A Comparative Analysis of CABRI CIP-1 and NSRR VA-2 Reactivity Initiated Accident tests M. PETIT*, V. GEORGENTHUM*, T. SUGIYAMA**, M. QUECEDO***, J. DESQUINES* * IRSN, DPAM/SEMCA, BP 3, 13115 Saint-Paul-lez-Durance Cedex, France ** Japan Atomic Energy Agency, Tokai-mura, Ibaraki-ken 319-115, Japan *** ENUSA Industrias Avanzadas S.A., Santiago Rusiñol 12, 284 Madrid, Spain Abstract: Zirlo is an improved fuel cladding material that was developed to accommodate constraints associated with increasing burn-ups in a more severe duty operation. The behavior of Zirlo irradiated above 75 GWd/t under RIA conditions was studied in CABRI CIP-1 and NSRR VA-2 tests. The samples were initially similar but VA-2 failed at relatively low enthalpy whereas CIP-1 did not fail. Because test conditions were different in terms of initial temperature, coolant nature and power pulse width, it is necessary to use a transient fuel code to compare the two results. Using SCANAIR and CLARIS it was shown that the concentration of hydrogen precipitates at the temperature of the test is of primary importance to assess both experiments in a consistent manner. The analysis demonstrated that room temperature is a testing condition much more severe than that expected at typical PWR conditions. 1 INTRODUCTION Economic considerations have lead utilities to propose burn-up increases of the fuel used in Light Water Reactors. However, experiments performed during the 9 s to simulate Reactivity Initiated Accidents (RIA), in particular in the experimental reactors CABRI in France and NSRR in Japan, have shown that Zr-4, the cladding material most commonly used in Pressurized Water Reactor, has a resistance to the loadings resulting from Rod Ejection Accidents that decreases with increasing burn-ups [1, 2, 3]. It was then commonly admitted that Zr-4 was not suitable for the targeted burn-up levels. During reactor operation, fuel clads are submitted to a high temperature water environment. Under these conditions, the cladding material oxidizes and a zirconia layer forms on the outer part of the CIP-1 clads. Part of the hydrogen Zircaloy-4 father rod generated during oxidation VA-2 diffuses into the clad base father rod metal and then precipitates Zirlo as zirconium hydrides if the solubility limit is reached. Different typical hydride morphologies, i.e. hydride distribution and orientation are observed. When the hydrogen mean concentration is low, below the solubility limit during Figure 1: Zr-4 and Zirlo oxidation versus burnup (from R.L. Kesterson et al. TOPFUEL 26)
irradiation, the hydrides observed at room temperature are more or less randomly distributed through the clad thickness. When the mean concentration increases, under the thermal gradients existing during operation, the soluble hydrogen migrates to the cooler clad outer diameter and accumulates as hydrides below the zirconia layer, to form a hydride rim. Besides, if a cold spot exists on the clad, for example resulting from zirconia layer spalling, hydrogen can move towards this cold spot resulting in an hydride blister, which is a massive zirconium hydride. It is well known that hydrogen embrittles metals. This deleterious effect of hydrogen was confirmed on Zr-4 as failures were observed in RIA experiments with clads containing hydride blisters (e.g. CABRI REP-Na1, Na8 and Na1) or hydride rims (e.g. NSRR HBO-1 and HBO-5). New zirconium alloys were then developed with better resistance to water corrosion and hydriding. Zirlo is one of these new alloys [4]. Figure 1 shows that corrosion resistance of Zirlo is globally better than that of Zr4 [5]. The objective of the CABRI CIP-1 and NSRR VA-2 tests was to characterize the behavior of Zirlo rods under RIA conditions. It can be seen in Figure 1 that both tested samples were taken from fuel rods that were representative of higher corrosion values observed on Zirlo rods. 2 DESCRIPTION OF THE TEST RODS Both CIP-1 and VA-2 rodlets were refabricated from the fifth (from bottom) span of fuel rods that were inserted in a fuel assembly irradiated in the Vandellos 2 reactor in Spain in the frame of a joint Japanese-Spanish R&D program [6]. This program aimed at studying the behavior of UO 2 fuel at high burn-up. The power histories of both rodlets were quite similar as shown in Figure 2. They had a low power 4 th cycle, even lower for CIP-1 than for VA-2. 25 2 Linear power (w/cm) CIP-1 VA-2 15 1 5 5 1 15 2 days Figure 2: Power histories of CIP-1 and VA-2 The maximum local burn-up was respectively 75 and 79 GWd/t for CIP-1 and VA-2. The fissile length was respectively 54 and 11 cm. The mean zirconia thickness was 8 µm for CIP-1, varying between 5 and 1 µm over the length of the sample. For VA-2, the mean value was 7 µm and variation was +/-2 µm. The state of the rods before test was characterized by examining radial cuts on the father rod. Typical metallographies are show in Figure 3. 2
ZrO2 Hydride rim Figure 3: Metallographies of father rods. Left: CIP-1. Right: VA-2 The hydrogen content of the rods was relevant, with a mean hydrogen concentration of about 1 ppm for CIP-1 and about 8 ppm for VA-2. The hydrides were long and oriented in the circumferential direction, as expected for Zirlo. In both cases, no hydride blisters were found and hydride rims with a similar thickness of 5 µm were observed. In conclusion, with respect to hydrogen content, the two rods were very similar. Thus their mechanical properties are also expected to be close to each other. 3 TESTS CHARACTERISTICS The main differences between CIP-1 and VA-2 are the test conditions. CIP-1 was tested on November 29, 22, in the former sodium loop of the CABRI reactor. Typical conditions in CABRI are: sodium at 28 C, f lowing at 4 m/s under a.3 MPa pressure. These are close to PWR hot zero power PWR conditions, except for the nature of the coolant and the channel pressure. VA-2 was performed on August 2, 25, in the test capsule of the NSRR reactor. In this case, the coolant was stagnant water at 2 C and. 1 MPa pressure. Because of their different designs, CABRI and NSRR have quite different characteristics with respect to energy injection into the test rodlet. For CIP-1, the injected energy after 1.2 s was 99 cal/g. The full width at half maximum (FWHM) of the power pulse was 32.4 ms. In the VA- 2 test, the 4.4 ms FWHM pulse resulted in an injected energy after.2 s of 138 cal/g. The core power traces and injected energy as a function of time are shown in Figure 4. Core power (MW) 4 Power 35 3 25 2 15 1 5 Energy Enthalpy Core power Enthalpy Injected energy cal/g 12 1 8 6 4 2 Core Power (MW) 25 Power 2 15 1 5 cal/g 15 Energy 12 Enthalpy 9 Core power Injected energy (cal/g 6 Enthalpy 3.3.4.5.6.7.8.9 1 time (s).5.1.15.2.25.3.35.4.45.5 time (s) Figure 4: Core power, Injected energy and Fuel enthalpy versus time. Left: CIP-1, Right VA-2 Because safety criteria for RIA are usually expressed in terms of fuel enthalpy or fuel enthalpy variation, it is usual to characterize RIA tests by these values. However, these values are not measurable and must be computed with a fuel code. 3
The SCANAIR code [7], developed by IRSN for RIA applications, was used to determine fuel enthalpies versus time for both tests. Enthalpy evolutions are plotted in Figure 4. In CIP-1, the maximum fuel enthalpy Hmax was 93 cal/g (enthalpy increase H=76 cal/g). In VA-2, Hmax= H=128 cal/g because H is conventionally equal to at 2 C. 4 TESTS RESULTS Results from the CIP-1 [8] and VA-2 [9] were already reported. They are briefly recalled below. 4.1 Results from CIP-1 On-line measurements recorded during the CIP-1 test did not show any noticeable event before reactor power shut down. However, one of the two microphones recorded a late event that remains unexplained to date. It was later confirmed, based on posttest exams that this late event did not correspond to a clad failure. Posttest exams also revealed important features of the test. The clad permanent hoop strain remained quite low, with a maximum value of.5%. Visual exams also showed that the rodlet experienced extensive zirconia spalling during the test, as exemplified in Figure 5. On metallographies of the clad performed on a radial cut (see Figure 6), one can see that part of the hydrides are oriented in the radial direction, whereas before test all hydrides were circumferential. This implies that a fraction of the hydrides were dissolved during the test and reprecipitated under sufficient hoop stress to form radial hydrides. Also visible on metallographies are radial cracks at the outer surface of the clad in spalled areas. These cracks extend through the hydride rim. Figure 5: Spalling in CIP-1 ID Circumferential H 1 µm Radial H Circumferential H Radial H Circumferential H 5 µm Outer Diameter Figure 6: Metallographies of CIP-1 after test. Left: hydrides morphology in the clad. Right: Detail showing radial cracks in the hydryde rim in spalled areas. 4
4.2 Results from VA-2 On-line measurements during the VA-2 test did record the failure of the rodlet at a time corresponding to a fuel enthalpy of 55 cal/g. The total uncertainty on this value was evaluated to be +/-1 cal/g. This event is clearly seen in Figure 7, which depicts the coolant pressure inside the capsule as a function of time. Failure time Integrated power (MJ) Figure 7: Capsule internal pressure versus time Visual examinations after test (see Figure 8) showed that a crack propagated all along the rodlet. A SEM examination of the fracture surface was performed at a location defined in Figure 8. SEM images are given in Figure 9. The appearance of the fracture surface reveals that the outer half of the clad experienced a brittle failure, whereas the crack propagated in a ductile manner in the inner half. Figure 8: Visual appearance of VA-2 after test Figure 9: SEM examinations of VA-2. Left: Clad thickness. Right: details showing a brittle fracture in the outer half (top) and a ductile fracture in the inner part (bottom) 5
5 ANALYSIS WITH SCANAIR-CLARIS Because test conditions in CIP-1 and VA-2 are quite different in terms of initial temperature, it is not possible to make a direct 6 comparison between both results. Critic al c rac k (µm ) It is then necessary to make use of a computer code to assess whether it is possible to analyze 5 4 CIP -1 both tests with a consistent set of V A -2 CIP-1 3 V A -2 failure hypotheses. The SCANAIR code was used for that purpose. It has been shown 2 previously that SCANAIR 1 modeling is adequate to 5 VA-2 1 2 3 4 5 6 7 8 9 mechanical behavior of the CIP- E nthalpy inc reas e (c al/g) reproduce the thermal and 1 11 12 13 1 rodlet as calculation results in Figure 1: Critical crack length versus enthalpy increase in terms of coolant temperature and CIP-1 and VA-2 permanent strain were in very good agreement with experimental results. Because hydrided rods fail due to the induced embrittlement of the clad, a post-processing module called CLARIS was developed for SCANAIR. CLARIS is based on elastic plastic fracture mechanics theory and enables to evaluate a critical crack size (the size at which the crack propagates) evolution during RIA transients. The critical crack size evaluated as a function of fuel enthalpy for CIP-1 and VA-2 is plotted in Figure 1. For CIP-1, the minimal critical crack size equals to 1 µm. Assuming that the crack size is equal to the hydride rim thickness of 5 µm, it gives a result consistent with experimental observation that the rod did not fail. With regards to VA-2, the critical crack size of 5 µm is reached for a fuel enthalpy value of 9 cal/g, which is far beyond the enthalpy at failure of 55 cal/g. Experimental uncertainty on enthalpy at failure can not explain this discrepancy because is amounts to +/- 1 cal/g only. If one looks at the enthalpy of 55 cal/g, SCANAIR-CLARIS calculation indicates that the critical crack size at that moment was about 28 µm. This is consistent with the result of SEM examination that revealed brittleness extending over half of the clad thickness (see Figure 9). Although the tested rods were nearly identical, the VA-2 result suggests that the apparent rim thickness to be considered in the fracture analysis is 28 µm instead of 5 µm as in CIP-1. This difference in behavior cannot be explained only by a difference in fracture toughness because the same fracture toughness model that includes the influence of temperature was used in both analyses. This toughness model, based on a large review of the literature is presented in Figure 11: Thoughness versus temperature for different hydrogen content 6
Figure 11. The difference can neither be explained by a substantial difference in the mechanical loading of the clad. The Figure 12 presents the mechanical strain at the outer part of the clad as a function of enthalpy increase computed by SCANAIR. At the time of VA-2 failure, the mechanical loading in both rodlets was not different. It is then necessary to look at the temperature evolution in the rodlets into more details. Radial temperature profiles at different times are plotted in Figure 13. It is important to note that at the time of failure of VA-2, the temperature of the outer half of the clad was still close to the initial temperature of 2 C. On the contrary, in CIP-1, the clad temperature goes very rapidly above 4 C, because of the initial temperature of 28 C, the large (compared to VA-2) pulse width and the good thermal conductivity of the sodium, which induces high thermal fluxes from the rod to the coolant. T ( C) 16 14 12 1 8 6 Hmax mechanical hoop strain 1.8% 1.6% 1.4% 1.2% 1.%.8%.6%.4%.2%.% fuel t=2s Pmax clad 4 Pm ax 2 Hm ax t=2s.1.2.3.4.5 r (m m ) CIP-1 1 2 3 4 5 6 7 8 9 1 11 12 13 Enthalpy increase (cal/g) T ( C) 3 25 2 15 1 5 55cal/g Hmax t=2s failure fuel Hmax t=2s clad.1.2.3.4.5 r (mm) Figure 13: Radial temperature profile in the rodlet at different times. Left: CIP-1, right VA-2 This difference in temperature has an impact on the hydride content in the clad during the test. Referring to curves defining hydrogen solubility 1 limits versus temperature in zirconium alloys (see Figure 14), one can evaluate that between 2 and 1 ppm 1 were dissolved during the CIP- Precipitation 1 transient (consistent with the fact that radial hydrides were 1 Csd - Kearns 67 evidenced after the test, see 4.2 Dissolution Csp - Zhang 93 above). Consequently, apart from the 5 µm outer rim, few 1 hydrides precipitates remained. 1 2 3 On the contrary, at the time of Temperature ( C) 4 5 6 failure of VA-2, no hydrogen Figure 14: Hydrogen dissolution and precipitation limits in dissolution had occurred in the zirconium alloys versus temperature [H](ppm) VA-2 VA2 CIP-1 VA-2 failure Figure 12: Mechanical hoop strain versus enthapy increase in CIP-1 and VA-2 7
outer part of the clad. In this situation, the VA-2 results indicates that the length to be compared to the critical crack size is not only the depth of the hydrogen rim but must also include the underlying zone containing significant hydrogen concentration. This is represented in a schematic way in Figure 15. Brittle zone ZrO2 Hydride rim ZrO2 Hydride rim Brittle zone DuctileZry4 DuctileZry4 Figure 15: Schematic representation of hydrides morphology at time of maximal mechanical loading Left: CIP-1, Right: VA-2 As a conclusion, the comparative analysis of CIP-1 and VA-2 shows that both temperature and hydrogen repartition have a strong impact on the clad resistance during RIA transients. The absence of hydrogen dissolution at room temperature as in the VA-2 test implies that the testing conditions were much more severe than those expected at typical PWR conditions. 6 CONCLUSION The behavior of Zirlo irradiated up to more than 75 GWd/t was tested under fast thermo mechanical loading representative of RIA in the CABRI CIP-1 and NSRR VA-2 tests. Examinations on the father rods showed that the two tested samples were very similar. The CIP-1 rod reached a 93 cal/g enthalpy without failure; the VA-2 rod failed at 55 cal/g, which is a relatively low value. Because test conditions were different in terms of initial temperature, coolant nature and power pulse width, it is necessary to use a transient fuel code to compare the two results. Using SCANAIR and CLARIS it was shown that the concentration of hydrogen precipitates at the temperature of the test is of primary importance to assess both experiments in a consistent manner. At low temperature, the initial crack length to be used in failure analysis is much wider than only the hydride rim. Room temperature is thus a testing condition much more severe than that expected at typical PWR conditions. New experiments on Zirlo foreseen in both the CIP program in CABRI and the ALPS program in NSSR will help to refine this analysis. REFERENCES 1. Summary and Interpretation of the CABRI REP-Na Program, J. Papin, B. Cazalis, J.M. Frizonnet, J. Desquines, F. Lemoine, V. Georgenthum, F. Lamare, M. Petit, Nuclear Technology, vol. 157, number 3, (27) 2. The Status of the RIA Test Program in the NSRR, T. Fuketa, T. Nakamura, K. Ishijima, Proc. 25 th Water Reactor Safety Meeting, Bethesda, MD, Oct. 2-22, 1997, NUREG/CP 162, Vol. 2, p. 179 3. Behavior of High-Burnup PWR Fuels with Low-Tin Zircaloy-4 Cladding Under Reactivity- Initiated Accident Conditions, T. Fuketa, H. Sashajima, T. Sugiyama, Nuclear Technology, vol. 133, number 1, (21) 4. Zirlo : An Alloy Development Success, G.P. Sabol, ASTM STP 1467, pp.3-24, (26) 5. Cladding Optimization for Enhanced Performance Margins, R.L. Kesterson, H.K. Yueh, H.H Shah, J.P. Foster, D.J. Colburn, L. Hallstadius, I. Arana, Proc. of TopFuel 26, Salamanca, Spain, Oct. 22-26, 26 8
6. Post Irradiation Examinations on 67-75 GWd/t rods for Confirmation of the Integrity and Appropriate Performance of the Claddings for Future, S. Watanabe, S. Abeta, J.J. Serna, J.M. Alonso, T. Sendo, P. Gonzalez, Proc. Of the 25 Water Reactor Fuel Performance Meeting, Kyoto, Japan, October 2-6, 25 7. The SCANAIR Code Version 3.2: Main features and Status of Qualification, E. Federici, F. Lamare, V. Bessiron, J. Papin, IAEA TCM on Fuel Behavior Under Transient and LOCA Conditions, Halden, Norway, September 1-14, 21 8. IRSN R&D Studies on High Burn-up Fuel Behaviour Under RIA and LOCA Conditions, J. Papin, M. Petit, C. Grandjean, V. Georgenthum, Proc. of TopFuel 26, Salamanca, Spain, Oct. 22-26, 26 9. PCMI Failure of High Burnup Fuels Under RIA Conditions, T. Sugiyama, Fuel Safety Research Meeting, Tokai, Japan, May 16-17, 27 9