Large-scale Simulations Research Articles

A methodology for appropriate withdrawal of tsunami warnings based on numerical simulations

This study performed large-scale numerical simulations for predicting the attenuation of tsunamis caused by the 2011 Tohoku, Japan, and 2010 Maule, Chile earthquakes, recorded at Japanese tide gauges. Tsunami amplitude waveforms were generated by computing the moving root-mean-square of the data for quantitative analyses. Sensitivity analysis showed that tsunami nonlinearity and computational grid intervals near the tide gauges significantly impact the prediction of tsunami attenuation. The predicted withdrawal times of tsunami warnings agreed with the observations; however, time discrepancies were observed for advisory withdrawals at several stations. Using the proposed method, we predicted the warning period of a great interplate earthquake in the Nankai Trough to be approximately one day. These findings can provide critical information for disaster prevention because the withdrawal of warnings is directly related to permission to enter coastal areas affected by the tsunami, whereas unnecessarily long warnings hinder rescue operations.

How Does the Ocean Melt Antarctic Ice Shelves?

The present-day state and future of the Antarctic Ice Sheet depend on the rate at which the ocean melts its fringing ice shelves. Ocean heat must cross many physical and dynamical barriers to melt ice shelves, with the last of these being the ice-ocean boundary layer. This review summarizes the current understanding of ice-ocean boundary-layer dynamics, focusing on recent progress from laboratory experiments, turbulence-resolving numerical simulations, novel observations, and the application to large-scale simulations. The complex interplay between buoyant meltwater and external processes such as current shear leads to the emergence of several melting regimes that we describe, as well as freezing processes. The remaining challenges include developing new parameterizations for large-scale ice-ocean models based on recent advances and understanding the coevolution of melt and basal topography.

Domain Decomposition for Data-Driven Reduced Modeling of Large-Scale Systems

This paper focuses on the construction of accurate and predictive data-driven reduced models of large-scale numerical simulations with complex dynamics and sparse training datasets. In these settings, standard, single-domain approaches may be too inaccurate or may overfit and hence generalize poorly. Moreover, processing large-scale datasets typically requires significant memory and computing resources, which can render single-domain approaches computationally prohibitive. To address these challenges, we introduce a domain-decomposition formulation into the construction of a data-driven reduced model. In doing so, the basis functions used in the reduced model approximation become localized in space, which can increase the accuracy of the domain-decomposed approximation of the complex dynamics. The decomposition furthermore reduces the memory and computing requirements to process the underlying large-scale training dataset. We demonstrate the effectiveness and scalability of our approach in a large-scale three-dimensional unsteady rotating-detonation rocket engine simulation scenario with more than 75 million degrees of freedom and a sparse training dataset. Our results show that compared to the single-domain approach, the domain-decomposed version reduces both the training and prediction errors for pressure by up to 13% and up to 5% for other key quantities, such as temperature, and fuel, and oxidizer mass fractions. Lastly, our approach decreases the memory requirements for processing by almost a factor of four, which in turn reduces the computing requirements as well.

QSOFEED: Relationship between star formation and active galactic nuclei feedback

Context. Large-scale cosmological simulations suggest that feedback from active galactic nuclei (AGN) plays a crucial role in galaxy evolution. More specifically, outflows are one of the mechanisms by which the accretion energy of the AGN is transferred to the interstellar medium (ISM), heating and driving out gas and impacting star formation (SF). Aims. The purpose of this study is to directly test this hypothesis utilising SDSS spectra of a well-defined sample of 48 low-redshift (z < 0.14) type 2 quasars (QSO2s). Methods. By exploiting these data, we were able to characterise the kinematics of the warm ionised gas by performing a non-parametric analysis of the [OIII]λ5007 emission line. We also constrained the properties of the young stellar populations (YSP; tysp < 100 Myr) of their host galaxies via spectral synthesis modelling. Results. These analyses revealed that 85% of the QSO2s display velocity dispersions in the warm ionised gas phase greater than that of the stellar component of their host galaxies, indicating the presence of AGN-driven outflows. We compared the gas kinematics with the intrinsic properties of the AGN and found that there is a positive correlation between gas velocity dispersion and 1.4 GHz radio luminosity – but not with the AGN bolometric luminosity or Eddington ratio. This either suggests that the radio luminosity is the key factor driving outflows or that the outflows themselves are shocking the ISM and producing synchrotron emission. We found that 98% of the sample host YSPs to varying degrees, with star formation rates (SFRs) of 0 ≤ SFR ≤ 92 M⊙ yr−1, averaged over 100 Myr. We compared the gas kinematics and outflow properties to the SFRs to establish possible correlations that could suggest that the presence of the outflowing gas could be impacting SF, but we found that no such correlation exists. This leads us to the conclusion that on the scales probed by the SDSS fibre (between 2 and 7 kpc diameters), AGN-driven outflows do not impact SF on the timescales probed in this study. However, we find a positive correlation between the light-weighted stellar ages of the QSO2s and the black hole mass, which might indicate that successive AGN episodes lead to the suppression of SF over the course of galaxy evolution.

