Skip to main content

Joint Advanced Severe accidents Modelling and Integration for Na-cooled fast neutron reactors

Final Report Summary - JASMIN (Joint Advanced Severe accidents Modelling and Integration for Na-cooled fast neutron reactors)

Executive Summary:
In the frame of the 7th FrameWork Programme of the European Commission (EC) and inspired by the Gen. IV target of designing innovative reactors that intrinsically prevent severe accidents from occurring and/or drastically reduce the expected consequences, the JASMIN project was launched. The generic aim was set to be the enhancement of the current capability of analysis of severe accidents in Na-cooled reactors and, particularly, to develop the capabilities to evaluate the consequences of unprotected accidents with fuel pin failure on materials relocation and primary system loads and to predict fission product and aerosols behavior once released. To do so, the ASTEC platform, developed by IRSN and GRS, has been chosen to be adapted and extended to the environment of Na-cooled fast reactors, the result being what has been called ASTEC-Na. As a consequence, JASMIN perfectly fits what stated in the ESNII (European Sustainable Nuclear Industrial Initiative) roadmap by developing and benchmarking European computer codes for ESNII Fast Neutron Reactors, in particular relating to safety performance, leading to establish a common platform for modelling and simulation.
The project addressed four main areas: Thermal-hydraulics, pin behavior, source term and neutronics. In each area code and models development and validation have been performed. In addition to the test matrices built within the frame of the project and used as references for the model validation, the adequacy of ASTEC-Na models have been evaluated through the application of other suitable codes for benchmarking purposes. The main takeaways from the validation work were withdrawn in the form of a SWOT analysis (Strengths Weaknesses Opportunities Threats) that allows clearly identifying the main needs for future model developments and outlining what is considered as the pathway to fully make ASTEC-Na a reliable analytical tool for severe accident in Na-cooled reactors.
In term of opportunities ASTEC-Na is one of the few codes nowadays supported (in contrast with old SFR safety codes with no development support anymore). Though the in-vessel phenomenology modelling for the entire initiation phase of an accidental scenario is not yet fully completed it already includes a point kinetics model, single and two-phase sodium thermal-hydraulics, precise fuel-pin thermo-mechanical models for describing the behaviour of the pin and associated fission gases during accidental conditions. In addition, ASTEC-Na is devoted to model the in-containment phenomenology. Though the in-containment source term modelling in SFRs during severe accident still need to be extended in ASTEC-Na, such further developments will make the code able to evaluate the source term produced by the migration of activated fission products inside the reactors and likely to be released to the environment in case of accidental situations. The integrated features of the ASTEC software platform will thus allow simulating in a single code what is generally today simulated in separate codes (i.e. SAS-SFR and CONTAIN-LMR...). The high modularity of ASTEC-Na (that characterizes the code compared to similar codes) allows to work separately with each of its module and to integrate it back in the code at a later stage, so that any development, implementation and assessment is more straightforward than once implemented in the integral code. In addition, the flexibility in defining the core geometry, materials composition and reactor components makes ASTEC-Na ready to study new SFR designs with fertile layers in outer radial or inner axial core regions (such as in ASTRID design), subassemblies with an inner duct channel to induce fast fuel axial relocation (such as FAIDUS design), or new safety systems to shut-down the core power.
The JASMIN project has represented a unique opportunity to bring together different actors of nuclear safety in Europe around ASTEC-Na code. The analytical work performed during the project has highlighted a major need: to build a robust and extensive database that can be used both to support individual model development and to validate more “integral” tools, like ASTEC-Na.

Project Context and Objectives:

In the current Nuclear Power Plants (NPPs) existing in Europe, the principles of defence in depth aim at protecting the public against radioactivity release for a series of postulated accidents. Nevertheless, in some very low probability circumstances, severe accident sequences may result in core melting and plant damage leading to dispersal of radioactive material into the environment and thus constituting a health hazard to the public. For Gen.IV advanced and innovative concepts, safety is a cross-cutting issue: it must be continuously enhanced, and thus reactor designs will aim at intrinsically preventing severe accidents from occurring or at least reducing the expected consequences on the environment and the population. Given the unfeasibility of carrying out a whole campaign of experiments at real scale to prove the expected design strengths against severe accidents, the availability of validated simulation tools is of utmost interest at this point of research and development (since designs of several prototypes are under definition). In other words, it is crucial that reliable computer codes are developed to simulate such severe accidents and validated. This statement is applicable to all Gen.IV designs including transmutation, and particularly to Sodium-cooled Fast neutron Reactors (SFR).
The JASMIN project was related directly to the Topic Fission 2011-2.2.1 dedicated to “Support for ESNII” in the Call 2011 of the European Commission (EC). It was then intended to support the ESNII (European Sustainable Nuclear Industrial Initiative) roadmap, elaborated in the general frame of SNETP (Sustainable Nuclear Energy Technology Platform), here in the following area: “development and benchmarking of European computer codes for ESNII Fast Neutron Reactors, in particular relating to safety performance, leading to establish a common platform for modelling and simulation”.
It also came as part of the support to SNETP’s Strategic Research Agenda (SRA) and Deployment Strategy. Indeed the SRA identified the R&D challenge to enhance safety “especially towards a higher resistance to severe accidents...” and “practically precluding large energy release in case of severe accident (even hypothetical) while investigating core designs with moderate sodium void effect and other favourable reactivity feedback effects, and core designs and reactor vessel internal structures likely to disperse core debris and minimize risks of core compaction”. In particular, the SRA recommended “re-establishing a set of modern simulation tools for severe accidents analyses and design studies of robust core catcher”.
In SFR, a major safety concern is the potential risk of large mechanical energy release to the vessel structures as SFR cores are highly sensitive to any modification of the core configuration (for instance, as a result of fuel melting). Moreover, with sodium as coolant and in case of loss of cooling and sodium boiling, the positive reactivity feedback (in some core regions) of the sodium void effect can generate fast power transients with subsequent significant fuel damage and core disruption. A large data base has been gained from past R&D experimental programmes that were dedicated to mixed oxide (UPuO2) fuel behaviour all over the world (i.e. CABRI in France, TREAT in USA,...). However new and specific features of future Gen.IV systems (i.e. higher fuel enrichments, ~15–20 at%; fuel compounds other than oxides; inclusion of minor actinides, etc..) and the remaining uncertainties make it necessary to keep on investigating before relying on past research to extrapolate its results to the present situation. Severe accidents (also called Core Disruptive Accidents) can be the consequences of various initiating events like unprotected slow power transients due to control rod withdrawal, unprotected loss of cooling and loss of flow due to pump coast-down or subassembly blockage, or propagation of more local faults.
The current accident simulation capability of SFRs dates back at the 80’s. In addition, the very few tools available lack some flexibility to integrate new physical models, especially for advanced designs. Therefore, based on the importance of considering severe accidents in the design phase of next SFRs, on the limitations of the available computation tools and their intrinsic weaknesses, the JASMIN project aims at developing a new computer code system, named ASTEC-Na, capable in the future of encompassing all the phases of a hypothetical severe accident and based on a robust advanced simulation platform.
The present project concerns only the initiation phase of SFR severe accidents i.e. before any significant molten fuel relocation occurred. At the end of the project, the developed simulation tool (ASTEC-Na) should be both able to:
- evaluate the consequences of unprotected accidents with fuel pin failure on materials relocation and primary system loads, due to either reactivity-driven transients, such as sodium voiding, or mechanically or thermally-initiated transients, such as the blockage of a coolant in a subassembly,
- evaluate the source term produced by the migration of activated fission products inside the reactors and likely to be released to the environment in case of accidental situations.
The project was not aimed to address the modelling of the part of the accident that may follow the initiation phase, i.e. the transition phase and the expansion phase, up to the final state of the molten materials (e.g. those located on the debris tray or any new core-catcher design) and the evaluation of the mechanical consequences of the accident on the structures. Then, it was not aimed for instance to address the modelling of sodium fires, sodium-water interactions and sodium-concrete interactions. Nevertheless, it was planned that investigations on the requirements for such SFR extensions of ASTEC-Na scope will be performed by all partners before the end of the project. These investigations then could serve for the launching of a further project.
The ASTEC-Na development was based on the ASTEC code system (Van Dorsselaere J.P. et al., “The ASTEC integral code for severe accident simulation” in Nuclear Technology, Volume 165, p.293-307 March 2009), jointly developed by IRSN and GRS, extensively validated through European projects (like EVITA, SARNET and SARNET2) and which is now considered as the European reference code for severe accidents in water-cooled NPPs. This code system has already been extended to cover severe accidents in diverse types of nuclear reactors: PWR, VVER, BWR and CANDU (the two latter yet partly for the moment) for Gen.II water-cooled reactors, some Gen.III reactors (like EPR), HTR (partly) and ITER for fusion facilities.
From a software point of view, the development was intended to be based on existing modules of the ASTEC V2 code that was re-used for ASTEC-Na. Specific physical models was intended to be developed on SFR phenomena, using as a basis the models of existing codes (i.e. SAS-SFR for in-vessel phenomena and CONTAIN-LMR fir in-containment phenomena). The outputs of the SP3-6 task (“Evaluation of modelling capabilities of accident scenarios”) of the CP-ESFR current FP7 project on the status of code models was intended to be used as a basis. Hence, the JASMIN project was complementary to the more reactor-engineering focused CP-ESFR project since it goes further in enlarging the simulation capability of severe accidents in SFRs.
More specifically the main objectives of the project that addressed four main areas (Thermal-hydraulics, fuel pin behaviour, source term and neutronics) were:
- to elaborate general specifications for the ASTEC-Na code development about its application to SFR initiation phase, keeping in mind all requirements linked to future reactor designs,
- to build adequate validation matrices for both in-vessel and in-containment phenomena based on either past available test or on-going new experiments, and to identify, if necessary needs of new experimental data,
- to develop new models and implement them in ASTEC-Na and provide the corresponding code documentation,
- to perform the code verification and validation work by comparing the ASTEC-Na calculations results to experimental data and performing benchmarks with other available codes,
- to identify the needs for ASTEC code extension to the other periods of the SFR severe accidents (i.e. transition phase, expansion phase, secondary phase, sodium fires...) and to LFR severe accidents.
The project was then entailed an extension of the ASTEC domain as a European reference code for Gen.II and Gen.III to the fast reactors domain. This strategy allowed getting the maximum benefit from a computation tool already available that has been significantly supported by EC and, at the same time, it supported ESNII to keep the European leading role in nuclear technology by developing a computation tool needed for the indispensable safety assessments and safety enhancements of the fast neutron systems.

Project Results:
Several experiments from the general ASTEC-Na validation matrix have been calculated in order to assess the capability of the code and the reliability of the results. Tests used for thermal-hydraulic model validation as well as codes used for comparison with ASTEC-Na are listed in Table 1.

