SCANAIR a transient fuel performance code Part two: Assessment of modelling capabilities

SCANAIR a transient fuel performance code Part two: Assessment of modelling capabilities

G Model ARTICLE IN PRESS NED-7870; No. of Pages 9 Nuclear Engineering and Design xxx (2014) xxx–xxx Contents lists available at ScienceDirect Nuc...

1MB Sizes 81 Downloads 137 Views

G Model

ARTICLE IN PRESS

NED-7870; No. of Pages 9

Nuclear Engineering and Design xxx (2014) xxx–xxx

Contents lists available at ScienceDirect

Nuclear Engineering and Design journal homepage: www.elsevier.com/locate/nucengdes

SCANAIR a transient fuel performance code Part two: Assessment of modelling capabilities Vincent Georgenthum ∗ , Alain Moal, Olivier Marchand Institut de Radioprotection et de Sûreté Nucléaire (IRSN), Pôle Sûreté des Installations et des Systèmes Nucléaires, Service de Maîtrise des Incidents et des Accidents, Laboratoire de Physique et de Thermomécanique des Matériaux, BP3, 13115 Saint-Paul-les-Durance Cedex, France

h i g h l i g h t s • • • • •

The SCANAIR code is devoted to the study of irradiated fuel rod behaviour during RIA. The paper deals with the status of the code validation for PWR rods. During the PCMI stage there is a good agreement between calculations and experiments. The boiling crisis occurrence is rather well predicted. The code assessment during the boiling crisis has still to be improved.

a r t i c l e

i n f o

Article history: Received 1 October 2013 Received in revised form 17 April 2014 Accepted 21 April 2014

a b s t r a c t In the frame of their research programmes on fuel safety, the French Institut de Radioprotection et de Sûreté Nucléaire develops the SCANAIR code devoted to the study of irradiated fuel rod behaviour during reactivity initiated accident. A first paper was focused on detailed modellings and code description. This second paper deals with the status of the code validation for pressurised water reactor rods performed thanks to the available experimental results. About 60 integral tests carried out in CABRI and NSRR experimental reactors and 24 separated tests performed in the PATRICIA facility (devoted to the thermalhydraulics study) have been recalculated and compared to experimental data. During the first stage of the transient, the pellet clad mechanical interaction phase, there is a good agreement between calculations and experiments: the clad residual elongation and hoop strain of non failed tests but also the failure occurrence and failure enthalpy of failed tests are correctly calculated. After this first stage, the increase of cladding temperature can lead to the Departure from Nucleate Boiling. During the film boiling regime, the clad temperature can reach a very high temperature (>700 ◦ C). If the boiling crisis occurrence is rather well predicted, the calculation of the clad temperature and the clad hoop strain during this stage have still to be improved. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Since the early 1990s, several changes occurred in the basic management of nuclear reactor core: burn-up increase, introduction of MOX fuel and use of new cladding materials. Such changes have significantly modified the behaviour of nuclear fuel and raise the question of the current safety criteria relevance and the assessment of safety margins involving the proper behaviour of the fuel cladding. In this context and in the frame of its research programme on fuel safety, the French Institut de Radioprotection et de Sûreté

∗ Corresponding author. +33 442199536. E-mail address: [email protected] (V. Georgenthum).

Nucléaire (IRSN) studies the high burn-up fuel rod behaviour during a reactivity initiated accident (RIA) (Latché et al., 1995; Papin et al., 2005; Sartoris et al., 2010). IRSN has then initiated an important experimental programme mainly consisting in integral tests on irradiated fuel rods in CABRI reactor in parallel to the development of the SCANAIR code. A RIA is characterised by a very rapid increase of reactivity and power in some rods of the reactor. After the initiating event, rod control ejection in pressurised water reactor (PWR), the accidental sequence can be schematically represented in two main stages. First, after the control rod ejection the energy deposition leads to a rapid rise of the fuel temperature which induces thermal swelling of the fuel pellets. During this stage the clad temperature is still very close to the initial temperature. The pellet clad mechanical interaction (PCMI) leads to a clad deformation and

http://dx.doi.org/10.1016/j.nucengdes.2014.04.030 0029-5493/© 2014 Elsevier B.V. All rights reserved.

Please cite this article in press as: Georgenthum, V., et al., SCANAIR a transient fuel performance code Part two: Assessment of modelling capabilities. Nucl. Eng. Des. (2014), http://dx.doi.org/10.1016/j.nucengdes.2014.04.030

G Model NED-7870; No. of Pages 9

ARTICLE IN PRESS V. Georgenthum et al. / Nuclear Engineering and Design xxx (2014) xxx–xxx

