General-purpose Simulation Research Articles

A Fast History-Matching and Optimization Tool and Its Application to a Full Field with More than 1,000 Wells

Summary In this work, we study a waterflood field containing more than 1,000 wells, and the modern field management techniques with full-fidelity 3D geocellular reservoir models become computationally prohibitive. To overcome the difficulty, we developed a novel flow-network data-driven model—the general-purpose simulator-powered network (GPSNet) model—and used it for rapid history matching and optimization. GPSNet includes physics, such as mass conservation, multiphase flow, and phase changes, while maintaining a good level of efficiency. To build such a model, a cluster of 1D connections among well completion points is constructed and forms a flow network. Multiphase fluid flow is assumed to occur in each 1D connection, and the flow in the whole network is simulated by our in-house general-purpose simulator. Next, to effectively reduce the uncertainty, a hierarchical history-matching workflow is adopted to match the production data. Ensemble smoother with multiple data assimilation (ESMDA) plays an important role in reducing the error at each history-matching step. After that, the best-matched candidate is selected for numerical optimization to maximize field oil production with constraints satisfying field conditions. Excellent history-matching results have been achieved on the field level, and good matches have also been observed for key producers. It is also worth mentioning that the history-matching process took a mere 4 hours to finish 1,100 simulation jobs. The successful application of the GPSNet to this waterflood field demonstrates a promising workflow that can be used as a fast and reliable decision-making tool for reservoir management.

Two is better than one: digital siblings to improve autonomous driving testing

Simulation-based testing represents an important step to ensure the reliability of autonomous driving software. In practice, when companies rely on third-party general-purpose simulators, either for in-house or outsourced testing, the generalizability of testing results to real autonomous vehicles is at stake. In this paper, we enhance simulation-based testing by introducing the notion of digital siblings—a multi-simulator approach that tests a given autonomous vehicle on multiple general-purpose simulators built with different technologies, that operate collectively as an ensemble in the testing process. We exemplify our approach on a case study focused on testing the lane-keeping component of an autonomous vehicle. We use two open-source simulators as digital siblings, and we empirically compare such a multi-simulator approach against a digital twin of a physical scaled autonomous vehicle on a large set of test cases. Our approach requires generating and running test cases for each individual simulator, in the form of sequences of road points. Then, test cases are migrated between simulators, using feature maps to characterize the exercised driving conditions. Finally, the joint predicted failure probability is computed, and a failure is reported only in cases of agreement among the siblings. Our empirical evaluation shows that the ensemble failure predictor by the digital siblings is superior to each individual simulator at predicting the failures of the digital twin. We discuss the findings of our case study and detail how our approach can help researchers interested in automated testing of autonomous driving software.

Rotorcraft stability analysis using Lyapunov characteristic exponents estimated from multibody dynamics

Stability analysis of complex, nonlinear dynamical systems is a challenge. The use of Lyapunov Characteristic Exponents through a Jacobian-less method is proposed as a means to identify the Maximum Lyapunov Characteristic Exponent, namely the fundamental stability indicator of a generic problem, solely from time series obtained through general-purpose multibody dynamics simulations of complex rotorcraft aeromechanics models. The method is first applied to a relatively simple scenario concerning the identification of ground resonance. Then, its application to more complex models is addressed by studying the aeroelastic stability and identifying the whirl flutter of the XV-15 tiltrotor using a comprehensive aeroelastic model.

ЗАГАЛЬНА ХАРАКТЕРИСТИКА СИСТЕМИ ІМІТАЦІЙНОГО МОДЕЛЮВАННЯ GPSS

The article examines the general characteristics of the GPSS simulation modeling system. Researching the topic is relevant because it is essential for comprehending and effectively utilizing a fundamental tool in the realm of simulation modeling. GPSS (General Purpose Simulation System) has been a prominent simulation language and environment for several decades, serving as a key instrument for modeling and analyzing complex systems across various domains. It is noted that the GPSS World simulation system is a modern and powerful tool for research and analysis of various systems, in particular in the field of mass service systems. Through its development history from the original GPSS to the latest version GPSS World, it has acquired a number of improvements that make it unique and highly effective. This system has a high level of interactivity, extensive functionality and the ability to model both discrete and continuous processes. Visualization and animation tools built into the system facilitate convenient observation and analysis of simulation results. Exploring the general characteristics of GPSS is essential for several reasons. Investigating the fundamental characteristics of GPSS provides insights into its underlying principles, methodologies, and computational mechanisms. By comprehensively studying the features and capabilities of GPSS, researchers can identify opportunities for enhancing its performance, scalability, and usability. GPSS offers a range of modeling constructs, simulation controls, and statistical analysis tools. Exploring its general characteristics allows researchers to uncover the full spectrum of modeling capabilities and adapt them to diverse application scenarios. As simulation modeling evolves, GPSS continues to evolve alongside new methodologies, techniques, and technologies. Investigating its general characteristics facilitates integration with advanced modeling approaches such as agent-based modeling, discrete-event simulation optimization, or hybrid simulation paradigms. Researching the general characteristics of GPSS contributes to educational initiatives in simulation modeling and operations research. It provides valuable content for instructional materials, textbooks, and courses aimed at introducing students to simulation concepts, programming languages, and best practices.

