Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “performance model”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,333 records · Page 74

A Model-Free Voltage Control Approach to Mitigate Motor Stalling and FIDVR for Smart Grids

Electric power networks are large and highly nonlinear dynamical systems that present unique challenges to control design. Though there is a large number of dynamic models for power system stability and control, many models are only useful with right assumptions and wrong for other tasks. Moreover, the dynamic behavior of the grid is increasingly complex under the banner of smart grids. These lead to the difficulty of developing appropriate dynamic modeling, and thus an efficient control strategy. To avoid such modeling challenges, here we present a novel dynamic voltage control strategy based on a model-free control (MFC) approach, requiring no modeling procedure. In particular, it focuses on fault-induced delayed voltage recovery (FIDVR) events, which require complex and accurate dynamic load models to replicate such events. This work utilizes MFC as an online controller to achieve the desired voltage stability under the FIDVR event. The proposed MFC strategy allows simple implementation and low computational cost for efficient mitigation of FIDVR. For benchmarking, a reasonably accurate dynamic performance model is explored. Simulation results with the IEEE 57 bus test network demonstrate the enhanced dynamic voltage profile for load buses having induction motors with the support of reactive power resources.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Multi‐Model Ensembles in Ecosystem Modeling: Challenges and Best Practices for Decision‐Making

Ecosystem models are increasingly central to the decision-making for environmental policy, conservation planning, and climate-related investments. Yet, the growing reliance on Multi-Model Ensembles (MMEs) of ecosystem models by practitioners and policymakers, sometimes under tight timelines and imperfect information, has frequently outpaced the scientific rigor required to ensure ensemble reliability. Here, MMEs refer to approaches that combine targeted predictions from multiple models with the expectation of improving robustness and quantifying predictive uncertainty. Poorly designed MMEs may create a false sense of confidence and lead to suboptimal policy and market decisions. This perspective argues that robust decision-making-relevant MMEs must be grounded on two pillars: (1) rigorous Model Intercomparison Projects (MIPs), which identify inter-model agreement and disagreement, characterize model uncertainties, and evaluate robustness with observationally based benchmarks—MIPs' diagnostic evaluation is so critical that it must be needed to drive MME's decision in model selection and weighting, especially when only a limited number of models available; and (2) co-design by both stakeholders and scientists to ensure that scenarios, metrics and uncertainty requirements provide decision-relevant information. Building upon the past success and lessons from the existing MIPs-MMEs efforts (e.g., climate/Earth system/crop), we derived the theoretical basis for MMEs, addressed their specific challenges in ecosystem modeling, and highlighted proper consideration of model numbers and diversity, risk of model inter-dependence, effective calibration of model parameters, possible overdue of some ecosystem model development, critical roles of open benchmark data across a wide range of conditions, and suggested use of Artificial Intelligence to support MIPs-MMEs. We highlighted the under-recognized opportunity for MIPs and MMEs to drive scientific progress and innovation through identifying better performing models, systematic benchmarking, feedback loops, and targeted model improvement. By following actionable best practice guidelines, MMEs can evolve from ad hoc aggregation of models into a trusted backbone of environmental policy and decision-making.

ecosystem modeling↗

The DOE Advanced Gas Reactor Fuel Development and Qualification Program

The high outlet temperatures and high thermal-energy conversion efficiency of modular High Temperature Gas-cooled Reactors (HTGRs) enable an efficient and cost effective integration of the reactor system with non-electricity generation applications, such as process heat and/or hydrogen production, for the many petrochemical and other industrial processes that require temperatures between 300°C and 900°C. The Department of Energy (DOE) has selected the HTGR concept for the Next Generation Nuclear Plant (NGNP) Project as a transformative application of nuclear energy that will demonstrate emissions-free nuclear-assisted electricity, process heat, and hydrogen production, thereby reducing greenhouse-gas emissions and enhancing energy security. The objective of the DOE Advanced Gas Reactor (AGR) Fuel Development and Qualification program is to qualify tristructural isotropic (TRISO)-coated particle fuel for use in HTGRs. The Advanced Gas Reactor Fuel Development and Qualification Program consists of five elements: fuel manufacture, fuel and materials irradiations, post-irradiation examination (PIE) and safety testing, fuel performance modeling, and fission-product transport and source term evaluation. An underlying theme for the fuel development work is the need to develop a more complete, fundamental understanding of the relationship between the fuel fabrication process and key fuel properties, the irradiation and accident safety performance of the fuel, and the release and transport of fission products in the NGNP primary coolant system. An overview of the program and recent progress is presented.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗

Thermal Model Heat Rate Predictions of the AGR-5/6/7 Experiment

Several fuel and material irradiation experiments have been planned for the U.S. Department of Energy Advanced Gas Reactor Fuel Development and Qualification Program, which supports the development and qualification of tristructural isotropic (TRISO) coated particle fuel for use in high-temperature gas-cooled reactors. The goals of these experiments are to provide irradiation performance data to support fuel process development, qualify fuel for normal operating conditions, support development of fuel performance models and codes, and provide irradiated fuel and materials for post-irradiation examination and safety testing. Originally planned and named as separate fuel experiments, but subsequently combined into a single test train. AGR-5/6/7 will test low-enriched uranium oxycarbide (UCO – a heterogeneous mixture of uranium oxide and uranium carbide) TRISO fuel. The AGR-5/6 portion of the experiment will provide data to support qualification of the selected reference fuel design, while the AGR-7 portion will serve as a margin test, irradiating the fuel beyond normal operating conditions. During irradiation, the temperature in each capsule is controlled by varying the helium / neon gas mixture flowing through the capsule until the desired thermocouple temperature is reached. Further adjustments can be made by changing the lobe power and the neutron filter. Three neutron filters are available to use as the experiment burns up the fissionable fuel. A thermal finite element model has been created for the five capsules comprising the AGR-5/6/7 experiment. Heat rates from a physics analysis are imported into the model along with fast neutron fluence. Graphite shrinkage due to the fast neutron fluence is incorporated into the model. Gas gaps change as a function of fast neutron fluence. Radiation heat transfer is a major contributor in this model. This is a large model with more than 1 million finite element brick elements. More than 150 different parts are modeled in the finite element model and communicate with each other from a heat transfer sense. More than 50 thermocouples are used in the experiment and are compared to actual measurements. The experiment is composed of five separate stainless steel capsules all welded together. This paper presents a method used to predict the gas mixture in the event that all of the thermocouples fail in a particular capsule. Curve fitting of heat rates from previous cycles and projecting them into the future is the basis of the method.

36 MATERIALS SCIENCE↗

A new method for predicting hurricane rapid intensification based on co-occurring environmental parameters

Abstract Tropical cyclones (TCs) that undergo Rapid Intensification (RI) can pose serious socioeconomic threats and can potentially result in major damaging impacts along coastal areas. Considering the complexity of various physical mechanisms that play a role in RI and its relatively low probability of occurrence, predicting RI remains a major operational challenge. In this study, we propose a simple deterministic binary classification model based on the co-occurrence of environmental parameters (MCE) to predict an RI event. More specifically, the model determines the possibility of RI based on a simple count of the number of environmental predictors deemed favorable and unfavorable. We compare our model results to logistic regression (LR) and decision tree (DT) models, well-trained using the same set of environmental predictors. Results reveal that at an RI threshold of 30 kt, the MCE exhibits a critical success index score of 0.233 which is 14% higher than DT and LR model performances. When tested at multiple RI thresholds, the MCE displays relatively higher skill scores across multiple metrics. By simultaneously evaluating the favorability of predictors, the MCE is able to comparatively reduce the number of false alarms predicted when certain predictors are unfavorable toward RI. Interpreting these model results to gain a physical understanding of how co-occurring environmental parameters can affect RI, we highlight future directions for using models based on the MCE approach to understand and predict TC RI as well as other meteorological extremes.

54 ENVIRONMENTAL SCIENCES↗

Comparison of fixed charge and polarizable models for predicting the structural, thermodynamic, and transport properties of molten alkali chlorides

Results from extensive molecular dynamics simulations of molten LiCl, NaCl, KCl, and RbCl over a wide range of temperatures are reported. Comparison is made between the “Polarizable Ion Model” (PIM) and the non-polarizable “Rigid Ion Model” (RIM). Densities, self-diffusivities, shear viscosities, ionic conductivities, and thermal conductivities are computed and compared with experimental data. In addition, radial distribution functions are computed from ab initio molecular dynamics simulations and compared with the two sets of classical simulations as well as experimental data. Here, the two classical models perform reasonably well at capturing structural and dynamic properties of the four molten alkali chlorides, both qualitatively and often quantitatively. With the singular exception of liquid density, for which the PIM is more accurate than the RIM, there are few clear trends to suggest that one model is more accurate than the other for the four alkali halide systems studied here.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Deployment of a Full-Scope Commercial Nuclear Power Plant Control Room Simulator at the Idaho National Laboratory

