The international program Phebus FP (fission

1The safety of nuclear reactors 1 6 Results of initial Phebus FP tests FPT-0 and FPT-1 S. BOURDON (IRSN) D. JACQUEMAIN (IRSN) R. ZEYEN (JRC/PETTEN) The international program Phebus FP (fission products) involves performance of in-pile experiments for the purpose of studying key physical phenomena associated with a severe PWR accident. This specifically encompasses fuel rod degradation, release to and transport of radioactive materials in the primary system and the containment, as well as their physico-chemical behavior. Three tests simulating typical low-pressure loss-of-coolant accident (LOCA) phenomena have been successfully completed, two of them in a steam-rich atmosphere (FPT-0 and FPT-1) and a third with a significant reducing phase (FPT-2). FPT-0 was performed in December 1993 with trace-irradiated fuel, FPT-1 in July 1996 and FPT-2 in October 2000, using irradiated fuel. Objectives set for the first two tests were achievement of advanced bundle degradation about 20% of the fuel by mass and release of 70 to 80% volatile FPs. Their results provide an experimental basis of vital importance to understanding fuel degradation and FP behavior. Analysis of FPT-0 and FPT-1 results has now ended, with that of FPT-2 data still underway. Findings from the first two tests are described on the following pages. Phebus program. Design of test bundles and experimental circuits The experimental fuel bundles used in FPT-0 and FPT-1 tests were of similar design. Each bundle contained twenty PWR fuel rods and a control rod made up of a neutron absorber alloy (silverindium-cadmium) with stainless steel cladding and a zircaloy guide tube (figure 1, page 46). While the FPT-1 bundle incorporated 18 PWR fuel rods pre-irradiated to an average burnup of 23.4 GWD/tU, the FPT-0 bundle comprised fresh fuel. Fuel rods measured 1 m in length and contained a fuel mass of 11 kg with a zircaloy cladding. The bundles were surrounded by an insulating shroud and placed in a tube cooled by a pressurized water circuit. The test fuel occupied a cell at the center of the Phebus reactor core. Various elements installed downstream of the bundle simulated the components of a standard PWR: RCS hot leg segment (700 C), steam generator U- tube, RCS cold leg segment (150 C) and containment (figure 2, page 46). SCIENTIFIC AND TECHNICAL REPORT 2002 45

Figure 1 Structure of experimental fuel bundles. Figure 2 Diagram of the Phebus experimental facility. Experimental sequence The bundle degradation phase lasted five to six hours. During this phase, bundle power and steam flow were increased by increments (figure 3) to gradually raise fuel rod temperature to the point where the fuel cladding failed. This was followed by control rod degradation and relocation of the absorber materials, oxidation of the zircaloy cladding, hydrogen production, fuel liquefaction and buildup of liquefied mixtures, molten pool formation, and release of fission products, structural and fuel rod materials. In the degradation process, steam was injected into the bundle at a pressure of approximately 2 bar and a flow of 0.5 to 2.2 g/s, thus generating an oxidizing or steam-rich environment. Following degradation, in-containment test phases lasted for five days, to allow investigation of aerosol deposition and the radiochemical behavior of iodine in sump water and the atmosphere. Bundle degradation results Figure 3 Principal events identified during the FPT-1 test. At the start of degradation phases in both FPT-0 and FPT-1experiments, zircaloy clad rupture due to increase in pressure inside the rods took place at about 800 C (maximum temperature measured at bundle mid-height). Such rupture was clearly confirmed by detection of FP in the containment and aerosols in the circuit. Fuel rod rupture was observed in both tests immediately prior to runaway of the zircaloy oxidation reaction by detection of In 116m in the circuit for a maximum bundle temperature between 1160 and 1330 C. These results indicate that control rod rupture took place at a temperature below the stainless steel melting point. Premature degradation of the control rod was probably caused by chemical interaction of its stainless steel cladding with the zircaloy guide tube. The liquefied mixture (of silver-indium-cadmium, stainless steel and zircaloy) then probably relocated to the lower grid during the oxidation period and solidified in the lower part of the bundle, forming a metallic mass rich in Ag and Zr. Hydrogen from zircaloy oxidation was detected in the FP circuit when fuel temperatures exceeded 1100 C in the upper bundle. An increase in fuel rod heatup rate was measured in 46 INSTITUT DE RADIOPROTECTION ET DE SÛRETÉ NUCLÉAIRE