Validation of the single-event method for low-energy electron transport via stopping power calculations with MCNP

Monte Carlo simulations of low-energy (<50 keV) electron transport in matter are essential for a broad range of application fields. Several Monte Carlo codes have developed specialized treatments for this case, but a comprehensive validation of low-energy electron transport for general-purpose simulations remains lacking in the literature. One approach to accomplish this validation is calculation of stopping power using low-energy electron transport physics, as stopping power is a fundamental radiation transport quantity which must be simulated accurately for nearly any application. In this work, we use the Monte Carlo N-Particle (MCNP) radiation transport code with the single-event method for electron transport to calculate stopping powers of low-energy electrons (50 eV to 30 keV) in 41 elemental solids, 14 compound solids, and five rare gas solids, comparing simulation results to published semi-empirical stopping power calculations from optical measurements. In general, the simulations give good agreement (typically within ±10%) with semi-empirical stopping power values at higher energies: 300 eV and above for most elemental solids, 1 keV and above for compound solids, and 400 eV and above for rare gas solids. Agreement between MCNP and semi-empirical values is worse below these energies. The most significant source of error is the EPRDATA14 cross section data, which does not account for changes in electronic structure due to solid-state bonding, particularly in compound materials. The simplistic model of atomic excitation used to generate the EPRDATA14 cross sections is another key source of error. Additionally, the breakdown of the continuous slowing-down approximation introduces significant uncertainty at low energies, although this is a limitation of the calculation method and not of the simulation procedure. Accounting for these and other uncertainty sources, the single event method in MCNP is robust and able to give good accuracy for a variety of low-energy electron transport problems through diverse kinds of materials.

Simulating animal space use from fitted integrated Step‐Selection Functions (iSSF)

Abstract A standing challenge in the study of animal movement ecology is the capacity to predict where and when an individual animal might occur on the landscape, the so‐called, utilisation distribution (UD). Under certain assumptions, the steady‐state UD can be predicted from a fitted exponential habitat selection function. However, these assumptions are rarely met. Furthermore, there are many applications that require the estimation of transient dynamics rather than steady‐state UDs (e.g. when modelling migration or dispersal). Thus, there is a clear need for computational tools capable of predicting UDs based on observed animal movement data. Integrated Step‐Selection Analyses (iSSAs), which integrates movement of the animal into habitat selection analyses, are widely used to study habitat selection and movement of wild animals, and result in a fully parametrised individual‐based model of animal movement, which we refer to as an integrated Step Selection Function (iSSF). An iSSF can be used to generate stochastic animal paths based on random draws from a series of Markovian redistribution kernels, each consisting of a selection‐free, but possibly habitat‐influenced, movement kernel and a movement‐free selection function. The UD can be approximated by a sufficiently large set of such stochastic paths. Here, we present a set of functions in R to facilitate the simulation of animal space use from fitted iSSFs. Our goal is to provide a general purpose simulator that is easy to use and is part of an existing workflow for iSSAs (within the amt R package). We demonstrate through a series of applications how the simulator can be used to address a variety of questions in applied movement ecology. By providing functions in amt and coded examples, we hope to encourage ecologists using iSSFs to explore their predictions and model goodness‐of‐fit using simulations, and to further explore mechanistic approaches to modelling landscape connectivity.

