes of different software design, code languages, and spatial and temporal integration schemes are coupled together with relatively complex data passing interfaces. The second multiphysics software approach is to employ a “cohesive” framework where all physics applications are developed with a common software design, i.e., data structures, syntax, input format, integrated spatial and temporal discretization schemes, etc. Here we present the Multiphysics Object-Oriented Simulation Environment (MOOSE) development and runtime framework and describe the framework’s cohesive modeling and simulation multiphysics approach. Then, a “cohesive-like” extension of the MOOSE framework is presented where MOOSE-based physics software applications are efficiently coupled to non-MOOSE (external) physics codes to form multiphysics applications using MOOSE’s unique interface capabilities. Finally, several examples of MOOSE’s cohesive and cohesive-like multiphysics applications will be demonstrated. These multiphysics demonstrations will incorporate both MOOSE-based applications and external codes, including Nek5000, RELAP-7, TRACE, BISON, and Pronghorn.

Faster and More Accurate Time Domain Electromagnetic Simulation Using Space Transformation

A novel finite difference time domain (FDTD) method based on space transformations is developed that overcomes the inherent obstacles of the conventional FDTD algorithm in a spatially complex domain. Our method leads to an adaptive mesh for the investigated structure based on its geometrical shape without adding additional numerical problems such as late-time instability. In this method, mesh boundaries can follow arbitrary geometrical shapes precisely meaning that discretization errors are minimized. Such errors can be considerable when dealing with large material differences or boundary conditions within the simulation domain. Different meshing and transformation techniques for a variety of different scenarios are presented. We show how boundaries discontinuity can be handled using our method without resulting artificial singularities or zeros. Unlike previous works no dispersive medium has been used so the simulation speed is similar to the standard FDTD method. The usability and superiority of this method in terms of simulation run time and accuracy are shown through a couple of scattering and plasmonic problems. Also, the result of any simulations are validated using finite element method.

A rapid, accurate image simulation strategy for mega-voltage cone-beam computed tomography

Intensive computation time is required to simulate images of electronic portal imaging device (EPID) using Monte Carlo (MC) technique, limiting the development of applications associated with EPID, such as mega-voltage cone-beam computed tomography (MV-CBCT). In this study, a fast, accurate simulation strategy for MV-CBCT utilizing the FastEPID technique has been developed and validated. During FastEPID simulation, photon detection was determined by pre-calculated photon energy deposition efficiency (η) and particle transport within the EPID was replaced with a pre-calculated optical photon spread function. This method is capable of reducing the time required for EPID image simulation by a factor of 90–140, without compromising image quality. MV-CBCT images reconstructed from the FastEPID simulated projections have been validated against measurement in terms of mean Hounsfield unit (HU), noise, and cupping artifact. These images were obtained with both a Catphan 604 phantom and an anthropomorphic pelvis phantom, under treatment beam energies of 2.5 MV, 6 MV, and 6 MV flattening filter free. The agreement between measurement and simulation was excellent in all cases. This novel strategy was capable of reducing the run time of a full scan simulation of MV-CBCT performed on a CPU cluster to a matter of hours, rather than weeks or months required by a conventional approach. Multiple applications associated with MV-CBCT (e.g. imager design optimization) are anticipated to gain from the implementation of this novel simulation strategy.

Hybrid numerical modeling of Friction Assisted Joining

Abstract A hybrid non-linear numerical model of the thermal field produced during Friction Assisted Joining process of Metal-polymers hybrid joints is developed. The model uses experimental processing loads (plunging load and torque) measured during an experimental campaign as inputs. A hybrid approach was used to simulate the contact between the tool and the metal sheet. Indeed, the plunging load was simulated as a prescribed load acting on the punch while the friction heat owing to the interaction of the rotating tool with the metal sheet was simulated as a distributed heat flux over the tool-metal contact surface. This enabled a dramatic reduction of the complexity of the model along with a short simulation runtime. During experimental tests, an IR camera was used to measure the real temperature evolution. These data were used to calibrate and validate the FE model. The results indicated that the developed model can accurately predict the temperature field on the upper metal surface. Thus, it can be readily used to forecast the temperature evolution and distribution and the metal-polymer interface, which is the most influencing factor that determines the mechanical behavior of this type of joints.