2

potentially to failure depending on the fuel enthalpy increase and on the level of clad embrittlement due to oxidation and hydriding. After the PCMI phase, the increase of cladding temperature may lead to the departure from nucleate boiling (DNB) and boiling crisis occurrence. In such a case, the clad-to-coolant heat transfer becomes very low and the clad temperature can reach a high level (>700 ◦ C). Depending on the internal gas pressure, the ductile clad can undergo an important deformation and a possible failure. The objective of SCANAIR is to describe the thermo-mechanical behaviour of irradiated fuel rod (including UO2 and MOX fuels) during these two phases until the possible clad failure. The general modelling of SCANAIR is described in a first paper (Moal et al., 2014). The objective of this current paper is to present the validation status of the SCANAIR V 7 3 code for PWR rods. After the description of the assessment matrix and a short description of the code, the comparison between code and available experimental results is presented for the integral tests carried out on PWR rods in CABRI and NSRR experimental reactors and the separate effect tests performed in the out-of-pile PATRICIA facility. 2. Assessment matrix The SCANAIR code assessment has been done on experimental results obtained on about 60 integral tests carried out in CABRI and NSRR experimental reactors and on 24 separate effect tests performed in the PATRICIA facility in PWR and NSRR conditions. The rod characteristics, the test conditions and the available experimental measurements are detailed hereafter for the different test facilities. 2.1. CABRI tests In 1992, IRSN, in partnership with Electricité de France (EDF), initiated the CABRI REP-Na research programme devoted to study RIAs for both UO2 and MOX fuels (Papin et al., 2005). The UO2 part of the programme was also supported by the US Nuclear Regulatory Commission (NRC). Single irradiated test rods have been subjected to fast power transients in the CABRI sodium loop, already extensively used in the frame of the fast breeder reactors safety studies. Use of sodium coolant allowed investigation of rod behaviour during the first phase of an RIA transient, when PCMI is predominant without significant clad heat-up. In 2002, in the framework of the CABRI International Programme (CIP), two tests with high burnup UO2 fuel (∼75 GWd/t, maximum local pellet) have been already performed in the CABRI sodium loop facility before the renewal of the facility and the implantation of the pressurised water loop. From 1993 to 2002, ten tests were performed on UO2 fuel rods and four others on MOX fuel rods. Test specimens consisted of commercially manufactured reconditioned rodlets from EDF rods (except for REP-Na2 which was a full length rod from BR3 reactor, REP-Na3 which was a segmented rod and CIP0-1 which was irradiated in Vandellos reactor in Spain). All test rods charged with UO2 fuel had a 235U enrichment of 4.5 wt%, except for REP-Na2, which had an enrichment of 6.85 wt%. The MOX fuel rods had fuel pellets of MIMAS (micronised masterblend) AUC (ammonium uranocarbonate) type. The test rods were filled with helium at low pressure (0.3 MPa, 0,1 MPa in REP-Na1) to simulate the fuel rod internal pressure balance with the coolant at the end of in-reactor irradiation. Fluid flow velocity and temperature conditions (4 m/s and 280 ◦ C respectively) were those corresponding to reactor hot shutdown. The rod characteristics and test conditions are summarised below: - fuel burn-up (maximum pellet value of the test rod): ranged from 33 GWd/tM (REPNa-2) to 75 GWd/tM (CIP0-1),

- clad material: most of the tests have been performed with standard Zircaloy-4, except REP-Na3 and REP-Na8 (low tin Zircaloy-4), REP-Na2 (BR2 cladding), REP-Na11 and CIP0-2 (M5), CIP0-1 (Zirlo), - clad corrosion: oxide thickness ranged from 4 ␮m in REP-Na2 up to 130 ␮m in REP-Na8, with more or less initial spalling of the oxide layer (REP-Na1, REP-Na8 and REP-Na10), - fuel pellet type: UO2 and MOX fuels, the later being of MIMAS type, - coolant conditions: sodium coolant at 3 bar, 280 ◦ C and 4 m/s, - energy deposit: from about 100 cal/g in most experiments to more than 200 cal/g in case of low burn-up tests (REP-Na2, REPNa9), - power pulse width: 9.5 ms to larger values (30–75 ms) resulting in different energy injection rates.

2.2. NSRR tests From 1989 to 2008, Japan Atomic Energy Agency (JAEA) performed thirty four tests in the Japanese NSRR reactor with UO2 fuel rods pre-irradiated in PWR located in Japan, in Europe and in the USA. Except five UO2 tests (MH and GK series) performed with 14 × 14 rods, all the UO2 rods have been reconditionned from 17 × 17 rods (Fuketa et al., 1997). The first test series (MH, GK, OI and HBO series) were done with standard Zircaloy-4, while the ten rods in the TK test series had low-tin Zircaloy-4 cladding (1.3 wt% Sn) (Fuketa et al., 2001). In the last test series (second series of OI test and ALPS tests) fuel with other cladding materials than Zircaloy-4 (Zirlo, MDA, NDA and M5) have also been tested (Sugiyama, 2006, 2010). All the tests were performed in stagnant water, at 1 bar and 20 ◦ C in most of the cases, at ∼70 bar and ∼280 ◦ C in some of the recent tests in the high temperature high pressure (HTHP) capsule. Three tests in the BZ series were done with MOX fuel refabricated from 14 × 14 fuel assembly irradiated in Beznau PWR. The BZ-1 test rod was charged with short binderless route (SBR) fuel pellets, while BZ-2 and BZ-3 rods used MIMAS AUC type pellets. To study the clad-to-coolant heat exchange conditions during RIA transient, the surface effect tests series (the so called NSRR-Ox tests) were carried out in the NSRR reactor with Zircaloy-4 cladding and fresh UO2 pellets (Sugiyama and Fuketa, 2004). Each rod was instrumented with 6 thermocouples spotwelded on the cladding at different axial locations. Three surface conditions were prepared: without oxide layer, with 1 ␮m oxide layer, and with 10 ␮m oxide layer. The tests covered the following rod characteristics and test conditions:

- fuel burn-up: ranged from fresh fuel (Ox-tests) to 77 GWd/tM (VA2, VA4), - clad material: Zircaloy-4 (standard, or low tin), M5, Zirlo, MDA and NDA, - clad corrosion: up to 80 ␮m in VA4 test, - coolant conditions: stagnant water, at 1 bar and 20 ◦ C or at ∼70 bar and ∼280 ◦ C, - energy deposit, from about 50 cal/g to more than 180 cal/g, - power pulse width: most of the tests performed with a very narrow pulse (4.4–4.6 ms).