Solidification analysis for variable thickness aluminum castings: simulation and chill design insights

Manufacturing high-quality casting parts with complex geometries requires high engineering skill and precision. One essential quality concern is isolated hot spots within the castings, often in thick sections. Each hot spot must be consistently fed or mitigated through directional solidification techniques. The impact of various mold sands and the geometry of chill parts on solidification direction was investigated using specialized casting and general-purpose simulation programs. A parametric optimization method was employed to analyze directional solidification to adjust the geometry of the chill parts. The results indicate that employing diverse mold sands to enhance cooling in the thick sections was a viable strategy for achieving directional solidification in parts where the feeding pathway is obstructed due to changes in cross-section. Furthermore, the study revealed that intricate details in the chill part’s geometry are not critical; however, a minimum volume (or weight) was necessary for adequate directional solidification. Lastly, an easily applicable mathematical model has been developed to determine the required volume of chill parts to ensure successful directional solidification.

ELOISE - Reliable background simulation at sub-keV energies

CaWO_44 and Al_22O_33 are well-established target materials used by experiments searching for rare events like the elastic scattering off of a hypothetical Dark Matter particle. In recent years, experiments have reached detection thresholds for nuclear recoils at the 10 eV-scale. At this energy scale, a reliable Monte Carlo simulation of the expected background is crucial. However, none of the publicly available general-purpose simulation packages are validated at this energy scale and for these targets. The recently started ELOISE project aims to provide reliable simulations of electromagnetic particle interactions for this use case by obtaining experimental reference data, validating the simulation code against them, and, if needed, calibrating the code to the reference data.

A Physics-Informed Recurrent Neural Network for RRAM Modeling

Extracting behavioral models of RRAM devices is challenging due to their unique “memory” behaviors and rapid developments, for which well-established modeling frameworks and systematic parameter extraction processes are not available. In this work, we propose a physics-informed recurrent neural network (PiRNN) methodology to generate behavioral models of RRAM devices from practical measurement/simulation data. The proposed framework can faithfully capture the evolution of internal state and its impacts on the output. A series of modifications informed by the RRAM device physics are proposed to enhance the modeling capabilities. The integration strategy of Verilog-A equivalent circuits, is also developed for compatibility with existing general-purpose circuit simulators. The Verilog-A model can be easily adopted into the SPICE-type simulator for the circuit design with a variable step that differs from the training process. Numerical experiments with real RRAM devices data demonstrate the feasibility and advantages of the proposed methodology.

Numerical Simulations in Ultrasonic Guided Wave Analysis for the Design of SHM Systems—Benchmark Study Based on the Open Guided Waves Online Platform Dataset

Structural health monitoring (SHM) strategies based on ultrasonic guided waves are very promising regarding thin-walled lightweight structures. To study the performance of such systems, validated numerical analysis tools have to be used. For that procedure, a benchmark between two numerical methods will be presented. The first promising approach is the elastodynamic finite integration technique (EFIT). Miscellaneous research shows that its capability of capturing wave characteristics and interactions is advanced in various media and structures, including thin-walled composites. The second approach employs conventional shell-type finite elements following the Reissner–Mindlin theory for modelling layered composite structures. The advantage of using such finite element methods (FEM) is their high availability in general purpose simulation tools. As a reference model, the measurement data coming from the Open Guided Waves Project (OGW) was taken into account. The OGW dataset provides the experimental data of ultrasonic guided wave propagation in carbon fiber composite plates with an additional omega stringer. By using this contribution, this experiment was reproduced by simulation. The paper presents the results of a validation and motivates further research, such as in research related to the probability of detection analysis and numerical performance.

Development and application of a user-friendly general-purpose predictive simulation tool for two-dimensional flow analysis

Development and application of a user-friendly general-purpose predictive simulation tool for two-dimensional flow analysis

Güç Elektroniği Sistemleri için HIL (Hardware In The Loop) Simülatör Kartı Tasarımı ve Aktif Güç Filtreleri için Uygulanması

In this study, firstly, a general-purpose test simulator circuit was developed that could test the response of the circuits used in power electronics on hardware under various operating conditions (hardware in the loop). In the next step, a control board to be used for Shunt Active Power Filters (SPAF) was designed. The control software for SPAF was prepared using CubeMX IDE compiler for two different harmonic extraction methods using instantaneous reactive power theory (IRPT) and synchronous rotating reference frame (SRF) methods. Finally, before connecting the prototype SPAF control board and software to the real system, it had been verified to work safely using the designed test system. With this study, both the SPAF prototype circuit was prepared and a cost-effective system that could be found in the market was designed.

