Abstract: Lead-cooled fast reactor (LFR) is one of six advanced nuclear energy systems recommended by Gen-4 International Forum. With the significant development in materials and equipment, LFR is more and more closed to engineering applications. Right now, plenty of research work has been carried out over the world and many types of small sized LFRs have completed the core design. Core physics calculation is the most fundamental question in core design, which will provide the parameters such as effective multiplication factor, power density distributions, operation life and reactivity control worth. The essence of core physics calculation is solving the neutron transport equation, with the deterministic method and Monte Carlo method. Currently, the Monte Carlo method is mainly chosen because of the high accuracy as the core physics calculation in small sized LFR by modeling the whole core. While this method with low efficiency will not satisfy the huge demand of calculation in core design with the progress of engineering applications. For the strong heterogeneities in energy and space in the core of small sized LFR, the approximations of spatial homogenization and energy condensation must be introduced. The open-source Monte Carlo calculation software OpenMC was employed through Python script language to fulfill the Monte Carlo homogenization function for fuel lattice and assembly. Then based on C++ language, a neutron diffusion equation solver DESOF was developed by calling the solvers from open-source finite volume method CFD software OpenFOAM, and functionally validated by diffusion benchmarks. At last, by Python script language Monte Carlo homogenization and diffusion solver were coupled to a complete core physics calculation software MCDESOF and were validated by the experimental results from Venus LBE zero power fast reactor. The equivalent of Monte Carlo homogenization was fully considered. Transport correction was made to treat the anisotropy of scattering. Neutron spectrum, neutron flux density distribution, reaction rate, sensitivities of cross section were analyzed, and the effects of energy group structure and space mesh partition to calculation results were discussed. Super homogenization factor was introduced to the correction of cross section. In reflective boundary conditions, the biases of effective multiplication factors between MCDESOF and Monte Carlo code for typical fuel lattice and assembly are less than 50 pcm. The Venus LBE zero power fast reactor was modeling by MCDESOF and the effective multiplication factor and control rod worth were calculated. Comparing with the experiment results, the bias of reactivity worth of safety rod was less than 200 pcm. The accuracy of MCDESOF is closed to full core Monte Carlo modeling but with less than 25% calculation time consumptions.