Table 1. Experimental test matrix for TH model validation and benchmarking codes.
Facility / Pant Test analysed with ASTEC-Na Benchmarking codes Organization TH investigation
Pre-test SS, ULOF, ULOHS, NC
(Pre-test analysis) CATHARE, RELAP5-3D, RELAP5-Na ENEA, KIT, CIEMAT Single-phase,
KNS N02 RELAP5-Na ENEA, KIT Single-phase,
PHENIX Ultimate natural circulation RELAP5-Na, CATHARE CIEMAT, IRSN Plant application,
Single-phase, NC
SPX Stabilization & Natural convection DYN2B IRSN

Both in-pile single fuel rod (CABRI) and fuel rod bundle (SCARABEE) experiments were calculated. Pre-test analyses of KASOLA experiments have also been performed while a KNS test has been calculated as a backup to delayed KASOLA experiments which were not made available in the time frame of the project. Finally, reactor calculations (Phenix, SPX) have been performed in order to evaluate the ASTEC-Na code capability to model a whole pool-type reactor and investigate the transition from forced to natural circulation conditions. The adequacy of ASTEC-Na TH models as been also investigated through the use of other suitable codes.
The next table compiles experiments and codes used in the JASMIN benchmark for the fuel thermos-mechanical model validation together with the organization which ran the codes.

Table 2. Experiments, codes and organisations in WP2.2 benchmark.
Test analysed Benchmarking codes Organization

Those tests were issued from the ASTEC-Na validation matrix built during the project (Girault, 2013). Few CABRI and SCARABEE tests were included in it where the CABRI tests were chosen in order to validate the ASTEC-Na models over a wide range of experimental conditions (Table 3).

Table 3. Characteristics and objectives of the experiments analysed in WP2.2 benchmark.
Test analysed AGS0 E7 E9 LT2
Burn-up (at.%) 2.9 4.6 4.6 12.4
Geometry solid annular Annular Annular
Clad 316 cw SS 316 cw SS 316 cw SS 15-15 Ti stabilized SS
Test Transients TOP Structured TOP Slow Ramp Structured Ramp
Duration (s) 0.6 0.8 118 0.9
Max. Power (P/PN) 12 150 2.2 25
Main results Partial melting Cavity pressure built-up
Clad rupture Extensive melting
PCMI Fuel squirting
Model Assessments Fuel melting
FG behavior Clad rupture Fuel swelling
Clad deformation In-pin fuel motion

All the transients were initiated at nominal cooling conditions with restrained clad. For each test were analyzed the fuel thermal-mechanics, axial expansion, melting limits, the clad deformation/ rupture, the fission gas release/retention and the gap heat transfers.
In order to assess the enhanced ASTEC-Na CPA version, a literature survey was carried out and as a result more than 20 experiments related to in-containment source term have been reviewed and properly stored in a data base. A selection of suitable experiments has been conducted. Four Na-pool fire experiments have been chosen (Table 4): ABCOVE-AB1 and –AB2, FAUNA-F2 and –F3 and EMIS-10b.

Table 4. Experimental test matrix.
AB1 AB2 F2 F3 EMIS 10b
Type Cylindrical Cylindrical Cylindrical Cylindrical Cylindrical
Volume (m3) 852 852 220 220 4.4
Initial Conditions
O2 (%) 19.8 20.9 17-25 15-25 20
Temperature (K) 299.65 293.65 298.15 298.15 294.05
RH (%) 35.5 43.3 - - -
Steam Addition NO YES NO NO NO
Sodium Spill
Initial Na Temp. (K) 873.15 873.15 773.15 773.15 554
Burning Area (m2) 4.4 4.4 2 12 0.125
Fire duration (s) 3600 3600 12600 4800 6000

Finally, as far as neutronics is concerned, the lack of experimental data - considering the large power excursions that some transients can experience – is a major drawback for a full validation of any neutronics model in sodium fast reactors. The validity of the ASTEC-Na neutronics model was, therefore, assessed through a benchmark with TRACE and SAS-SFR codes under a selected accident scenario which was an unprotected loss of flow accident (ULOF) in a pool-type sodium cooled fast reactor. As the ULOF accident represents the enveloping case for the consequence of some of postulated severe accidents in sodium cooled fast reactors (i.e. gas bubble migration, local fuel melting...) in which the neutronics feedback plays a crucial role, the insights of the benchmark can be usefully extrapolated to other scenarios under the domain of validity of the point kinetics model.
As a benchmark reactor model serves the “Reference Oxide core design” concept of the CP-ESFR project under Beginning of Life (BOL) core load conditions.

In this section is summarized the main outcomes from WP2 benchmark exercises (“Code and Models Development and Validation”), which addressed four main areas: thermal-hydraulics, pin behavior, source term and neutronics.
Sodium property tables extended to two-phase conditions and suitable heat transfer and pressure loss correlations have been implemented in the ASTEC-Na code, including interphase heat transfer and friction correlations to better represent single and two-phase thermal-hydraulics within the core and the primary circuit of the reactor. Validation of thermal-hydraulic models has been based on suitable experiments and through extended benchmarking with other thermal-hydraulic codes (see table 1). The main results and conclusions from the ASTEC-Na validation and benchmarking work are presented in the following. A more detailed description of this work and the results obtained are reported in the Deliverable 2.5 (Bandini et al., 2014) and the Technical Report in Bandini et al. (2016).
In all three CABRI tests investigated, the LOF resulted in boiling onset after about 20 s. For all the simulation tools, including ASTEC-Na, the sodium flow coast-down before boiling onset was predicted in accordance with flowmeter measurements. Moreover, the calculated sodium temperatures at the LOF onset and just before the boiling start at t = 20 s are within the uncertainty range of thermocouple measurements, as shown in Figure 1 for the BI1 test. The largest discrepancies are in the upper part of the test section, above the fissile column, mainly due to uncertainty in the modelling of heat losses. In all three tests also the sodium temperature rise up to saturation is well predicted by ASTEC-Na and most of other codes at the different elevations as shown in 2 for the BI1 test

Figure 1. BI1 test: Sodium temperature just before boiling onset around t = 20 s.
Figure 2. BI1 test: Sodium temperature evolution at 77 cm from bottom fissile column.

After the onset of boiling, the inlet flow rate fluctuations are well calculated by all the codes including ATEC-Na as observed in the tests, but generally the differences in the computed amplitude and frequency of oscillations are rather large (Figure 2). However, the largest uncertainties are encountered in the calculation of the boiling front propagation for both upper and lower two-phase front evolution, as illustrated in Figure 3 for BI1 test. An important user-effect on the obtained two-phase upper front evolution was observed where some ASTEC-Na calculations significantly underestimated it as others largely overestimated it. The results of sensitivity studies performed on the more uncertain parameters, such as heat losses (Figure 4) or test section upper part meshing, show that they might have a significant impact in the prediction of boiling front propagation.

Figure 3. BI1 test: Evolution of Na boiling front. Figure 4. BI1 test: Evolution of Na boiling front (sensitivity to heat loss in test section upper part).

In conclusion, ASTEC-Na gives satisfactory results for sodium single-phase of the simulated CABRI experiments (BI1, E8 and EFM1). Coolant heat-up along the fissile column and coolant heat-up during the single phase of the LOF transient are in good agreement with the experiments. Boiling onset is predicted by ASTEC-Na with a small difference with respect to the observed one. It can be partly explained by the difficulty to consider the fuel pin distortion observed during the experiments affecting strongly the local boiling inception. Calculated sodium two-phase flow behavior are rather poor using ASTEC-Na. Inlet and outlet flow rates are underestimated in ASTEC-Na calculations after boiling onset in all simulated tests. The lower boiling interface is in a sufficient agreement with experimental data while the upper front is underestimated for all tests.
Sensitivity studies on BI1 and EFM1 tests showed that a fine meshing of the test section above the fuel region, as well as the adjustment of heat losses in this region, improve the ASTEC-Na prediction during the boiling phase. Recalculation with the latest version of ASTEC-Na (v2.0) showed minor differences, except for a better robustness in two-phase flow calculations.

Among the available SCARABEE tests, the flow blockage BE+3 test and the LOF APL1 and APL3 tests have been considered for validation and benchmarking purposes as representative of both single and two-phase flow regimes for fuel rod bundle geometry. A 2D-meshing of the fuel rod bundle, discretized in four radial channels, has been used in both ASTEC-Na and SIMMER-III simulations, while a single-channel model has been used with SAS-SFR. The ASTEC-Na calculations have been extended up to the transient phase, after significant clad melting and relocation were observed in the tests, and just before the onset of fuel melting.
Figure 5 shows the sodium axial temperature profile at the onset of the transient phase (t = 0 s) in which calculated temperatures in the inner (R1) and outer (R4) radial channels are compared with available thermocouple measurements. According to the experimental evidence, radial temperature profile calculated by ASTEC-Na and SIMMER-III near the top of fissile column (+30 cm) shows outer temperatures lower than inner temperatures, despite the higher outer ring power, due to 2D flow redistribution effects inside the fuel rod bundle and the larger flow area ratio of the outermost ring.
Generally, all codes, including ASTEC-Na, predict well the rise in sodium temperature up to reaching the saturation point (1323 K in ASTEC-Na, 1315 K in SIMMER-III and 1334 in SAS-SFR) with start of boiling at different elevations (Figure 6). Also sodium temperature evolution at the top of the fissile zone is rather well predicted (see Figure 7) where the sudden rise in temperature at the onset of boiling (2.5 s) is much more uncertain leading to some uncertainty in the evaluation of the timing of clad dry-out. This discrepancy might be mainly related to 2D radial cross-flow effects that cannot be properly simulated with both ASTEC-Na and SIMMER-III codes.
As in CABRI tests, some important deviations from the experimental trend are evidenced in all code calculations, in the two-phase front evolution (Figure 8) where only the lower boiling front evolution is rather well predicted.

Figure 5. BE+3 test: Axial temperature profile of sodium at t = 0 s in radial channels 1 & 4. Figure 6. BE+3 test: Evolution of sodium temperature at middle of fissile column in channel 1.

Figure 7. BE+3 test: Evolution of sodium temperature at the top of fissile column in channel 1. Figure 8. BE+3 test: Two-phase front evolution in radial channel 2.
In flow coast-down tests, both the evolution of sodium flow rates, sodium and clad temperatures until reaching the boiling onset are well predicted by ASTEC-Na. The uncertainty in the prediction of clad dry-out is once more rather large, as in the prediction of boiling front propagation. As for the BE+3 test, the largest discrepancy is found in the evolution of the upper two-phase front. In this loss of flow test, the uncertainty in the heat losses and steam condensation phenomena which might occur above the fissile zone can play an important role.

In conclusion, the analysis of SCARABEE tests confirms the good performance of ASTEC-Na code in the calculations of single-phase conditions, while the largest deviations are found during the boiling phase. Some observed 2D effects in the bundle can reasonably well captured by the code during the single-phase flow, but most probably 2D effects amplify the deviations in code results and against experimental data during two-phase flow conditions. It must be underlined that ASTEC-Na results become unreliable in the late transient phase of analyzed tests after significant clad melting, since clad relocation and melt interaction with coolant is not yet computed by the code.

