Studies on calculation models of ASTRA critical facility benchmark using OpenMC. (August 2021)

• Record Type: Journal Article
• Authors: Zhang, Tengfei; Yin, Han; Li, Xiangyue; She, Ding; Pan, Qingquan; He, Donghao; Liu, Xiaojing
• Is Part Of: Annals of nuclear energy. Volume 158(2021)
• Publication Date: 2021-08
• Subjects: OpenMC -- Validation -- ASTRA critical facility benchmark -- Homogenization scheme -- Cross section library; Nuclear energy -- Periodicals; Nuclear engineering -- Periodicals; 621.4805
• DOI: 10.1016/j.anucene.2021.108291
• Languages: English
• ISSNs: 0306-4549

Abstract

Highlights: This paper reports studies on calculation models of ASTRA using openMC. The accuracies of ENDFB-7.0 and ENDFB-7.1 are compared. Spatial homogenization models for deterministic codes are examined. Abstract: In this paper we report validation studies of the Monte Carlo code OpenMC by the ASTRA critical facility benchmark. A high-fidelity reactor model is built using the regular lattice coated fuel particle and the body centered cube pebble spheres representations. Two nuclear data libraries, ENDFB-7.0 and ENDFB-7.1 are considered to investigate the validity of nuclear cross section libraries. In addition, three homogenization models namely the core-hom model, the ref-hom model and the CR-hom model are built to provide references for analyzing the spatial approximations employed in deterministic codes. The core-hom model equivalents the octahedral core by a cylindrical zone, the ref-hom model homogenizes the side reflector into a cylindrical region, and the CR-hom model approximates the tube-cluster-shaped control rods (CR) into a concentric annulus. Results demonstrate that ENDFB-7.1 library outperforms ENDFB-7.0 by predicting criticality more accurately. The difference between the two libraries is 1224 pcm for the criticality calculation at the experimental pebble bed height (268.9 cm). With ENDFB-7.1 the high-fidelity model well-predicts the experimental data by 620 pcm error in k eff and less than 6.5% difference in CR worth. The homogenization of the reflector brings about a pronounced error of 1143 pcm. In contrast, the core-hom model causes a comparably minor discrepancy of 241 pcm to the k eff, while its influence on the CR worth appears most significant for CRs located close to the core zone. The CR-hom model yields a −2.1% maximum accuracy loss in CR worth from the high-fidelity model. After all, both the high-fidelity model and the core-hom model succeed in capturing the CR interference effect and in predicting the trend of CR differential worth. This work is not only meaningful for demonstrating the validity of OpenMC, but also useful to investigate the numerical accuracy of codes used in emerging work related to PBHTGR.