LOCAL RECTANGULAR REFINEMENT WITH APPLICATION TO AXISYMMETRICAL LAMINAR FLAMES

Within realistic combustion devices, physical quantities may change by an order of magnitude over an extremely thin flamefront, while remain ing nearly unchanged throughout large areas nearby. Such behaviour dic tates the use of adaptive numerical methods. The recently developed lo cal rectangular refinement (LRR) solution-adaptive gridding method pro duces robust unstructured rectangular grids, utilizes novel multiple-s cale finite-difference discretizations, and incorporates a damped modi fied Newton's method for simultaneously solving systems of governing e lliptic PDEs. Here, the LRR method is applied to two axisymmetric lami nar flames: a premixed Bunsen flame with one-step chemistry and a diff usion flame employing various complex chemical mechanisms. The Bunsen flame's position is highly dependent upon grid spacing, especially on coarse grids; it stabilizes only with adequate refinement. The diffusi on flame results show excellent agreement with experimental data for f lame structure, temperature and major species. For both dames, the LRR results on intermediate grids are comparable to those obtained on equ ivalently refined conventional grids. Solution accuracy on the final L RR grids could not be achieved using conventional grids because the la tter exceeded the available computer memory. In general, the LRR metho d required about half the grid points, half the memory and half the co mputation time of the solution process on conventional grids.