2. Main results of the research projects in 2018
2.2 Reactor safety
2.2.11 USVA - Uncertainty and sensitivity analyses for reactor safety
• HEXTRAN-OpenFOAM and OpenFOAM-SMABRE interface routines have been prepared so that OpenFOAM can be used together HEXTRAN and SMABRE instead of the PORFLO code.
• Several transients and accidents have been simulated using fully two-way coupling between HEXTRAN, SMABRE and PORFLO:
o OECD/NEA dynamic benchmark V1000CT-2 - scenario 1 VVER-1000 main steam line break (MSLB)
o OECD/NEA dynamic benchmark V1000CT-2 - scenario 2 VVER-1000 main steam line break (MSLB)
o OECD/NEA Kalinin-3 MCP benchmark
Scenario specific inputs have been prepared for all these cases. Interface routines have been completed for reversed flows. Results have been compared to with those of HEXTRAN-SMABRE.
• OpenFOAM input files were prepared for VVER-1000 pressure vessel. A transient simulation of the OECD/NEA dynamic benchmark V1000CT-2 - scenario was computed with the new analysis framework HEXTRAN-OpenFOAM-SMABRE.
Results were compared with other modelling approaches.
• The project included also participation in the AER working group D meeting in Dresden, Germany, where the presentation was given on development of a VTT’s new nodal code ANTS. Also kick-off meeting of OECD/NEA Rostov-2 benchmark in Lucca Italy was participated.
With CFD Internal coupling Parallel coupling Typical system code nodalization T (K)
Time 56s, min 515.5 K 71.3 s, min. 517.5 K 71.2 s, min. 517.4 K 85.1 s, min 539.1 K P (MW)
Time 56.8 s 72.9 s 72.5 s 85.5s
Figure 2.32. Core inlet temperature at time of minimum temperature, and assembly-wise fission power at time of maximum fission power during the main steam line break of VVER- 1000 reactor with distinct coupling methods between HEXTRAN, SMABRE and PORFLO.
OECD/NEA V1000CT-2 benchmark.
also trained. USVA promoted activities at the interfaces of the different disciplines in reactor safety.
Many of the tasks in USVA were related to the topics of the OECD Nuclear Energy Agency (NEA) Benchmark for Uncertainty Analysis in Modelling (UAM) for the Design, Operation and Safety Analysis of LWRs.
Two publications made during the previous years of USVA project were used in an article based doctoral dissertation defended in 2018 (Arkoma, Modelling design basis accidents LOCA and RIA from the perspective of single fuel rods, Aalto University).
Specific goals in 2018
WP1, methods and analyses, had three subtasks in 2018.
Task 1.1 continued the development of sensitivity analysis methodology for statistical fuel failure simulations of EPR large break loss-of-coolant accident (LB-LOCA) started in 2015. In 2016, it was demonstrated that support vector machines (SVMs) can be used as a surrogate model to replace some of the computationally expensive rod-level FRAPTRAN-GENFLO simulations in estimating the number of failed fuel rods in one global accident scenario.
Applicability of SVMs to sensitivity analysis was demonstrated in 2017. In 2018, the work was extended to the full array of global simulation data and its sensitivity analysis. Sensitivity indices between global parameters and cladding maximum hoop stress were calculated, and that way the importance of global input parameters was studied. The global parameters that were varied in the original statistical analysis included input parameters of the system code APROS, and model parameters of the steady-state fuel behaviour code FRAPCON and the subchannel thermal hydraulics code GENFLO. The results showed that the global parameters had actually negligible effect compared to the local parameters on cladding maximum hoop stress in LB-LOCA.
Task 1.2 was a three-year task that started in 2016 with a literature review of potential methods for determining input uncertainties of thermal hydraulic codes. Due to personnel changes, a new expert on the subject was trained in 2018. The focus was on drawing up an updated plan on the application of the best-suited methodologies to the quantification of uncertainties related to physical models of APROS. Methods were reviewed and a path forward suggested.
Task 1.3 utilized the sensitivity/perturbation calculation capability implemented in Serpent 2.1.29 in USVA in 2017. In 2018, Serpent was further extended to enable reading, processing and interacting with the multi-group nuclear covariance data. This data was applied with the above mentioned sensitivity calculation capabilities to propagate nuclear data uncertainties into several different group constants that can be expressed as reaction rate or detector tally value ratios. The capabilities were demonstrated in calculations where detectors were set up to tally reaction rates, and fluxes and the group constants were calculated as a post processing step. As an example, Fig. 2.33 shows the contributions to the total uncertainty of the homogenised one group fission cross section in a hot-full-power pin cell for the Three Mile Island (TMI) 1 fuel type from UAM. In the future, a sampling based method should be used to verify the propagated nuclear data uncertainties.
Figure 2.33.Top contributors (as a fraction of the total covariance) to the uncertainty of the macroscopic fission cross section homogenized by Serpent in a TMI1 (PWR) pin-cell.
WP2 was dedicated for uncertainty and sensitivity analyses of reactor dynamics codes, and it had one subtask in 2018. The objective was to model UAM benchmark Phase 3 problems, with the primary area of interest being the VVER-1000 task. The final specifications for this phase have not been finalized but the group constant data containing propagated
uncertainties has already been prepared for the VVER case. Preliminary simulations with this data were done in USVA in 2018. For this work, the HEXTRAN-SMABRE model of Kalinin-3 VVER-1000 plant was utilized, and 500 simulation runs of the benchmark problem, switching off one main coolant pump (MCP) of working four MCPs, were completed (Fig. 2.34).
Additionally, the effect of nodalization of the pressure vessel on coolant mixing was studied.
Figure 2.34.HEXTRAN-SMABRE simulation of Kalinin-3 benchmark, switching off one main coolant pump (MCP) of working four MCPs using 500 separate group constant sets.
Deliverables in 2018
• Research report on large break loss-of-coolant accident sensitivity analysis related to uncertainties of global parameters.
• Research report containing a literature review of potential methods for determining input uncertainties in thermal hydraulic codes.
• Research report on uncertainty propagation capabilities augmented to Serpent Monte Carlo code.
• Research report on preliminary uncertainty simulations performed for the OECD/NEA UAM benchmark on VVER-1000 plant with HEXTRAN-SMABRE code.