Abstract: In-vessel retention (IVR) was widely recognized as the key mitigation strategy for severe reactor accidents in the new-generation PWR. The research on the effectiveness of IVR is a hot spot presently. Experiments like COPO, BALI, SEMICO and COPRA were carried out to solve problems above, and many numerical simulations were also conducted to explore heat transfer characteristics of IVR. However, in previous experiments and simulations, lack of real-time volumetric heat source renders results uncertainty. Previous experiments mostly use a constant heat source to simulate the molten pool heat generation. However, the heat released consists of fission power and decay heat, which is closely related to neutron physics and nuclide evolution in the molten pool. Artificial constant heat source cannot simulate the physical process in the molten pool, and therefore cannot accurately reflect the fine real-time pattern of heat released, rendering simulation and heat flux measurement uncertainties. To accurately calculate the power generated by physical analysis, Tsinghua University proposed a new molten pool decay heat calculation method based on the Reactor Monte Carlo program, which was verified to be accurate. At distinct core operation times, the nuclide composition, heat release pattern and physical evolution of molten pool are different. The previous work managed to explore the molten pool decay power distribution, but failed to consider how different core operation times and other nuclear reactions affect power distributions, which is necessary for safety analysis and design. Therefore, it is important to establish the fine time-series model of heat source in the molten pool and analyze the corresponding physical mechanism. RMC is a Monte Carlo neutron transport program with functions of critical calculation and burnup calculation. DEPTH module is a built-in burnup calculation module of RMC with the functions of large-scale burnup calculation and decay heat calculation. The code can be used to simulate the fission process driven by delayed neutron and change of nuclides composition in the molten pool. BEAVRS benchmark was taken as the simulated reactor core to offer nuclide information at different core operation times. The initial boundary conditions of molten pool formed at different time spots were confirmed accordingly. Based on RMC and BEAVRS benchmark, the fine modeling of molten pool was carried out. The Monte Carlo method was used to simulate the neutron transport and related nuclear reactions in the molten pool. The linear sub-chain method was used to solve the burnup equation and simulate the decay chain of fission products. The purpose is to establish an accurate time-series model of heat source in molten pool and evaluate the pattern from the perspective of physical analysis, to provide reference for the safety and design of IVR. The results show that the model can reflect the changes of heat generated in the molten pool, and the obtained time-series data is of great significance for further study of IVR.