Rapid 3D Grain Burnback Simulation Based on a Minimum Distance Field Method and Hybrid Surface Reconstruction

A GPU-based minimum distance field method is presented for rapid three-dimensional geometric burnback analysis of homogeneous propellant grains under spatially uniform and isotropic regression. An adapted three-dimensional jump flooding algorithm accelerates distance-field construction. A boundary-constrained hybrid Marching Cubes–Marching Tetrahedra procedure uses hexahedral reconstruction in single-field-active voxels and tetrahedral extraction and clipping in dual-field-active voxels. Each prescribed burned-web state is evaluated independently from the initial fields. An analytical star grain and an independent CAD benchmark assess geometric outputs for the tested shapes. A complex-grain example demonstrates reconstruction and quantifies the JFA approximation using brute-force distances. For the finocyl benchmark at a grid resolution of (128,128,128), the sampled burning-area–web curve was computed in 281 ms, with mean and maximum relative errors of 0.8360% and 1.9838%, respectively, against published CAD-derived area data. Same-platform comparisons with a GPU-based distance-field baseline show reduced computation times at comparable burning-area errors. Remaining-grain geometric and mass properties are evaluated, and the burning-area curves are used in two zero-dimensional quasi-steady internal-ballistic applications. The proposed framework provides an efficient computational tool for three-dimensional grain burnback evaluation, supporting repeated geometric design assessment and zero-dimensional internal-ballistic analysis.