The injected energy and the enthalpy at failure in HBO and TK series performed in NSRR reactor have been tentatively reevaluated in this paper according to a recent reassessment of maximal enthalpy made by JAEA (Udagawa et al., 2011).

Please cite this article in press as: Georgenthum, V., et al., SCANAIR a transient fuel performance code Part two: Assessment of modelling capabilities. Nucl. Eng. Des. (2014), http://dx.doi.org/10.1016/j.nucengdes.2014.04.030

G Model

ARTICLE IN PRESS

NED-7870; No. of Pages 9

V. Georgenthum et al. / Nuclear Engineering and Design xxx (2014) xxx–xxx

3

puncturing process and the total amount of fission gas in the fuel before the test, calculated by an irradiation code.

2.3. PATRICIA tests The PATRICIA facility located at CEA/Grenoble (France) consists in a water loop that can operate in PWR conditions and also in pool conditions similar to those of the RIA experiments in the NSRR reactor. A tubular rod is centred in the test section. The inner part of the rod is filled with air. The outer part is in contact with the water. The rod length is 600 mm and the inner and outer diameters are 8.8 mm and 9.5 mm respectively. The channel cold wall diameter is 14.2 mm so that the PWR-subchannel cross-section is conserved. Thermocouples are welded on the inner side of the cladding, near the top end of the heated length (respectively 580 mm and 595 mm from the rod bottom). The cladding is made of Inconel instead of Zircaloy 4 because of difficulties to weld the thermocouples on the inner clad surface. The cladding is electrically heated up by Joule effect. The estimation of the clad outer surface temperature and of the clad-to-coolant heat-flux from the measurement of the inner clad temperature is achieved in two steps: - correction of the thermocouple response time (order of 20 ms), - smoothing and inverse conduction calculation to estimate the temperature field in the cladding and the clad-to-coolant heatflux. 2.4. Experimental measurements The experimental results come mainly from (Papin et al., 2005; OCDE/NEA/CSNI, 2010) and (Bessiron, 2007). The physical quantities measured in the RIA tests and that can be used for the validation process are summarised in Table 1: The measurement of sodium temperatures during CABRI tests was made by thermocouples located along the test rod at different azimuthal location, through spacer grids in the annular section between the clad and the channel wall. The only integral tests instrumented with thermocouples welded on the outer part of the clad have been performed in NSRR reactor. For irradiated rods thermocouples were welded on the sound clad after oxide removal. This tricky operation has been done in JAEA hot cell for rods with an oxide thickness lower than 35 ␮m. In CABRI tests the clad and fuel elongation were respectively measured by the displacement transducer and the hodoscope device. In some NSRR tests the fuel and clad elongation have been also measured using magnetic iron cores and LVDT sensors. For all non failed tests the experimental residual hoop strain is deduced from profilometries performed along the fissile column for different azimuthal locations before and after test in hot cell. The failure occurrence and the failure time have usually been estimated by microphones analysis in CABRI tests and coolant pressure sensors in NSRR tests. The fission gas release (FGR) corresponds to the ratio between the amount of fission gas measured after the test through

3. SCANAIR code 3.1. General description of the code SCANAIR is a 1.5D code designed to model a single rod surrounded by a coolant channel and possibly limited by an external shroud. To describe properly the complex phenomena occurring during the fast power transient, SCANAIR takes into account the following physical phenomena and their strong coupling: - thermal-dynamics and thermal-hydraulics including clad-tocoolant heat transfer modelling in sodium or water conditions, - structural mechanics for the rod constitutive elements, - fission gas transient behaviour. SCANAIR is thus a set of three main modules that communicate with each other through a database. A failure module has been added to the three main modules to evaluate the possibility of clad failure during the transient. A schematic view of SCANAIR general processing is shown in Fig. 1. The initial rod state is an input data of the SCANAIR calculation, usually given by an irradiation code. Interfaces with several irradiation codes are currently available. The power transient is also an input data computed by neutron kinetics codes or measured from experimental tests. The clad failure occurrence can be computed online during the SCANAIR computation. Four different approaches are available to predict the failure. In this paper, only two failure approaches will be tested: - the fracture mechanics approach, called CLARIS model (Georgenthum et al., 2008). This approach relies on the hypothesis that the PCMI failure is resulting from the propagation in the outer part of the cladding of an existing incipient crack (due to the presence of a brittle area with dense hydride). This model allows to calculate, at each time during the PCMI phase, the critical flaw depth that would lead to the clad failure. The failure is detected when, in one axial slice, the critical flaw depth calculated by SCANAIR becomes lower than the incipient crack size given in the input data deck. - The ‘CSED approach’ based on the calculation of the strain energy density (SED) (Bernaudat, 2009). The clad failure is detected when, in one axial slice, the SED becomes higher than the critical strain energy density (CSED) given in the input data deck.

Table 1 Experimental measurements used for SCANAIR validation. CABRI

NSRR

PATRICIA

Thermics Coolant temperature Clad temperature

Yes No

No Yes

No Yes

Mechanics Clad elongation Clad hoop strain Fuel elongation Clad failure

Yes Yes Yes Yes

No Yes Yes Yes

No No No No

Gas Fission gas release

Yes

Yes

No

Fig. 1. Schematic view of SCANAIR general processing.

