Abstract: Cold crucible melter (CCM) technology holds broad application prospects in radioactive waste treatment due to its advantages of long equipment service life, wide adaptability to waste types, and high safety. However, the traditional startup process relies on special startup glass beads and requires multiple replacement procedures. Moreover, solidified glass in the crucible after abnormal shutdown leads to difficult restart and easy furnace scrapping, which restricts the efficiency of industrial continuous operation. To address these bottlenecks, a novel startup and restart process was developed in this study. The process used basic glass beads (particle size: 1-2 mm) as the startup substrate and calcined products (oxides) of simulated high-level radioactive liquid waste (HLLW) as the fluxing agent. The effects of calcined product mass fraction (2%-10%) and solidified glass mass in the crucible (400-460 kg) on startup/restart time were systematically investigated. The results show that compared with the traditional startup glass bead process (average time of 152 min), the average startup time of the novel process is shortened to 84 min, with an efficiency improvement of 44.7% (a reduction of 68 min). Startup time is significantly negatively correlated with the calcined product content: For each 1% increase in mass fraction, the startup time is shortened by an average of 4.8 min. The novel process enables stable restart with solidified glass in the crucible. When the solidified glass mass increases from 400 kg to 460 kg, the restart time extends from 116 min to 223 min (an increase of 89.7%), indicating that restart difficulty rises significantly with the increase in solidified glass mass. This process eliminates the subsequent glass bead replacement step, simplifies the workflow, and reduces the generation of non-contained vitrified bodies. Meanwhile, it overcomes the restart challenge after abnormal shutdown, providing key technological support for the large-scale, continuous, and safe operation of CCV technology.