Discrete differential geometry-based model for nonlinear analysis of axisymmetric shells

In this paper, we propose a novel one-dimensional (1D) discrete differential geometry (DDG)-based numerical method for geometrically nonlinear mechanics analysis (e.g., buckling and snapping) of axisymmetric shell structures. Our numerical model leverages differential geometry principles to accurately capture the complex nonlinear deformation patterns exhibited by axisymmetric shells. By discretizing the axisymmetric shell into interconnected 1D elements along the meridional direction, the in-plane stretching and out-of-bending potentials are formulated based on the geometric principles of 1D nodes and edges under the Kirchhoff–Love hypothesis, and elastic force vector and associated Hessian matrix required by equations of motion are later derived based on symbolic calculation. Through extensive validation with available theoretical solutions and finite element method (FEM) simulations in literature, our model demonstrates high accuracy in predicting the nonlinear behavior of axisymmetric shells. Importantly, compared to the classical theoretical model and three-dimensional (3D) FEM simulation, our model is highly computationally efficient, making it suitable for large-scale real-time simulations of nonlinear problems of shell structures such as instability and snap-through phenomena. Moreover, our framework can easily incorporate complex loading conditions, e.g., boundary nonlinear contact and multi-physics actuation, which play an essential role in the use of engineering applications, such as soft robots and flexible devices. This study demonstrates that the simplicity and effectiveness of the 1D discrete differential geometry-based approach render it a powerful tool for engineers and researchers interested in nonlinear mechanics analysis of axisymmetric shells, with potential applications in various engineering fields.

High proportion of new energy power system frequency and voltage support and regulation technology based on doubly fed induction generator

Abstract With the increasing proportion of new energy in power systems, issues regarding system voltage, inertia, and primary frequency regulation have become increasingly severe. In response to control and grid safety issues, a high proportion of new energy power system frequency and voltage support and regulation technology based on a doubly fed induction generator (DFIG) is proposed. The voltage-magnetic link equation of DFIG is theoretically derived, and the control principles of active and reactive power of DFIG are proposed. The support and regulation characteristics of improving the voltage of new energy substations are studied. Verification is conducted through large-scale grid simulation and on-site measurements.

An analytic traction-displacement model for a reinforcing ligament bridging a crack at an arbitrary angle, including elastic, frictional, snubbing, yielding, creep, and fatigue phenomena

A micromechanical model is developed that generates analytic expressions for the crack displacement vector u given an arbitrary far-field stress state σa for a crack that is bridged by an array of ligaments oriented at an arbitrary angle with respect to the crack plane. The model is applicable to various materials, e.g., fibrous ceramic composites, or polymer composites reinforced by stitches or z-pins or woven tows, and deals with interfacial friction, enhanced friction due to increased contact pressure (“snubbing”), and the possibility of ligament deflection enabled by yield or damage. The model also conveniently incorporates ligament failure and rate dependent phenomena (fatigue or creep). Adaptability of the model is enabled by the definition of a standard Reference Model, which generates analytic expressions for the crack displacement for given possible yield, ligament deflection, and friction and snubbing effects and is invariant for all geometrical and material choices. The switching on or off and the strengths of all phenomena are governed by assigning values to a handful of material parameters. The material parameters will generally be calibrated against data in a top-down strategy, the model thereby mapping material selection onto engineering fracture via the predicted bridging relationship u[σa]. The relationship u[σa] can depend strongly on bi-angular ligament orientation. Yield and deflection can change u[σa] qualitatively, e.g., by creating fracture surface contact even when σa includes substantial opening tension. Snubbing has significant effects, including possible stabilization of the pullout of a finite ligament. Since model output is computed via analytic expressions, its speed will support the model's use in large-scale material simulations or as constraining physical information in machine learning algorithms.

