It has been reported that crevice corrosion had been created on the bearing pads of fuel bundles and also on the inner surface of pressure tubes in CANDU nuclear power plants. The reasons of the crevice corrosion might be that small local boiling may be occurred in the small cavities between the surfaces of bearing pad and pressure tube, therefore, Lithium, which had been added to coolant for proper chemical characteristics of the coolant, might be concentrated in the cavities.
Following have been considered as the methods to solve this crevice corrosion problem.
1. Changes of coolant chemical characteristics.
2. Improvement of coolant flow in the cavities.
3. Reduction of heat flux from bearing pad surface to pressure tube inner surface by changing the shape of bearing pad.
But the first two methods (1 and 2) are thought to affect various effects to primary coolant system and may be difficult to actually perform. And the third method has been considered a desirable and practical method to reduce or eliminate the crevice corrosion, especially because it is known that below a certain threshold temperature, the crevice corrosion may not be occurred.
Some intuitional methods to optimize the bearing pad shape have been suggested but any analytical and systematic research has not been reported. In this paper, optimal design of bearing pad shape using finite element method has been done by simultaneously considering the reduction of the heat flux from bearing pad surface to pressure tube internal surface and the limitation of stress on the bearing pad within the allowable stress.
On the optimal designed bearing pad shape, the average temperature of the surface is lower than the average temperature of currently using shape's surface by about 10℃, any stress concentration is not occurred and all stresses on the bearing pad are fallen within allowable stress limit.
In this optimal design work, matrix variational method and adjoint variable technique were used to reduce computing time and computer memory capability, to simplify the variables mathematical relationships and to achieve high accuracy on design sensitivity analysis. The results are very satisfactory.
More accurate results are expected if isoparametric element and auto-mesh generation are adopted for the performance of finite element analysis, while triangle elements and uniform mesh generation method were adopted in this work.