Recursively Coupled FEM-Analysis Of Pressure Vessel Creep Failure Experiments

The hypothetical scenario of a severe accident with core meltdown and formation of a melt pool in the lower plenum of a Light Water Reactor Pressure Vessel (RPV) can result in the failure of the RPV and the discharging of the melt to the containment. One accident management strategy could be to stabilize the in-vessel debris or melt pool configuration in the RPV as one major barrier against uncontrolled release of heat and radionuclides into the containment of the plant.
To obtain an improved understanding and knowledge of the melt pool convection, the vessel creep, possible failure processes and modes occurring during the late phase of a core melt down accident the FOREVER-experiments have been performed at the Division of Nuclear Power Safety of the Royal Institute of Technology, Stockholm. These experiments were simulating the behavior of the lower head of the RPV under the thermal loads of a convecting melt pool with decay heating.
An axisymmetric Finite Element (FE) model was developed to simulate these experiments. First the temperature field within the melt pool and within the vessel wall was calculated with a CFD-Model. But due to the lack of a turbulence model for very high internal Rayleigh numbers of up to 1017 in the prototypic case and due to the very time consuming CFD-solution, the Effective Conductivity Convectivity Model (ECCM) as proposed by Bui has been implemented. Once the temperature field in the vessel wall is evaluated, the transient structural mechanical calculations are then performed applying a creep model which takes into account the large temperature, stress and strain variations. The creep model includes the primary, secondary and tertiary creep stages.
In the prior work only a one-way coupling between the thermal and the mechanical model was applied: first the transient temperature field was calculated and then the transient mechanical calculation was performed applying the appropriate temperature field at each time step. To take into account the melt level drop by thermal expansion the initial level was lowered “by hand”. The feedback from the viscoplastic deformation to the temperature field was not correctly considered in the one-way coupling. On the other hand the earlier investigations showed that slight temperature shifts of only 10 K at an overall temperature level clearly above 1000 K had significant effects concerning the failure time of the vessel.
To overcome the disadvantages of the single coupling model a new recursively coupled model is developed. This paper deals with the new results and so far the results show that a recursively coupled thermal and mechanical solution is closer to reality than a single coupled model where the mechanical deformations have no impact to the thermal model. Due to the scaling effect it is expected, that the fully coupled model is necessary to be able to perform reliable best estimate calculations for prototypic light water reactor scenarios.