Overview of Serpent related activities at HZDR E. Fridman

Outline “Typical” applications Generation of few-group constants

EU/FP7 FREYA FREYA - Fast Reactor Experiments for hYbrid Applications Support for design and licensing of MYRRHA/FASTEF Modeling of MYRRHA and ALFRED critical mock-ups VENUS-F facility at SCK•CEN, Mol

Very detailed VENUS-F facility data and model V&V of Serpent against MCNP and experimental data k-eff, β-eff, control rod worth Radial and axial traverses Detector responses and spectral indices (e.g. F8/F5)

Serpent vs. MCNP: Integral parameters Difference value error k-eff 1.00485 0.00003 1.00498 0.00002 13 (pcm) Gen. time, sec 4.018E-07 0.2% 4.010E-07 0.1% -0.2% Beta-eff, pcm 728 2 731 1 3 (pcm) Performance Executed using 4 MPI tasks with 16 OpenMP threads Serpent runs 9.3 times faster than MCNP

Serpent vs. MCNP: difference in flux spectra

EU/FP7 ESNII+ Modeling of ASTRID SFR Evaluation of neutronic parameters and safety coefficients Code-to-code benchmarks Radial layout Axial layout

OECD/NEA: SFR Benchmark Task Force Analysis of feedback behavior of next gen. of SFRs Code-to-code benchmarks Large oxide SFR core

Common issues Nuclear data uncertainties Δρ 500 – 600 pcm between JEFF-3.1 and ENDF/B-VII For both ASTRID and large oxide SFR Uncertainties propagation to low-worth reactivity coefficients

Common issues Nuclear data uncertainties Δρ 500 – 600 pcm between JEFF-3.1 and ENDF/B-VII For both ASTRID and large oxide SFR Uncertainties propagation to low-worth reactivity coefficients Example of Serpent setup for ASTRID analysis: 200 skipped + 1k active cycles + 500k neutrons= 0.5×109 histories σk-eff = ±3 pcm (0.003%) Q: Is it enough to get “good” reactivity coefficients?

Uncertainty in reactivity coeff.: ASTRID core

Proposal for follow-up benchmark Benchmark for uncertainty analysis in modelling (UAM) Within OECD/NEA WPRS Evaluation of uncertainties on reactivity feedback coefficients Analyses of principal unprotected transients Using the results from the previous step First SFR-UAM meeting is planned for 21-22.5 (UPM, Madrid) In conjunction with other OECD/NEA workshops

Generation of SFR few-group constants

Generation of SFR few-group constants MC is too expensive for full-scale reactor calculations Neutronics + TH + BU + kinetics Two-step procedure still dominates reactor analysis Deterministic 2D lattice codes  homogenized constants Deterministic 3D coarse mesh core simulators Increasing interest in using MC for homogenization Improved computer performance Flexibility - not limited to any particular technology Especially useful for the modeling of innovative reactor concepts

Application example: large oxide SFR core OECD/NEA SFR Benchmark Fuel: 225 inner and 228 outer subassemblies Control: 18 primary and 9 secondary subassemblies

Verification of XS generation methodology 3D full core calculations At BOL DYN3D Multi-group diffusion solution Serpent: Reference solution Few-group XS for DYN3D Compared parameters: k-eff Doppler constant Coolant void reactivity Control rod worth Radial power distribution

3D single-assembly model for fuel assemblies Reflective radial and black axial boundary conditions

2D super-cell models for non-multiplying regions Control rods, control rod channels, reflectors, etc

Results: integral core parameters Serpent, pcm DYN3D vs. Serpent, pcm Reactivity - CR out 1059 -128 Reactivity - CR in -4988 -255 Total CR worth -6046 -127 Doppler constant -852 -15 Na void reactivity 1864 87

Results: integral core parameters Serpent, pcm DYN3D vs. Serpent, pcm Reactivity - CR out 1059 -128 Reactivity - CR in -4988 -255 Total CR worth -6046 -127 Doppler constant -852 -15 Na void reactivity 1864 87

SPH factors for CRs and CR channels SPH = Super-homogenization Typically used for pin-wise homogenization in LWRs Requires equivalent transport/diffusion model SPH = μ = ϕSerpent ϕDYN3D and Σ* = μ*·Σ For every homogenized region and energy group Q: How to make equivalent model for hexagonal supercell?

SPH factors for CRs and CR channels SPH = Super-homogenization Typically used for pin-wise homogenization in LWRs Requires equivalent transport/diffusion model SPH = μ = ϕSerpent ϕDYN3D and Σ* = μ*·Σ For every homogenized region and energy group Q: How to make equivalent model for hexagonal supercell?

SPH factors for CRs and CR channels DYN3D has a trigonal diffusion solver SPH were generated as follows: → → Serpent model Homogenized regions Σ and ϕSerpent DYN3D Δ diffusion ϕDYN3D

Results: integral core parameters Serpent, pcm DYN3D vs. Serpent, pcm Reactivity - CR out 1059 -128 Reactivity - CR in -4988 -255 Total CR worth -6046 -127

Results: integral core parameters Serpent, pcm DYN3D vs. Serpent, pcm DYN3D + SPH vs. Serpent, pcm Reactivity - CR out 1059 -128 -64 Reactivity - CR in -4988 -255 -107 Total CR worth -6046 -127 -43

Ring-wise power distribution CR out CR in w/o SPH: max. diff. 0.6% w/o SPH: max. diff. 4.8%

Ring-wise power distribution CR out CR in w/o SPH: max. diff. 0.6% w/ SPH: max. diff. 0.4% w/o SPH: max. diff. 4.8% w/ SPH: max. diff. 2.1%

Assembly-wise power distribution

Re-homogenization with Serpent

Thank you!