Development of an open-source data-driven simulator for the unit-load multi-aisle automated storage and retrieval systems

ABSTRACT Discrete-event simulations are widely used to research automated storage/retrieval systems (AS/RS). However, using commercial general-purpose simulators for this purpose has limitations such as lack of specific functionalities to capture the peculiarities of AS/RS, low model reusability, and lack of access to source code. Consequently, researchers have developed their own bespoke programs to meet their specific needs. These programs are specific to their research objective and are not meant for easy adoption, modification, or extension. As a result, there has been a lot of duplication of efforts across different studies. Motivated by this requirement for a customisable special-purpose simulator for AS/RS, this paper develops an open-source data-driven discrete-event simulator that allows its user to create and run simulation models of multi-aisle AS/RS without needing to write any code. The data-driven approach allows the quick creation of models of different multi-aisle AS/RS configurations and control policies. The simulator is developed in Python programming language, leveraging the functionalities of various libraries in its ecosystem. The simulator’s architecture is kept modular to facilitate its management, modification, and extension. The simulator’s features and ability to adapt to changes in input data are demonstrated through three example scenarios.

Explicit Learning of Derivatives with the KREG and pKREG Models on the Example of Accurate Representation of Molecular Potential Energy Surfaces.

The KREG and pKREG models were proven to enable accurate learning of multidimensional single-molecule surfaces of quantum chemical properties such as ground-state potential energies, excitation energies, and oscillator strengths. These models are based on kernel ridge regression (KRR) with the Gaussian kernel function and employ a relative-to-equilibrium (RE) global molecular descriptor, while pKREG is designed to enforce invariance under atom permutations with a permutationally invariant kernel. Here we extend these two models to also explicitly include the derivative information from the training data into the models, which greatly improves their accuracy. We demonstrate on the example of learning potential energies and energy gradients that KREG and pKREG models are better or on par with state-of-the-art machine learning models. We also found that in challenging cases both energy and energy gradient labels should be learned to properly model potential energy surfaces and learning only energies or gradients is insufficient. The models' open-source implementation is freely available in the MLatom package for general-purpose atomistic machine learning simulations, which can be also performed on the MLatom@XACS cloud computing service.

Neural simulation pipeline: Enabling container-based simulations on-premise and in public clouds.

In this study, we explore the simulation setup in computational neuroscience. We use GENESIS, a general purpose simulation engine for sub-cellular components and biochemical reactions, realistic neuron models, large neural networks, and system-level models. GENESIS supports developing and running computer simulations but leaves a gap for setting up today's larger and more complex models. The field of realistic models of brain networks has overgrown the simplicity of earliest models. The challenges include managing the complexity of software dependencies and various models, setting up model parameter values, storing the input parameters alongside the results, and providing execution statistics. Moreover, in the high performance computing (HPC) context, public cloud resources are becoming an alternative to the expensive on-premises clusters. We present Neural Simulation Pipeline (NSP), which facilitates the large-scale computer simulations and their deployment to multiple computing infrastructures using the infrastructure as the code (IaC) containerization approach. The authors demonstrate the effectiveness of NSP in a pattern recognition task programmed with GENESIS, through a custom-built visual system, called RetNet(8 × 5,1) that uses biologically plausible Hodgkin-Huxley spiking neurons. We evaluate the pipeline by performing 54 simulations executed on-premise, at the Hasso Plattner Institute's (HPI) Future Service-Oriented Computing (SOC) Lab, and through the Amazon Web Services (AWS), the biggest public cloud service provider in the world. We report on the non-containerized and containerized execution with Docker, as well as present the cost per simulation in AWS. The results show that our neural simulation pipeline can reduce entry barriers to neural simulations, making them more practical and cost-effective.

GEM: A Generalized Memristor Device Modeling Framework Based on Neural Network for Transient Circuit Simulation

Conventional physics-based memristor device modeling methods highly rely on human expertise, which results in a long development period. To address the aforementioned challenges, we propose a new generalized memristor (GEM) device modeling framework based on the artificial neural network (ANN) technique, which has a minimum dependency on the underlying physics, resulting in a fast turn-around development time for customized memristor devices. GEM framework models the switching and conducting behaviors of the memristor devices separately, avoiding the signal-dependence issue in the prior time-series data modeling method. The result of the GEM framework is a compact model that supports general-purpose circuit simulators. Experimental results show that our compact model achieves a ratio of root-mean-square error to peak-to-peak (RMSE/PP) of 3.6% compared to the physics-based device model. Performance analysis of memristor-based logic and memristor crossbar circuits are conducted to demonstrate the effectiveness of our proposed GEM framework for the design and analysis of memristor-based circuits.

