Conference Agenda

The core interfacial friction model plays a dominant role in the prediction of core cooling and maximum clad temperature in PWR and BWR accident sequences. Due to a lack of precise void fraction data in such complex geometry, the code models of core interfacial friction still have a rather high uncertainty band. The FONESYS network of system code developers initiated an activity to compare the models of the ATHLET, CATHARE, LOCUST, RELAP5 and SPACE system codes. A first comparison of models is made in the domain of low flowrate in conditions where wall friction is small and void fraction depends only on a balance between buoyancy force and interfacial friction. Only the pre-CHF bubbly-slug-churn and annular flow regimes without drop entrainment are considered in the domain of void fraction (0 < α < 0.8). The effects of pressure (0.1 MPa < P < 12 MPa) and hydraulic diameter are investigated. The paper presents first the origin of the models. The results show that the various codes agree on the qualitative pressure effects with some differences on the hydraulic diameter effect.
However, significant differences are found in the very low-pressure domain. Possible effects of the wall heat flux are discussed. Further work is planned including validation and possible model improvements.

The LWR thermalhydraulic behavior in accidental transients encounters many types of bifurcating events (BE) with cliff-edge effects where some phenomena disappear and other appear. They are due to automatic control or operator actions such as SCRAM, ECCS actuation, pump start or stop, valve opening, etc, or to transitions between different flow regimes or heat transfer regimes. Such BE have a major impact on the main parameters of interest such as primary (and secondary) pressure and fluid mass inventory, and on the figures of merit such as a peak clad temperature. The prediction of all BEs and of the right occurrence time of BEs is the main challenge of experimental and numerical simulation tools. In the Phenomena Identification and Ranking Table (PIRT), the successive time phases of a transient or Phenomenological Windows (PhW) are first identified and they can be defined and bounded by some BE. In the scaling analysis performed to design Integral Tests Facilities (ITFs) and Integral Effect Tests (IETs), most scaling methods use dimensional analysis of scaling equations (mass, momentum, energy) at system scale with acceptance criteria on the ratio of non-dimensional numbers at reactor and experimental scales. This forgets the dominant role of BEs. One may use a mature system code to perform a more detailed scaling analysis of a transient and to focus on the respect of occurrence and timing of occurrence of major BEs as acceptance criteria for the design of an ITF and of IETs. Examples are given on some PWR LOCA analyses.

FONESYS is an international network of system code developers created in 2010 to share information on R&D, to benchmark codes, to discuss the Validation and Verification as well as the code scalability and uncertainty quantification. APROS, ARIANT, ATHLET, CATHARE, CATHENA, COSINE, MARS, LOCUST, MARS-KS, RELAP5, RELAP5-3D, SPACE, TRACE are the codes that were involved in the FONESYS activities: updating the state of the art, identifying issues, discussing envisaged solutions, sharing experience in 3-field models, transport of interfacial area, numerical issues and well-posedness, code uncertainty evaluation. Code benchmarking were performed on boiling channel with CHF and Post-dryout, two-phase critical flow, flow regime transitions in horizontal flow, core interfacial friction, core 3D-mixing effects and two-phase singular pressure losses. Many code improvements were implemented in the various codes following the benchmark activities. When the need of new experimental data was identified, FONESYS discussed with SILENCE experimentalists to define requirements of new instrumentation and new experiments. Future activities will focus on code scalability, core 3D modelling and validation, two-phase pressure losses, use of system codes for scaling analysis and applications to passive systems, SMRs and AMR. The present paper presents the major achievements and the motivations for the future activities.

The system codes can model 3D flow in a core either with 3D solvers in porous medium, or with sub-channel models, or even using cross-flow junctions between parallel 1D models. Such tools rise many questions on the modelling of mass momentum and energy transfers including diffusion and dispersion processes. The extrapolation of many closure laws from 1D to 3D flow and the formulation of wall friction and interfacial friction when the flow is not parallel to fuels rods require some attention. The FONESYS network of system code developers initiated an activity to compare the 3D-core models of the ATHLET, CATHARE, COSINE, LOCUST, RELAP-3D and SPACE codes in very simple situations. One considers first two adjacent fuel rod assemblies with different power and one calculates the vapor flow above a swell level at two pressures (1 and 7 MPa) to obtain either friction-driven crossflows or buoyancy driven crossflows (chimney effect). Other calculations include the two-phase region, a swell level and a dry zone above, the assemblies being connected to an upper plenum and a lower plenum, allowing different inlet flowrates as in a reactor situation. The impact of axial friction pressure losses and spacer-grid form losses on radial crossflows is shown. The homogenization of void fraction below a swell level seems very efficient. The sensitivity on radial pressure losses in non-axial flow is not very high but the uncertainty on the radial wall friction and interfacial friction is very high. From the analysis of these results, further investigations are planned.