KASOLA pre-test analyses investigated several operating conditions including steady-state at different power levels, in the range 200–400 kW, and ULOF and ULOHS transients. Furthermore, the stabilization of natural circulation in the primary loop following an ULOF transient starting from different power levels has also been studied. Only the CESAR module of ASTEC-Na, which computes the 1D thermal-hydraulics of primary and secondary circuits, has been taken into consideration in this analysis and compare to RELAP5-3D, CATHARE and RELAP5-Na results. For ULOF transients (starting from different power levels) all calculations predict similar behavior with natural circulation flow rate in the range 6-7 % of the nominal value. Some discrepancies observed during the initial transition phase mainly depend on the different pump model used. As for the time evolution of sodium temperature at the heated section outlet, all codes predict similar temperature behavior, while the main differences are induced by the different steady-state temperatures. The largest difference in two ASTEC-Na results is mainly induced by different assumptions in heat removal by the secondary circuit through the Na-Air heat exchanger.
The unprotected loss of heat sink transient (ULOHS) in the KASOLA base loop has been investigated to verify the right prediction of whole primary circuit thermal inertia (including pipe walls and insulation materials). CATHARE, ASTEC-Na and RELAP5-3D by ENEA predict very similar heat-up rate of the primary sodium at the heated section outlet. The main reason for the discrepancies evidenced in two ASTEC-Na calculations was the large difference in the code input regarding the heat exchanger external shell mass, which strongly affect the overall thermal inertia of the system.
In conclusion, the ASTEC-Na code has been successfully applied in the simulation of KASOLA base loop for the evaluation of thermal-hydraulic behavior of both primary and secondary systems. Some significant differences in the calculation results have been observed between different users of the ASTEC-Na code. The reason for these differences concerns mainly different assumptions in the modelling of the Na-Air heat exchanger. The benchmarking with other codes has highlighted a generally good agreement in code results under the different steady-state and transient conditions. In particular, the differences between ASTEC-Na and CATHARE results in ENEA calculations have been minimized thanks to a good harmonization of code input decks, even if a different number of meshes is used to represent the loop in the two simulations.

One natural circulation test of the N-series of KNS tests (N02) was selected as a backup of the KASOLA experiment which was not performed in the time frame of the project due to a delay in the construction of the facility. In this test, the natural circulation in the loop and in the bypass was driven by the progressive heating of the sodium in the test section. The main conclusion was that the simulation of the N02 test has been inevitably affected by the lack of data on KNS loop. Anyway, the agreement between ASTEC-Na results (without and with loop simulation) and available experimental data can be considered satisfactorily enough. Observed discrepancies between calculated and measured temperatures in steady state conditions can be, at list in part, explained by existing uncertainties on sodium physical properties. The available experimental data don’t clarify the significance of 2D/3D effects on fluid behavior within the rod bundle and this does not allow a reliable assessment of the proposed ASTEC-Na 2D model. Anyway, the adoption of a 2D model of the fluid behavior in the rods bundle (4 radial channels) does not lead to an undeniable and evident improvement of ASTEC-Na performances with respect to 1D approach (1 channel) to simulate the fluid. Finally, the results obtained with RELAP5-Na and ASTEC-Na 1D model using the same geometry, meshing, material properties and physical assumptions, are in a quasi-perfect agreement.
During the project some reactor applications have also been performed. Results from natural circulation tests performed in the pool-type Phenix and SuperPhenix (SPX) reactors have been used to verify the capability of the ASTEC-Na code to simulate the overall plant behavior under both steady-state and transient conditions. The Phenix end-of-life natural circulation test, aimed at studying the establishment of natural circulation in the primary circuit by tripping the three primary pumps, has been calculated with ASTEC-Na and the results have been compared with CATHARE and RELAP5-Na.The Phenix study showed the capability as well as the present limits of the ASTEC-Na system-code in simulating the thermal hydraulic behavior of a sodium-cooled fast reactor during a natural convection transient. The relevant results observed from the natural circulation test are well calculated by ASTEC-Na. Discrepancies are mainly observed during the transient short term, when natural convection is settling and local phenomena are of prime importance. Choices in modelling, meshing, pressure drop and heat transfer correlations as well as others are identified as possible explanations for such discrepancies. Meanwhile, the calculation of the natural convection test in the SPX reactor with ASTEC-Na and DYN2B codes (where temperatures and mass flow rates are imposed at the inlet of IHX secondary side) showed that the establishment of natural circulation in the primary circuit, after stop of primary pumps (t = 0 s) and emergency shutdown of the reactor (t = 3 s), is predicted in reasonable agreement with DYN2B results (see Figure 9 and Figure 10).

Figure 9. SPX test: total primary mass flow rate
(comparison with DYN2B code). Figure 10. SPX test: core outlet temperature
(comparison with DYN2B code).

This section presents the results of the four CABRI tests calculated with the latest version of the ASTEC-Na code released within the project (v2.1) and compared both against experimental data and against SAS-SFR and SIMMER-III results. All the transients were initiated at nominal cooling conditions with restrained clad. For each test were analyzed the fuel thermal-mechanics, axial expansion, melting limits, the clad deformation/ rupture, the fission gas release/retention and the gap heat transfer through the Na coolant temperatures.
For CABRI test interpretation, the conditions of fuel pin pre-irradiation in reactor and of fuel pin conditioning during the power levels in CABRI facility are of peculiar importance and were thus studied. While in SAS-SFR code, the calculation of the irradiation phase is an integral part of the modelling, ASTEC-Na relies on an interfacing to the GERMINAL code for the fuel pre-irradiation state. The models simulating the irradiated fuel behavior have been validated upon non-destructive Post-Irradiation Examinations and Post-Test Analyses of sibling pins. The main conclusion from the benchmark performed on the fuel irradiation was that GERMINAL 1.4 calculation results were found suitable to be used as entry data for CABRI test simulation with ASTEC-Na. The only major deviation from experimental data concerns the fission gas release in RIG2 pin, significantly overestimated, which could impact the calculated pin thermo-mechanical behavior during the AGS0 transient.
Similarly, for the CABRI steady-states (during which the fuel was conditioning before the transient) the comparison of the code calculation results (SAS-SFR and ASTEC-Na) to the experimental values was performed for identifying where the code results differ from the likely state of the pins as they enter into the transient. Unfortunately, for validation of the selected test calculations, only a very few data are available as for AGS0 where the code predictions can be compared to AG0 (same steady-state as AGS0 but without subsequent transient) experimental data. The ASTEC-Na and SAS-SFR CABRI steady-state calculation results were found rather similar except for AGSO. For this test, the comparison of ASTEC-Na calculation results and available experimental data shows a rather good agreement (on Na coolant heat-up and fuel inner radius) but some discrepancies were observed for the fission gas behavior. Due to the specificity of this test however, the observations made for the fission gases can’t be extended to the other tests. That is why, despite some differences in details, the confidence in ASTEC-Na predictions regarding the parameters that are important in modelling the subsequent CABRI transients, is rather good.
For transient calculations, the model parameters in the different ASTEC-Na calculations have been in order to avoid any users’ effect. This especially concerns the gap heat transfer where the clad roughness was set to 5 µm and the fission gas release in the RIA model where recommended values for key parameters (DP-LIMIT, DP-PORO, INTER-SA and PERMEA) have been agreed. More details on ASTEC-Na models and RIA parameters can be found in previous JASMIN deliverables (Perez-Martin et al., 2016; Cloarec and Moal, 2012).
2.2.1 Fission Gas Release
In ASTEC-Na/RIA model, the fission gas initialization at the transient onset time is a key point, in calculating the fuel grain surface saturation by inter-granular bubbles, the amounts of fission gases in intra-, inter- granular bubbles and in porosities. Inter-granular gases can be released in porosities either by grain boundary saturation or rupture. The rupture occurs when the strength generated by inter- granular bubble over-pressurization overcomes the grain boundary mechanical resistance. When one of these criteria is reached, all the quantity of inter-granular gases is released immediately into the porosities and then to the free volumes provided that the open pores would be connected to them. From that time onwards, the quantity of inter-granular gases is left to zero and the intra-granular gas bubbles, which continue to migrate towards grain boundaries, go directly into the porosities. In all the ASTEC-Na calculations however, the rupture of the grain boundary by over-pressurization was preventing by choosing a very high value for the grain boundary material mechanical resistance (DP-limit parameter). This saturation occurring by intra-granular bubble diffusion towards the grain boundary is reached at some places in the pin at the end of the CABRI steady-states.
During fast power transients, the behavior of fission gases is expected to be simplified as compared to slow power transients because the fission gases have a smaller time to evolve. The accurate prediction of the initial fission gas content retained in the fuel and its distribution is then even more critical. The calculated fission gas retention close to experimental results for AGSO was found strongly over estimated for E9 (Figure 11). The difference between the two ASTEC-Na AGS0 calculations is explained by different inter-granular gas fraction at the end of the steady-state.

a) b)
Figure 11. Axial fission gas retention profiles at end time of AGS0 (a) and E9 (b) tests.
In ASTEC-Na, the intra and inter- granular gases as well as the gases in porosities (regardless they are connected or not to free volumes) are considered as retained in the fuel. For AGS0, most of the retained fission gases are calculated to be in intra-granular bubbles at end time of transient whereas for E9 the retained fission gases are equally distributed between intra-granular bubbles and open porosities. Taking into account only fission gas intra-granular fraction in E9 the retained concentration would be strongly decreased and better fit with the experimental measurements.
The data-code comparisons for fission gas releases are gathered in the following table.

Table 5. Fission gas releases at end time of transients.
Tests Exp. Measurements* SAS-SFR ASTEC-Na****
AGS0 33.1 (+17,5)** 39 (+21,5) 51.3-52.7 (+23,1-+24,4)
E9 83.16 (+30.1) 82.7 (+29,5) 78-80.5 (+27,9-30,6)
E7 n.a (clad failure test) 84 (+30,8) 57.3-60.5 (+7,2-+10,6)***
LT2 82 (pin piercing) (+10,4) 93.8 (+24,2) 87.8-88. (+12,3-+12,5)***
in brackets: overall fission gas release during CABRI steady-states and transients
*: relative experimental uncertainties are about ±10%
**: for AGS0, some additional release occurred during the transient overpower (+11.4%).
***: at calculation end respectively 0.45 s for E7 (vs 1.2 s) and 0.79-0.9 s for LT2 (vs 1.2 s)
**** ASTEC-Na predictions correspond to the maximum fission gas releases as the initial inventory of gases in the free volumes have not been deduced and because all the fission gases in open pores were assumed to be released.

In conclusion, the overall fission gas behavior during both fast and slow transients is rather satisfactorily calculated by ASTEC-Na, the peculiar AGS0 test being excepted. In that test, the great overestimation of the fission gas release mainly comes from the GERMINAL irradiation calculations at low power heating rates. In all other tests more than 80 % of the fission gases are predicted to be released at end time of the transients despite the slower time for fission gas to evolve in fast transients. It is quite in a good agreement with the experimental data considering the measurement uncertainty range and the fact that the open porosities were assumed to be all connected to the free volumes. Though the fission gas radial distribution and form were found to be in agreement with some of the available data, more tests should be investigated before concluding on that issue. The extension of the porosity network is indeed a key parameter determining the fission gas flow towards the free volumes.