MOSFET Physics-Based Compact Model Mass-Produced: An Artificial Neural Network Approach

The continued scaling-down of nanoscale semiconductor devices has made it very challenging to obtain analytic surface potential solutions from complex equations in physics, which is the fundamental purpose of the MOSFET compact model. In this work, we proposed a general framework to automatically derive analytical solutions for surface potential in MOSFET, by leveraging the universal approximation power of deep neural networks. Our framework incorporated a physical-relation-neural-network (PRNN) to learn side-by-side from a general-purpose numerical simulator in handling complex equations of mathematical physics, and then instilled the "knowledge'' from the simulation data into the neural network, so as to generate an accurate closed-form mapping between device parameters and surface potential. Inherently, the surface potential was able to reflect the numerical solution of a two-dimensional (2D) Poisson equation, surpassing the limits of traditional 1D Poisson equation solutions, thus better illustrating the physical characteristics of scaling devices. We obtained promising results in inferring the analytic surface potential of MOSFET, and in applying the derived potential function to the building of 130 nm MOSFET compact models and circuit simulation. Such an efficient framework with accurate prediction of device performances demonstrates its potential in device optimization and circuit design.

Real-time high-resolution CO2 geological storage prediction using nested Fourier neural operators

Nested FNO is a machine learning framework that offers a general-purpose numerical simulator alternative to provide high-resolution CO2 storage predictions in real time.

Simulation-based evaluation of the effectiveness of fiber-optic sensing in monitoring and optimizing water circulation in next-generation enhanced geothermal systems

The success of multi-stage hydraulic fracturing technology in unconventional oil and gas reservoir development has attracted much attention from the geothermal community. Some pilot field experiments have been conducted to investigate the multi-stage hydraulic stimulation along horizontal or high-deviated wellbores in geothermal reservoirs (denoted as next-generation enhanced geothermal systems). One common output of such stimulation is that the created hydraulic fractures show various conductive capabilities, which could potentially lead to undesirable flow localization, early thermal breakthrough, and low cold water sweep efficiency. Monitoring the water circulation between horizontal wellbores connected through hydraulic fractures becomes important in avoiding early severe producing temperature reduction in practical operations. In this study, we investigate the effectiveness of distributed temperature sensing (DTS) and distributed strain sensing (DSS) in identifying dominant flow paths in geothermal reservoirs with multi-stage hydraulic stimulation and horizontal well completion designs. A coupled thermo-poro-elastic model for general-purpose geothermal reservoir simulation is developed and solved using a combined finite volume/finite element method, together with a fully implicit sequential coupling scheme. The temperature and strain responses along horizontal production wells are simulated and analyzed during water circulation. Reservoir temperature reduces in the area that is close to the water flow paths, which induces thermal contraction. These thermoelastic characteristics can be monitored by fiber-optic sensing, forming distinct signatures for flow-localization identification. DTS data may be dominantly influenced by the fractures with higher conductivity, ‘masking’ the signals related to fractures in the downstream, but DSS data can provide additional information for accurate identification of high conductive fractures. Successful mitigation strategies, for example plugging the dominant fractures illustrated in this study, could substantially increase the producing fluid temperature. The analyses in this study show that fiber-optic sensing is an effective technology to monitor and optimize water circulation in next-generation enhanced geothermal systems.

Applying Massively Parallel Interface for MPFA scheme with advanced linearization for fluid flow in porous media

We apply Massively Parallel Interface for MPFA-O scheme with state-of-the-art Operator-Based Linearization (OBL) approach for multiphase flow in porous media. The implementation of MPFA-O scheme enhances the modelling capabilities for non-K-orthogonal mesh. A fully implicit scheme is applied to guarantee the stability of solutions when a mass-based formulation is involved to keep the flexibility of the framework for general-purpose reservoir simulation. As the MPFA-O introduces more non-zeros elements in the Jacobian matrix than the traditional TPFA, massively parallel computations via Message Passing Interface (MPI) in this work help to guarantee competitive computational efficiency for high-fidelity geological models. Concerning the Jacobian assembly hassle, we apply the OBL approach which introduces operators combining the fluid and rock properties in the conservation equations and discretizes the operators in the physical parameter space. By computing values and derivatives of the operators via a multilinear interpolation, the assembly of Jacobian matrix and residual vector could be drastically simplified. Another benefit of the OBL is that by only evaluating operator values on the predefined nodes in the physical parameter space, the overhead related to complex phase behavior and property evaluation is significantly reduced. In the end, we present several benchmark cases to rigorously demonstrate the accuracy, convergence, and robustness of the framework and two challenging field-scale cases to further prove its computing performance and parallel scalability.