The current system codes use 2-fluid models or 3-field models and predict wall friction and singular pressure losses by models developed for a unique mixture momentum equation. Two-phase multipliers exist that can extrapolate 2-phase pressure losses from 1-phase models but the repartition between phases is not modeled although it may play a very significant role on the result. This may result in a rather high uncertainty of predictions not only in high velocity flow but also in low flow situations encountered in natural circulation. The FONESYS network of system code developers initiated an activity to compare the predictions in a few basic singular geometries for which one can simply evaluate the single-phase pressure loss coefficient. In a reactor circuit, there are several locations with an abrupt area restriction or an abrupt area enlargement or even some plates behaving as a diaphragm. Some code prediction comparisons in such basic geometries are presented and analyzed in both vertical upward and vertical downward flow and in five flow regimes: 1-phase liquid, two-phase bubbly flow, low velocity and medium velocity slug-churn flow and high velocity annular-mist flow. The calculations were made with five codes: ATHLET, CATHARE, MARS-KS, RELAP5 and RELAP5-3D. Some differences are found between codes. The effects of nodalization are investigated and the impact on void fraction perturbation are analyzed. First conclusions are drawn on the reliability of predictions and some requirements for future well-instrumented pressure loss experiments are defined.

The Atomic Energy and Alternative Energies Commission (CEA) is a French Research and Development institution that plays a major role in the French nuclear energy program. CEA has been developing the CATHARE code for 45 years. It is an extensively validated thermal-hydraulic system code based on the 2-fluid 6-equation model. The CEA is currently developing new tools in CATHARE to facilitate its use for scaling analysis, which contribute significantly in the identification of dominant phenomena occurring during reactor accidental transients and in the design of experiments able to simulate major phenomena with minimal distortions.

Previous published work on scaling analysis with the CATHARE code focused on the primary mass and primary pressure equations to identify the dominant terms controlling the mass inventory and the system pressure. These terms were calculated “by hand” during the post-processing phase. A new tool enables this analysis to be carried out automatically during a calculation, by fetching the thermal-hydraulics quantities at code execution. It relies on a Python supervisor, based on the ICoCo (Interface for Code Coupling) standard, which extracts the required fields at each calculation time step. The fields are then manipulated using the MEDCoupling open source library. The integrated momentum equation along cooling loops is also added to the analysis.

This work describes the way this tool is applied to a SB-LOCA transient using the equations of mass, momentum and pressure. Figures of merit and available post-processing are presented, which enables the dominant phenomena to be quantified. Leads for future developments are given.

The models in system codes for LWR core interfacial friction still have a rather high uncertainty in some conditions due to a lack of void fraction data in such complex geometry. Usually, only axial pressure differences are available to test the interfacial friction. The FONESYS network of system code developers initiated an activity to compare the models of several system codes and found significant differences at very low pressure. Recently a new void fraction measurement technique based on the wire mesh sensor was implemented by CRIEPI in a 5x5 rod bundle and produced real void data. There are steady state data at 1, 3 and 7.2 MPa, with a wall heat flux from 5 to 45 kW/m² and a mass flux in the range 90 to 400 kg/m²/s. CRIEPI also performed boil-off tests with an outlet atmospheric pressure and heat fluxes in the range 2.4 to 7.2 kW/m². EDF and CEA performed validations calculations with the CATHARE code on these data. Several sensitivity tests are presented to investigate some possible reasons of the code-to data differences. Previous validations on data in boil-up tests and on air-water non-heated rod bundles data are used to investigate the various possible effects of fluid properties, of pressure, and of the wall heat flux. Preliminary conclusions are drawn on the validity and limitations of the current interfacial friction model and on possible ways to improve it.