Reservoir water quality simulation with data mining models.

Water pollution is a concern in the management of water resources. This paper presents a statistical approach for data mining of patterns of water pollution in reservoirs. Genetic programming (GP), artificial neural network (ANN), and support vector machine (SVM) are applied to reservoir quality modeling. Input data for GP, ANN, and SVM were derived with the CE-QUAL-W2 numerical water quality simulation model. A case study was carried out using measured reservoir inflow and outflow, temperature, and nitrate concentration to the Amirkabir reservoir, Iran. Data mining models were evaluated with the MAE, NSE, RMSE, and R2 goodness-of-fit criteria. The results indicated that using the SVM model for determining nitrate pollution is time saving and more accurate in comparison with GP, ANN, and particularly CE-QUAL-W2. The SVM model reduces the runtime of nitrate concentration simulation by 581, 276, and 146 s compared with CE-QUAL-W2, GP, and ANN, respectively. The goodness-of-fit results showed that the highest values (R2 = 0.97, NSE = 0.92) and the lowest values (MAE = 0.034 and RMSE = 0.007) corresponded to SVM predictions, indicating higher model accuracy. This study demonstrates the potential for application of data mining tools to solute concentration simulation in reservoirs.

Improving confidence in complex ecosystem models: The sensitivity analysis of an Atlantis ecosystem model

There is growing interest in using mechanistic ecosystem models for ecosystem-based management, as they have the advantage of capturing both bottom-up and top-down processes as well as system interactions from food web structure, spatial constraints, and human activities. However, they have the disadvantage of requiring many parameters, many of which are unknown and must be estimated or calibrated to available data. Sensitivity analysis (SA) is an important part of simulation model development in order to understand model uncertainty and which parameters are more or less influential, but has been relatively neglected with Atlantis models due to the large number of parameters and long simulation run time. The Atlantis Eastern English Channel (Atlantis-EEC) model has been applied to investigate ecosystem dynamics and processes as well as fishery management scenarios. Here we present the results of a SA of growth, mortality, and recruitment parameters, which are parameters particularly difficult to measure and thus commonly tuned through model calibration. To manage the large number of parameters in the model, we used a Morris screening approach. This method can efficiently provide information on parameter main effects and interactions/non-linear effects with relatively few simulations. We performed an initial SA including all groups on 90 parameters, where we found that the most important drivers of system dynamics and biomass across groups were: (1) plankton growth and mortality rates and (2) top predator's fixed recruitment and juvenile mortality rates. We then performed a follow-up SA on a subset of 61 parameters, excluding top predators and plankton groups from the analysis. We found that all parameters were important for system stability, while individual groups’ biomass were generally most influenced by their own parameters and a subset of benthic invertebrates. Nonlinear/interaction effects were widespread, demonstrating the prevalence of feedback loops in the trophic structure, and the importance of bottom-up effects and, to a lesser extent, top-down effects. The information gained from this SA provided a better understanding of the model structure. It also allowed us to make recommendations on the general Atlantis model calibration process as well as suggesting which parameters may be most important for propagation of uncertainty in model scenarios.

Bounded Input Dissipativity of Linearized Circuit Models

We introduce the concept of Bounded Input Dissipativity (BID) for the characterization from an energy perspective of linearized models of nonlinear circuit blocks. Such linearized models are commonly employed in the design of large systems to approximate circuit blocks that operate in the neighborhood of some well-defined and asymptotically stable bias point and lead to a dramatic reduction in circuit simulation runtime. Even if the original circuit block always behaves as a dissipative system, its linearized model may behave as dissipative or non-dissipative depending on the amplitude of its small-signal port voltages and currents, compared to the corresponding constant bias or supply signal components. This paper presents a theoretical framework for the analysis of such a conditional dissipativity and introduces BID criteria based on the feasibility of Bilinear or Linear Matrix Inequalities. These criteria allow to estimate the maximum small-signal input amplitude that guarantees the dissipativity of the linearized model. Various examples demonstrate the validity of the proposed theory, including a test case showing how spurious and physically inconsistent instabilities may arise if the proposed BID conditions are not verified.

