[en] In this paper, three power ramp tests performed on high burn-up Recrystallized Zircaloy2 - UO₂ BWR fuel rods (56 to 63 MWd/kgU) within the SCIP project have been simulated with METEOR and ALCYONE 3D. Two of the ramp tests are of staircase type up to Linear Heat Rates of 420 and 520 W/cm and with long holding periods. Failure of the 420 W/cm fuel rod was observed after 40 minutes. The third ramp test consists in a more standard ramp test with a constant power rate of 80 W/cm/min till 410 W/cm and a short holding time. The tests were first simulated with the METEOR 1D fuel rod code leading accurate results in terms of profilometry and fission gas releases. The behaviour of a fuel pellet fragment and of the cladding piece on top of it was then investigated with ALCYONE 3D. The size and the main characteristics of the ridges after base irradiation and power ramp testing were recovered. Finally, the failure criteria validated for PWR conditions and fuel rods with low to medium burn ups were used to analyze the failure probability of the KKL rodlets during ramp testing

[en] This study is concerned with the modelling of fuel behaviour and of pellet-cladding interaction (PCI). A new fuel software (PLEIADES) is currently co-developed by the Atomic Energy Commission (CEA) and Electricite de France (EDF). This software includes a multi-dimensional FE program (ALCYONE) devoted to Pressure Water Reactors (PWR) fuel rods. PCI studies are mainly undertaken with the 3D model of ALCYONE. The objectives of this work are twofold: first, to propose a constitutive model for the fuel pellet which accounts for the stress relaxation of the material resulting from cracking and creep, second, to estimate the impact of the pellet cracking oil PCI. In this paper, a mathematical formulation which couples a viscoplastic law for creep with a multi-surface plastic softening law for cracking is detailed, leading a two inelastic strains model. Mesh dependency is overcome thanks to a material parameter related to the finite element size. The 3D calculations of PCI presented in this paper show that the considered modelling of fuel cracking is consistent with the experimental knowledge available on crack development under irradiation. A parametric study is then presented which leads to the conclusion that the tangential stresses at the pellet cladding interface and hence the risk of PCI failure are significantly reduced when the fuel tensile strength is divided by two. (authors)

[en] Thanks to an integrated approach, the CYRANO3 fuel performance code allows a complete description of PWR fuels behaviour from the early stages of in-core irradiation to decades of storage. To this end, a modular approach is used, including highly coupled multi-physics models (e.g. neutronics, physics, mechanics...), each of them being developed and identified on the basis of dedicated experimental investigations. In the present paper, a focus on spent fuel modelling during transport and dry storage is proposed. In CYRANO3, a spent fuel rod model is determined thanks to a first in-core computation run; its description thus includes several physical and mechanical parameters that might affect the rod.s behaviour during transport and storage. Then, a second computation run is performed in order to determine the evolution of the rod.s clad creep strain during specific transport/storage conditions. The maximum creep strain value can then be compared to an experimentally determined failure criterion. Taking advantage of the broad validation range of CYRANO3 code, computations can be applied to a large variety of in-core and out-of-core scenarios, including design and safety analyses. In this paper, the problem of claddings creep excessive deformation during transport or dry storage of spent fuel assemblies is addressed. Typical evolutions of the spent fuel rods properties are detailed, such as geometric or mechanical changes, showing the importance of a good knowledge of spent fuel rods inner pressure during their whole life - which relies on a correct prediction of the free volumes and the fission gas release in the whole rod - and irradiated cladding materials creep properties in the different transport/storage expected conditions (temperature and stress ranges). Taking all these aspects into account, it is shown that, even with conservative assumptions regarding temperature levels of the different phases, important margins usually exist regarding clad creep excessive deformation during transport or dry storage of spent fuel assemblies

[en] This study is concerned with structural integrity assessment of Pressure Water Reactor's (PWR) fuel rods under pellet-cladding interaction (PCI) loading condition. An important experimental and research cooperative program between EDF, AREVA-NP and the Atomic Energy Commission CEA is achieved in order to get a better understanding of the mechanisms possibly leading to PCI failure, as well as to qualify a PCI resistant rod design. The objectives of this work are: first, to improve the understanding of the pellet mechanical properties impact on cladding local loading with 3D simulations results, and second, to propose a new phenomenological rupture criterion for a better assessment of the failure risk. In this study fuel behaviour modelling under nominal and transient loading conditions is achieved with a multi-dimensional simulation tool called ALCYONE, included in the new fuel software PLEIADES currently co-developed by the CEA and EDF. Cladding loading due to mechanical interaction during power transient stage is first analysed through pellet-cladding interfacial stresses computed in the 3D simulation. Then, a 2D model is proposed in order to establish a correlation between interfacial loading and stress concentration in the cladding. In order to assess the failure risk under PCI a phenomenological criterion based on the membrane circumferential stress in the cladding and shear stresses at pellet-cladding interface is proposed. To compute the shear loading at pellet-cladding interface a new parameter (called W_rθ) is introduced. Based on 3D calculations of PCI, it is shown in this paper that pellet fracture properties can have a significant effect on PCI loading

