Continuous-Energy Monte Carlo on GPUs in Shift
Explore the source record for details and available documents.
Hamilton, Steven P.↗
Engineering topics
Hamilton, Steven P.
Publications and source records attributed to Hamilton, Steven P..
Deterministically estimated fission source distributions for Monte Carlo k-eigenvalue problems
The standard Monte Carlo (MC) k-eigenvalue algorithm involves iteratively converging the fission source distribution using a series of potentially time-consuming inactive cycles before quantities of interest can be tallied. One strategy for reducing the computational time requirements of these inactive cycles is the Sourcerer method, in which a deterministic eigenvalue calculation is performed to obtain an improved initial guess for the fission source distribution. This method has been implemented in the Exnihilo software suite within SCALE using the SPNSPN or SNSN solvers in Denovo and the Shift MC code. The efficacy of this method is assessed with different Denovo solution parameters for a series of typical k-eigenvalue problems including small criticality benchmarks, full-core reactors, and a fuel cask. Here it is found that, in most cases, when a large number of histories per cycle are required to obtain a detailed flux distribution, the Sourcerer method can be used to reduce the computational time requirements of the inactive cycles.
73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
Multigroup Monte Carlo on GPUs: Comparison of history- and event-based algorithms
This article presents an investigation of the performance of different multigroup Monte Carlo transport algorithms on GPUs with a discussion of both history-based and event-based approaches. Several algorithmic improvements are introduced for both approaches. By modifying the history-based algorithm that is traditionally favored in CPU-based MC codes to occasionally filter out dead particles to reduce thread divergence, performance exceeds that of either the pure history-based or event-based approaches. The impacts of several algorithmic choices are discussed, including performance studies on Kepler and Pascal generation NVIDIA GPUs for fixed source and eigenvalue calculations. Single-device performance equivalent to 20–40 CPU cores on the K40 GPU and 60–80 CPU cores on the P100 GPU is achieved. Last, in addition, nearly perfect multi-device parallel weak scaling is demonstrated on more than 16,000 nodes of the Titan supercomputer.
97 MATHEMATICS AND COMPUTING↗
Variance Estimation in Monte Carlo Eigenvalue Simulations Using Spectral Analysis Method
The Monte Carlo (MC) method is the 'gold standard' for performing nuclear reactor calculations and criticality safety analyses because of its ability to handle complex geometries and physics. Reactor and criticality safety analyses require solving eigenvalue problems in complex nuclear systems. Once a stationary fission source distribution is obtained in MC simulations, the sample mean of many stationary cycles is calculated. The MC simulation sample mean is typically referred to as a tally. Variance, or standard deviation, of the sample mean/tally is needed to indicate the level of statistical uncertainty in the simulation and to understand the convergence of the sample mean. Current MC codes typically use sample variance to estimate the statistical uncertainty of the simulation and assume that the MC stationary cycles are independent. However, it is known that cycles are correlated, and estimators of the variance that ignore these correlations systematically underestimate the variance. In this paper, we propose a novel method in the Fourier frequency domain to estimate the variance of the sample mean in MC eigenvalue calculations. Results indicate that the proposed estimator can be unbiased compared to the real variance. The proposed estimator requires minimal additional computing effort.
42 ENGINEERING↗
Use of Generalized Davidson Eigenvalue Solver for Coarse Mesh Finite Difference Acceleration
This paper summarizes improvements in the coarse mesh finite difference (CMFD) performance by focusing on the eigenvalue solution methodology. The current solution technique uses a shifted power iteration with the PETSc solver package. A generalized Davidson methodology has been implemented into MPACT using the Trilinos solver package. The new generalized Davidson solver shows significant improvements in the overall runtime for the CMFD acceleration scheme and the overall MPACT performance. The performance improves between 5 and 27 times for the CMFD runtime. This results in a speedup of between 2.5 to 5 times on the overall solution time for MPACT.
Collins, Benjamin↗
Analysis of Monte Carlo accelerated iterative methods for sparse linear systems: Analysis of Monte Carlo accelerated iterative methods for sparse linear systems
Here, we consider hybrid deterministic-stochastic iterative algorithms for the solution of large, sparse linear systems. Starting from a convergent splitting of the coefficient matrix, we analyze various types of Monte Carlo acceleration schemes applied to the original preconditioned Richardson (stationary) iteration. We expect that these methods will have considerable potential for resiliency to faults when implemented on massively parallel machines. We also establish sufficient conditions for the convergence of the hybrid schemes, and we investigate different types of preconditioners including sparse approximate inverses. Numerical experiments on linear systems arising from the discretization of partial differential equations are presented.