2.2.2 Fuel Thermal-Mechanical behavior, melting limits, axial fuel expansion
In this section only the calculated results for fuel melting limits and transient fuel axial expansion will be displayed to be compared with experimental data. However for a clear understanding of the fuel thermal-mechanical behavior many other phenomena were also investigated (i.e. transient behavior of average fuel temperature, gap size and inner fuel radius). Due to numerical problems, the ASTEC-Na calculations couldn’t be run up to the end time of LT2 and E7 transients. The comparisons of ASTEC-Na calculations with experimental data were in that case limited and comparisons with SAS-SFR were made.
At PPN, the calculated fuel thermal behavior by SAS-SFR and ASTEC-Na was generally similar regardless the fast transients considered, with almost identical average fuel temperature increase during the power injection. The largest discrepancies were observed at the beginning of the transient, probably coming from the differences during steady state. For LT2, when the FOAM development is calculated by ASTEC-Na to start, lower average fuel temperatures are calculated as compared to SAS-SFR. Regardless the code, the gap was calculated to be closed at PPN throughout all transients. Some discrepancies between SAS-SFR and ASTEC-Na calculations were however observed where ASTEC-Na predicts a gap reopening during AGSO cooling phase and when fuel is progressively converted in FOAM in LT2 (potentially explained by the fact the pressurized foam is not taken into account in the calculation of the fuel stresses).
The size evolution of the central hole predicted by ASTEC-Na and SAS-SFR were found rather different with in case of fast transients a slower central hole closure of the annular fuel pellets in ASTEC-Na compared to SAS-SFR and a later reopening of the central hole of the solid fuel pellets during the cooling phase. In the slow power transient E9, the central hole closure at PPN as the fuel temperature is increasing, is predicted to be faster by ASTEC-Na compared to SAS-SFR.
In fast power tests, the transient axial fuel expansion is significantly over-predicted by ASTEC-Na while SAS-SFR reaches a better agreement with the experimental results (Figure 3.3). For instance, for LT2, the calculated axial fuel expansion by ASTEC-Na roughly doubles the experimental measurement (with curves still rising at the end of the calculation time). However, for LT2 this may again be explained by the fact that presently in the ASTEC-Na modelling the calculation of the fuel stresses is made as no fuel movement towards foam had occurred. For the slow power transient E9, a maximum axial expansion of about 4 mm was calculated by ASTEC-Na (Figure 12). Though it overestimates the experimental measurement of 3 mm, it is closer than the SAS-SFR result.
The comparison of the calculated melting limits (Figure 12) forAGS0 and E9 shows a good agreement of ASTEC-Na results with the experimental estimates, although for AGS0 the maximum radial extension around the axial position of the power peak is under-predicted. For LT2 and E7, ASTEC-Na calculated radial meting limit comparison with experimental data is not directly possible because calculations stopped prematurely. Common to all LT2 and E7 code predictions is the relatively large maximum radial extension reaching almost the fuel outer radius and leaving only a thin solid shell.
a) a)



Figure 12. Fuel melting limits (left) and transient axial fuel expansion (right) for AGS0 (a), E9 (b).

The new FOAM model for in-pin fuel relocation developed within the JASMIN project was successfully applied and partly tested for LT2. About 2.4 g of squirted fuel was found in the upper plenum, 0.2 g in the central hole, 3.8 g in the lower fertile and 7.8 g in lower plenum. This experimental phenomenon was not reproduced in ASTEC-Na calculations. No relocation of the FOAM component above or below the fissile column was obtained.
In conclusion, the fuel thermal behavior seems to be rather well predicted by ASTEC-Na where the calculated fuel melting limits for slow and fast power transients are reasonably well calculated. Only the axial fuel melting limits were found slightly underestimated while the radial limits are always in a good agreement with the experimental results. This in turn means that the gap behavior is also probably well calculated by ASTEC-Na. In all the calculations the gap was calculated to be closed at PPN in hot conditions. However, the fuel axial expansion was significantly overestimated in all the transient calculations. The overestimation was found greater for the fast power transients. The fuel mechanics strongly improved in this new version of ASTEC-Na allows running the calculations up to the end time of the transients for AGS0 and E9. However, for fast and energetic transients the calculations still stop too early. For low smear density fuel, the central hole size was calculated to decrease as the fuel temperature is increasing. In case of the slow power transient E9 where the fission gas induced swelling is also expected to be significant, the code predicts a complete and definitive closure of the hole. Contrarily to SAS-SFR, the code didn’t calculate a re-opening of this central hole even in cold conditions. At this stage, no conclusion about the fuel creep model validation nor about the fuel hydrostatic behavior at high temperatures could be drawn.

2.2.3 Clad Mechanical behavior (Clad Deformation) and pin failure for E7 test
The clad mechanical behavior was analysed through the final axial profile of clad deformation calculated as the difference between the clad outer radius profile at the end of the transient calculation and the nominal initial (as fabricated) clad outer diameter. For comparison of experimental measurements with calculated values it has to be considered that the first have been measured at room temperature, while latter are given at elevated temperatures corresponding to the respective conditions at the end of the transient calculation. Thus, the analysis of the clad deformation calculated by ASTEC-Na for LT2 and E7 tests were not relevant as, due to early stop of the calculations during the transient, the clad temperatures at the calculation end was very high (~1100-1200K).
For AGS0 (Figure 3.4) the two ASTEC-Na calculations predicting relatively close deformations significantly overestimate the experimentally measured one, while SAS-SFR is better in agreement with it. Also, as SAS-SFR, ASTEC-Na fails to predict the top-biased shape of the experimental profile which shows no deformation below 30 cm, while the shape of the calculated deformation profiles more or less follows the shape of the axial power profile. In E9 (Figure 13) only a small additional deformation (compared to post-irradiation state) of maximum 1.5% was created during the transient. The SAS-SFR results practically returns to the pre-transient profile and is in quite good agreement with the measured profile while ASTEC-Na overestimates the experimental deformation significantly. Part of the deviation between SAS-SFR and ASTEC-Na may however be attributed to the higher temperatures (difference of 100-150K) at the end of the ASTEC-Na calculations compared to SAS-SFR.
a) b)
Figure 13. Axial profiles of clad deformation in AGS0 (a), E9 (b) at end time of calculations.
Data about clad mechanical failure in E7 are given in the next table.
Table 6. Clad failure in E7.
Exp. Measurements SAS-SFR ASTEC-Na*
Time of failure (ms) 467 468 444-448
Failure site (cm BFC) 53 47.7 39.4-41.3
Axial power factor (%) 85 93 99-100
* Results obtained by two different calculations

ASTEC-Na predicts an earlier clad failure than in the experiment. Both ASTEC-Na and SAS-SFR codes fail in predicting the axial failure location. In ASTEC-Na the predicted axial failure location is close to the maximum of the power profile, while in the experiment failure occurred further upwards.
The analysis of calculated inner clad temperatures, cavity pressurization and clad deformation during the E7 transient where there are no experimental data suggests different causes for predicted clad failure. In ASTEC-Na, failure occurs at lower temperatures and higher deformation (strain) than in SAS-SFR. Also, SAS-SFR predicts a strong increase of the cavity pressure due to massive release of fission gases just prior to clad failure. It seems that clad burst occurs by the pressurization of the large molten pool kept inside only by the cladding (insignificant contribution from residual solid fuel shell) whereas ASTEC-Na rather predicts clad failure due to PCMI when still a sufficiently thick solid fuel shell exists which pushes the cladding. Then SAS-SFR is not only closer to the experiment concerning failure time and location, but also concerning the physical failure mechanism.

2.2.4 Gap heat transfer (Coolant Temperature)
For extracting conclusions on the gap heat transfer model, the calculated coolant temperatures (transient coolant heat-up at the top of fissile column and axial coolant temperatures profiles at different times) were analyzed by comparing against experimental data and SAS-SFR results.
The transient sodium temperatures calculated by ASTEC-Na at TFC are quite quickly under-predicted for AGS0 (Figure 14) and LT2. Depending on the end time of ASTEC-Na calculations this leads to more or less significant under-predictions of the Na coolant temperatures at the top of fissile column (~50K for AGS0 at the end of the cooling phase). Though part of the underestimation in AGS0 could be explained by a re-opening of the gap calculated in ASTEC-Na from about 0.8 s after the transient onset, from 0.4 s the calculated kinetics of sodium heat-up is less than measured. ASTEC-Na is however better than SAS-SFR in predicting the Na heat-up kinetics and the time at which the maximum coolant temperatures are reached. In E9, the calculated ASTEC-Na coolant temperature is in good agreement with the lower measured transient temperature (Figure 14). The temperature rise is almost linear as the power increases and rather similar to the measured one. A small difference in Na temperatures at the top of fissile column between ASTEC-Na calculations and measurements can however be observed (~10-25K) due to the deviations in experimental measurements from 40 s after the transient onset.
Except for LT2, the ASTEC-Na axial coolant profiles at different times are in good agreement with experimental results with a predicted sodium temperature corresponding at the top of fissile column to the lowest measurements. For E9 the calculated temperatures were found to deviate from the experimental ones from the top of fissile column.
In conclusion, in fast transients the kinetics of sodium heat-up is slightly underestimated by ASTEC-Na leading rather quickly during the transient to lower calculated sodium temperatures as compared to measurements. In slow power transients however a very slight underestimation was observed in sodium heat-up kinetics with only 15 K difference in sodium temperatures observed at top of the fissile column close to the transient end. Some discrepancies between different ASTEC-Na calculations at top of the fissile column were observed that are explained by a different modelling of both the heat loss through the shroud and the test section (i.e. whether fissile pin axial region or the whole test section was considered).

a) a)


Figure 14. Transient coolant temperatures at TFC and axial coolant temperature profiles
for AGSO (a), E9 (b).

ASTEC-Na is conceived as an integral Beyond Design Basis Accident (BDBA) code for sodium fast reactors. As such, in-containment source term behavior is addressed to be included by adapting the LWR containment module (CPA) to the Na environment. Within the JASMIN project, two models have been developed and implemented in ASTEC-Na CPA module: particle generation from pool fires (PG model) and aerosol chemical ageing (AERAGE model). The modelling capabilities of the new module have been tested on containment thermal-hydraulics, aerosol behavior and ageing both by code-to-experiment and code-to-code comparisons. A synthesis of the code benchmarking conducted and reported elsewhere (Herranz et al., 2016) is given in this section.

2.3.1 Thermal-hydraulics
In AB1 and AB2 tests (Figure 14 and 15) from the experimental data, gas temperature evolution may be described in two main phases: heat-up and cool-down. The heat-up phase is, in turn, split into 4 stages: a first period (about 60 s) in which a sharp increase of temperature occurs; a second one along which temperature hardly changes for about 500 s; a third one in which temperature increase proceeds gently for some 2000 s; and a last period until the fire quenching, where heat-up slightly slows down.
The cooling down shows two steps: a fast one due to chemical reactions brought to a halt and the isolation of the hot sodium by covering the burn pan, and a longer and more progressive one, as heat is steady lost to the vessel walls by natural convection.
Consistently with major experimental observations, CPA* and other calculations also show the two main experimental phases and differences in magnitude are roughly between 10 and 20 K. However, differences show up when focusing in each of them. On one side, only two periods are predicted in the heating phase and, regardless if they match the maximum temperature at the end of the phase, all in all measured heating rates are generally poorly captured. This is not surprising as none of the models took into account the Na-injection effects discussed above and set key parameters on arbitrary basis requiring validation. On the other, cooling rate due to fire quenching is largely overpredicted by CPA* at the beginning of the cooling phase, which might indicate a too high energy source during the fire period (0-3600 s).
The two major points discussed above were more or less found also applicable in the three other tests analyzed (F2, F3and EMIS10b).

Figure 14. Atmosphere temperature (AB1).
Figure 15. Atmosphere temperature (AB2).
From the modeling perspective some takeaways have been withdrawn:
- Energy source into the gas atmosphere during Na pool burning should be further investigated, so that CPA* models are based on a sort of mechanistic understanding instead of on empirical and sometimes arbitrary settings.
- The single-node approach seems to be capable of following data trends under soft transients; however, under fast transient conditions driving to atmosphere non-uniformity for a while, it should be expected to undergo significant deviations.