Development and deployment of data-driven turbulence model for three-dimensional complex configurations

Abstract In recent years, the synergy between artificial intelligence and turbulence big data has given rise to a new data-driven paradigm in turbulence research. Data-driven turbulence modeling has emerged as one of the forefront directions in fluid mechanics. Most existing studies focus on feature construction, selection, and the development of modeling frameworks, often overlooking the practical deployment and application of trained models. This paper examines the entire process from model construction to real-world deployment, using data-driven turbulence modeling for high Reynolds number flows over complex three-dimensional configurations as a case study. Key stages include data generation, input-output feature construction, model training, model compilation and optimization, deployment, and validation. We successfully implemented the entire workflow in a heterogeneous supercomputing environment and, through mixed programming techniques, integrated the resulting turbulence model into the Platform for Hybrid Engineering Simulation of Flows (PHengLEI) open-source software framework. This allowed for mixed-precision simulations, with the main equations solved in double precision and the turbulence model in half precision. The new computational framework was validated through large-scale parallel numerical simulations on grids with tens of millions of elements for three-dimensional complex configurations. The results highlight the efficiency of our model deployment, with overall computational efficiency improving by 13.35% and the turbulence model’s solution speed increasing by approximately 3.9 times. The accuracy of the computations was also confirmed, with the average relative error in the lift and drag coefficients calculated by the data-driven turbulence model within 3%. Across various computing nodes, the relative error in the computed aerodynamic coefficients remained within 1%, demonstrating the framework’s scalability. Notably, our contributions have been incorporated as a case study in the latest PHengLEI open-source project5 5 https://forge.osredm.com/PHengLEI/PHengLEI-TestCases/tree/master/Y02_ThreeD_M6_Unstruct_Branch_Ascend. .

JuMonC: A RESTful tool for enabling monitoring and control of simulations at scale

As systems and simulations grow in size and complexity, it is challenging to maintain efficient use of resources and avoid failures. In this scenario, monitoring becomes even more important and mandatory. This paper describes and discusses the benefits of the advanced monitoring and control tool JuMonC, which runs under user control alongside HPC simulations and provides valuable metrics via REST-API. In addition, plugin extensibility allows JuMonC to go a step further and provide computational steering of the simulation itself. To demonstrate the benefits and usability of JuMonC for large-scale simulations, two use cases are described employing nekRS and ICON on JURECA-DC, a supercomputer located at the Jülich Supercomputing Centre (JSC). Furthermore, a large-scale use case with nekRS on JSC’s flagship system JUWELS Booster is described. Finally, the interplay between JuMonC and LLview (a standard monitoring tool for HPC systems) is presented using a simple and secure JuMonC-LLview plugin, which collects performance metrics and enables their analysis in LLview. Overall, the portability and usefulness of JuMonC, together with its low performance impact, make it an important application for both current and future generations of exascale HPC systems.

Artificial neural network enabled photovoltaic-thermoelectric generator modelling and analysis

Photovoltaic-Thermoelectric Generator (PV-TEG) system has emerged as a promising approach to significantly enhance the efficiency of conventional PV cells. However, optimizing the performance of these hybrid systems presents a formidable challenge due to their complex structure and multitude of design parameters. This study tackles such challenge by developing a machine learning based Artificial Neural Network (ANN) model which comprises two sub-ANN models that can work independently for PV and TEG modules or in combination through a cyclic approach for the hybrid PV-TEG system. The model demonstrates remarkable versatility, allowing control over various parameters such as PV coating, device geometry, and environmental conditions. Compared to COMSOL simulations, the ANN model achieves over 97.6 % accuracy with a 6000-fold increase in simulation speed, enabling extensive parameter sweeps and insightful system analysis. Within 18 min, the model conducted a real-time simulation using 8712 weather data entries from Singapore in 2022 and predicted that the hybrid PV-TEG system would generate a total power of 265 kWh/m2, 6.4 % more than that of the standalone PV system with an average system temperature reduction of 7 K. The model's rapid processing capabilities and high accuracy are particularly beneficial for large-scale simulations and practical applications in renewable energy technology.