Building control virtual test bed and functional mock-up interface standard: comparison in the context of campus energy modelling and control

ABSTRACT This paper compares Building Control Virtual Test Bed (BCVTB) and Functional Mock-up Interface (FMI) standard to conduct coupled energy modelling and control for academic and research campuses. The intent is to systematically compare BCVTB and FMI. As proof-of-concept demonstration, building energy models of Colorado School of Mines are developed in EnergyPlus and coupled through BCVTB and FMI. The coupled models in BCVTB and FMI represent a virtual campus and they are compared from different perspectives such as simulation runtime, easiness of employing simple if–then–else control, easiness of integrating other programs and feasibility to employ advanced control and optimization. Overall, FMI standard is more promising due to its (a) capability to be programmed in a multithreading manner, (b) capability to be implemented as Gauβ–Seidl and Jacobi methods of loose coupling and (c) higher feasibility to implement advanced control and optimization programs.

Selection of random number generators in GATE Monte Carlo toolkit

Pseudo-random number generators are the heart of Monte Carlo simulations. Generally, the quality of random number generators may affect the output of the simulation. Any GEANT4-based Monte Carlo tool has different random number generators. The GATE toolkit, a widely used package in nuclear medicine, also provides three generators for the users to select. In this study, we used the random number generators in identical simulations in order to check their possible effect on the outputs of simulations.Simulations were performed considering two point radiation sources, including electrons and photons at the center of water phantoms, for eight monoenergies between 10 and 1000 keV. Paired t-test, normal probability plot and Bland–Altman plots were used to compare pairs of corresponding results.In all the simulations, the average values of data derived using different random number generators were almost the same with small differences. However, statistical variation was considerably high, particularly in the estimation of track length of electrons and the spectra of X-ray radiation due to the interaction of electron with atoms. At high energies noticeable differences were observed in simulation runtimes using different generators.There is no systematic difference between average parameters derived using different random number generators. Statistical fluctuations in corresponding data derived using random number generators are considerably high. It is recommended to use a large number of histories in simulation if the individual points in a series of data points are important. Due to simulation runtimes considerations, selection of Mersenne Twister generator is advocated.

Primary atomization of a turbulent liquid jet in crossflow: a comparison between VOF and FGVT methods

The primary atomization of a turbulent liquid jet in crossflow was investigated using a mathematical, numerical and computational model. A comparison between the standard volume of fluid (VOF) method and the fine grid volume tracking (FGVT) method was reported. The FGVT method advects the interface between two fluids using a finer grid than the employed by the standard VOF method, and according to the literature, it provides a better interface resolution. Simulations were performed using adaptive mesh refinement in a three-dimensional domain subjected to gravitational field using the in-house code MFSim. The flow was modeled using an Eulerian–Lagrangian approach to capture the interface. The interface was tracked initially with VOF or FGVT methods until the initial breakup. Broken off, small-scale nearly spherical drops were transferred into the Lagrangian point particle description. Column breakup and shear breakup modes were observed on the liquid jet. Drops were small as one-hundredth the size of the injector diameter. The model was validated against experimental correlations for the liquid jet column trajectory, and the droplet size distribution was compared to a previous numerical study from the literature. In addition, the breakup mechanisms predicted were qualitatively compared to those in previous reports. The results of the liquid column trajectory from the simulations performed presented low differences with the literature for both methods tested. According to the numerical results obtained from the computational simulations, the liquid column trajectory was well captured and the droplet size distribution was similar to the literature; however, the FGVT method provided higher accuracy compared to VOF method. The two main breakup modes were identified, namely the column breakup with Kelvin–Helmholtz instabilities and the surface breakup with the formation of multiple ligaments which later lead to droplet formation. The FGVT method provided a more detailed interface contour and improved the number of droplets converted from the Eulerian to the Lagrangian approach compared to the standard VOF method. On the other hand, the FGVT method presented relatively higher computational costs compared to VOF. Therefore, the FGVT method presented a higher interface quality and allowed a larger number of droplet conversion to the Lagrangian approach compared to the VOF method, even though the simulation run time using VOF was lower than with FGVT.

