The design of PWR emergency core cooling systems is based on the Design Requirement that the fuel retains its structural integrity after a guillotine break of the main coolant pipe work adjacent to the vessel. This design-basis fault is termed the Large-break Loss-of Coolant Accident (LBLOCA). The Design Requirement leads to Design Criteria that constrain the limiting fuel-pin clad temperatures in the fault and require that the fuel should maintain a coolable geometry despite the possibility of the internal gas pressure causing ballooning of the clad.
Extensive experimental work has been carried out to simulate the performance of fuel pins under temperature transients typical of PWR LOCA. These experiments included both studies of single fuel pins and bundles representative of parts of fuel assemblies. Supplementary studies were performed on the flow and heat transfer in bundles of deformed rods. The conclusions of these studies were that individual pins can credibly reach very high diametric strains, but that these appear to be sufficiently misaligned axially to prevent total blockage of the flow. The heat transfer during core reflood is also able to accommodate flow blockages of up to 90% without substantial deterioration in cladding surface temperatures. Most safety cases involve some engineering judgement about the likelihood of a tolerable geometry based upon these experiments.
Most of the relevant experiments were performed in the 1980s and therefore were principally carried out on the various forms of Zircalloy 4 cladding available at the time. Much of the fuel used was either fresh or of modest burnup compared to the discharge irradiations achievable today. Since then, single pin experiments have been carried out with new cladding material and (to a limited extent) with high-burnup fuel. There is a need to interpret the performance of this fuel in the context of the wider body of evidence.
As the fuel heats up in dry steam, hydride precipitates dissolve and irradiation damage is annealed from the cladding material. Ductility values of over 100% are achieved. The balloon diameter is only limited by the fact that the dislocation creep process is very sensitive to temperature and also to stress, so that strain is rapidly concentrated at the hottest and thinnest part if the cladding. The texture of the cladding is also such that hoop strain is accommodated by axial shrinkage, leading to a tendency for the hot side of the cladding to remain in contact with the heat source while the cold side lifts off. The diametric strain at failure is therefore determined by a complex interaction between the cladding material and its thermal environment.
The thermal environment in which the cladding deforms is determined partly by the presence of neighbouring fuel (or RCCA guide tubes) but mostly by the performance of the fuel pellets, which is only partly understood and is modelled empirically.
Modern cladding materials (such as the 1% Nb alloys) have slightly different stress exponents for creep, their phase-change transition temperatures are different and their yield stresses vary compared to Zircaloy material. Single-pin tests indicate that these differences are not significant, but these tests need to be interpreted in the context of the LBLOCA design transient and the rod bundle geometry in reactor.
The fuel pellets are now irradiated sufficiently for the rim of the pellet to experience significant microstructural changes due to the presence of fission gas bubbles within the material. These changes potentially affect the integrity of the pellet in the LBLOCA transient and this could in turn influence the thermal environment in which the cladding deforms.
British Energy has followed the general trend in the use of fuel by adopting new cladding materials to reduce the level of in-reactor corrosion and to permit further increases in cladding irradiation.
The approach to licensing these changes has been to separate the effect of the change in material from the change in the irradiation level.
The change in material from Zircaloy to a 1% Niobium alloy was examined by consideration of the results of ballooning tests on cladding tubes, together with a detailed modelling of the fuel performance in the LBLOCA fault. The model was developed from an existing fuel performance code. The details of the model and the results of the modelling studies are outlined below.
The pellet rim performance was not addressed directly. Instead, credit was taken for the reduction in reactivity of uranium oxide fuel with irradiation: Analysis is used to demonstrate that the bulk of this fuel assembly will not balloon to failure and therefore does not need to be considered in the context of flow blockages. This approach places constraints on permitted power levels of “high burnup” fuel, but can currently be accommodated in reload core designs.
Development of the modelling consisted of two sets of changes to the existing model: Firstly, the empirically data and model formulations used to represent the performance of Zircaloy cladding were replaced by those appropriate to the particular 1% Niobium alloy envisaged; secondly, the physical model of the fuel pin was interfaced with both a thermalhydraulic code and, other instances of the fuel pin model, to permit the thermal environment surrounding the the fuel pin to be better represented.
The fuel pin model includes a three-dimensional representation of the cladding and fuel pellet, but recognise that knowledge of the relative movement of the cladding and the fuel pellet stack is incomplete. The model therefore requires some empirical data to represent this relative movement (which is both central to determining the final balloon diameter and we believe highly stochastic). The approach adopted was to represent the offset of the pellet relative to the cladding centroid as a single constant value of pellet eccentricity. For each fuel rod in the bundle, this value was tuned to reproduce an individual diametric strain at failure in a bundle ballooning experiment. The data was then validated by applying them to a new bundle under different coolant flow conditions. Finally, the model derived was applied to conditions representative of a high-powered fuel assembly in reactor. The fuel response in the design-basis fault was calculated for both the existing fuel and the proposed replacement.
The assessment concluded that the proposed replacement alloy was more creep hard at high temperature and therefore was expected to fail later in the transient. This potentially introduces benefits in the number of rods expected to fail, but also affects the thermal conditions under which ballooning occurs: Failure later in the transient could be expected to occur under improved heat transfer conditions, with implications for the diameter of balloon achieved and also for the tendency of the balloon to extend axially. However in practice, the main difference between the two claddings was a reduced stress at which local plastic instability occurs in the 1% Niobium material, leading to generally lower diametric strain at failure under the particular conditions of the fault considered.
The modelling also demonstrated that it is possible to represent the effect of the differences in confinement likely to be experienced by fuel pins in test bundles and by those in reactor.