Laser Powder Bed Fusion (LPBF) offers great design flexibility and fast production but is prone to defects like porosity, residual stresses, and anisotropy, which are major concerns affecting part quality. In LPBF, melt pool dynamics are sensitive to process parameters. Exceeding a threshold energy density can form a significant vapor depression, leading to keyhole-induced porosity and potentially compromising the build quality. This study uses numerical methods to investigate keyholing during the LPBF of Inconel 718. By integrating powder distribution from the discrete element approach with computational fluid dynamics model, it analyses the formation and evolution of keyhole-induced porosity. The multiphysics model tracks surface dynamics using a volume of fluid approach while incorporating key phenomena like evaporation, Marangoni convection, recoil pressure, multiphase flow, laser reflection, and absorptivity to understand the mechanism. The simulated results indicate that a narrow and deep melt pool is formed during keyholing, implying a shift in the primary direction of heat transfer to a vertical orientation. Because of the high energy density, a vapor column forms, and the trapped bubbles in the keyhole under the transient melt flow cause pores under the laser scan track. The simulated pore morphology is in good agreement with the experimental results found in the existing literature. Additionally, the characteristics of keyholes arising from varying scan speeds and laser power are examined. Parametric results indicated that with the same energy density, a higher power and scan speed combination resulted in more turbulent melt pools, which reduced the occurrence of pores and their size. These findings significantly contribute to understanding laser additive manufacturing, facilitating robust computational models for reliable process development.