A Behavioral Model of Digital Resistive Switching for Systems Level DNN Acceleration

The deployment of IoT has brought on the generation of massive amounts of data in need of analysis. In recent times, resistive switching-based crossbar arrays have been presented as a viable candidate for the acceleration of neural network inference, in pushing beyond the limit of CMOS process scaling so as to keep pace with the ever-growing complexity of computation. While plenty of empirical and physically descriptive models exist, their simulation run times become inconvenient for users when used in large scale crossbar arrays. In this brief, we first present a behavioral model of digital resistive switching devices, demonstrated on experimental PCM data to exhibit generality which can see useful implementation in circuit analysis methods for compute-in-memory applications. This model is based on a pair of nonlinear ordinary differential equations that request switching time and threshold voltage inputs from the user, which are the most important concerns for binarized weights in crossbar arrays. By stripping the model of detailed physical characteristics that is not required at the systems level, we demonstrate an improvement of computational run time of up to 20-fold over state-of-the-art physics-based models, and 1.3 times over the most commonly used empirically driven models.

Development of a bond graph model for electromechanical actuators

Abstract This article addresses the bond graph model that allows for better comprehension of what the physical and mathematical concepts involved in electromechanical actuators are. In this study a disc type electromechanical actuator is modeled according to the bond graph method. Nonlinear effects such as flux path permeances, leakage loss and material saturation in a magnetic circuit are taken into account. The model is run on 20-sim 4.7 software, which allows for working directly with bond graph concepts. Simulation run-time is approximately 0.3 s for the simulation time of 12 ms. Results achieved with this model are compared with an MATLAB/Simulink model prepared using magnetic circuit algebraic equations. It was determined that the results of both models are almost identical. The static and dynamic model results are also verified by test results. As a consequence, a simple, fast running, accurate and easy-to-understand comprehensive bond graph model with magnetic circuit characteristics was developed. The model can be adapted to any type of electromechanical actuators with proper arrangement.

Pore-Scale Modeling of the Fuel Cell Catalyst Layer: The Role of Nafion Thin Films

In this work, a multiphysics pore-scale model of electrochemical reactive transport inside a fuel cell catalyst layer (CL) is presented and is used to elucidate the role of Nafion thin films on the overall performance of the CL. One of the main challenges of modeling transport in CL is the hierarchy of length scales and the multiphysics nature of transport. Consequently, traditional continuum methods, in which volume-averaged quantities are used to encapsulate the microstructural features, cannot give a clear picture of the multiphysics transport processes inside the CL. Alternatively, pore-scale methods, which take into account the true geometry down to the pore level, are suited to capture these processes for such systems. Previously, researchers have mainly employed direct numerical simulation (DNS) techniques for pore-scale modeling of the CL. While these techniques are highly accurate, they are very computationally expensive. This limits the domain size that can be simulated in a reasonable time. In this work, pore network modeling, a computationally cheap technique to simulate pore-scale transport, is used to study transport in the CL. Using this approach, we were able to speed up the simulation runtime up to 20-50x compared to DNS. Consequently, simulation of much larger domains of the CL was made possible, leading to more representative results. Finally, the developed multiphysics pore-scale model was used to study the role of Nafion thin films on the overall performance of the CL. It is notable that the network approach was only applied to the void phase, and the Nafion region was resolved using DNS. Therefore, yet further speed up can potentially be obtained by applying the network approach to the Nafion phase.

