Large Component Simulation of an Inboard Sector of a Dual-Coolant Lead-Lithium Blanket

Advances in high-performance computing now enable the engineering evaluation of large blanket components from a systems perspective at the beginning of a normal design cycle. As an example, we discuss the computational fluid dynamics (CFD) involved in characterizing the thermal performance of an entire 22.5  toroidal sector of a dual-coolant lead-lithium blanket with particular attention to the integrated manifolding and flow distributions. The inboard sector is roughly 7 m tall, has 321 first-wall helium-cooled channels, and five parallel PbLi breeder channels complete with insulating SiC flow channel inserts. A finite volume model was developed with various degrees of mesh refinement from 36 to 187 million cells to perform the flow calculations on disparate fluids with conjugate heat transfer in the solid. Steady-state CFD calculations were performed using a realizable k-ε turbulence model. Simplifications include a uniform applied surface heat flux and a one-dimensional radial volumetric neutron heating profile. Constant material properties were used for the F82H reduced-activation ferritic martensitic (RAFM) steel walls, SiC flow channel inserts, and the PbLi breeder. Ideal gas behavior was assumed for the helium, which includes compressibility. Helium mass flows of 54 kg/s at 8 MPa and PbLi flows of 3 kg/s at 101 kPa supply the sector, both at an inlet temperature of 350°C.

This early model is not optimized; however, it reveals important features not obvious in a conglomeration of smaller independent models. For example, all the manifolding was included to evaluate flow distributions throughout the full component. Submodels were only used to obtain the convective heat transfer coefficients (HTCs) inside the helium first-wall channels equipped with enhancement vanes that allow the first wall to handle 0.8 MW/m² of plasma heat flux with the aim of keeping bulk RAFM steel temperatures near the creep-fatigue limit of 550°C. Average HTCs obtained from these detailed submodels were then used in the large model to predict thermal performance without incurring the meshing overhead from the thousands of internal vanes. Although much progress was made, initial results indicate that more must be done to further reduce hot spots on the first wall, minimize pressure drops, and provide optimal flow distributions.