The INL operates the HSSL to conduct research in the design and evaluation of advanced reactor control rooms, integration of intelligent support systems to assist operators, development and assessment of advanced human performance models, and visualizations to assess advanced operational concepts across various infrastructures. This advanced facility consists of a reconfigurable simulator and a virtual reality capability (known as the Computer-Aided Virtual Environment (CAVE)) (Figure 2). It supports human factors research, including human-in-the-loop performance, HSI, and analog and digital hybrid control displays. It can be applied to the development and evaluation of control systems and displays for complex systems such as existing and advanced NPP control rooms, command and control systems, and advance emergency operations centers. The HSSL incorporates a reconfigurable control room simulator, which is currently housed in the Center for Advanced Energy Studies (CAES), a joint venture of the DOE and the Idaho University System. The simulator is a platform- and plant-neutral environment intended for full-scope and part-task testing of operator performance in various control room configurations. The simulator is not limited to a particular plant or even simulator architecture. It can support engineering simulator platforms from multiple vendors using digital interfaces. Due to its ability to be reconfigured, it is possible to switch the HSI - not just to digital panels but also to different control modalities such as those using greater plant automation or intelligent alarm filtering. The simulator currently includes three operator workstations, each capable of driving up to eight 30-inch monitors. The size and number of monitors varies depending on the particular front-end simulator deployed for a simulator study. These operator workstations would typically be used for the shift supervisor or senior reactor operator, reactor operator, and assistant reactor operator in current US NPPs. In addition to the three workstations, information can be shared between the workstations and further displayed on a large-screen overview display or a panel mimic. An 82-inch high-definition display is commonly used for the overview display.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

Plant-Level Modeling and Simulation of Used Nuclear Fuel Dissolution

Plant-level modeling and simulation of a used nuclear fuel prototype dissolver is presented. Emphasis is given in developing a modeling and simulation approach to be explored by other processes involved in the recycle of used fuel. The commonality concepts presented in a previous communication were used to create a model and realize its software module. An initial model was established based on a theory of chemical thermomechanical network transport outlined previously. A software module prototype was developed with the required external behavior and internal mathematical structure. Results obtained demonstrate the generality of the design approach and establish an extensible mathematical model with its corresponding software module for a wide range of dissolvers. Scale up numerical tests were made varying the type of used fuel (breeder and light-water reactors) and the capacity of dissolution (0.5 t/d to 1.7 t/d). These tests were motivated by user requirements in the area of nuclear materials safeguards. A computer module written in high-level programing languages (MATLAB and Octave) was developed, tested, and provided as open-source code (MATLAB) for integration into the Separations and Safeguards Performance Model application in development at Sandia National Laboratories. The modeling approach presented here is intended to serve as a template for a rational modeling of all plant-level modules. This will facilitate the practical application of the commonality features underlying the unifying network transport theory proposed recently. In addition, by example, this model describes, explicitly, the needed data from sub-scale models, and logical extensions for future model development. For example, from thermodynamics, an off-line simulation of molecular dynamics could quantify partial molar volumes for the species in the liquid phase; this simulation is currently at reach for high-performance computing. From fluid mechanics, a hold-up capacity function is needed for the dissolver device; this simulation is currently at reach for computational fluid mechanics given the existing CAD geometry. From chemical transport phenomena, a simulation of the particle-scale dissolution front is needed to derive an improved solid dissolution kinetics law by predicting the local surface area change; an example was provided in this report. In addition, the associated reaction mechanisms for dissolution are presently largely untested and simplified, hence even a parallel experimental program in reaction kinetics is needed to support modeling and simulation efforts. Last but not least, a simple account of finite rates of solid feed and transfer can be readily introduced via a coupled delayed model. These are some of the theoretical benefits of a rational plant-level modeling approach which guides the development of smaller length and time scale modeling. Practical, and other theoretical benefits have been presented on a previous report.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗

The ENSO-Australian rainfall teleconnection in reanalysis and CMIP5

Australian rainfall is strongly influenced by the El Niño-Southern Oscillation (ENSO). The relationship between ENSO and rainfall in eastern Australia is non-linear; the magnitude of La Niña events has a greater effect on rainfall than does the magnitude of El Niño events, and the cause of the non-linearity is unclear from previous work. The twentieth century reanalysis succeeds in capturing the asymmetric ENSO-rainfall relationship. In the reanalysis the asymmetry is strongly related to moisture availability in the south-west Pacific whereas wind flow is of less importance. Some global climate models (GCMs) in the CMIP5 archive capture the asymmetric nature of the ENSO-rainfall relationship whilst others do not. Differences in thermodynamic processes and their relationships with ENSO are the primary cause of variability in model performance. Analysis of an atmosphere-only run of a GCM which fails to capture the non-linear ENSO-rainfall relationship is also conducted. The atmospheric run forced by observed SSTs shows no significant improvement in the ENSO-rainfall relationship over the corresponding coupled model run in the CMIP5 archive. This result suggests that some models are failing to capture the atmospheric teleconnection between the tropical Pacific and Australia, and both this and a realistic representation of oceanic ENSO characteristics are required for coupled models to accurately capture the ENSO-rainfall teleconnection. These findings have implications for the study of rainfall projections in the region.

Meteorology & Atmospheric Sciences↗

gLaSDI: Parametric physics-informed greedy latent space dynamics identification

A parametric adaptive physics-informed greedy Latent Space Dynamics Identification (gLaSDI) method is proposed for accurate, efficient, and robust data-driven reduced-order modeling of high-dimensional nonlinear dynamical systems. In the proposed gLaSDI framework, an autoencoder discovers intrinsic nonlinear latent representations of high-dimensional data, while dynamics identification (DI) models capture local latent-space dynamics. Here, an interactive training algorithm is adopted for the autoencoder and local DI models, which enables identification of simple latent-space dynamics and enhances accuracy and efficiency of data-driven reduced-order modeling. To maximize and accelerate the exploration of the parameter space for the optimal model performance, an adaptive greedy sampling algorithm integrated with a physics-informed residual-based error indicator and random-subset evaluation is introduced to search for the optimal training samples on the fly. Further, to exploit local latent-space dynamics captured by the local DI models for an improved modeling accuracy with a minimum number of local DI models in the parameter space, a -nearest neighbor convex interpolation scheme is employed. The effectiveness of the proposed framework is demonstrated by modeling various nonlinear dynamical problems, including Burgers equations, nonlinear heat conduction, and radial advection. The proposed adaptive greedy sampling outperforms the conventional predefined uniform sampling in terms of accuracy. Compared with the high-fidelity models, gLaSDI achieves 17 to 2,658× speed-up with 1 to 5% relative errors.

97 MATHEMATICS AND COMPUTING↗

PV_LIB Toolbox v. 1.3

PV_LIB comprises a library of Matlab? code for modeling photovoltaic (PV) systems. Included are functions to compute solar position and to estimate irradiance in the PV system's plane of array, cell temperature, PV module electrical output, and conversion from DC to AC power. Also included are functions that aid in determining parameters for module performance models from module characterization testing. PV_LIB is open source code primarily intended for research and academic purposes. All algorithms are documented in openly available literature with the appropriate references included in comments within the code.

Source record↗

DME-Propane Ignition Delay Time Measurements at Mixing Controlled Compression Ignition Engine-Relevant Conditions

The blend of dimethyl ether (DME, CH 3 OCH 3 ) and propane (C 3 H 8 ) is a potentially renewable fuel mixture that has the potential to replace diesel in compression ignition engines. The combination can potentially reduce particulate and greenhouse gas emissions compared to a conventional diesel engine operating under similar conditions. However, detailed conceptual and simulation studies must be conducted before adopting a new fuel on a compression ignition engine. For these simulations, accurate chemical kinetic models are necessary. However, the validity of chemical kinetic mechanisms in the literature is unknown for mixing controlled compression ignition (MCCI) engine operating conditions. Hence, in this work, we studied the ignition of dimethyl ether (DME) and propane blends in a shock tube at MCCI engine conditions. Ignition delay time (IDT) data was collected behind the reflected shock for DME-propane mixtures for heavy-duty compression ignition (CI) engine parameters. Undiluted experiments spanning temperatures of 700 to 1100 K and pressures of 55 to 84 bar for various blends (100% CH 3 OCH 3 / 0% C 3 H 8, 100% C 3 H 8 / 0% CH 3 OCH 3 , 60% CH 3 OCH 3 / 40% C 3 H 8 ) of DME and propane were combusted in synthetic air (21% O 2 / 79% N 2 ). Some experiments were conducted at higher pressures (90-120 bar) to understand model performance at these conditions. Comparisons of IDT were made with the predictions of recent chemical kinetic mechanisms for DME-propane mixture, including the Aramco3.0, NUIG, and Dames et al. mechanisms. All mechanisms overpredicted IDT compared to experimental values. Sensitivity analysis was conducted with Dames et al. model, and critical reactions sensitive to IDT of DME-propane mixture near 100 bar are outlined.

