Application of ADVANTG to the JET3 – NEXP streaming benchmark experiment

• Published in: Fusion engineering and design Vol. 147; p. 111252
• Main Authors: Kos, B., Mosher, S.W., Kodeli, I.A., Grove, R.E., Naish, J., Obryk, B., Villari, R., Batistoni, P.
• Format: Journal Article
• Language: English
• Published: Amsterdam Elsevier B.V 01.10.2019; Elsevier Science Ltd
• Subjects: ADVANTG; Benchmarks; Computer simulation; Experiments; Fluence; JET; Joint European Torus; MCNP; NEXP; Parameters; Reduction; Research facilities; Simulation; Statistical tests; Toruses; Transport; Variance; Variance reduction; Workflow; JET; ADVANTG; NEXP; MCNP; Variance reduction
• ISSN: 0920-3796; 1873-7196
• DOI: 10.1016/j.fusengdes.2019.111252

•Application of ADVANTG to an extremely challenging neutron streaming experiment in a realistic fusion geometry.•Comparison of ADVANTG accelerated MCNP simulations from three different instutions to analog simulations.•ADVANTG input parameter variation analysis to find optimal parameters for a complex fusion streaming problem.•Comparison of statistically converged ADVANTG accelerated MCNP results to experimental results from TLD measurements.•Speed-ups of analog simulations more than 1100 times for the farthest locations from the plasma source. An application of the ADVANTG (AutomateD VAriaNce reducTion Generator) code, which combines the Denovo deterministic transport solver with the MCNP (Monte Carlo N-Particle) Monte Carlo code, to the JET3-NEXP (Joint European Torus work package 3) streaming benchmark experiment is presented in this paper. An ADVANTG input parameter variation analysis was performed in order to find optimal input parameters for the hybrid two-step workflow. ADVANTG-accelerated calculations from three different institutions (JSI (Jožef Stefan Institute), ORNL (Oak Ridge National Laboratory), and CCFE (Culham Centre for Fusion Energy)) were compared to analog MCNP simulations confirming no bias is introduced due to the use of ADVANTG. Additionally, ADVANTG-accelerated MCNP numerical simulations of the neutron fluence were compared to experimental results performed in 2016 at JET using thermo-luminescence detectors. Calculation/Experiment values from 0.7 to 13 were calculated for the experimental positions in the SW (southwest) labyrinth and SE (southeast) chimney. Using ADVANTG-generated variance reduction parameters, speed-up factors of up to 1100 relative to analog calculations were achieved. The MCNP statistical tests on track length estimator volume averaged tallies were consistently passed for all experimental locations with ADVANTG-generated variance reduction parameters. These results demonstrate that ADVANTG is capable of accelerating tally convergence at locations far from the source in streaming-dominated transport simulations of complex fusion facility models.