97 MATHEMATICS AND COMPUTING↗
Demonstrate VERA Core Simulator Performance Improvements L2:PHI.P13.03
This report describes the performance improvements made to the VERA Core Simulator (VERA-CS) during FY2016. The development of the VERA Core Simulator has focused on the capability needed to deplete physical reactors and help solve various problems; this capability required the accurate simulation of many operating cycles of a nuclear power plant. The first section of this report introduces two test problems used to assess the run-time performance of VERA-CS using a source dated February 2016. The next section provides a brief overview of the major modifications made to decrease the computational cost. Following the descriptions of the major improvements, the run-time for each improvement is shown. Conclusions on the work are presented, and further follow-on performance improvements are suggested.
21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗
GPU Acceleration of History-Based Multigroup Monte Carlo
Accurate solutions of the radiation transport equation play an important role in the analysis of many nuclear systems. Because there is no need to discretize the problem, Monte Carlo methods offer the most accurate possible solutions to the transport equation. However, significant computational effort is required to reduce the stochastic noise inherent to Monte Carlo methods. Current trends in high performance computing are moving towards vectorized, single instruction multiple data (SIMD), architectures such as Intel Xeon Phis and graphics processing units (GPUs) due to their favorable ratio of performance to power consumption. The challenge of adapting Monte Carlo transport to vectorized computing architectures was first addressed by Brown with an algorithm known as event-based Monte Carlo. In this approach, rather than simulating individual particle histories from birth to death, groups of particles are processed one event at a time. This allows each event batch to perform the same operations in lockstep, enabling use of the vector hardware. Although modern vectorized hardware is more versatile than the early vector machines, reducing so-called thread divergence is nonetheless still an important consideration for algorithm design. Many of the algorithms proposed so far for Monte Carlo transport on advanced architectures have focused on variants of event-based transport. Some evidence, however, suggests that low-level thread divergence, resulting from short-lived branching statements, may not be detrimental to overall performance on GPUs as long as the effect on achievable memory bandwidth is considered. This paper characterizes the performance of a traditional history-based Monte Carlo algorithm on GPUs. We show that with only minor modifications to traditional algorithms, significant performance gains relative to CPUs are possible. (authors)
73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
Accelerated Monte Carlo Fission Source Convergence with Fission Matrix and Kernel Density Estimators
Accurate solution of the k-eigenvalue form of the Boltzmann radiation transport equation is required in many application areas, including nuclear reactor analysis and criticality safety. Monte Carlo calculations are free of the discretization errors inherent to deterministic methods and therefore offer the most accurate solutions. However, converging the fission source can be difficult for many problems, resulting in either a substantial computational overhead or potential inaccuracies that negate many of the benefits of using Monte Carlo methods. Difficulty in converging the fission source occurs for two distinct reasons. The first is the existence of a high dominance ratio in the problem (i.e. the largest and second largest eigenvalues have approximately the same magnitude). The second difficulty arises due to undersampling of the fission source because an insufficient number of particle histories are simulated. The first difficulty is shared by deterministic solvers when using power iteration, while the second difficulty is unique to Monte Carlo solvers. In practice, both difficulties may be present in the same problem, compounding the difficulty in achieving a converged distribution. In this paper, we consider a new approach to accelerate the convergence of the fission source based on combining the use of fission matrix methods with kernel density estimators (KDEs). It has been observed that deterministic fission matrix methods show the potential to accelerate the convergence of the fission source in Monte Carlo criticality calculations but may introduce significant additional noise for difficult problems. In this paper, we have demonstrated the use of a KDE approach in combination with the fission matrix acceleration. Although the addition of the KDE initially slows the convergence of the fission matrix acceleration, it decreases the magnitude of the noise while still offering significantly improved convergence relative to analog Monte Carlo. (authors)
73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
Hot zero power reactor calculations using the Insilico code
In this paper we describe the reactor physics simulation capabilities of the Insilico code. A description of the various capabilities of the code is provided, including detailed discussion of the geometry, meshing, cross section processing, and neutron transport options. Numerical results demonstrate that Insilico using an SP{sub N} solver with pin-homogenized cross section generation is capable of delivering highly accurate full-core simulation of various pressurized water reactor problems. Comparison to both Monte Carlo calculations and measured plant data is provided.