Mohammed, Zuhayr Pasha↗

PV_LIB Toolbox v. 1.3

PV_LIB comprises a library of Matlab? code for modeling photovoltaic (PV) systems. Included are functions to compute solar position and to estimate irradiance in the PV system's plane of array, cell temperature, PV module electrical output, and conversion from DC to AC power. Also included are functions that aid in determining parameters for module performance models from module characterization testing. PV_LIB is open source code primarily intended for research and academic purposes. All algorithms are documented in openly available literature with the appropriate references included in comments within the code.

Source record↗

DOE Advanced Gas Reactor Fuel Development and Qualification Program Overview

Slides discussing the DOE Advanced Gas Reactor Fuel Development and Qualification Program Overview to include focus on LEU UCO TRISO fuel in cylindrical compacts, with data for fuel qualification in support of reactor licensing and a path to establish a domestic commercial TRISO fuel fabrication capability. Timelines from 2004 through 2029. Major program highlights include: Developed fuel performance models based on past observed behavior, Study to understand differences between US and German (“good”) fuel, Developed and improved fuel fabrication and characterization methods, Fabricated lab-scale fuel (AGR-1) with very high quality; excellent in-pile and post-irradiation high-temperature fuel performance, Scale-up coating fabrication (AGR-2) with excellent in-pile performance, AGR-3/4 fuel fabrication and irradiation experiment to assess FP transport behavior, Scale-up matrix fabrication (AGR-5/6/7); & fuel performance evaluation in progress.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗

Implementation of a Pressure Correction Term in the CTF Subchannel Code

CTF is a modernized and improved version of the legacy subchannel code COBRA-TF, being jointly developed and maintained by Oak Ridge National Laboratory (ORNL) and North Carolina State University. The code was adopted by ORNL for use in the Consortium for Advanced Simulation of Light Water Reactors (CASL) in 2012 to help address CASL challenge problems. Since that time, activities related to CTF have included implementing software quality assurance measures, implementing new features and models, performing validation and verification testing, establishing and supporting a CTF user group, and developing the code for use in coupled applications. CTF is applicable to single- and two-phase flows in light water reactor geometries at normal and accident operating conditions. The two-phase flow model consists of the six-equation two-phase flow model augmented by equations for the droplet field, and it also includes appropriate source terms. This work focuses on the numerical stability of the six-equation two-phase flow model and is part of a recent effort to improve CTF capabilities for simulating flows in boiling water reactors (BWRs). The six-equation two-phase flow model is not strongly hyperbolic, meaning that some of its eigenvalues are either not real or not distinct. In a Cauchy sense, and for an initial-value problem, strong hyperbolicity was found to be equivalent to well-posedness, which ensures the uniqueness of the solution in the space-time used. This concept is also of importance when designing high-order numerical methods and also when implementing boundary conditions that rely on real eigenvalues such as Riemann solvers. A tremendous amount of research is available in the literature on how to hyperbolize the six-equation two-phase flow model by adding ad hoc terms to the system of equations. Such terms are often referred to as virtual mass or interfacial pressure correction terms. From a mathematical perspective, this approach adds terms to the phasic momentum equation containing first-order partial derivatives of the unknowns to modify the Jacobian matrix, thus the imaginary part of the eigenvalues. For this work, an interfacial pressure correction (IPC) term on the same model as what was used for the CATHARE system code, is implemented in CTF. This allows for recovery of a two-pressure formulation. In the following, the six-equation two-phase flow model is recalled, along with the IPC term and its influence on the eigenvalues. Then, numerical results and convergence studies are presented for a water-faucet problem at a pressure of 10 bar. Finally, conclusions and future work are addressed.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