A Novel Approach to Atomistic Molecular Dynamics Simulation of Phenolic Resins Using Symthons.

Materials science is beginning to adopt computational simulation to eliminate laboratory trial and error campaigns—much like the pharmaceutical industry of 40 years ago. To further computational materials discovery, new methodology must be developed that enables rapid and accurate testing on accessible computational hardware. To this end, the authors utilise a novel methodology concept of intermediate molecules as a starting point, for which they propose the term ‘symthon’ (The term ‘Symthon’ is being used as a simulation equivalent of the synthon, popularised by Dr Stuart Warren in ‘Organic Synthesis: The Disconnection Approach’, OUP: Oxford, 1983.) rather than conventional monomers. The use of symthons eliminates the initial monomer bonding phase, reducing the number of iterations required in the simulation, thereby reducing the runtime. A novel approach to molecular dynamics, with an NVT (Canonical) ensemble and variable unit cell geometry, was used to generate structures with differing physical and thermal properties. Additional script methods were designed and tested, which enabled a high degree of cure in all sampled structures. This simulation has been trialled on large-scale atomistic models of phenolic resins, based on a range of stoichiometric ratios of formaldehyde and phenol. Density and glass transition temperature values were produced, and found to be in good agreement with empirical data and other simulated values in the literature. The runtime of the simulation was a key consideration in script design; cured models can be produced in under 24 h on modest hardware. The use of symthons has been shown as a viable methodology to reduce simulation runtime whilst generating accurate models.

Improving Forecast Uncertainty Quantification by Incorporating Production History and Using a Modified Ranking Method of Geostatistical Realizations

Abstract The majority of the geostatistical realizations ranking methods disregard the production history in selection of realizations, due to its requirement of high simulation run time. They also ignore to consider the degree of linear relationship between the “ranks based on the ranking measure” and “ranks based on the performance parameter” in choosing the employed ranking measure. To address these concerns, we propose an uncertainty quantification workflow, which includes two sequential stages of history matching and realization selection. In the first stage, production data are incorporated in the uncertainty quantification procedure through a history matching process. A fast simulator is employed to find the realizations with consistent flow behavior with the production history data in shorter time, compared to a comprehensive simulator. The selected realizations are the input of the second stage of the workflow, which can be any type of the realization selection method. In this study, we used the most convenient realization selection method, i.e., ranking of the realizations. To select the most efficient ranking measure, we investigated the degree of the linear correlation between the ranks based on the several ranking measures and the performance parameter. In addition, due to the shortcomings of the traditional ranking methods in uncertainty quantiles identification, a modified ranking method is introduced. This modification increases the certainty in the probability of the selected realizations. The obtained results on 3D close-to-real synthetic reservoir models revealed the capability of the modified ranking method in more accurate quantification of the uncertainty in reservoir performance prediction.

Simulation of seismic wave propagation in poroelastic media using vectorized Biot’s equations: an application to a CO$$_{2}$$ sequestration monitoring case

Wave propagation through porous media allows us to understand the response and interaction that occur between the elastic rock matrix and the fluid. This interaction has been described by Biot in his theory of poroelasticity. Seismic wave simulation using Biot’s formulations is computationally expensive when compared with the acoustic and elastic cases. This computational burden can be reduced by reformulating the numerical derivative operators to improve the efficiency. To achieve this, we used a staggered-grid finite difference operator to discretize 2D velocity stress equations as given by Biot’s theory. A vectorized derivative is applied on the staggered grid by shifting the coordinates. The reformulated equations were applied to compute the seismic response of a reservoir, where $$\hbox {CO}_2$$ is being injected and the effect of injected $$\hbox {CO}_2$$ in the formation is clearly seen in the synthetic data generated. The algorithm was coded in Python and to test its efficiency, the simulation run-time was compared for both serial and vectorized equations, and the speed-up ratio was calculated. Our results show a decrease in the simulation run-time for the vectorized execution with over a factor of a hundred percent (100%). We further observed that the amplitudes of the events increase with an increase in $$\hbox {CO}_2$$ saturation in the formation. This matches well with the real data.