1 both tests at a maximum bundle temperature of 1550-1650 C. Temperature peaks of 2500 C and 2200 C were observed in FPT-0 and FPT-1, with heatup rates of 10 to 15 C/s. Most hydrogen production took place during the oxidation runaway. Total steam starvation did not occur at bundle outlet in either of the tests. In test FPT- 1, a second lower hydrogen production spike was observed in the last twenty minutes of the transient. This delayed oxidation probably resulted from downward displacement of the molten materials to a colder zone where cladding had not yet been fully oxidized. The total mass of released hydrogen represented some 115 g in FPT-0 and 96 g in FPT-1; this corresponded to 77% and 64% oxidation of the bundle s zircaloy inventory. Analysis of these results shows that, in both tests, there was significant displacement of materials at the start of heatup, with maximum fuel temperature at 2000 to 2300 C. Low temperature relocation of fuel (i.e. at about 500 C below the melting point of pure UO 2 ) could be explained by formation, in the upper half of the bundle, of eutectic mixtures made up of absorber rod materials (with iron, nickel and chromium oxides) and partially oxidized cladding. During heatup, the resulting mixtures progressed downward, through the upper grid. At bundle center, in hotter zones, these mixtures may have interacted with absorber materials and fuel rods. This process probably led to fuel rod degradation in the upper half of the bundle with partial accumulation of the interacting ( Figure 4 Radiography and tomography of test bundles. materials near the lower grid. Analysis of measured bundle and shroud temperatures showed that fuel rod degradation and relocation took place gradually throughout the heatup period, which led to formation of a homogeneous molten pool at the lower grid. Tests ended with reactor trip when rapid, significant temperature increases were detected in the lower portion of the shroud. The safety of nuclear reactors In both tests, there was significant displacement of materials at the start of heatup, with maximum fuel temperature at 2000 to 2300 C. SCIENTIFIC AND TECHNICAL REPORT 2002 47

Figure 5a Post-test examinations and final bundle status Gamma scanning, radiography, tomography and destructive testing subsequently confirmed the advanced stage of FPT-0 and FPT-1 fuel bundle degradation achieved during the experiments (figure 4, page 47). Analysis of bundle post-irradiation examinations (PIEs) showed accumulation in the lower bundle of a (U-Zr)O 2 corium containing 1 to 2% iron and chromium oxides (by weight). PIE also identified a solidified, Ag-rich metallic mass just below the UO 2 -rich molten pool. Interpretation of tomograms enabled evaluation of total UO 2 mass in this pool (about 2.6 kg in FPT-0 and 2 kg in FPT-1). Moreover, corium melting temperature in the lower bundle, as measured during PIE, was around 2500 C. Transport of iodine-131 (volatile FP) in the cold leg (150 C) during the FPT-1 transient. Figure 5b Transport of ruthenium-103 (low volatility FP) in the hot leg (700 C) and cold leg (150 C) during the FPT-1 transient. Based on PIE results for the FPT-1 bundle, it was clear that fresh fuel rods had sustained significantly less damage than the pre-irradiated rods. FP and material behavior in the test circuits RELEASE AND TRANSPORT OF FISSION PRODUCTS, FUEL, BUNDLE STRUCTURES AND CONTROL ROD MATERIALS Released bundle fractions were comparable for FPT-0 and FPT-1, with slightly lower volatile FP values in FPT-1. These lower values can be attributed to a lesser degree of bundle degradation in the second test. In FPT-1, fuel material (U) release was similar to that measured in FPT-0. The difference in type of fuel used in the two tests (trace-irradiated for FPT-0, irradiated for FPT-1) did not seem to sharply impact U release. FP elements can be classified as follows, based on the fractions released: high release (approx. 90% of inventory): rare gases Kr and Xe and volatile FPs I, Cs and Te; significant release (20-60% of inventory): Mo, Tc and Sb, and Sn originating from bundle structures; low release (5-15%): Ag and In from the control rod and Re from the thermocouples; very low release (less than 1% of inventory): low volatility FPs Ba(La), Ru, Sr, Zr-Nb, Nd and Sm (in FPT-1), fuel material U and Zr from bundle structures. In short, results of tests FPT-0 and FPT-1 confirmed the release fractions obtained in analytical experiments, except for Ba (lower release in the Phebus series). In both tests, the release and transport of materials in the primary circuit correlated closely with bundle degradation events (figures 5a and 5b). Volatile FP release generally reached a maximum during the early and late zircaloy oxidation phases. Their rate of release decreased in the corium propagation phase. Release of fuel and structural materials, as well as silver (from the silver-indium-cadmium control rod) peaked during the late oxidation phase. Except for iodine and cadmium, most elements were already in condensed form in the circuit hot leg (700 C). Cs behavior was more complex than expected, since a significant fraction of this element was probably transported as vapor and another fraction in condensed form. Materials transported as vapor in the hot leg condensed in 48 INSTITUT DE RADIOPROTECTION ET DE SÛRETÉ NUCLÉAIRE