Nucleation-dependent early growth of dendritic grains in Al-Cu alloys: The real-time observations and large-scale phase-field simulations

Formation of a dendritic grain starts from nucleation and then growth propagation occurs. Through the in situ and real-time solidification experiments of Al-Cu alloys observed by synchrotron X-ray imaging, it is found that the early-stage free growth rate of dendritic grains goes far beyond the concept described by the classical crystal growth theory. The rate gradually drops down even though the liquid is continuously cooled down. Quantitative 3D phase-field simulations demonstrate that this abnormal early-stage growth behavior depends strongly on the nucleation. A critical nucleus can grow rapidly to a peak rate driven by the initial nucleation undercooling, and then its rate gradually drops down to a minimum before it approaches the steady-state free growth regime. This strong dependence of the early-stage growth on the nucleation is then supported by an analytical model, and this correlation enables the accurate identification of the nucleation undercooling for each grain in the experiment and thus allows for a large-scale quantitative simulation of the real-time observed polycrystalline growth. This progress provides a comprehensive understanding on the crystal growth kinetics from a critical nucleus to the growth end, and thus will provide a new theoretical framework to design novel technology to control the solidification microstructures.

Learning dislocation dynamics mobility laws from large-scale MD simulations

By dispensing with all the atoms and only focusing on dislocation lines, the computational method of Discrete Dislocation Dynamics (DDD) gains greatly over Molecular Dynamics (MD) in simulation efficiency of metal plasticity. But whereas in MD dislocations follow natural dynamics of atomic motion, DDD must rely on a dislocation mobility function to prescribe how a dislocation line should respond to the driving force exerted on it. However, reflecting our still incomplete understanding of ways in which dislocations move, mobility functions presently employed in DDD simulations entail simplifications and approximations of limited or, worse still, unknown accuracy and applicability. Here we introduce a data-driven approach in which the dislocation mobility function is modeled as a graph neural network (GNN) trained on large-scale MD simulations of crystal plasticity. We apply our proposed approach to predicting plastic strength of body-centered-cubic (BCC) metal tungsten and show that, once implemented in a DDD model, our GNN dislocation mobility function accurately reproduces the challenging tension/compression asymmetry of plastic flow observed both in ground-truth MD simulations and in experiment. Furthermore, subsequently validated by MD simulations, the same function accurately predicts plastic response of tungsten under conditions not previously seen in training. By demonstrating its ability to learn relevant physics of dislocation motion, our DDD+ML approach opens a promising avenue to bringing fidelity of DDD models closer in line with direct MD simulations at a much reduced computational cost.

Martini 3 OliGo̅mers: A Scalable Approach for Multimers and Fibrils in GROMACS.

Martini 3 is a widely used coarse-grained simulation method for large-scale biomolecular simulations. It can be combined with a Go̅ model to realistically describe higher-order protein structures while allowing the folding and unfolding events. However, as of today, this method has largely been used only for individual monomers. In this article, we describe how the Go̅ model can be implemented within the framework of Martini 3 for a multimer system, taking into account both intramolecular and intermolecular interactions in an oligomeric protein system. We demonstrate the method by showing how it can be applied to both structural stability maintenance and assembly/disassembly of protein oligomers, using aquaporin tetramer, insulin dimer, and amyloid-β fibril as examples. We find that addition of intermolecular Go̅ potentials stabilizes the quaternary structure of proteins. The strength of the Go̅ potentials can be tuned so that the internal fluctuations of proteins match the behavior of atomistic simulation models, however, the results also show that the use of too strong intermolecular Go̅ potentials weakens the chemical specificity of oligomerization. The Martini-Go̅ model presented here enables the use of Go̅ potentials in oligomeric molecular systems in a computationally efficient and parallelizable manner, especially in the case of homopolymers, where the number of identical protein monomers is high. This paves the way for coarse-grained simulations of large protein complexes, such as viral protein capsids and prion fibrils, in complex biological environments.