71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
Performance Improvements for Coarse Mesh Finite Difference Acceleration L3:RTM.PRT.P13.02
The development of VERA-CS in recent years has focused on developing the capability to simulate multiple cycles of operating commercial nuclear power plants. Now that these capabilities have advanced to the point where it is being deployed to users, the focus is on improving the computational performance of various components in VERA-CS. In this work, the focus is on the Coarse Mesh Finite Difference (CMFD) solution in MPACT. CMFD serves multiple purposes in the 2D/1D solution methodology. First, it is a natural mechanism to tie together the radial MOC transport and the axial SP3 solution. Because the CMFD system solves the multigroup three-dimensional core in one system, it pulls together the global response of the system. In addition, the CMFD solution provides a framework to accelerate the convergence of the eigenvalue problem.
21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗
Implementation, capabilities, and benchmarking of Shift, a massively parallel Monte Carlo radiation transport code
This paper discusses the implementation, capabilities, and validation of Shift, a massively parallel Monte Carlo radiation transport package developed and maintained at Oak Ridge National Laboratory. It has been developed to scale well from laptop to small computing clusters to advanced supercomputers. Special features of Shift include hybrid capabilities for variance reduction such as CADIS and FW-CADIS, and advanced parallel decomposition and tally methods optimized for scalability on supercomputing architectures. Shift has been validated and verified against various reactor physics benchmarks and compares well to other state-of-the-art Monte Carlo radiation transport codes such as MCNP5, CE KENO-VI, and OpenMC. Some specific benchmarks used for verification and validation include the CASL VERA criticality test suite and several Westinghouse AP1000 ® problems. These benchmark and scaling studies show promising results.
22 GENERAL STUDIES OF NUCLEAR REACTORS↗
A parallel multi-domain solution methodology applied to nonlinear thermal transport problems in nuclear fuel pins
This paper describes an efficient and nonlinearly consistent parallel solution methodology for solving coupled nonlinear thermal transport problems that occur in nuclear reactor applications over hundreds of individual 3D physical subdomains. Efficiency is obtained by leveraging knowledge of the physical domains, the physics on individual domains, and the couplings between them for preconditioning within a Jacobian Free Newton Krylov method. Details of the computational infrastructure that enabled this work, namely the open source Advanced Multi-Physics (AMP) package developed by the authors is described. Details of verification and validation experiments, and parallel performance analysis in weak and strong scaling studies demonstrating the achieved efficiency of the algorithm are presented. Furthermore, numerical experiments demonstrate that the preconditioner developed is independent of the number of fuel subdomains in a fuel rod, which is particularly important when simulating different types of fuel rods. Finally, we demonstrate the power of the coupling methodology by considering problems with couplings between surface and volume physics and coupling of nonlinear thermal transport in fuel rods to an external radiation transport code.
71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
Efficient solution of the simplified P N equations
We show new solver strategies for the multigroup SPN equations for nuclear reactor analysis. By forming the complete matrix over space, moments, and energy a robust set of solution strategies may be applied. Moreover, power iteration, shifted power iteration, Rayleigh quotient iteration, Arnoldi's method, and a generalized Davidson method, each using algebraic and physics-based multigrid preconditioners, have been compared on C5G7 MOX test problem as well as an operational PWR model. These results show that the most ecient approach is the generalized Davidson method, that is 30-40 times faster than traditional power iteration and 6-10 times faster than Arnoldi's method.
97 MATHEMATICS AND COMPUTING↗
Identification of Leakage Spectra from Neutron Activation in Glass/Cherenkov Detectors - Paper 96
Oak Ridge National Laboratory has developed a new detector system concept that can be deployed in major US cities to identify the leakage spectrum of a nuclear weapon detonation. The detector is based on a small piece of glass that has been doped with an element that has a high activation cross section and creates a radioactive activation product with a reasonable length half-life. The decay of the activation product(s) after exposure creates Cherenkov light which can propagate to a photomultiplier and can be turned into a count-rate signal. The signal from each glass in each detector in a city can be individually analyzed to determine the amount of activation products produced in the initial exposure. Several activation products can be determined in a single glass if the half-lives are not the same. With a known source location and a detailed 3D model of a city, neutron transport simulations can be made by using a library of known sources to predict the activation amounts each source would have made in the glass detectors. A simple comparison of the measured activation amounts and the predicted activation amounts can identify the source spectrum contained in the library that is the best match. Choosing dopants with a large range in the threshold energy for the activation reactions can help to differentiate the spectra in the library. The method of matching the measured activation amounts to the predicted activation amounts to identify the source spectrum is similar to the analysis used for the passive Activation Foil Integrated Detector System (AFIDS), also developed at Oak Ridge National Laboratory. In AFIDS, wires or foils of known mass and isotopic composition are deployed before and then collected after exposure. Counting gammas in a nearby laboratory determines the activation amounts. The glass/Cherenkov system is active - it uses a battery-powered detector to count flashes of Cherenkov light and to communicate its count rates to a remote location for further analysis. Work has been performed in several areas, including glass making, automated detection and communication systems and software tools for analysis of the signal and source spectrum identification. Bell and Boatner have developed phosphate glasses that incorporate elements with high capture cross sections, and Hayward has simulated the response with Geant4 and measured the response with isotopic gamma sources. This paper will focus on the software tools for determining the activation amounts from the Cherenkov count-rate signal and finding the best matching source spectrum. A test problem in an urban environment is used to demonstrate the potential of a spectrum identification system using neutron activation. The analysis methods developed to identify the neutron leakage spectra from Cherenkov light given off by neutron activation in doped glasses work well. The calculation used to unfold the count-rate information into the number of actual activation products uses robust algorithms that can also compute associated uncertainties. These uncertainties and the known uncertainties of the glass tube material compositions are incorporated into the matching analysis to determine which of the source spectra in the library best matches the measured activation data. Any measurement method used to identify a neutron source spectrum relies at some point on a comparison against known sources; no identification method is definitive. Detection systems measure some physical aspect of the source, not its serial number. Many systems measure the physical aspect to determine that the source has some property, and then leave it to experts to determine which potential sources share that property. The analysis system presented here incorporates the comparison into the analysis itself. The statistical tests show which of the known library sources match the measured data the best and whether that match is a good match. If the best match still does not fit the measured data well (a very high χ{sup 2} value), that indicates the true source is not one of the sources in the library. This is still valuable information for decision makers. (authors)
46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
A Parallel Multi-Domain Solution Methodology Applied to Nonlinear Thermal Transport Problems in Nuclear Fuel Pins
This paper describes an efficient and nonlinearly consistent parallel solution methodology for solving coupled nonlinear thermal transport problems that occur in nuclear reactor applications over hundreds of individual 3D physical subdomains. Efficiency is obtained by leveraging knowledge of the physical domains, the physics on individual domains, and the couplings between them for preconditioning within a Jacobian Free Newton Krylov method. Details of the computational infrastructure that enabled this work, namely the open source Advanced Multi-Physics (AMP) package developed by the authors is described. Details of verification and validation experiments, and parallel performance analysis in weak and strong scaling studies demonstrating the achieved efficiency of the algorithm are presented. Furthermore, numerical experiments demonstrate that the preconditioner developed is independent of the number of fuel subdomains in a fuel rod, which is particularly important when simulating different types of fuel rods. Finally, we demonstrate the power of the coupling methodology by considering problems with couplings between surface and volume physics and coupling of nonlinear thermal transport in fuel rods to an external radiation transport code.
11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
Operational Reactor Model Demonstrations with VERA: Watts Bar Unit 1 Cycle 1 Zero Power Physics Tests (Revision 1)
The Consortium for Advanced Simulation of Light Water Reactors (CASL) is developing a collection of methods and software products known as VERA, the Virtual Environment for Reactor Applications. VERA will provide coupled, high fidelity nuclear reactor analysis capabilities in multiple areas of physics such as neutronics, thermal-hydraulics, fuel mechanics, and others. A key milestone for this endeavor is to validate VERA against measurements from operating nuclear power reactors. This document demonstrates the capabilities of VERA to accurately simulate the initial startup physics tests of Watts Bar Nuclear Power Station, Unit 1 (WBN1), demonstrating the many achievements that have been made by the CASL partners over the past year.
21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