2.3.2 Aerosol behavior
The three main metrics used in the code benchmarking exercises are the suspended mass concentration, the aerodynamic mass median diameter (AMMD), and the deposition of material on different surfaces at the end of the tests.
During the aerosol injection period it was generally hard to identify a clear trend especially due to the large uncertainties associated with the experimental data. In AB1 (Figure 16) during the burning phase (0-3600 s), a quasi-steady state between 0.02 and 0.03 kg/m3 was measured, so that there was a balance between particle generation and aerosol deposition; afterwards, once the Na fire is over, concentration decayed progressively. Despite CPA* and other calculations results fall within experimental uncertainties during the quasi-steady state period, none of them follows the observed steady trend before the sodium pool fire ending. Besides, the experimental depletion rate during the first 1000 s of the depletion phase is about twice higher than the predictions.
Regarding AMMD evolution (Figure 17) the data experienced a jump from about 6 μm to approximately 8 μm between 4000 and 5000 s. This change is nothing to do with a sort of agglomeration enhancement, but it is the result of the end of injection of small particles from the fire, so that naturally the size distribution shifted to bigger sizes. Then, a progressive decrease of the particle size can be observed as a consequence of preferential deposition of large particles by sedimentation. CPA* and also CONTAIN results reasonably follow measurements; however, the other calculations noticeably under predicts particle size. Even though the reasons for this different behavior should be investigated further, what is important is confirming the CPA* capability to capture particle size changes.
Finally, regarding the final mass distribution (Figure 18), even though some tools got it qualitatively, noticeable differences in CPA* results are noted in magnitude. In general, it seems that CPA* and the other codes overestimate the importance of thermophoresis as a deposition mechanism, predicting a much higher mass fraction on walls and vertical surfaces than observed.
In AB2 (Figures 19-21) a soft growing trend from around 0.02 kg/m3 to 0.035 kg/m3 during the aerosol injection period while during the depletion phase, although qualitative similar to AB1 one, an initial decay rate nearly 3 times slower was observed. These variations with respect to AB1 make CPA* calculation resembles concentration measurements more closely. Anyway, the uncertainty associated to the data point at the end of the Na fire period might well be on the root of that difference. Regarding AMMD evolution and final mass distribution the main insights from AB1 can be applied due to similar experimental results.
For the other tests investigated the comparison between calculations and measurements were found most of the times highly likely unsuitable and thus were not reported here.

Figure 16. Airborne concentration (AB1).
Figure 17. AMMD (AB1).
Figure 18. Na mass distribution (AB1).

Figure 19. Airborne concentration (AB2). Figure 20. AMMD (AB2). Figure 21. Na mass distribution (AB2).

2.3.3 Particles ageing
The moderate O2 concentration reduction during the burning time of sodium is rather well captured by CPA* with a reasonable accuracy. As for aerosol speciation, experimental data is only available for AB1 test as mass fraction of the involved species at five times (16, 46, 190, 610 and 7200 min); calculations extended up to 160 min though, so a limited number of comparisons are set below.
As for specific observations at given times, at 16 min (Figure 22) the aerosol was predominantly NaOH (82 %) coming from the reaction of sodium-oxides aerosol particles with the ambient humidity in the atmosphere; the rest distributes mainly between Na2O2 (15 %) and Na2CO3 (3%). Somewhat later (46 min), most airborne aerosols consisted of Na2O2 (56%) and NaOH (43%), which indicates the total depletion of the water vapor content in the atmosphere while aerosol formation is still ongoing. From 46 min. on the experimental observations look controversial as they indicate the availability of some steam of an unknown source; anyway, those later times are out of the period in which predictions-data comparisons can be set. As for CPA* predictions, none of them showed remarkable similarities to data at the two times (16 min and 46 min) comparisons are feasible (Figure 22). According to those comparisons, as long as steam is available the ageing model produces less NaOH than observed; in other words, given that the reactions above are diffusion controlled, this means that steam diffusion coefficient is noticeably underestimated. Differences between CIEMAT and GRS indicate that f3 and f4 values assumed by GRS fit what experimentally observed (no Na2O measured airborne), whereas CIEMAT’s calculations resulted in a fraction of particles in monoxide form. Additionally, discrepancies in the initial atmosphere humidity in both simulations might slightly affect the composition share. As carbonaceous species were hardly formed, no conclusion can be derived as for the model performance.
As for the settled aerosols (Figure 23), data show that particles consist mostly of NaOH (61%) and a good fraction of Na2O2 (36%) and a minor share of Na2CO3 (3%). As pinpointed above, the ageing model substantially underestimate the transformation to NaOH and to Na2CO3 (although this is less noticeable due to little formation, even experimentally). These evidences might also indicate a too low diffusion coefficient in the model; however, the sensitivity calculations performed within the project do not support this potential explanation. In addition to pure chemical considerations, it is noticeable that GRS calculations underestimate sedimentation in about a factor 1.7.
Unfortunately, no data on airborne aerosol composition are available for AB2. Nevertheless, huge discrepancies have been found in settled particles composition: whereas measurements indicated more than 80% of particles in the form of carbonates (Na2CO3), GRS estimated that particles would mostly consist of NaOH (>95%).

Figure 22. Suspended species (AB1).
Figure 23. Settled mass percent of species (AB1).
Due to the lack of experimental data, the validity of the ASTEC-Na neutronics model was assessed through a benchmark with TRACE and SAS-SFR codes under a selected accident scenario which is an unprotected loss of flow accident (ULOF) in a pool-type sodium cooled fast reactor. As a benchmark reactor model serves the “Reference Oxide core design” concept of the CP-ESFR project under Beginning of Life (BOL) core load conditions.
Based on the point-kinetics theory, the ASTEC-Na neutronics model has currently two different methods to compute the reactivity feedbacks: one model (called LOCAL) uses, for normal and incidental conditions, coefficients related to temperature (pcm/K), the other model (LOCALM) is based on mass variation and allows extending the scope of the ASTEC-Na neutronics model to post boiling onset analysis. More specifically, the model (called LOCALM) is able to predict reactivity evolution resulting from sodium boiling, and limited fuel in-pin relocation. The work of assessment of LOCAL model performed with the benchmark exercise in the framework of WP2.4 activities and described in Ammirabile et al. (2016a) has been complemented by the analysis of the LOCALM model on the base of the same benchmark exercise (Ammirabile et al., 2016b).
The first phase of the ULOF simulation covers the single phase coolant heat-up up to coolant boiling onset considering relevant contributing reactivity feedbacks. The second phase deals with the boiling phase of the transient up to the occurrence of first fuel pin failure, again considering relevant contributing reactivity feedbacks. The specification for this ULOF benchmark exercise is given in Ammirabile et al. (2016a) including the scope and stages of the simulation, the characteristics of the reference core, the core physics and thermal hydraulic conditions as well as the channel grouping of the subassemblies (SA) in the core. This ULOF specification is based on the SAS-SFR analysis performed in the CP-ESFR project where the neutron physics calculations providing the power profiles and the reactivity coefficients were done with KANEXT code (Ammirabile et al., 2016a).
The results show a relatively good agreement among the codes both qualitatively and quantitatively in computing the progress of the main core parameters. The initial benchmark specifications have been sufficiently accurate to avoid larger discrepancies due to user interpretation. The steady-state results of calculations show a very good agreement between SAS-SFR and ASTEC-Na with some discrepancies observed for TRACE (Table 17).

Table 17. Comparison steady-state values.
Total core pressure drop [bar] 10.61 5.23 4.80 4.11
Total core mass flowrate [kg/s] 20700.00 18775.60 18741.10 18906.00
Mean Na core outlet temperature [°C] 545.88 545.00 544.98 545.00
Avg core fuel temperature [°C] 1218.72 1202.68 1228.18 1201.00
Avg core cladding temperature [°C] 496.78 484.62 N/A 484.10

The reactor power evolution during the first 35 seconds of the transient calculated by SAS-SFR and ASTEC-Na (respectively using LOCAL and LOCALM model) is shown in Figure 24. The calculations of ASTEC-Na/LOCAL and SAS-SFR are in good agreement. However, the ASTEC-Na/LOCALM shows a lower peak due to its slightly lower total reactivity feedback.
By analyzing the individual reactivity feedbacks, it appears that ASTEC-Na predicts similar evolutions of the coolant and Doppler reactivity feedbacks while over-predicts the contribution of the fuel expansion and under-predicts that of the cladding expansion to the total reactivity feedback. Given the good agreement of ASTEC-Na results using the LOCAL reactivity feedback model (based on temperature), it appears that by using the LOCALM reactivity feedback model ASTEC-Na finally computes a different fuel and cladding expansion than SAS-SFR with the same material temperature evolution.
After the onset of boiling, the results show first a small decrease of total reactivity calculated by all codes and then the sharp increase due to the coolant feedback. In this calculation the maximum positive total reactivity shown by SAS-SFR is over 340 pcm, and then the negative fuel relocation feedback plays the major role in the reduction of the total reactivity. This results in a total power excursion of about 50 times the nominal power.
The calculation performed with ASTEC-Na beyond the boiling onset and using LOCALM model shows the capability of the code to compute boiling conditions together with a large power excursion due to the large positive reactivity insertion (Figure 25). Due to the sharp power excursion the calculation stops abruptly caused by a convergence issue in the thermal-hydraulic module CESAR. This impedes to assess the maximum power peak reached during the transient. The overall behavior is qualitatively in agreement among the codes. Discrepancies are observed in the progression (in time and value) of the power excursion also among the same codes. The reason of such discrepancies needs to be investigated further, but it seems strongly linked to the evolution of the boiling front progression in the individual codes. Indeed, if all the codes predict the initial position of the boiling front at the top of the fissile zone, significant differences in the evolution of upper boiling front between ASTEC-Na(that used a simplified modelling of the core layout) and SAS-SFR that matches the delay in the power excursion .

Figure 24. Power evolution in transient first phase. Figure 25. Power evolution after Na boiling onset.

As the main goal of the benchmark was to test the new neutronics model but not to predict boiling front, more detailed model was not considered necessary at this stage. Nevertheless, it is important to stress that it is practically impossible to assess any neutronics model once we move from reactivity coefficient effects and power to thermal-hydraulic variables (temperatures, pressures, mass flow rates). Therefore, any neutronics assessment requires attention on the thermal-hydraulics behavior to clearly discriminate thermal-hydraulics’ uncertainties from neutronics’ ones.
The neutronics model implemented in ASTEC-Na was found able to compute the adequate power evolution taking into account the main reactivity feedbacks but provided that the reactivity coefficients are calculated based on the characteristics of the model that currently imply a fixed axial meshing. The overall exercise has proven the validity of the point kinetics model implemented in ASTEC-Na in providing adequate feedback in terms of power evolution during transient when compared to other codes also using point kinetics. Hence the insights of the benchmark (e.g. the influence of thermal-hydraulics features) can be usefully extrapolated to other scenarios provided that they fall under the domain of validity of the point kinetics model.