1 For FPT-1, the aerosol mass transiting through the circuit was 150 g in the hot leg and 130 g in the cold leg (for an initial fuel inventory of 11 kg). ( The safety of nuclear reactors the steam generator, so that all of them, other than a small fraction of iodine, were in condensed form in the circuit cold leg. Condensed materials were transported as multicomponent aerosols in both hot and cold legs. The aerosol mass predominantly comprised control rod, structural and fuel materials. For FPT-1, the aerosol mass transiting through the circuit was 150 g in the hot leg and 130 g in the cold leg (for an initial fuel inventory of 11 kg). Aerosol mass concentrations were highest in the late oxidation phase. Aerosol composition (by mass) was determined by degradation events, with large Ag, In, Cd and Sn contributions during the first oxidation phase, large Ag and Re input and significant Cs and Mo contributions at start of meltdown, and predominantly Ag, Re and U content in the late oxidation phase. CIRCUIT RETENTION OF FISSION PRODUCTS, AND BUNDLE/STRUCTURAL/CONTROL ROD MATERIALS The quantity of material deposited at bundle outlet seems to correlate with the degree of volatility of the various elements involved. The vapors of low volatility elements were probably the first to condense. This was confirmed for Ru and Zr; in both cases, some 45% of the freed fraction was deposited in the upper plenum (at degraded bundle outlet), with only low deposits in the vertical line. Similar behavior was observed for Ag. There was little deposition of volatile elements in the upper plenum and more in the vertical line. In the steam generator tube, iodine and cadmium deposits, due essentially to steam condensation, were more significant than aerosol deposits, and represented some 25% of the mass injected into the tube. Aerosol deposits resulting mostly from thermophoresis represented only 14% of the mass entering the tube. Most deposition (85%) took place in the SG hot leg. AEROSOL BEHAVIOR IN THE CONTAINMENT Settling was the main mechanism of aerosol deposition in the containment: some 65 to 70% of total inventory was deposited by gravity on the containment floor. Diffusiophoresis (aerosol entrainment in condensed steam) was the second most important deposition mechanism. As a result, 25% to 28% of containment inventory was entrained by condensation towards the painted surfaces of the condenser. Containment wall deposits were of minor significance: only 2% of the inventory was detected on walls at the end of the aerosol deposition phase. During containment floor washing, most of the containment inventory was entrained into the sump, where it appeared either as soluble matter in sump water or as surface deposits (having primarily settled by gravity to the sump floor). Closer analysis showed that Cs was present in soluble form, while Mo, Tc, Ba, Cd and Re were only partially dissolved. Volatile FPs I, Te and Sb, as well as low volatility FPs Ru, La, Sm and Nd, control rod elements Ag and In, actinides and elements from the fuel and its structures remained mostly insoluble and were present as particulate matter in the sump. After analysis, a complete overview of data for all of these elements was possible, as shown for iodine in figure 6 (page 50). SPECIFIC BEHAVIOR OF IODINE In both tests, nearly all of the fuel bundle iodine (about 87% of total inventory) was released in gaseous form. This gaseous iodine probably reacted partially with Ag, In, Cs or Rb in the vertical line above the bundle, to form a metal iodide vapor. In the circuit hot leg, most of the iodine flow was in gaseous or metal vapor form. Iodine was then transported to the SG tube, where it formed more deposits than any other element (23.5% of bundle inventory in FPT-0 and 19.2% in FPT-1). Deposition was due either to metal iodide condensation in the fluid and on SCIENTIFIC AND TECHNICAL REPORT 2002 49

Figure 6 Iodine-131 balance in the test circuits. tube walls, to chemisorption of gaseous iodine onto Cd deposits, or to iodine condensation on aerosols, which are subsequently deposited by thermophoresis. In the circuit cold leg, most of the iodine was apparently transported by aerosols. More than 60% of bundle iodine inventory was injected into the containment (63% in FPT-0 and 64% in FPT-1). Most of it had been transported by aerosols that either settled on the containment bottom or deposited on the containment surfaces (essentially by diffusiophoresis on the painted condenser surfaces). In both tests, gaseous iodine fractions measured in the containment reached their highest levels during zircaloy oxidation phases (for FPT-1, about 0.2% of bundle inventory, during the first phase of oxidation, which represented 4% of containment inventory at that point in time; and, for FPT-0, some 3% of bundle inventory, or 33% of containment inventory at the same point in time). In FPT-1, following the two oxidation phases, gaseous iodine fraction decreased by a factor of 2 in less than an hour. This is linked to diffusiophoresis of gaseous iodine at the painted condensing surfaces and/or to a fast chemical transformation of the iodine species considered. The volatile fraction of the iodine originating from the primary circuit also disappeared in a few hours in FPT-0, probably due to deposition on the containment surfaces (painted condensers). Evolution of the gaseous iodine fraction during the aerosol phase was different in the two tests (figure 7). In FPT-1, the gaseous iodine fraction increased significantly from 0.07% to an average fraction of 0.14% of bundle inventory immediately after containment isolation. It then remained more or less constant at this value for about 5 hours. The increase (from 0.07% to 0.14%) can be attributed to a measured desorption of gaseous iodine from the painted condensing surfaces. Measured data was consistent with 50 INSTITUT DE RADIOPROTECTION ET DE SÛRETÉ NUCLÉAIRE

1 release of organic iodides. In FPT-0, the gaseous iodine fraction decreased exponentially during the aerosol phase, dropping from 2.6% to 0.32% of bundle inventory. Unlike what was observed in FPT-1, there was no significant decrease during FPT-0 in iodine activity on the painted condenser surfaces. In both tests, before the washing phase, the gaseous iodine fraction was divided between I 2 and the organic iodides with, in the case of FPT-1, an increase in organic iodide fractions during the aerosol phase that followed containment isolation. During the washing phase, iodine in the sump behaved mainly like particulate matter and settled by gravity to the bottom of the sump, meaning that this iodine was transported to the sump and formed an insoluble species (e.g. particulate Agl). After washing, the gaseous iodine fraction was low in both cases, representing only 0.063% of bundle inventory in FPT-0 and 0.094% of this inventory in FPT-1 (figure 7). Speciation of gaseous iodine differed from one test to the other. In FPT-1, measured data was homogeneous and showed that I 2 was the predominant gaseous iodine species at the time of washing. After this phase, the I 2 contribution to the gaseous fraction diminished, while that of organic iodides increased. In FPT-0, data was not precise enough to detect any impact of washing on the gaseous iodine fraction; however, organic iodides were the major gaseous iodine species at that time. In both tests, measured data showed that in the long term, organic iodides predominated over other gaseous iodine species. Results indicated a small transfer of gaseous iodine from the sump to the atmosphere due to iodine trapping by Ag and the ensuing inhibition of volatile iodine formation by radiolysis. ( Figure 7 General behavior of the in-containment gaseous iodine fraction during Phebus tests FPT-0 and FPT-1. Conclusion Several of the key phenomena potentially occurring in a severe PWR accident were reproduced by Phebus FPT-0 and FPT-1 experiments. These included clad burst, zircaloy oxidation and hydrogen production, control rod failure and relocation of absorber materials, relocation of fuel rods, UO 2 buildup and molten pool formation and FP and aerosol release, and transport and deposition in the primary circuit and the containment. Findings derived from experimental results can now be used to address important questions about the source term in a severe accident situation. The control rod failure process and early accumulation of metallic materials in the lower fuel bundle (below the UO 2 pool) were clearly identified. Measurement of total hydrogen mass The safety of nuclear reactors Measured data showed that, over the long term, organic iodides were the predominating gaseous iodine species. SCIENTIFIC AND TECHNICAL REPORT 2002 51

1 Leaktight housing for remote handling in the Phebus facility. produced in both tests provided interesting insight into zircaloy oxidation kinetics. The relatively low temperature at which fuel rod relocation took place is a point worth considering in severe PWR accident assessments. The FPT experiments showed that relocation of materials and their buildup leads to formation of a UO 2 - rich molten pool and its gradual propagation through the lower part of the bundle. Post-irradiation examinations plainly demonstrated that, under identical conditions, irradiated fuel sustained more damage than fresh fuel. For fission products, the points of ongoing interest to severe PWR accident assessment include: the main phases of material transport through the circuit, which are correlated with bundle degradation events (zircaloy oxidation, dislocation of materials, etc.); deposition in the circuit, which in FPT experiments, affected mainly the hot leg (in piping above the degraded fuel bundle and at the SG inlet); injection of a significant quantity of gaseous iodine during zircaloy oxidation phases; aerosols that transported FPs in the circuit, which were the multicomponent type, mostly made up of fuel bundle structural and control rod materials; long-term predominance of organic iodides in the gaseous iodine fraction, in the containment model; containment sump chemistry, marked by trapping of iodine by silver (released on degradation of the silver-indium-cadmium control rod), which precludes formation of volatile iodine by radiolysis. Experiment FPT-2 was geared to the same types of phenomena as FPT-1, for steam-poor conditions in the primary circuit and different in-containment conditions impacting iodine volatility. The last test in the program, FPT-3, will study the effect of control rod absorber material (B 4 C) on fuel bundle degradation and the impact of its oxidation products on FP behavior. ( Relocation of materials and their buildup leads to formation of a UO 2 -rich molten pool and its gradual propagation through the lower part of the bundle. 52 INSTITUT DE RADIOPROTECTION ET DE SÛRETÉ NUCLÉAIRE