Please cite this article in press as: Georgenthum, V., et al., SCANAIR a transient fuel performance code Part two: Assessment of modelling capabilities. Nucl. Eng. Des. (2014), http://dx.doi.org/10.1016/j.nucengdes.2014.04.030

G Model NED-7870; No. of Pages 9

V. Georgenthum et al. / Nuclear Engineering and Design xxx (2014) xxx–xxx

3.2. Assumptions used for the SCANAIR calculations The SCANAIR input data decks result from the end-of-life state of the rod calculated with the FRAPCON code (Geelhood et al., 2011). For the CABRI tests, the available rod characteristics before irradiation and the irradiation histories were relatively exhaustive. An initial state calculation has then been performed in each case. For NSRR tests, as available data were not detailed enough, initial states have been deduced from the CABRI rods with the closest characteristics except for rodlets with a burnup close to 40 GWd/tM (MH, GK series, some TK and OI tests) for which a specific initial state calculation has been performed. The geometrical description and the clad outer oxide thickness have been corrected when some measurements before tests were available. In particular when oxide spalling has been observed (before or after the RIA in some CABRI tests), transient calculations have been done with initial oxide spallation. The outer fuel rugosity is supposed to be very low (0.1 ␮m) in order to model a quasi-perfect fuel-clad thermal contact. It is especially the case when the fuel is stuck to the cladding after the creation of an inner zirconia layer for intermediate burnups. The transient clad-to-water heat transfer is described by a heat transfer coefficient approach. The coefficients are estimated by correlations. During fast transient conditions, the radial temperature profile in the coolant can be much steeper than in steady state conditions and the shape of the boiling curve is different from the one in steady state conditions. Two different water boiling curves have been used: the first one is deduced from the interpretation of some NSRR experiments (Bessiron et al., 2007), the second one is deduced from PATRICIA-PWR experiments (Bessiron, 2007). The fuel is assumed to be cracked, before the beginning of the transient (i.e., unresisting to tensile stresses). As a consequence, no tensile stress can exist in the fuel. Considering the high strain rates reached in RIA conditions, the fuel mechanical behaviour has been modelled with a perfect elastoplastic law. The fuel yield stress has been calculated using a law derived from the Canon’s law (Canon et al., 1971) (for UO2 and MOX fuel). The clad behaviour has been modelled using a viscoplastic law with a Lemaitre formulation (Moal et al., 2014). The mechanical properties are mainly coming from the PROMETRA experimental programme (Cazalis et al., 2007). As mentioned before, to predict the possible clad failure and the enthalpy at failure during the PCMI phase, both the clad failure model CLARIS and CSED have been used. The determination of the incipient crack size for CLARIS approach has been done thanks to clad metallographic examinations performed before or after the tests. It was assumed that the incipient crack size corresponds to the clad hydride rim depth observed on metallographic examinations. When no examinations were available, it was assumed that the clad hydride rim depth was equal to the outer clad zirconia thickness, considering results from (Sartoris et al., 2010). The CSED value has been evaluated as a function of the outer clad zirconia thickness according to (Bernaudat, 2009). The fission gas initialisation in the fuel has been done using FRAPCON results for the total amount of retained gas and based on experimental results analysis that gives in particular an estimation of the repartition between gas in the grain and gas at the grain boundary location (Lemoine, 2005).

440 420

Temperature (C)

A new tool named SCAVALID has been developed to perform automatic computation and post-processing of a large set of RIA tests. The technical report is then automatically generated with tables and figures comparing experimental and computational results. The figures presented in the paper are generated thanks to this validation tool. A detailed description of the code version 7.3 can be found in (Moal et al., 2014).

400 380 360 340 320 300 280

0

0.5

1

1.5

2

2.5

3

3.5

4

Time (s) Fig. 2. Sodium temperature – SCANAIR (black curve) versus experiment (other curves) for CABRI REP-Na9 test at PPN location.

4. Thermal and thermalhydraulics assessment 4.1. Sodium temperature in CABRI tests Fig. 2 shows the comparison between SCANAIR calculations and three thermocouples measurements for the REP-Na9 test at PPN axial location. The maximal sodium temperatures calculated and measured for all the CABRI non failed tests are presented in Fig. 3 (the error bars represent the measurement deviations of the three thermocouples at the same location). A good agreement is found between calculated and measured sodium temperature, both in terms of maximal value and of time evolution provided that the heat transfer between the fuel and the clad is improved (assuming a very low fuel rugosity) and that the transient spallation of the cladding oxide is taken into account when it actually occurs during the test.

550

500

SCANAIR calculations (C)

4

ARTICLE IN PRESS

450

400

350 350

400

450

500

550

Exp. results (C) Fig. 3. Maximum sodium temperature – SCANAIR versus experiment for CABRI tests.

Please cite this article in press as: Georgenthum, V., et al., SCANAIR a transient fuel performance code Part two: Assessment of modelling capabilities. Nucl. Eng. Des. (2014), http://dx.doi.org/10.1016/j.nucengdes.2014.04.030

G Model

ARTICLE IN PRESS

NED-7870; No. of Pages 9

V. Georgenthum et al. / Nuclear Engineering and Design xxx (2014) xxx–xxx

5

1200

900 800

1000

600 500 400 300 200 100 0 -1

0

1

2

3

4

5

6

SCANAIR calculations (C)

Temperature (C)

700

800

600

400

Time (s) Fig. 4. Clad temperature – SCANAIR (black curve) versus measurements (other curves) for 10H test.

4.2. Clad temperature

0

4.2.1. NSRR tests The boiling curve has been adjusted for NSRR room temperature conditions (stagnant water at 20 ◦ C and 1 bar) based on fresh PWR type rods with pre-oxidation of the cladding, NSRR-Ox tests (Sugiyama and Fuketa, 2004; Bessiron et al., 2007). An example of comparison between measurements and SCANAIR calculation is presented in Fig. 4. The comparison between calculations and measurements on clad maximal temperature for all NSRR-Ox tests are gathered in Fig. 5. The agreement is very good for the cases during which the boiling crisis is reached. Nevertheless, the SCANAIR calculations tend to reach the boiling crisis slightly too early and, as a result, predict a too high clad temperature when the film boiling regime is not fully established (clad temperature measurement lower to 400 ◦ C).