Large Scale Simulation of 3D Fault Rupture Subjected to Far‐Field Loading With PDS‐FEM: Application to the 2018 Palu Earthquake

AbstractSimulating dynamic rupture of fault systems can be computationally demanding, as it requires reproducing complex fault geometry and accurately capturing waves propagating away from the rupture front. It is in particular challenging to predict an initial stress state consistent with fault geometries, heterogeneous distribution of surrounding materials, and far‐field tectonic loading. While standard techniques such as contact analysis and Lagrangian multipliers can be used to model the fault, it can lead to significant computational overhead in FEM. We extended Particle Discretization Scheme‐FEM, which provides numerically efficient crack treatment, without requiring contact analysis, to simulate dynamic fault rupture as a frictional crack propagating along a pre‐existing shear crack surface. Initial stress, which is consistent with initial frictional forces, material distribution and fault geometry, is derived using Coulomb friction and far‐field boundary conditions. The study first demonstrates the ability of the numerical method to reproduce a 2D ideal supershear scenario, and the underlying Burridge‐Andrew rupture mechanism. The methodology is then applied to the large scale simulation of the 2018 Palu earthquake on the Palu‐Koro fault. The simulation successfully reproduces the early and sustained supershear rupture which was observed for the Palu earthquake. Also, it indicates that the presence of an off‐fault damage zone can contribute to the low rupture velocity measured during the earthquake. Unlike sub‐Rayleigh earthquakes, the shockwave propagation was observed to lead to significant amplitudes of the ground motion even far from the fault.

Efficient digital design of ganglion cells in the retinal pathway

Spiking networks, the third generation of neural networks, are presented as low-power consumption machines with higher cognitive ability, one of the main concerns in intelligence machines. In fact, neuromorphic systems are hardware implementations of spiking networks with minimum resource, area, and power consumption while preserve maximum working frequency. Here, the focus is on the digital implementation of Retinal Ganglion Cell (RGC) based on the linear approximation of non-linear terms which is called Linear Retinal Ganglion Cell (LRGC). The low-cost hardware design of biological cells is acceptable when the digital model of the cell has the same phase and time domain behavior as the original model and follows the dynamic behavior of the original model accurately, which is discussed and confirmed with different analyzes in this paper. The low-cost hardware design of biological cells allows the optimal implementation of a neural population on the hardware, provided that the collective behavior of the digital model matches the original model which is approved by the large-scale simulation of RGC and LRGC models. Cognitive processes are performed in the nervous system at a very low cost, which neuromorphic systems are trying to achieve this important. In this regard, the behavior of RGC and LRGC models in the reconstruction of the image through the retina pathway was examined and a high agreement between the performance of the two models was achieved. Finally, the high functional compatibility of RGC, LRGC models proves that the proposed model is a good candidate of the main model in neuromorphic systems with low hardware cost.

Stable creation of a single plasma channel and collimated fast electrons using large-scale PIC simulations with new dynamic load-balancing technique

We developed a PIC code using new load balancing technique, in which the lower load processes help the higher load processes. A test calculation indicates more than 10 times faster than that without load balancing. Large scale 3-d calculations indicate the formation of central current whose density is close to critical density, supported by the magnetic field inside the channel.

Reduced-order reconstruction of discrete grey forecasting model and its application

Discrete grey forecasting models based on an accumulative operator have been widely used in many practical fields. With the development of grey forecasting models, it is a problem to be solved to further analyze internal mechanisms and unify the structures. This paper aims to reconstruct the model from a perspective of sequence characteristics and simplify the modeling steps under the condition of ensuring the accuracy of the model. First, this paper analyzes dynamic sequence evolution hidden and mines relationship between the structure and original sequence features contained in discrete grey forecasting model. Then, the reconstruction is carried out to prove the equivalence and quantitative relation between reduced-order model and original model. Under order recursive estimation, new parameters are addressed. Finally, theoretical correctness is verified by large-scale numerical simulation. Moreover, the reduced-order model is applied for prediction on the peak of battery incremental capacity and capacity degradation. Results show the effectiveness and superior prediction performance of the reduced-order model, where MAPEs of grey forecasting models have controlled under 4%.

