Exponential Characteristic Research Articles

Split-Cell Exponential Characteristic Transport Method for Unstructured Tetrahedral Meshes

The nonlinear, exponential characteristic (EC) method is extended to unstructured meshes of tetrahedral cells in three-dimensional Cartesian coordinates. The split-cell approach developed for the linear characteristic (LC) method on such meshes is used. Exponential distributions of the source within a cell and of the inflow flux on upstream faces of the cell are assumed. The coefficients of these distributions are determined by nonlinear root solving so as to match the zeroth and first moments of the source or entering flux. Good conditioning is achieved by casting the formulas for the moments of the source, inflow flux, and solution flux as sums of positive functions and by using accurate and robust algorithms for evaluation of those functions. Various test problems are used to compare the performance of the EC and LC methods. The EC method is somewhat less accurate than the LC method in regions of net out leakage but is strictly positive and retains good accuracy with optically thick cells, as in shielding problems, unlike the LC method. The computational cost per cell is greater for the EC method, but the use of substantially coarser meshes can make the EC method less expensive in total cost. The EC method, unlike the LC method, may fail if negative cross sections or angular quadrature weights are used. It is concluded that the EC and LC methods should be practical, reliable, and complimentary schemes for these meshes.

Exponential Characteristic Nonlinear Radiation Transport Method for Unstructured Grids of Triangular Cells

The exponential characteristic (EC) method is one of a family of nonlinear spatial quadratures for discrete ordinates radiation transport that are positive and at least second-order accurate and provide accurate results for deep-penetration problems using coarse meshes. We use a split-cell methodology to adapt the method to unstructured grids of arbitrarily shaped and oriented triangular cells that provide efficient representation of curved surfaces. Exponential representations of the flux entering through a cell edge and of the scattering source within a cell are constructed to match average values and first moments passed from the adjacent cell (or from the boundary conditions) or obtained from the angular quadrature of the directional flux spatial moments in the previous iteration (or from an initial guess). The resulting one- and two-dimensional nonlinear rootsolving problems are efficiently solved using Newton’s method with an accurate starting approximation. Improved algorithms, presented here, have increased the efficiency of the method by a factor of 10 as compared to an initial report. The EC method now costs only twice as much per cell as does the linear characteristic method but can be accurate with many fewer cells. Numerical testing shows the EC method to be robust and effective.

Exponential Characteristic Spatial Quadrature for Discrete Ordinates Radiation Transport with Rectangular Cells

The exponential characteristic (EC) spatial quadrature for discrete ordinates neutral particle transport previously introduced in slab geometry is extended here to x-y geometry with rectangular cells. The method is derived and compared with current methods. It is similar to the linear characteristic (LC) quadrature (a linear-linear moments method) but differs by assuming an exponential distribution of the scattering source within each cell, S(x) = a exp(bx + cy), whose parameters are rootsolved to match the known (from the previous iteration) spatial average and first moments of the source over the cell. Similarly, EC assumes exponential distributions of flux along cell edges through which particles enter the cell, with parameters chosen to match the average and first moments of flux, as passed from the adjacent, upstream cells (or as determined by boundary conditions). Like the linear adaptive (LA) method, EC is positive and nonlinear. It is more accurate than LA and does not require subdivision of cells. The nonlinearity has not interfered with convergence. The exponential moment functions, which were introduced with the slab geometry method, are extended to arbitrary dimensions (numbers of arguments) and used to avoid numerical ill conditioning. As in slab geometry, the method approaches O({Delta}x{sup 4}) globalmore » truncation error on fine-enough meshes, while the error is insensitive to mesh size for coarse meshes. Performance of the method is compared with that of the step characteristic, LC, linear nodal, step adaptive, and LA schemes. The EC method is a strong performer with scattering ratios ranging from 0 to 0.9 (the range tested), particularly so for lower scattering ratios. As in slab geometry, EC is computationally more costly per cell than current methods but can be accurate with very thick cells, leading to increased computational efficiency on appropriate problems.« less