Upscaling Technique for Highly Heterogeneous Reservoirs Based on Flow and Storage Capacity and the Lorenz Coefficient

Summary Field-scale representation of highly heterogeneous reservoirs remains a challenge in numerical reservoir simulation. In such reservoirs, detailed geological models are important to properly represent key heterogeneities. However, high computational costs and long simulation run times make these detailed models unfeasible to use in dynamic evaluations. Therefore, the scaling up of geological models is a key step in reservoir-engineering studies to reduce computational time. Scaling up must be carefully performed to maintain integrity; both truncation errors and the smoothing of subgrid heterogeneities can cause significant errors. This work evaluates the latter—the effect of averaging small-scale heterogeneities in the upscaling process—and proposes a new upscaling technique to overcome the associated limitations. The technique is dependent on splitting the porous media into two levels guided by flow- and storage-capacity analysis and the Lorenz coefficient (LC), both calculated with static properties (permeability and porosity) from a fine-scale reference model. This technique allows the adaptation of a fine highly heterogeneous geological model to a coarse-scale simulation model in a dual-porosity/dual-permeability (DP/DP) approach and represents the main reservoir heterogeneities and possible preferential paths. The new upscaling technique is applied to different reservoir-simulation models with water injection and immiscible gas injection as recovery methods. In deterministic and probabilistic studies, we show that the resulting coarse-scale dual-permeability models are more accurate and can better reproduce the fine-scale results in different upscaling ratios (URs), without using any simulation results of the reference fine-scale simulation models, as some of the current alternative upscaling methods do.

Multiresolution Grid Connectivity-Based Transform for Efficient History Matching of Unconventional Reservoirs

Summary Proper characterization of heterogeneous rock properties and hydraulic fracture parameters is essential for optimization of well spacing and reliable estimation of estimated ultimate recovery (EUR) in unconventional reservoirs. High resolution characterization of matrix properties and complex fracture parameters require efficient history matching of well production and pressure response. We propose a novel reservoir model parameterization method to reduce the number of unknowns, regularize the ill-posed problem, and enhance the efficiency of history matching of unconventional reservoirs. The proposed method makes a low-rank approximation of the spatial distribution of reservoir properties taking into account the varying model resolution of the matrix and hydraulic fractures. Typically, hydraulic fractures are represented with much higher resolution through local grid refinements compared to the matrix properties. In our approach, the spatial property distribution of both matrix and fractures is represented using a few parameters via a linear transformation with multiresolution basis functions. The parameters in transform domain are then updated during model calibrations, substantially reducing the number of unknowns. The multiresolution basis functions are constructed by using Eigen-decomposition of an adaptively coarsened grid Laplacian corresponding to the data resolution. Higher property resolution at the area of interest through the adaptive resolution control while keeping the original grid structure improves quality of history matching, reduces simulation runtime, and improves the efficiency of history matching. We demonstrate the power and efficacy of our method using synthetic and field examples. First, we illustrate the effectiveness of the proposed multiresolution parameterization by comparing it with traditional methods. For the field application, an unconventional tight oil reservoir model with a multistage hydraulic fractured well is calibrated using bottomhole pressure and water cut history data. The hydraulic fractures as well as the stimulated reservoir volume (SRV) near the well are represented with higher grid resolution. In addition to matrix and fracture properties, the extent of the SRV and hydraulic fractures are also adjusted through history matching using a multiobjective genetic algorithm. The calibrated ensemble of models are used to obtain bounds of production forecast. Our proposed method is designed to calibrate reservoir and fracture properties with higher resolution in regions that have improved data resolution and higher sensitivity to the well performance data, for example the SRV region and the hydraulic fractures. This leads to a fast and efficient history matching workflow and enables us to make optimal development/completion plans in a reasonable time frame.