Natural disasters, such as tsunamis or severe nuclear power plant incidents, can submerge safety–critical internal systems, posing significant risks. Flooding pathways play a crucial role in plant safety during these events. Conventional methods often overlook the complex effects of porous characteristics in internal structures on fluid dynamics, potentially causing deviations from actual flooding behavior. This study integrates a porous media model with the Least Squared Moving Particle Semi-Implicit (LSMPS) method to analyze the impact of local porous geometries on fluid flow variations. The model was validated through benchmark scenarios, including filtration, small glass bead, and crushed rock cases. Simulations of the filtration with varying porosities show that lower porosity increases flow resistance, reduces velocity, and alters overall flow behavior. Simulations for small glass bead and crushed rock cases accurately predicted pressure distribution and fluid surface height, matching experimental observations. The crushed rock flow simulation aligns with reactor building conditions during severe accidents. Comparing flow simulations for small glass bead and crushed rock shows that porous media with varying porosity significantly affect fluid movement, an important factor in severe accident analysis. In conclusion, the improved model enhances simulation accuracy and can predict fluid flow behavior under various porous media conditions. This provides a reliable tool for further accident analysis and reactor safety research.