The main outcomes of the performed work within the four different sub Work-Packages in evaluating and validating the ASTEC-Na code models are summarized and discussed in the following section. The models under consideration deal with sodium thermal-hydraulics, fuel pin thermos-mechanical behavior, in-containment source term and core neutronics. The development status of these models at the project end is far more advanced than before but reflects a high level of disparity in development status depending both on the complexity of the considered phenomena to be simulated, the availability of experimental data for validation and the state of the simulation tool when the project was launched. This is brought out clearly in the SWOT analysis (Strengths Weaknesses Opportunities and Threats) of the tool that has been conducted in order to assess the progress made in ASTEC-Na models and is presented in the following sections.
The ASTEC-Na code has evolved during the JASMIN project into a versatile tool that can be applied to simulate the Na thermal-hydraulics in the primary vessel, thermo-mechanical behavior of fuel pins up to cladding failure, the Na pool fires and related aerosol physic-chemistry of Na-bearing aerosols, the evolution of the main reactivity feedbacks and neutronics power during a transient.
3.1.1 Strengths
This section reports the strengths of ASTEC-Na regarding successively the sodium thermal-hydraulics, the fuel pin thermos-mechanical behavior, the in-containment phenomenology and the core neutronics. These strengths were pointed out during the performed validation work and code-to-code benchmarking.
As far as sodium thermal-hydraulics is concerned, it can be stated that:
- Good robustness of the code which allowed completing a large number of simulations with different degree of complexity, from the simple geometry of CABRI tests to the description of PHENIX and SPX power plants.
- In general, the validation against all experiments pointed out good ASTEC-Na performances on the simulation of single-phase thermal-hydraulics. Furthermore in the CABRI and SCARABEE experiments the onset of boiling is well predicted.
- The simulation of PHENIX NC tests and SPX stabilization and NC tests, pointed out that ASTEC-Na can simulate the whole power plant including a multi-channel core, the pool-type primary system and the secondary circuits. The predictions of steady-state and transition to natural circulation conditions are good enough.
- The quality of ASTEC-Na results on experiments simulation is similar to what exhibited by more mature and mechanistic thermal-hydraulic codes such as CATHARE, RELAP5-Na and RELAP5-3D. The results of severe accident codes such as SIMMER-III and SAS-SFR are also close to the ones of ASTEC-Na.
During the project, ASTEC-Na was applied to simulate the thermo-mechanical behavior of fuel pins up to cladding failure in tests representing a bandwidth of RIA accident conditions from slow ramps up to fast structured TOP. From the analysis of calculation results it can be stated:

- Prediction of steady-state prior to onset of the transient is in good agreement with experimental data and with the well-established code SAS-SFR.
- The overall fission gas release fraction during the considered fast and slow transients is rather satisfactorily calculated by ASTEC-Na (with exception of the AGS0 test where the overestimation of the fission gas release mainly can be attributed to the GERMINAL irradiation calculations at low power heating rates). In the other transients more than 80 % of the fission gases are predicted to be released at the transient end, which is quite in a good agreement with the experimental data considering the experimental data uncertainty range. Also the fission gas distribution and form were found to be in agreement with some of the available data.
- Concerning fuel thermal-mechanical behavior, melting limits and axial fuel expansion significant improvements have been reached in the calculations with ASTEC-Na V2.1 compared to the calculations with the previous version V1.1. Availability of the visco-plastic modelling if the fuel yielded much more realistic behavior partly in good agreement with experimental data. Especially in the tests AGS0 and E9 results on fuel melting limits and axial expansion are in satisfactory agreement with available experimental data. Functionality of the newly introduced FOAM model could be verified in the calculations for test LT2, although validation with respect to in-pin fuel motion was not yet possible.
- With respect to clad deformation the ASTEC-Na V2.1 results have also significantly improved compared to V1.1. Although still with some overestimation, results for final clad deformation profiles come close to PTE data. Pin failure was qualitatively correctly predicted for test CABRI E7 (although too early and at lower elevation than in the experiment), proving in principle that ASTEC-Na is able to describe the relevant physical processes up to and including pin failure.
Improved documentation especially regarding the RIA model help the user in running the codes and in obtaining more reliable results that comes closer to experimental data and results of other codes. Differences in calculated results obtained by different users are very small reflecting the robustness of the code and the few parameter values available in the code to be set by users.
Extension of existing and introduction of new models carried out during the JASMIN project, e.g. the FOAM model for in-pin fuel motion, have shown the feasibility to readily adapt ASTEC-Na, thanks to its modern code architecture, to new requirements that may arise from new materials and innovative fuel design features in next generation fast reactor concepts. Compared to existing codes, the modelling in ASTEC-Na follows a more mechanistic approach, especially concerning the fission gas behavior. Relying less on empirical correlations that were fitted to previous experimental data such as from the CABRI test series, it has the potential to be successfully applied in a predictive manner also to new designs.
The “in-containment source-term” analysis was focused such as the WP2.3 of JASMIN was formulated. It means that any other source term issue, like fission product scrubbing in Na pools, was left out of the scope of WP2.3 and it is, hence, beyond the scope of the present analysis. From the performed analysis of the in-containment models in ASTEC-Na, it can be stated:
- The present version of ASTEC-Na (CPA**) has models implemented for containment thermal-hydraulics, aerosol generation from Na pools and chemical ageing while airborne in the atmosphere in presence of steam and carbon dioxide. Even though still incomplete, the current ASTEC-Na capability to model in-containment source term has been noticeably enhanced with respect to what existed before the JASMIN project.
- Several approaches are presently implemented in CPA** for particle generation from Na-pool fires. Even though still preliminary, this diversity might provide ASTEC user with some flexibility to model any scenario.
- Comparisons to data have shown that the anticipated thermal evolution in SFR containment during a Na-pool fire might be captured whenever models parameters are adequately set and containment is meshed properly. In particular, the most sensitive parameters are those involved in the chemical energy distribution between Na pool, containment atmosphere and surrounding structures.
- Similarly, aerosol models show a promising response in terms of order of magnitude of airborne concentration, dominant depletion mechanism (i.e. sedimentation) and particle size variation.

Finally, regarding the core neutronics, it can be stated that:
- Compared to similar codes ASTEC-Na is characterized by a high level of modularity allowing a rather flexible use of the code.
- The point kinetics model implemented in ASTEC-Na is reliable to calculate the power evolution in those scenarios that fall under its domain of validity.
- The implementation of reactivity coefficients is also highly flexible allowing them to be defined over time intervals. In addition if the reactivity coefficients have been computed first on a dedicated meshing, the latest version of ASTEC-Na is capable to accommodate it in case its meshing is more refined than the ‘coefficients’ meshing.
- As the neutronics properties of the cores are provided to the neutronics modules of ASTEC-Na as external input data they can therefore be easily modified.
- The point kinetics models implemented in ASTEC-Na is much faster than space-time kinetics ones. It makes then ASTEC-Na very suited to perform sensitivity studies. This feature might be quite useful in the designing phase or when uncertainty effects want to be assessed.
- The evolution of reactivity is decomposed on different effects, which gives useful information to analyze the unfolding of the transients and makes the module evolutionary, since contribution of other effects (control rod expansion for instance) can directly be added.

3.1.2 Weaknesses
Some weaknesses in the ASTEC-Na models have been identified in sodium thermal-hydraulics, fuel pin thermos-mechanical behavior, in-containment phenomenology and core neutronics that will be reported in this section
Regarding the sodium thermal-hydraulics it mainly concerns the two-phase flow calculations:
- The simulation of two-phase thermal-hydraulics is a more difficult challenge for the code. In particular, CABRI and SCARABEE inlet and outlet flow rates after boiling onset are not well captured and also the evolution versus time of the upper boiling front is not well predicted.
- The calculation of radial propagation of two-phase front in SCARABEE experiments is also not satisfactory. It is thought that the 2D axial-symmetric model of ASTEC-Na, based on rough meshes including solid structures and fluid, is not sufficiently accurate to well reproduce the complex 3D phenomena during boiling.
- It must be however observed that the above mentioned ASTEC-Na weakness dealing with two-phase flow simulation are shared with the other applied severe accidents codes.
- In some cases, as in the simulation of CABRI BI1 test and KASOLA loop pre-test calculations, not negligible discrepancies have been remarked between ASTEC-Na calculations performed by different partners, indicating an important user effect on the results.
Regarding the fuel pin thermos-mechanical behavior though the ASTEC-Na code has significantly evolved during the JASMIN project especially regarding the simulation of the thermo-mechanical behavior, the modelling capabilities of the latest version of the code only extend up to pin failure. In the future, this has to be extended, either by incorporating additional models or by coupling to a code like SIMMER for the later phase. Although, the code has to be interfaced with a fuel irradiation code that will calculate the fuel pin conditioning before the transient. Within the JASMIN WP2.2 the fuel t0 state was calculated by GERMINAL v1.4. Though this version, which is quite old now, was fully validated by that time more recent versions of the codes have been developed now that includes improved modelling. Furthermore, the use of two different codes (one for the t0 state and the other one for the transient) requires the transformation of the output of GERMINAL to ASTEC-Na input data. Though this procedure is quite through the use of the dedicated TOSCAN interface, the two codes employ different physical models (i.e. for the fission gases) and this complicates the connecting procedure.
From the analysis of calculation results performed within the JASMIN WP2.2 some weaknesses have been identified:
- The new FOAM model has not yet the capability to correctly simulate the in-pin fuel relocation driven by the over-pressurization of the molten materials. The foam viscosity takes very high values even if only a very small part of the fuel is frozen in the foam. Then the model using a common velocity for the fuel and gas material, a freezing of the fuel materials leads to a very low foam velocity and to a total blockage of the central hole. Further development will be needed to correct this non-physical behavior of the foam.
- The pressure computed by the FOAM model is not yet taken into account in the modelling of the cladding mechanics. This can lead to a wrong prediction of the clad failure because the mechanisms responsible for it will not be adequate. Further developments are then needed in order to take into account the FOAM feedbacks into the RIA model.
- The RIA model implemented in ICARE for computing the fuel thermos-mechanics and issued form SCANAIR LWR code developed for RIA transients though greatly improved during the project till shows some deficiencies in correctly predicting both the axial fuel expansion and the clad deformation by PCMI. Also, the simulation of hydrostatic fuel at high temperatures prevents from adequately calculate the molten fuel pressurization and then the subsequent potential clad failure. The validation work will have to be extended to more experiments to more deeply check the fuel behavior at high temperatures. Also the robustness of this model responsible for the early end of calculations in fast energetic transients will have to be improved and the associated numerical problems solved.
- The more mechanistically based modelling approach (as compared e.g. to SAS-SFR) followed by ASTEC-Na in simulating the fission gas behavior requires the user to provide data that is not readily available in many cases (DP-LIMIT/DP-PORO for grain boundary/porosity rupture stress limit, fuel permeability for fission gas migration, INTER-SA for grain boundary initial saturation). In the absence of experimental data within fast reactor conditions, sensitivity analyses on their impact will have to be performed for providing recommended values.
- The gap heat transfer modelling, though quite satisfactorily sodium temperatures were simulated in the tests considered within the WP2.2 benchmark, will have to be improved by considered the JOG.