900

800

SCANAIR calculations (C)

200

700

0

200

400

600

800

1000

1200

Exp. results (C) Fig. 6. Maximum clad temperature – SCANAIR versus experiment for NSRR-PWR tests.

The departure from nucleate boiling (supposed when clad outer temperature exceeds significantly 100 ◦ C), which occurs in most cases for an enthalpy higher than ∼70 cal/g, is globally correctly predicted for NSRR PWR-rod tests (see Fig. 6). However, in some tests, boiling crisis is not reached for enthalpy equal to or higher than 100 cal/g. The boiling curve deduced from the results obtained on the fresh pre-oxidise tests gives the correct trend for irradiated rods, but with an over-estimation of the cladding temperature by about 300 ◦ C. Nevertheless, as recently mentioned in (Udagawa et al., 2013) the thermocouples measurement underestimates the actual temperature due to the fin-cooling effect of the thermocouples. The temperature drop is difficult to estimate and uncertainty is ranged from 0–250 ◦ C, depending on the cladding surface temperature, the coolant subcooling, the coolant pressure and the diameter of the thermocouple wire. The clad temperature and the boiling duration are overestimated in irradiated cases, showing possibly an impact of irradiation on the clad-to-coolant heat exchange enhancement and on the boiling curve. 4.2.2. PATRICIA tests The comparison of measured and calculated inner clad temperature for PATRICIA-PWR tests (065 and 112 series) are gathered in Fig. 7. The experimental uncertainties have been estimated to 30 K on the clad outer surface temperature (Bessiron, 2007). The calculated maximum temperatures are within the uncertainty range of the experimental ones. A rather good agreement is also obtained on the duration of the film boiling phase. Nevertheless, the use of non oxidised Inconel rods without fuel compared to irradiated and oxidised PWR claddings with fuel pellet raises questions about the representativity of the tests, especially regarding wettability and thermal inertia, and underlines the lack of thermal-hydraulics experimental data in PWR conditions.

600

500

400

300

200

5. Clad mechanical assessment 100 100

200

300

400

500

600

700

800

900

5.1. Clad elongation

Exp. results (C) Fig. 5. Maximum clad temperature – SCANAIR versus experiment for NSRR-Ox tests.

The comparison between SCANAIR calculation and measurement during REP-Na3 test is presented in Fig. 8. For a short time

Please cite this article in press as: Georgenthum, V., et al., SCANAIR a transient fuel performance code Part two: Assessment of modelling capabilities. Nucl. Eng. Des. (2014), http://dx.doi.org/10.1016/j.nucengdes.2014.04.030

G Model

ARTICLE IN PRESS

NED-7870; No. of Pages 9

V. Georgenthum et al. / Nuclear Engineering and Design xxx (2014) xxx–xxx

6

700

14

650

SCANAIR calculations (mm)

SCANAIR calculations (C)

12 600

550

500

450

10

8

6

400

4

350

2

350

400

450

500

550

600

650

700

2

4

6

Exp. results (C)

8

10

12

14

Exp. results (mm)

Fig. 7. Maximum clad temperature – SCANAIR versus experiment for PATRICIAPWR tests.

scale after the beginning of the transient, both deconvoluated (with a response time of 20 ms, blue dashed line) and non-deconvoluated signals are shown. The deconvoluated signal is to be used to compare with the calculated result at the beginning of the energy increase. However, the deconvolution generates a peaked value of elongation (which can be twice as high as the non-deconvoluated signal in the case of rapid transient) which is not physical, and that must be discarded in the interpretation. The agreement between SCANAIR results and experiment in term of maximal and residual value but also in term of kinetic validate the hypothesis of bonding between fuel and clad when the gap is closed. In NSRR tests, the fissile column is short (usually about 120 mm), the clad elongation is then relatively low. In CABRI tests the fissile column being higher, usually close to 560 mm, the clad elongation

Fig. 9. Maximal clad elongation – SCANAIR versus measurements for REP-Na tests.

is more significant and in all cases higher than the clad thermal swelling. The measured and calculated maximal clad elongation are gathered in Fig. 9 only for the CABRI tests. The clad maximal elongation is correctly predicted especially in the case of tests with low or medium energy injection (up to a fuel enthalpy of about 120 cal/g). 5.2. Clad hoop strain The comparison of calculated and mean measurements (for the different azimuths) along the entire fissile column in CABRI CIP0-2 and REP-Na9 tests are respectively presented in Figs. 10 and 11. The experimental residual hoop strain has been evaluated on the pre and post-test clad diameter measurements. In both cases, some fine irregularities have been observed on the curve. For the CIP02 test, the transient zirconia spalling led indeed to a noisy signal.