Quantum-critical properties of the one- and two-dimensional random transverse-field Ising model from large-scale quantum Monte Carlo simulations

We study the ferromagnetic transverse-field Ising model with quenched disorder at T = 0T=0 in one and two dimensions by means of stochastic series expansion quantum Monte Carlo simulations using a rigorous zero-temperature scheme. Using a sample-replication method and averaged Binder ratios, we determine the critical shift and width exponents \nu_\mathrm{s}νs and \nu_\mathrm{w}νw as well as unbiased critical points by finite-size scaling. Further, scaling of the disorder-averaged magnetisation at the critical point is used to determine the order-parameter critical exponent \betaβ and the critical exponent \nu_{\mathrm{av}}νav of the average correlation length. The dynamic scaling in the Griffiths phase is investigated by measuring the local susceptibility in the disordered phase and the dynamic exponent z'z′ is extracted. By applying various finite-size scaling protocols, we provide an extensive and comprehensive comparison between the different approaches on equal footing. The emphasis on effective zero-temperature simulations resolves several inconsistencies in existing literature.

Mixing prisoner’s dilemma games on higher-order networks

The prisoner’s dilemma game is a paradigmatic example of a plight situation in which different individuals vie for maximal payoffs. And it has been shown that interaction networks that host such individuals have an immense impact on individual strategy choices. In this paper, we advance this subject by mixing the two-person prisoner’s dilemma and various multiplayer prisoner’s dilemma games and explore the role of 2-order simplicial complexes on the evolutionary dynamics. In the studied model, the fraction of multiplayer prisoner’s dilemma games on the network can be indirectly controlled by adjusting the ratio of 2-simplex complexes in the network. Through large-scale simulations, we show that by adjusting this ratio, the level of cooperation can be enhanced by a mechanism that is unique to higher-order networks. In particular, the strategic unity of the triangles in the network can foster the stability of cooperation and its spreading through enhanced network reciprocity. These results thus reveal benefits of higher-order interactions on cooperation that have thus far, on classical networks, been reserved only for games played in groups, such as the public goods game.

Experimental study on identifying anomaly azimuth based on acoustic vector array of borehole acoustic reflection imaging

Accurate anomaly azimuth identification is a major challenge in borehole acoustic reflection imaging. We propose an azimuth estimation method using an acoustic vector array to address this issue. This method was rigorously tested through a physical simulation experiment in a water-filled tank using a monopole acoustic source and an acoustic receiver station comprised of eight azimuthal elements. The proposed method involves the synthesis of multiazimuth acoustic pressure waveforms recorded by the receiver into multicomponent particle velocity waveforms. These waveforms are paired and subjected to coordinate conversion, yielding multiple groups of orthogonal component particle velocity waveforms. Each group underwent polarization analysis to obtain the direction of particle polarization on the horizontal plane and isolate the principal component particle velocity waveform. Using the particle polarization direction, the amplitude of the multiazimuth acoustic pressure waveform, and the type of echo mode, we accurately determined the azimuth of the anomaly, facilitating the creation of a spatial distribution image of the anomaly. Further validation of this method was conducted through a large-scale physical simulation experiment in a deep-water lake using a borehole acoustic reflection imaging instrument. The obtained experimental data were used to validate the effectiveness of this method. Finally, considering an actual logging environment, the feasibility of this method in a borehole environment was verified using numerical simulations and field data of acoustic detection of nearby wells. The present findings demonstrate that this novel method considerably enhances the accuracy and stability of the measurement of anomaly azimuth based on the particle polarization direction. In addition, this method effectively harnesses both acoustic pressure and particle velocity in borehole acoustic reflection imaging, exhibiting superior noise robustness, mitigating noise, and improving the signal-to-noise ratio of the echoes.