End-to-End Models for Effects of System Noise on LIMS Analysis of Igneous Rocks

The ChemCam instrument on the Mars Science Laboratory will be the first extraterrestial deployment of laser-induced breakdown spectroscopy (UBS) for remote geochemical analysis. LIBS instruments are also being proposed for future NASA missions. In quantitative LIBS applications using multivariate analysis techniques, it is essential to understand the effects of key instrument parameters and their variability on the elemental predictions. Baseline experiments were run on a laboratory instrument in conditions reproducing ChemCam performance on Mars. These experiments employed Nd:YAG laser producing 17 mJ/pulse on target and an with a 200 {micro}m FWHM spot size on the surface of a sample. The emission is collected by a telescope, imaged on a fiber optic and then interfaced to a demultiplexer capable of >40% transmission into each spectrometer. We report here on an integrated end-to-end system performance model that simulates the effects of output signal degradation that might result from the input signal chain and the impact on multivariate model predictions. There are two approaches to modifying signal to noise (SNR): degrade the signal and/or increase the noise. Ishibashi used a much smaller data set to show that the addition of noise had significant impact while degradation of spectral resolution had much less impact on accuracy and precision. Here, we specifically focus on aspects of remote LIBS instrument performance as they relate to various types of signal degradation. To assess the sensitivity of LIBS analysis to signal-to-noise ratio (SNR) and spectral resolution, the signal in each spectrum from a suite of 50 laboratory spectra of igneous rocks was variably degraded by increasing the peak widths (simulating misalignment) and decreasing the spectral amplitude (simulating decreases in SNR).

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

The coastal streamflow flux in the Regional Arctic System Model

The coastal streamflow flux from the Arctic drainage basin is an important driver of dynamics in the coupled ice-ocean system. Comprising more than one third of the total freshwater flux into the Arctic Ocean, streamflow is a key component of the regional and global freshwater cycle. To better represent the coupling of the streamflow flux to the ocean, we have developed and applied the RVIC streamflow routing model within the Regional Arctic System Model (RASM). The RASM is a high-resolution regional Earth System Model whose domain includes all of the Arctic drainage basin. Here in this paper, we introduce the RVIC streamflow routing model, detailing its application within RASM and its advancements in terms of representing high-resolution streamflow processes. We evaluate model simulated streamflow relative to in situ observations and demonstrate a method for improving model performance using a simple optimization procedure. We also present a new, spatially and temporally consistent, high-resolution data set of coastal freshwater fluxes for the Arctic drainage basin and surrounding areas that is based on a fully coupled RASM simulation and intended for use in Arctic Ocean modeling applications. This data set is evaluated relative to other coastal streamflow data sets commonly used by the ocean modeling community. We demonstrate that the RASM-simulated streamflow flux better represents the annual cycle than existing data sets, especially in ungauged areas. Finally, we assess the impact that streamflow has on the coupled ice-ocean system, finding that the presence of streamflow leads to reduced sea surface salinity, increased sea surface temperatures, and decreased sea ice thickness.

54 ENVIRONMENTAL SCIENCES↗

An efficient mixed-precision, hybrid CPU-GPU implementation of a nonlinearly implicit one-dimensional particle-in-cell algorithm

Recently, a fully implicit, energy- and charge-conserving particle-in-cell method has been developed for multi-scale, full-f kinetic simulations [G. Chen, et al., J. Comput. Phys. 230, 18 (2011)]. The method employs a Jacobian-free Newton-Krylov (JFNK) solver and is capable of using very large timesteps without loss of numerical stability or accuracy. A fundamental feature of the method is the segregation of particle orbit integrations from the field solver, while remaining fully self-consistent. This provides great flexibility, and dramatically improves the solver efficiency by reducing the degrees of freedom of the associated nonlinear system. However, it requires a particle push per nonlinear residual evaluation, which makes the particle push the most time-consuming operation in the algorithm. This paper describes a very efficient mixed-precision, hybrid CPU-GPU implementation of the implicit PIC algorithm. The JFNK solver is kept on the CPU (in double precision), while the inherent data parallelism of the particle mover is exploited by implementing it in single-precision on a graphics processing unit (GPU) using CUDA. Performance-oriented optimizations, with the aid of an analytical performance model, the roofline model, are employed. Despite being highly dynamic, the adaptive, charge-conserving particle mover algorithm achieves up to 300 400 GOp/s (including single-precision floating-point, integer, and logic operations) on a Nvidia GeForce GTX580, corresponding to 20 25% absolute GPU efficiency (against the peak theoretical performance) and 50-70% intrinsic efficiency (against the algorithm s maximum operational throughput, which neglects all latencies). This is about 200-300 times faster than an equivalent serial CPU implementation. When the single-precision GPU particle mover is combined with a double-precision CPU JFNK field solver, overall performance gains 100 vs. the double-precision CPU-only serial version are obtained, with no apparent loss of robustness or accuracy when applied to a challenging long-time scale ion acoustic wave simulation.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