7 60

6 50

Axial elevation (cm)

Clad elongation (mm)

5 4 10

3

8 2

6 4

1

30

20

2

0

10

0 -1

40

0

1.06 2

1.08

1.1

4

1.12 6

1.14 8

10

Time (s)

0 -0.2

-0.1

0

0.1

0.2

0.3

0.4

0.5

Clad residual hoop strains (%) Fig. 8. Clad elongation – SCANAIR (black curve) versus measurements (raw: red curve, deconvoluated: blue dashed curve) for REP-Na3 test. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 10. Clad residual hoop strain – SCANAIR (black curve) versus measurements (red curve) for CIP0-2 test. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Please cite this article in press as: Georgenthum, V., et al., SCANAIR a transient fuel performance code Part two: Assessment of modelling capabilities. Nucl. Eng. Des. (2014), http://dx.doi.org/10.1016/j.nucengdes.2014.04.030

G Model

ARTICLE IN PRESS

NED-7870; No. of Pages 9

V. Georgenthum et al. / Nuclear Engineering and Design xxx (2014) xxx–xxx

60

7

10

40

8

30 20 10 0 -10 -1

0

1

2

3

4

5

6

7

8

SCANAIR calculations (%)

Axial elevation (cm)

50

6

4

Clad residual hoop strains (%) Fig. 11. Clad residual hoop strain – SCANAIR (black curve) versus measurements (red curve) for REP-Na9 test. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

For the REP-Na9 case, the concavities and convexities are resulting from an increase of the primary local ridges up to 100 ␮m in diameter during the test. This 3D phenomenon is not modelled by the 1.5D code SCANAIR. Nevertheless, there is an excellent agreement between the experiment and the code prediction in both cases. Fig. 12 gathers the results for the maximal clad residual hoop strain in all CABRI non failed tests. As for the elongation, for low or medium injected energy, the present modelling is satisfying, although there is a tendency to slightly underestimate the residual cladding hoop strains in MOX cases. For highly energetic tests, especially REP-Na2, the calculation is not accurate. This result shows that the fuel and clad mechanical laws should be adapted when high temperatures are reached: in particular the fuel mechanical behaviour could be more accurately modelled with a viscoplastic law than the perfect plastic law presently implemented in SCANAIR.

8

7

SCANAIR calculations (%)

2

4

6

8

10

Exp. results (%) Fig. 13. Clad residual hoop strain – SCANAIR versus measurements for NSRR tests.

The clad residual hoop strain in NSRR tests calculated by SCANAIR as a function of measurement are gathered in Fig. 13. For relatively low residual hoop strain (corresponding to cases until a maximal enthalpy of about 100 cal/g) there is a rather good agreement between calculations and measurements. For a higher enthalpy, that is to say when the film boiling regime is fully established, in some cases SCANAIR globally underestimates the residual hoop strain. The difficulty in these cases is that the clad loading is a mix of imposed pressure when the gap is opened (due to fission gas release and inner pressure increase) and imposed strain when the gap is closed (due to pellet swelling), the final hoop strain is thus not only linked to the fuel and clad mechanical behaviour but also to the clad temperature and to the gas inner pressure. The evolutions of all these data are coupled and complex to predict. 6. Fission gas behaviour assessment

6

5

4

3

2

1

0

2

0

1

2

3

4

5

6

7

8

Exp. results (%) Fig. 12. Clad residual hoop strain – SCANAIR versus measurements for CABRI tests.

The results obtained in NSRR and CABRI tests (Fig. 14) show a wide scattering. However, the calculation are more accurate in CABRI tests (red points) than in NSRR tests (black points) where SCANAIR globally overestimates the FGR results. The fission gases released during the RIA transient are mainly coming from the grain boundary cavities (intergranular bubbles and porosities, including the ones located in the fuel high burnup structure). When the strengths generated by the cavities overpressurisation exceed the material mechanical resistance, the rupture of grain boundaries occurs and all the GB gases are released and flow through the porosities network towards the free volume following a Darcy law with a constant coefficient of permeability (see Moal et al., 2014). For very energetic transient (>150 cal/g) where gas located in the intra-granular bubbles may be released. The estimation of initial amount of gas retained in the fuel, but also the repartition between intra and inter-granular gases is then essential for the fission gas release evaluation. NSRR tests results are relatively correct considering the fact that the initial states of the rods and in particular the gas initialisation have been done considering the gas repartition in the CABRI rods with the closest irradiation characteristics.

Please cite this article in press as: Georgenthum, V., et al., SCANAIR a transient fuel performance code Part two: Assessment of modelling capabilities. Nucl. Eng. Des. (2014), http://dx.doi.org/10.1016/j.nucengdes.2014.04.030

G Model

ARTICLE IN PRESS

NED-7870; No. of Pages 9

V. Georgenthum et al. / Nuclear Engineering and Design xxx (2014) xxx–xxx

8

Table 3 PCMI failure prediction using CSED and CLARIS model for NSRR-PWR tests.

35

SCANAIR calculations (%)

30

25

20

15

10

5

0

0

5

10

15

20

25

30

35

Exp. results (%) Fig. 14. Fission gas release – SCANAIR versus measurements for CABRI (red points) and NSRR (black points) tests. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

The kinetic of fission gas release is also a key issue for the rod thermomechanical behaviour mainly in the post-DNB phase. The instant of fission gas release is very difficult to validate; the future JAEA FGD (Fission gas Dynamics) programme that is foreseen in NSRR reactor in collaboration with IRSN should give valuable informations on the fission gas release kinetic. 7. Clad failure analysis The PCMI failure prediction for CABRI tests and NSRR tests are respectively gathered in Tables 2 and 3 for CLARIS and CSED models. The failure prediction has not been done for REP-Na1, REP-Na8 and REP-Na10 tests. In these three tests clad outer oxide spalling occurred during the base irradiation. This phenomenon led to a strongly non-uniform hydride distribution with large hydride blisters in the outer part of the cladding (Papin et al., 2005). As the blister depth is erratic, it is not possible to evaluate the size of the initial flaw depth required for the CLARIS model or the local CSED and the failure analysis has then not been carried out in these cases. The failure occurrence is correctly predicted in all the other CABRI tests with the fracture mechanics approach (CLARIS) while the CSED model seems to be more conservative: five intact tests have been estimated failed. Table 2 PCMI failure prediction using CSED and CLARIS model for CABRI tests. Reference

Test (cal/g)

CSED (cal/g)

CLARIS (cal/g)

REP-Na2 REP-Na3 REP-Na4 REP-Na5 REP-Na6 REP-Na7 REP-Na9 REP-Na11 REP-Na12 CIP01 CIP02

No No No No No Yes (112) No No No No No

Yes Yes No No Yes Yes (107) Yes No Yes No No

No No No No No Yes (128) No No No No No

Reference

Test (cal/g)

CSED (cal/g)

CLARIS (cal/g)

BZ-1 BZ-2 BZ-3 GK-1 GK-2 HBO-1 HBO-2 HBO-3 HBO-4 HBO-5 HBO-6 HBO-7 MH-1 MH-2 MH-3 OI-1 OI-2 OI-10 OI-11 OI-12 RH-1 RH-2 TK-1 TK-2 TK-3 TK-4 TK-5 TK-6 TK-7 TK-8 TK-9 TK-10 VA-1 VA-2 VA-3 VA-4 MR-1

Yes (75) Yes (127) No No No Yes (93) No No No Yes (110) No No No No No No No No Yes (118) No No No No Yes (88) No No No No Yes (130) No No No Yes (63) Yes (54) Yes (98) No No

Yes (111) Yes (135) No No No Yes (92) No Yes No Yes (77) Yes Yes No No No No No No Yes (123) Yes No No Yes Yes (104) No Yes Yes Yes Yes (117) No Yes No Yes (69) Yes (73) Yes (84) Yes No

No No No No No Yes (79) No Yes No Yes (55) Yes Yes No No No No No Yes Yes (78) Yes No No Yes Yes (75) No No No No Yes (83) No No No Yes (53) Yes (57) Yes (100) Yes No

With CLARIS and CSED approaches, the failure occurrence is correctly predicted respectively in 28 and 27 NSRR tests (on a total of 37 tests). The accuracy of enthalpy at failure is similar with the two approaches. These predictions are consistent considering the fact that the two approaches have been set up for tests performed at a temperature higher than 280 ◦ C. As discussed in (Georgenthum et al., 2008) the determination of the incipient crack size, necessary for CLARIS approach, is not easy in room temperature test due to the strong dependence of hydride solubility with temperature. The brittle zone at room temperature conditions is not only the depth of the hydride rim in the cladding but also an underlying zone containing a significant hydrogen concentration, which depth is difficult to evaluate. 8. Conclusions The calculations of more than 60 PWR-RIA tests performed in different reactors and 20 thermalhydraulic RIA tests performed in PATRICIA facility allowed to assess the SCANAIR code prediction with various fuel and clad types, with water or sodium coolant, at room temperature or at 280 ◦ C, at atmospheric pressure or in PWR pressure condition and with a wide scope of pulse widths. The development of an automatic tool allows such an exhaustive experiments/computations comparison for each new SCANAIR release. During the first stage of the transient, PCMI phase, there is a good agreement between SCANAIR calculations and experimental results especially in the case of CABRI tests for which the experimental data are exhaustive. The clad and fuel elongations and clad hoop strains are accurately calculated for all kind of rod characteristics and conditions when the injected energy is lower

Please cite this article in press as: Georgenthum, V., et al., SCANAIR a transient fuel performance code Part two: Assessment of modelling capabilities. Nucl. Eng. Des. (2014), http://dx.doi.org/10.1016/j.nucengdes.2014.04.030

G Model NED-7870; No. of Pages 9

ARTICLE IN PRESS V. Georgenthum et al. / Nuclear Engineering and Design xxx (2014) xxx–xxx

than ∼120 cal/g. The failure occurrence and failure enthalpy are also correctly predicted by the fracture mechanics approach or the CSED approach. After the PCMI phase, the increase of cladding temperature can lead to the departure from nucleate boiling and the clad temperature can reach a very high value. If DNB occurrence is rather well predicted the calculations of the maximal clad temperature and the clad hoop strain during the boiling crisis have still to be improved. In the next years the modelling effort will be enhanced to better describe the post-DNB phase: validation of viscoplastic laws for the fuel and clad mechanical behaviour at high temperatures, improvement of the thermal-hydraulics modelling to better calculate the clad temperature in all conditions, improvement of the modelling of the gas behaviour during the transient. As presented in (Moal et al., 2014), some of these developments are already in progress but need validation. Further tests in the frame of the CABRI International Programme and the Fission Gas Dynamics programme in NSRR reactor will provide valuable experimental data for the validation of the post-DNB models. Acknowledgement The work described in the present paper has been performed in the continuity of many years of development of the SCANAIR code at IRSN with the constant support of Electricité de France. The authors are thankful to the IRSN colleagues for their fruitful contribution to the code development and the physical modelling. Mr. Vissyrias from the Assystem E&OS Company is also thanked for informatics support. Appendix A. Acronyms

BWR CIP CLARIS CSED DBA DNB E110 EGBs FCI FEM FGD FGR FGs FVM GBs HZP IGBs LWR MDA M5 NDA NSRR OECD

Boiling Water Reactor CABRI International Programme CLAd failure RISk under RIA transients Critical Strain Energy Density Design Basis Accident Departure from Nucleate Boiling Cladding material used in VVER fuel rods (Zr-1.0Nb by wt%) intEr Granular Bubbles Fuel Coolant Interaction Finite Element Method Fission Gases Dynamics programme Fission Gases Release Fission Gases Finite Volume Method Grains Boundaries Hot Zero Power Intra Granular Bubbles Light Water Reactor Mitsubishi Developed Alloy (Zr-0.8Sn-0.5Nb-0.32Fe0.1Cr by wt%) Cladding trademark of Framatome ANP (Zr-1.0Nb-0.13O by wt%) New Developed Alloy (Zr-1.0Sn-0.27Fe-0.16Cr-0.1Nb0.01Ni by wt%) Nuclear Safety Research Reactor Organisation for Economic Co-operation and Development

9

PCMI Pellet Cladding Mechanical Interaction PWR Pressurised Water Reactor ODESSA Organisation of Data Exchanges in Scientific Software Architecture Russian type, graphite-moderated, light water power RBMK reactor Rod Ejection Accident REA RIA Reactivity Initiated Accident SCANAIR Système de Calcul et d’ANalyse d’Accident d’Insertion de Réactivité Strain Energy Density SED VVER Vodo-Vodianoï Energuetitcheski Reaktor ZIRLO Cladding trademark of Westingouse Electric Company (Zr-1.0Nb-1.0Sn-0.1Fe by wt%) References Bernaudat, C., 2009. An analytical criterion to prevent PCMI fuel rod cladding failure during RIA transients. In: CSNI workshop on Nuclear Fuel Behaviour during Reactivity Initiated Accidents, Paris, France. Bessiron, V., 2007. Modelling of clad-to-coolant heat transfer for RIA applications. J. Nucl. Sci. Technol. 44 (2), 211–221. Bessiron, V., Sugiyama, T., Fuketa, T., 2007. Clad-to-coolant heat transfer in NSRR experiments. J. Nucl. Sci. Technol. 44 (5), 723–732. Canon, R.F., Roberts, J.T.A., Beals, R.J., 1971. Deformation of UO2 at high temperatures. J. Am. Ceram. Soc. 5., 105–112. Cazalis, B., Desquines, J., Poussard, C., Petit, M., Monerie, Y., Bernaudat, C., Yvon, P., Averty, X., 2007. The PROMETRA program: fuel cladding mechanical behaviour under high strain rate. Nucl. Technol. 157 (3), 215–229. Fuketa, T., Sasajima, H., Sugiyama, T., 2001. Behaviour higher burnup PWR fuels with low-tin zircalloy-4 cladding under reactivity-initiated-accident conditions. Nucl. Technol. 133 (1), 50–62. Fuketa, T., Sasajima, H., Tsuchihuchi, Y., Mori, Y., Nakamura, T., Ishijima, K., 1997. NSRR/RIA experiments with higher burnup fuels. In: ANS Int. Topical Mtg. on LWR Fuel Performance, Portland, Oregon. Geelhood, K., Luscher, W., Beyer, C., 2011. FRAPCON-3.4: a computer code for the calculation of steady-state thermal-mechanical behaviour of oxide fuel rods for high burnup. In: Tech. Rep. Office of Nuclear Regulatory Research, NUREG/CR7022, v1. Georgenthum, V., Desquines, J., Sugiyama, T., Fuketa, T., Udagawa, Y., 2008. Fracture mechanics approach for failure mode analysis in CABRI and NSRR RIA tests. In: Water Reactor Fuel Performance Meeting, Séoul, Korea. Latché, J., Lamare, F., Cranga, M., 1995. Computing reactivity initiated accidents in PWRs. In: SMIRT, Porto Alegre, Brazil. Lemoine, F., 2005. Estimation of the grain boundary gas inventory in MIMAS/AUC MOX fuel and coherence with REP-Na tests results. In: Water Reactor Fuel Performance Meeting, Kyoto, Japan. Moal, A., Georgenthum, V., Marchand, O., 2014. SCANAIR: a transient fuel performance code – part one: general modelling description. Nucl. Eng. Des. (in press). OCDE/NEA/CSNI, 2010. Nuclear Fuel Behaviour Under Reactivity Initiated Accident (RIA) Conditions – State of the Art., Tech. Rep. OCDE NEA N6847. Papin, J., Cazalis, B., Frizonnet, J.M., Desquines, J., Lemoine, F., Georgenthum, V., Lamare, F., Petit, M., 2005. Summary and interpretation of the CABRI REP Na program. Nucl. Technol. 157 (3), 230–250. Sartoris, C., Taisne, A., Petit, M., Barre, F., Marchand, O., 2010. A consistent approach to assess safety criteria for reactivity initiated accidents. Nucl. Eng. Des. 240 (1), 57–70. Sugiyama, T., 2006. Results from recent RIA test with high burnup fuel. In: Fuel Safety research Meeting, Tokai Mura, Japan. Sugiyama, T., 2010. High burnup fuel behaviour under high temperature RIA conditions. In: Fuel Safety research Meeting, Tokai Mura, Japan. Sugiyama, T., Fuketa, T., 2004. Effect of cladding surface pre-oxidation on rod cool ability under reactivity initiated accident conditions. J. Nucl. Sci. Technol. 41 (11), 1083–1090. Udagawa, Y., Sugiyama, T., Suzuki, M., Amaya, M., 2013. Experimental analysis with RANNS code on boiling heat transfer from fuel rod surface to coolant water under reactivity initiated accident conditions. In: IAEA Technical Meeting on Modelling of Water-Cooled Fuel Including Design-Basis and Severe Accidents, Chengdu, China. Udagawa, Y., Sugiyama, T., Suzuki, M., Fumihisa, N., 2011. PCMI failure limit assessed by fracture mechanics approach based on NSRR high-burnup PWR fuel test. In: Technical Meeting on fuel behaviour and modelling under severe transient and LOCA conditions, Mito, Japan.

Please cite this article in press as: Georgenthum, V., et al., SCANAIR a transient fuel performance code Part two: Assessment of modelling capabilities. Nucl. Eng. Des. (2014), http://dx.doi.org/10.1016/j.nucengdes.2014.04.030