Regarding in-containment phenomenology it can be stated that:
- Major phenomena potentially affecting in-containment source term are still missing in ASTEC-Na, mostly due to lack of data. Some key examples are: particle generation from Na spray fires, fission product entrainment from Na-concrete interaction, thermal and chemical behavior of fission products in Na-containing aerosols.
- Key phenomena for in-containment source term are described by heavily parametrized models (i.e. chemical energy distribution and Na oxides partition between pool and atmosphere). In other words, a deep understanding of those phenomena that would allow a mechanistic modeling is missing.
- The current database does not support a sound and reliable definition of default values for the parameters embedded in the mentioned models. The comparisons to data set within WP2.3 are far from a thorough validation. A number of reasons underpins this statement: the number of such comparisons is a way too reduced due to scarcity of open data and project resources constraints; often variables recorded are insufficient to qualify any of the models developed (most data come from integral tests and separate effect tests data are either lacking or difficult to scale up); and, in addition, measurements are many times affected by large uncertainties that prevent even from soundly stating trends. Two specific examples are: the large experimental uncertainties existing in aerosol concentration during the Na-pool fire period, which prevent from withdrawing any specific insights regarding models of particle generation and/or aerosol mechanisms in codes; and the lack and/or low confidence of the available experimental data on H2Ov and CO2 consumption and or aerosol speciation, which hinders a proper AERAGE model validation.
This being said, the comparisons set have given meaningful information.
- Deviations from data trends have been observed both in the airborne concentration of aerosols and in their chemical composition. Although comparisons are inconclusive yet, it seems highly likely that the models developed need review, extension and further assessment. This is the case, for example, of particle agglomeration over the pool at high concentrations, the presence of Na as a vapor fluid in the containment module of the ASTEC-Na code or the Na2O reactions.
- Benchmarking with other codes has not given assisted to devise further developments in ASTEC-Na. On one side, ASTEC-Na containment module is at the forefront of in-containment source term modeling. On the other, the database uncertainties prevent often from highlighting a “best response” code.
- The single-node approach seems to be capable of following data trends under soft transients in which intense natural circulation is effective in making the gas phase uniform; however, under fast transient conditions driving to atmosphere non-uniformity for a while, it should be expected to undergo significant deviations.

Finally, regarding the core neutronics, the identified weaknesses are:
- The power produced in various sub-assemblies channels in a core, the axial power profiles and the spatial distribution of the reactivity coefficients cannot be computed by ASTEC-Na. So it is required to perform neutron physics evaluations of the core with dedicated codes and methods in order to provide that information to ASTEC-Na.
- The point kinetics is valid until the flux shape (spatially and in terms of energies) is not excessively perturbed. As soon as material relocation starts (as early as the beginning of the boiling phase), this assumption becomes already questionable. The assumption of validity of the point kinetics model up to hexcan failure needs to be confirmed by an estimate of the divergence of the two models (point-kinetics vs 3D space-kinetics model) as it will depend on the material relocation occurring before hexcan failure and thus on the accidental scenario considered. However, in case of extensive relocation of materials to the transition phase after the hexcan failure, the point kinetics approach is not anymore valid and application of space-time models is rather mandatory as materials might relocate radially. The ability of ASTEC-Na to use time-dependent reactivity coefficient might help to overcome this difficulty (provided that also axial power distribution may change with time), but it requires to perform a lot of iterations (to have adequate coefficients for a time period, the state of the core during this time period has to be known).
- The user misses a manual where recommendations for creating an input deck for the neutronics module are provided.
- Some reactivity feedbacks such as control rods drive lines expansion feedback and core diagrid radial expansion feedback that become of importance for the evolution of the reactor behavior in case of long time transients are still missing though they will be implemented in the future version of the code.

3.1.3 Opportunities
ASTEC-Na is one of the few codes nowadays supported (in contrast with old SFR safety codes with no development support anymore) which has precise fuel-pin thermo-mechanical models for describing the behavior of the pin during accidental conditions. Also ASTEC-Na is devoted to model the in-containment phenomenology. Though the in-containment source term modelling in SFRs during severe accident still need to be extended in ASTEC-Na, such further development will make the code able to evaluate the source term produced by the migration of activated fission products inside the reactors and likely to be released to the environment in case of accidental situations. The integrated features of the ASTEC software platform will thus allow simulating in a single code what is generally today simulated in separate codes (i.e. SAS-SFR and CONTAIN-LMR...). Finally, the code is capable to model the entire reactor system, including the primary, intermediate and secondary systems.
The code in assessing the impact of different transients has the potential to be a unique code able to support the safety assessment of the reactor behavior in different plant states ranging from normal operation to DBA and severe accident conditions. This provided that the code is interfacing to an irradiation code that calculates the fuel t0 state and to a neutron physics codes providing the reactivity coefficients. The opportunity would then be to develop an interface with an irradiation and a neutron physics codes, so that ASTEC running requires the less possible user work. All these facts make this code very promising for assessing future SFR designs. Indeed, the high modularity of ASTEC-Na (that characterizes the code compared to similar codes) allows to work separately with its module and to integrate it in the code at a later stage, so that any development, implementation and assessment is more straightforward than once implemented in the integral code. In addition, the flexibility in defining the core geometry, materials composition and reactor components makes ASTEC-Na ready to study new SFR designs with fertile layers in outer radial or inner axial core regions (such as ASTRID design), subassemblies with an inner duct channel to induce fast fuel axial relocation (such as FAIDUS design), or new safety systems to shut-down the core power (such as melting-induced insertion of absorber material into the core). The only requirement in all cases is the validation of the corresponding models; however, the fact that they are already implemented in ASTEC-Na is already a big advance.
The JASMIN project represented a unique opportunity to bring together different actors of nuclear safety in Europe around ASTEC-Na code. The analytical work performed during the project has highlighted a major need: to build a robust and extensive database that can be used both to support individual model development and to validate more “integral” tools, like ASTEC-Na.
3.1.4 Threats
Part of the identified threats was explained by the current status of the code which is still under development. In order for ASTEC-Na to meet the requirements for a versatile and reliably applicable tool for the simulation of SFR accident scenarios in the frame of reactor safety studies, it is of utmost importance that the numerical stability issues that have surfaced in some calculations can be solved. Further, taking into account the partly large calculation times that were required for single pin tests, calculation times could become prohibitively large for full plant calculations and should be reduced. The numerical issues as well as the computation times are very challenging for a code where both empirical and mechanistic model were embedded. If improvements concerning these aspects cannot be achieved in due time, ASTEC-Na may run into the risk to just become a code casually used by a few experts.
The more mechanistically based modelling approach followed by ASTEC-Na (as compared e.g. to SAS-SFR) requires to provide data that is not readily available in many cases. In order that ASTEC-Na becomes also applicable for user outside a small circle of true experts it will be necessary to provide guidance (recommendations, best practice guidelines). Such can only be derived from a broad validation basis. Up to now, the validation basis of ASTEC-Na is limited to calculations of a few experiments and must be definitely extended in the future. However, it cannot be guaranteed, even if the effort is significantly increased, that it is achievable to find consistent settings that lead to validated models with extrapolation capabilities especially regarding future SFRs designs and new types of fuels and materials.
Finally, the extension of the modelling capabilities of the code has to be established in time otherwise it might lose the opportunity that ventures like ASTRID will give in the near future and thus further development may lead to a dead end. For instance, the ability to correctly predict power evolution for heterogeneous cores such as ASTRID-like core (different fissile length, sodium plenum on top of the core, fertile plate inside the active zone) will be a key point to evaluate their safety. Several organisms are designing such cores (BN800, ASTRID) to decrease sodium void effect and the ability to model their neutronics properly with point kinetics model will be challenging. Continuity of ASTEC-Na development and assessment requires not-negligible resources. Lack of support would make ASTEC-Na be of little usefulness, since the tool is still far from being mature and, besides, other tools are being developed at this time (i.e. MELCOR-Na).
When performing the benchmark exercises some points were especially pointed out as, for example:
- Scarcity of data in the open literature handicaps any development and validation necessary for a safety analytical tool. This statement is not restricted to phenomena addressed within JASMIN but also to others potentially relevant for accidental scenarios.
- Confidentiality of some data gathered in the 1970’s and 1980’s and the access restriction to those who were not involved in the work at the time, is risking their preservation, as they were not usually stored in digital support. This makes highly likely repetition of experiments ran decades ago. In addition, these ‘old data’ are sometimes affected by a non-negligible uncertainty.
- Setting up new experiments requires a strong financial support and careful thinking to optimize resources. At present, there are few national and international bodies willing to do such an investment, without which any progress is hard to achieve and to believe in.
Part of identified needs can be explained by the current status of the code which is still under development. In order to have a standalone computational tool for safety assessments of SFRs, the code development still have to be extended, the models validated and an exhaustive documentation provided. More especially the identified needs were:
- There is the strong need to finalize the availability of an exhaustive documentation. This should in particular include a manual where the neutronics models are described together with a set of default or recommended values. In addition some sample neutronics input decks could also help beginners to build their first model. More precise users’ guide-lines, to be provided by code developers, might be useful to reduce the observed user effect.
- The robustness of the code has to be improved by the resolution of the numerical problems still existing i.e. in handling sharp power excursions both in thermal-hydraulic CESAR module and in thermos-mechanical RIA models.
- The computational time when the RIA model is used has to be reduced especially for sensitivity calculations to be performed. This is particularly important for integral reactor calculations when RIA and neutronics models are activated together. The algorithm to solve the different equations should then be revised with the goal to reduce the computational time by at least those of similar codes. This would allow to perform a consistent number of sensitivity calculations as well as to increase the complexity of the problems analyzed
- An interfacing of ASTEC-Na with the most recent versions of fuel irradiation codes (i.e. GERMINAL or TRANSURANUS) is certainly necessary as well as an interfacing with neutron physics codes to create a standalone computational platform for safety assessment.
- Even though the new models implemented in ASTEC-Na/ICARE and in ASTEC-Na/CPA* have enhanced the code capability to address severe accidents in Na-cooled reactor primary vessels and containments, some work still have to be performed. Regarding the fuel thermos-mechanical behavior, some phenomena still have to be taken into account and modelled: JOG, cavity pressurization, molten fuel ejection and potential interaction with the Na coolant after pin clad rupture. Regarding the new models implemented in CPA* a thorough review of their fundamentals and implementation would be desirable. A less parametrized approach to Na-fires and particle generation should be pursued and uncertainties in the gas diffusivity in the ageing model should be ascertain. Also, major phenomena potentially affecting in-containment source term are still missing in ASTEC-Na: Na spray fires, fission product entrainment from Na-concrete interaction, fission product behavior in Na-containing aerosols and Na-concrete interaction.
- The validation work needs to be taken forward regarding sodium thermal-hydraulics (especially for the two-phase flows), fuel pin behavior, source term and neutronics. For instance, the RIA model still needs further validation (especially on fuel behavior at high temperatures and fission gas releases) where experimental data in representative conditions and for prototypic MOx fuels are sometimes scarce. In addition, the existing the in-containment database should be extended to support a thorough validation and further development of ASTEC-Na. The new experiments should take good care of their representativeness, scalability and measurements accuracy. The validity of the neutronics models against experimental data also still needs to be assessed to the possible extent considering the large power excursions that some transients can experience, or at least to space-time kinetics codes.
- There is a need to perform an integral calculation where the neutronics model is activated together with all the relevant models, including fuel pin mechanics and moderate relocation.

In the sections above the main accomplishments achieved by JASMIN has been synthesized in each of the areas addressed within the project. As a result of all of them, ASTEC-Na has been developed and validated as much as feasible under the technical (i.e. data availability) and project constraints (i.e. work force and deadlines). In short, ASTEC-Na is an analytical tool which predictive capability goes farther than previous generation of analytical tools and, no less important, its strengths and shortcomings are clearly identified so that a number of needs have been formulated to enhance its robustness and to extend its scope. As a major cross-cutting need of the four areas one might highlight data availability. Finally, the sound bases of ASTEC-Na (i.e. ASTEC code system) and the existing similarities with Pb-cooled and Pb-Li cooled reactors, turn it to be a good option to develop an ASTEC-LM (Liquid Metal) version capable of addressing also severe accidents in this type of reactors.

