Consortium for Advanced Simulation of LWRs

CASL-U-2015-0247-000

The OpenMC Monte Carlo Particle

Transport Code

Pablo Ducru, Jon Walsh Will Boyd, Sam Shaner, Sterling Harper, Colin Josey, Matthew Ellis, Nich Horelik,

Benoit Forget, Kord Smith Massachusetts Institute of Technology

Bryan Herman

Knolls Atomic Power Laboratory

Paul Romano Argonne National Laboratory

July 7, 2015

CASL-U-2015-0247-000

The OpenMC Monte Carlo Particle Transport CodePablo Ducru1, Jon Walsh1, Will Boyd1, Sam Shaner1, Sterling Harper1, Colin Josey1,

Matthew Ellis1, Nick Horelik1, Bryan Herman2, Paul Romano3, Benoit Forget,1 and Kord Smith1

1Massachusetts Institute of Technology, Department of Nuclear Science and Engineering

2Knolls Atomic Power Laboratory

3Argonne National Laboratory

Resolved Resonance Region

IntroductionAn open source Monte Carlo code called OpenMC is beingdeveloped at MIT to address shortcomings in legacy Monte Carlo codes. OpenMC was developed with a focus on [1]:• High performance scalable algorithms• Modern software design practices• Optimization for reactor design and analysis

Doppler Broadening

Windowed Multipole Algorithm

• On-the-fly Doppler broadening of cross sections using the windowed multipole algorithm is implemented in OpenMC [4]

• For each energy, the poles and residues that will have a strong influence on the cross section value are selected and Doppler broadened, while the rest is curve-fitted as a background term and treated with a developped recurrence algorithm.

Multiphysics

• To incorporate more detailed feedback, OpenMC has been coupled to the Multiphysics Object-Oriented Simulation Environment (MOOSE)

• Functional Expansion Tallies (FETs) have been implemented in OpenMC to provide detailed power distributions to unstructured finite elements [8]

Resonance Scattering

References

Domain Decomposition

OpenMC has exact treatments of the energy and temperature dependencies of free gas elastic scattering kernels:• Doppler broadening rejection correction (DBRC)

Fig: (top) Visualization of multipole.(bottom) U-238 Doppler broadened cross section at 300K via windowed multipole method compared to BROADR values• Accelerated resonance elastic

scattering kernel sampling (ARES)• ARES reduces the runtime

overhead with DBRC by 30-90% in thermal reactor simulations [6] Fig: Free gas elastic scattering kernels

• Domain decomposition algorithms have been explored to address per-node memory limitations that preclude high-fidelity LWR simulations [9]. Successful decomposition of both material and tally data has been demonstrated on BlueGene/Q.

Fig: Assembly power distribution generated in OpenMC and used in MOOSE

[1] P. K. Romano and B. Forget, “The OpenMC Monte Carlo particle transport code,” Ann. Nucl. Energy 51, pp. 274–281 (2013). [2] L. Abel and W. Boyd, “Interactive Visualization of Nuclear Reactor Geometries in OpenCG,” Proc. of the ANS Stud. Conf, College Station, TX, USA, 2015.[3] W. Boyd, B. Forget and K. Smith. “OpenCG: A Combinatorial Geometry Modeling Tool for Data Processing and Code Verification,” Proc. of Int’l Conf. on Math. and Comp., Nashville, TN, 2015.[4] C. Josey, P. Ducru, B. Forget, and K. Smith, “Windowed Multipole for Cross Section Doppler Broadening,” Journal of Computational Physics, Submitted. (2014).[5] J. A. Walsh, B. C. Kiedrowski, B. Forget, K. S. Smith, and F. B. Brown, “Direct, On-the-fly Calculation of Unresolved Resonance Region Cross Sections in Monte Carlo Simulations,” Proc. of Int’l Conf. on Math. and Comp., Nashville, TN, 2015.[6] J. A. Walsh, B. Forget, and K. S. Smith, “Accelerated sampling of the free gas resonance elastic scattering kernel,” Ann. Nucl. Energy (2014).[7] B. R. Herman, B. Forget, and K. Smith, “Progress toward Monte Carlo–thermal hydraulic coupling using low-order nonlinear diffusion acceleration methods,” Annals of Nuclear Energy (2014).[8] M. Ellis, D. Gaston, B. Forget and K. Smith. “Preliminary Coupling of the Monte Carlo Code OpenMC and the Multiphysics Object-Oriented Simulation Environment (MOOSE) For Analyzing Doppler Feedback in Monte Carlo Simulations,” Proc. of Int’l Conf. on Math. and Comp., Nashville, TN, 2015.[9] N. Horelik, A. Siegel, B. Forget, K. Smith. “Monte Carlo domain decomposition for robust nuclear reactor analysis.” Parallel Computing, 40, pp. 646-660 (2014).

Fig: Load imbalance penalty with different load balancing strategies

Fig: Timing results for full-core analog fission tallies. 10 Batches (5 active) with 5.12 million particles per batch (512 nodes)

• Incorporating multiphysics feedback was initially investigated using the low-order CMFD operator in OpenMC [7]

Multipole Representation

• The conversion of resonance parameters into a set of poles and residues has been implemented for both Multi-Level Breit-Wigner and Reich-Moore formalisms.

• In Multipole representation, the free-gaz model Doppler broadening kernel yields analytical expressions for cross sections temperature dependance.

Unresolved Resonance Region

OpenMC is being used to develop nuclear data processing methods in the unresolved resonance energy region (URR) [5]:• Oon-the-fly generation of continuous-energy, temperature-

dependent cross sections• Excellent agreement between the OTF and probability table cross

section generation methods• Models resonance structure of inelastic scattering cross sections

Fig: Realizations of URR resonance structure

• The windowed multipole algorithm greatly outperforms other Doppler broadening methods in the resolved resonance region.

• For neutral particles, angle-integrated cross section formulas from quantum R-matrix theory can be represented in an equivalent Multipole representation.An emphasis on certain features of OpenMC’s nuclear data treatment

and parallelization and acceleration work is here presented:• Exact, continuous energy cross section treatment.• Domain decomposition and CMFD acceleration.

OpenMC is available at: https://github.com/mit-crpg/openmc

CASL-U-2015-0247-000