[en] In order to survey the supplier's fuel rod design studies, to establish the operation technical specifications, to perform safety studies for new fuel management and to define the fuel initial conditions of accidental situations, EDF needs a reliable code to simulate the PWR thermomechanical fuel behaviour. CYRANO3 is the actual computer code developed and used by EDF to perform all these calculations. It has to be able to simulate a large range of fuel rods: pellets can be made of UO₂, MOX, UO₂+Gd₂O₃, or UO₂ + additives, with a surrounding cladding made of Zircaloy 4, M5, Zirlo... Within the PWR safety studies framework, the verification of the fuel safety criteria, the integrity of the fuel rod (first safety barrier) has to be guaranteed in any class I and II operating conditions. Furthermore, in a strongly competitive environment, the industrial fuel behaviour modelling code CYRANO3 has to be updated continuously to challenge the licensing requirements resulting from the new fuel managements and to optimise plant operating strategies independently of the fuel suppliers. The validation of the code is then supported by a large database of measurements providing by EDF fuel suppliers commercial rods irradiated during survey programmes in French and international PWR as well as rods irradiated in CEA experimental reactors or within international programmes. This database has to cover the entire materials and loadings range and so each code evolution requires to performing a large number of calculations. Even if the processing time cost is quite short (up to 30 minutes for high burnup simulations), about 3 days are necessary to perform a complete validation of a CYRANO3 version. From a R and D point of view, if calculations time was not act as a break upon calculation number, a systematic run of the entire database could be operated in relation with a model updating, and multi-parameters calibration of a model could be more largely envisaged within a parametric study framework. The use of a reference version of CYRANO3 for industrial studies can also be restricted by calculation time cost since CYRANO3 has to allow performing several thousand fuel rods calculations simultaneously, corresponding to 1/8 of the core, and this number is directly multiply by various input parameters or loadings. That is the reason why EDF is more and more interesting in new solutions which allow to expanding the limits imposed by calculation time. IBM offers an innovative high performance computing solution through its IBM Blue Gene/P system. The IBM Blue Gene/P is a massively parallel platform built around high density racks of PowerPC 450 chips, with low-consumption 850 MHz frequency. Each processor is squeezed onto a single socket Compute Card along with 2 GB RAM and network connectors. Compute Cards are grouped into packages of 32 units, called Node Cards. Each rack stacks 32 Node Cards, i.e. 1024 physical processors, i.e. 4096 physical cores. Compute Cards handle user application processing on top of a very lightweight Linux-compatible Kernel, the Compute Node Kernel (CNK). EDF R and D current Blue Gene/P configuration, named 'Frontier 2', is made of 8 computation racks served by a 500 TB shared, high-performance file system. It is ranked 24. most powerful supercomputer in the world, according to the latest TOP500 list issue from November 2008. It is hosted in the IBM facilities in Montpellier, France. As a massively parallel architecture, the Blue Gene/P platform is specifically dedicated to highly scalable parallel applications that are able to take advantage of tens of thousands of processing cores. However, it also offers a specific execution mode named High Throughput Computing (HTC) designed to address the growing need for massive serial, concurrent executions. In HTC mode, the serial jobs are submitted concurrently on every physical core inside the partition, running independently 'in parallel' but with no inter-communication between the processes, the submission being handled by an HTC-specific scheduler. The HTC feature, taking advantage of the high-density design of the Blue Gene/P system, would allow performing up to 32768 CYRANO3 concurrent executions on the 'Frontier 2' platform, thus minimizing the overall time to perform whole simulation campaigns. It is worth mentioning that the effort to port the CYRANO3 code - originally developed for Linux cluster environments - onto the Blue Gene/P platform has been quite acceptable despite the complexity of the code mixing Fortran 77, C and C++. The compatibility to Linux standards and the existence of a Blue Gene/P variant of the well-known IBM XL compilers family tend to make it fairly straightforward to port onto the Blue Gene/P platform. One of the main obstacles indeed comes from the physical memory limitation to 512 MB RAM per core that needs to be taken care of, and might require some adjustments to be made to the code to ensure its run-time consumption fits into this range. (authors)

[en] Within the PLEIADES simulation platform, built to model any reactor concept fuels behavior, ALCYONE is the multi-dimensional application dedicated to study fuel rods for Pressurized Water Reactors. This paper will focus on a standard UO₂ fuel rod, and present simulations of a ramp test through four approaches (1D multi-slices, 2D axisymmetric, 2D strain plane and 3D). The aim of this study is to underline the interest of each approach, in relation with their corresponding hypotheses, showing their advantages and limits to simulate the whole rod behavior as well as the phenomena taking place at the pellet-cladding scale. The paper will present a comparison between the results obtained at the different scales, and thus its capability to offer to a user all the elements to realize a full study. (authors)