Ammirabile, L., et al. 2016a. Status of validation of ASTEC-Na neutronics models, FP7 JASMIN project deliverable D2.9 March 2016.
Ammirabile, L., et al., 2016b. Technical Note in Annex to D2.9 FP7 JASMIN project deliverable D2.9 June 2016.
Bandini G., et al., 2014. WP2.1 Sodium Thermal-Hydraulics - Status of validation of ASTEC-Na sodium thermal-hydraulic models. Deliverable D2.5 of JASMIN Project.
Bandini G., et al., 2016. WP 2.1 Sodium Thermal-Hydraulics – Assessment of ASTEC-Na Thermal-Hydraulic Models on Experiments and Benchmarking with Other Codes. ENEA Technical Report.
Cloarec, L., Moal, A., 2012. Documentation for RIA models. Rapport PSN-RES/SAG/2012-00331.
Girault, N., 2013. ASTEC-Na validation test matrix. JASMIN-INT-D3.2 May 2013.
Herranz L.E. Garcia M., Lebel L., Mascari F., Spengler C., 2016. WP 2.3 Source Term – Status of validation of ASTEC-Na containment and source term models. Deliverable D2.8 of JASMIN Project.
Perez-Martin, S., et al., 2016. Status of validation of ASTEC-Na v1.1p1 pin behavior models. JASMIN-MODELLING-D2.7 Jan. 2016.

Potential Impact:

The current project relates directly to the Topic Fission 2011-2.2.1 “Support for ESNII” in the EC calls 2011. The expected impact, as written in this EC calls, was: “Progress towards realising SET-Plan goals, in particular as they relate to the ESNII Implementation Plan, as agreed and published in the SET-Plan Information System (SETIS), and/or more robust treatment of safety-related issues through the development of common European approaches to modelling and simulation of safety-relevant phenomena for Fast Neutron Reactors”.
ESNII is, inside Europe, the demonstration programme for fast neutron technology with closed cycle, and it has required improvements in particular in terms of neutronics and safety characteristics. Five members of the ESNII Task Force are partners of the present JASMIN project: AREVA, CIEMAT, EDF, ENEA and JRC. And a majority of the other members of the ESNII Task Force have endorsed the JASMIN project as being consistent with the ESNII roadmap. The aim of the project was to get a European validated reference code for simulation of SFR severe accidents. The architecture of this simulation toll should be flexible enough to allow investigations of new SFR core designs with favourable inherent and passive safety features in the initiation phase of such severe accidents: “investigating core designs with moderate sodium void effect and other favourable reactivity feedback effects, and core designs and reactor vessel internal structures likely to disperse core debris and minimize risks of compaction”. Despite the big differences between the NPPs, the recent events in Fukushima again emphasise the importance of these aspects. When defining priorities for the code development and validation, lessons learned from Fukushima events has been taken into account as far as possible.
The JASMIN project supported ESNII in keeping the European leading role in nuclear technology by developing a computation tool needed for the indispensable safety assessments of the SFRs, while, in that field like in some other ones, the reference codes were initially mostly developed in the United States.
The project concerned cross-cutting aspects:
- First the nature of safety as a fundamental and inherent cross-cutting issue,
- Secondly the possibility to extend the ASTEC platform to another Gen.IV nuclear system such as LFR have been investigated,
- Thirdly the interest of several stakeholders and national programs in Europe (and beyond), as shown at least by the endorsement of a large majority of members of the ESNII Task Force.
A European approach was a great opportunity to achieve the expected impacts. During the project, links were maintained with the following FP7 projects at the MT level (some details on the way these links were managed are given just below):
- CP-ESFR (“Collaborative Project on European Sodium Fast Reactor”) that started on January 2009 and ended in July 2013. More especially, within the frame of this project, the SP3-6-1 Task investigated the present modelling capabilities of SFR core damage and provided recommendations for modelling improvement/development to be initiated. In addition, the SP3-6-2 Task addressed in a similar way the source term behaviour. The inputs of both Tasks were used to build ASTEC-Na detailed specifications. Note that most JASMIN partners (all except GRS and USTUTT) were also participating to these two SP3-6 tasks.
- SARGEN-IV (“Safety Assessment for Reactors of Generation IV”), coordinated by IRSN, that started in January 2012 for 18 month duration, and included the elaboration of a “roadmap and preliminary deployment plan for the fast reactor safety-related R&D”. All JASMIN partners (except USTUTT) also participated to SARGEN-IV.
- THINS (“Thermal-hydraulics for Innovative Systems”) on sodium two-phase thermal-hydraulics phenomena and modelling. Some reactor transients that were analysed in this project were considered for ASTEC-Na validation work. Three JASMIN partners (IRSN, GRS and ENEA) participated to THINS.
- ADRIANA (“Advanced Reactor Initiative and Network Arrangement”) that ended in September 2011, in particular about an identification of the infrastructures development needs for SFR, LFR and GFR. Most JASMIN partners are participating to ADRIANA. The interest concerned the experimental programmes that could be used to validate ASTEC-Na, without prejudging of the access rights of data that should be probably addressed case by case.
- SARNET2 (“Severe Accident Network of Excellence”) on R&D on severe accidents in water-cooled reactors where ASTEC played a central role as integrator of the whole knowledge that is obtained through new physical modelling. Here the link was obvious since IRSN is coordinating SARNET2 and since IRSN-GRS were fully involved in ASTEC deployment and support to users in SARNET2. All generic code improvements done for the sake of SARNET2 were then automatically beneficial for JASMIN. All JASMIN partners were also network’s members.
- ESNII+ (“European Sustainable Nuclear Industrial Initiative”) a 4-year project that started in 2013 for preparing ESNII for Horizon 2020 in which more especially various ASTRID-related analytical and experimental R&D studies are conducted and dedicated to Core and Fuel Safety. Most of the JASMIN partners participated to this project. More especially, KIT contributed to the WP in charge of establishing the roadmap for future R&D work on SFRs.
- Finally if any of the initially planned link with the LFR community through the MYRRHA and ALFRED consortia and the FP7 associated projects LEADER (“Lead-Cooled European Advanced Fast Reactor) and FREYA (“Fast Reactor Experiments for hYbrid Applications”) was finally established, at least some actors of the LFR community were invited at the last JASMIN workshop to discuss about the project results and the way forward for adapting ASTEC-Na to LFRs.
As to the links with SFR end-users, most JASMIN key players were already “internal” end-users. Potential users of the code in the future could come also from organisations/institutes currently involved in SFR safety studies (such as CEA, PSI, NRG, FZD, UPM, RSE, EA, ANSALDO...). Additionally, TSOs in countries where ESNII prototypes are planned to be build (Belgium, Romania ...) may be interested. The link was also maintained within the project with the French SFR research programmes devoted mainly to the ASTRID prototype development, which must be operational in the early 2020’s, where AREVA and EDF are actors.
Beyond Europe, the project was in support of the EURATOM contribution to the Generation IV International Forum (GIF), where pre-competitive R&D is being performed on key aspects and technologies relevant to future nuclear fission systems. Selected project deliverables has been identified during the project to be provided as EURATOM contribution to the GIF Safety and Operation project of the SFR that began in 2009. The discussion with JASMIN partners led to the following proposal that seems to be actually relevant from the technical standpoint:
- Deliverable 1.4 “Project presentation report” (PU dissemination level),
- Deliverable 2.10 “Overall synthesis of ASTEC-Na validation and benchmarking” (CO dissemination level)
- Deliverable 3.1 “General specifications of ASTEC-Na code for the SFR initiation phase” (PP dissemination level, i.e. restricted to project participants),
- Deliverable 3.7 “Requirements for the extension of ASTEC-Na beyond SFR initiation phase” (PP level),
- Deliverable 3.8 “Requirements for ASTEC-Na adaptation to LFR severe accidents” (PP level),
- Deliverable 4.7 “Final open workshop on ASTEC-Na main features” (PU level).
This will further help to stimulate the collaborative R&D activities on SFRs within the GIF and will be part of a fruitful exchange of knowledge and information with international partners, in particular Japan, Korea, and China.
For measuring the Consortium’s success, the Deliverable 1.3 “Project performance indicators” was issued at the beginning of the project where several indicators for assessing and monitoring the project’s progress measuring the implications of PhDs and young researchers were set.

By addressing the general public, industry and policy makers, the project also made a meaningful impact:
­ on economics by providing a safety demonstration of the reliability and efficiency of the optimized safety provisions envisaged for severe accident prevention and mitigation in advanced reactor designs;
­ on society by improving the public acceptance of the future nuclear power in providing a robust and reliable computational tool that will be used in Safety Assessments of new types of sodium-cooled fast reactors with advanced designs. By participating in the safety demonstration of this new generation of sodium-cooled reactors, characterized by a significant higher safety level compared to past reactors and then their promotion, the project also indirectly contributed to the opportunity to reduce the legacy stockpiles of Pu, to an improved utilization of uranium resources and to the recycling of nearly all long-lived radioactive waste within the reactor system rather than relying on long term storage in permanent geological repositories;
­ on EU policy by helping European countries to meet the safety requirements of advanced sodium-cooled fast reactors to be built in the coming years then indirectly promoting the nuclear energy as an alternative to fossil fuels for decarbonized electricity production.

Two main Deliverables for dissemination were initially foreseen:
- The Project presentation, released at the very beginning of the project where the project goal, objectives, activities expected outcome and impact were presented,
- The Communication Action Plan also prepared at the very beginning of the project, which indicated activities (international conferences, national meetings, publications, training activities, etc.) with their dates over the lifetime of the project.
The project results were on one hand a new European simulation code and on the other hand a database of the most relevant experiments on severe accidents in SFR. Both types of products are opened to the partners of the Consortium.
Access rights and property rights over the Background and Foreground were preliminary defined in Annex II Part C of the Grant Agreement and by the Consortium Agreement, and are subject to the following:
- Notwithstanding the foregoing, the further development of ASTEC-Na will remain under the control of IRSN. During the project, IRSN has granted a free access right for use of the ASTEC-Na code to each Party. This didn’t include any sub-licence right;
- Beyond the project timeframe, any access grant to the ASTEC-Na code by the partners, subject to the conclusion of a license agreement with each Party asking this access right, should be required to IRSN. This license agreement shall define all necessary conditions (duration, price, possible in-kind contribution, type of granted use...);
- The implementation in the ASTEC-Na code of models using Foreground shall not modify IRSN’s Intellectual Property Rights on ASTEC code and its updates, as well as on ASTEC-Na. IRSN in any cases remains owner of the ASTEC and ASTEC-Na codes and any of their updates or new versions.
The decision to disseminate or to disclose any version of the ASTEC and ASTEC-Na codes (the latter including foreground generated by the present project), or parts thereof, and related documentation shall be at the absolute discretion of IRSN. This will be in accordance with the terms of the Consortium Agreement of the project.
The dissemination of the results has taken place through different ways:
• Workshops: the last workshop, organized at the end of the project, was open to external organizations where the main features of the code ASTEC-Na were presented and discussions on possible future exchange of information or collaboration have taken place.
• Education and training programme: several PhDs participated to the periodic workshops, and were seconded in the laboratories or offices of the project’s partners for training.
• Publications in scientific journals and participations to conferences.
• Public web site.

List of Websites: