Scram and nonlinear reactor system seismic analysis for a liquid metal fast reactor

Abstract

Earthquake vibrations cause large forces and stresses that can significantly increase the scram time required for safe shutdown of a nuclear reactor. The horizontal deflections of the reactor system components cause impact between the control rods and their guide tubes and ducts. The resulting frictional forces, in addition to other operational forces, delay the travel time of the control rods. To obtain seismic responses of the various reactor system components (for which a linear response spectrum analysis is considered inadequate) and to predict the control rod drop time, a non-linear seismic time history analysis is required. Nonlinearities occur due to the clearances or gaps between various components. When the relative motion of adjacent components is large enough to close the gaps, impact takes place with large impact accelerations and forces. This paper presents the analysis and results for a liquid metal fast rector system which was analyzed for both scram times and seismic responses such as bending moments, accelerations and forces. The reactor system was represented with a one-dimensional nonlinear mathematical model with two degrees of freedom per node (translational and rotational). The model was developed to incorporate as many reactor components as possible without exceeding computer limitations. It consists of 12 reactor components with a total of 71 nodes, 69 beam and pin-jointed elements and 27 gap elements. The gap elements were defined by their clearances, impact spring constants and impact damping constants based on a 50% co-efficient of restitution. The horizontal excitation input to the model was the response of the containment building at the location of the reactor vessel supports. It consists of a 10 sec safe shutdown eathquake (SSE) acceleration-time history at 0.005 sec intervals and with a maximum acceleration of 0.408 g. The analysis was performed with two Westinghouse special purpose computer programs. The first program calculated the reactor system seismic responses and stored the impact forces on tape. The impact forces on the control rod driveline were converted into vertical frictional forces by multiplying them by a coefficient of friction, and then these were used by the second program for the scram time determination. The results give time history plots of various seismic responses, and plots of scram times as a function of control rod travel distance for the most critical scram initiation times. The total scram time considering the effects of the earthquake was still acceptable but about four times longer than that calculated without the eathquake. The bending moment and shear force responses were used as input for the structural analysis (stresses, deflections, fatigue) of the various components, in combination with the other applicable loading conditions.