[en] This paper is concerned with the modeling of fuel behaviour and of Pellet Cladding Mechanical Interaction (PCMI) in Light Water Reactors (LWR). Failures due to Pellet-Cladding Interaction (PCI), as discovered in the early 1970's, can be avoided in Light Water Reactors (LWRs) thanks to optimized plant operational procedures and fuel management schemes. However, research and development programs on this topic are still undertaken worldwide in order to improve the understanding of the mechanisms possibly leading to PCMI failure. To model PCMI in LWRs, two distinct fuel codes called METEOR and TOUTATIS have been developed at the Atomic Energy Commission (CEA) during the last two decades. METEOR is based on a one-dimensional axisymmetric description of the radial dimension of the fuel element, associated to a discrete axial decomposition of the fuel rod in stacked independent fuel 'slices'. The one-dimensional METEOR code is used to assess the global geometrical changes of the fuel rod during irradiation and to estimate the quantity of fission gases generated by irradiation, up to high burn-up levels. However, it is not sufficient to estimate precisely the local stresses in the cladding resulting from PCMI. Only a more detailed description of the thermo-mechanical behaviour of the fuel pellet and of the cladding, based on 2D or 3D Finite Element analyses can provide detailed information on this point. This has prompted the development of the fuel application TOUTATIS which is based on the finite element code CAST3M. Recently, in order to benefit simultaneously of the chemo-physical models available in the METEOR application and of the detailed thermomechanical description of the fuel pellet proposed in the TOUTATIS code, the multi-dimensional PWR fuel application ALCYONE has been developed in the framework of the PLEIADES environment. Of particular interest in this study, the fission gas models of the METEOR 1D code have been recently introduced in ALCYONE and a specific iterative loop developed to couple fission gas swelling (which depends on the hydrostatic pressure) to the thermo-mechanical solution of the 2D and 3D problem (which depends on gas-induced swelling). The fission gas models account for intragranular bubbles growth, coalescence and migration, for inter-granular bubbles growth and migration, for gas transport in the porosity and for gas release in the plenum. The usefulness of 1D fuel codes relies on their ability to model correctly the experimental data available. Generally, they consist in profilometries after base irradiation and ramp tests, corrosion thickness after base irradiation, rod elongation, fission gas release and internal pressure. To assess the code ability to predict with reasonable accuracy the behaviour of fuel rods during base irradiation and power ramp testing, extensive databases are required. ALCYONE 1D predictions are validated on experimental data concerning 80 base irradiations and 30 ramp tests performed on UO₂-Zy₄, UO₂-M5^R and MOX-Zy₄ fuel rods with mean burn-ups up to 60 GWd/tU. While most of the models used in the simulations are usually based on laboratory tests independent of the experimental measures performed after irradiation in commercial or experimental reactors, a best-estimate of the fission gas swelling model parameters (mainly transport parameters between the different type of voids initially existing in the fuel pellet or formed during irradiation) must be deduced from the 1D simulations of the database. In this paper, the extension of the validation process to 2D and 3D analyses of a fuel pellet - cladding fragment with ALCYONE V1.1 is presented in details. First, the main aspects of the 1D, 2D and 3D schemes of ALCYONE, i.e., the geometry of the meshed fuel pellet fragment or fuel rod, the boundary conditions, the loading applied to the pellet and the cladding, the thermo-mechanical coupling, the material laws used to describe fuel cracking and cladding creep-plasticity, the fission gas swelling model, the friction at the pellet-cladding interface are briefly recalled. Second, the application of the 2D and 3D schemes to a database consisting of 30 base irradiations and ramp tests performed on UO₂-Zy₄ and UO₂-M5^R with burn-ups up to 45 GWd/tU is described. The ability of the 3D scheme to predict the behaviour of a single fuel pellet - cladding element situated at the maximum Linear Heat Rate (LHR) during ramp testing is demonstrated by comparing the following experimental and calculated data: residual clad diameter after base irradiation and ramp test, height of inter-pellet and mid-pellet ridges after base irradiation and ramp test, dish filling after ramp testing. Results obtained on the same database with the 2D scheme which models the mid-pellet plane of the 3D fragment (also at the maximum LHR) are then presented. It is shown that the calculated diameter increases during ramp test are very close to those obtained with the 1D scheme but usually greater than the 3D estimates at inter-pellet level in spite of the fact that the 1D, 2D and 3D schemes share the same fission gas swelling model. The reasons for these differences are then discussed in the paper. It is shown that the hydrostatic pressure state in the pellet during ramp testing is notably different between the inter-pellet and mid-pellet planes because of dish filling and that a model tuned on the mid-pellet behaviour might not be adequate to model the phenomena occurring at inter-pellet level. (authors)