Abstract: Large break loss of coolant accident is the most severe design basis accident. To study the uncertainty of droplet and bubble behavior models, an evaluation was conducted on the independent software COSINE multiphase field subchannel code to calculate the results of large break loss of coolant accidents. In response to the mechanism of large break loss of coolant accidents, its specialized software serves as an analysis tool, which puts higher demands on the accuracy and suitability of key process law analysis models. The above indicators depend on the depth of the model’s understanding and mechanism analysis of important physical processes. In order to improve the accuracy and robustness of COSINE multiphase field code for optimal estimation of large break loss of coolant accidents, the code was used for modeling and numerical calculations based on thermal hydraulic experiments in this paper. And two sensitivity quantification analysis methods were used to provide and analyze the calculation bias of temperature and water inventory. In addition, quantitative analysis was conducted on the influencing parameters and the interaction between the parameters was discussed. The results show that the calculation band for the water inventory results in the downcorner section of the refilling stage is very wide, with a maximum relative error of 33.15%, the interphase interaction between vapor and liquid has a significant impact on the mass flow rate. The calculation range of the maximum temperature of the cladding results during the reflooding stage is very narrow, and the relative error of the peak temperature of the cladding is within 3.5%, the uncertainty error of calculation is continuously amplified on the time scale. The limitation of droplet diameter plays a major role in the influence of water inventory in the downcorner section, with a correlation coefficient of 0.27. As the dimensionless velocity of the upward steam formed in the descending section changes, the diameter of the droplet directly affects its interfacial area, resulting in changes in interphase heat transfer and friction. The friction coefficients of gas and liquid droplets have the most significant impact on the maximum temperature of the cladding during the reflooding stage, with a correlation coefficient of 0.54. The relative velocity of gas and droplets determines the enhancement factor of gas convection in dispersed flow film boiling, and shows a positive correlation. In addition, the diameter limitations of droplets and bubbles have a significant impact on numerical calculations. The friction between liquid and droplets has a significant relationship with the heat transfer between droplets and the wall.