Laser direct metal deposition (DMD) can supply a new method in the fields of surface modification and near-net forming. The powder flow behavior and its convergence characteristics play a crucial role in the deposition quality during the DMD process. In this research, the k-ε turbulence model based on the Computational Fluid Dynamics (CFD) modeling method was innovatively employed to establish the numerical model of the gas-powder flow. Then, the Dense Discrete Phase Model (DDPM) was utilized in this gas-powder coupling model to accurately calculate the collision between particles and between particle and inner wall of the nozzle. Afterward, the Response Surface Method (RSM) was carried out to design the numerical simulation scheme, analyze a series of simulation results, explore the correlation between the process parameters and the responses, and establish the prediction model of powder convergence characteristics. Furthermore, the process parameters were optimized by considering the influence of defocusing amount, with smaller powder spot diameter and higher maximum powder mass concentration as optimization objectives. It was found that the prediction model of responses demonstrated a high degree of accuracy and reliability. The single deposition track exhibited better deposition quality fabricated with the optimized process parameters. The research method and results mentioned in the present study were expected to provide significant theoretical guidance for the selection and application of process parameters during the laser direct metal deposition process.