The link between the geostatistical modeling of multiscale carbonate heterogeneities and representative simulation flow models remains a challenge because conventional upscaling procedures often disregard the complex dynamic behavior that links the geostatistical static properties, dynamic rock/fluid data and reservoir operational conditions. This work proposes a methodology to build robust simulation models for naturally fractured carbonate reservoirs with multiscale geological characterizations. The development of this work follows three main steps: (1) hierarchical upscaling procedure by flow units, (2) integration of flow units into a reservoir-scale simulation model, (3) validation of simulation model. The hierarchical upscaling procedure was applied to three inter-well regions and extrapolated to three flow units. The proposed workflow is applied to a reservoir based on a combination of real and synthetic data from a Brazilian offshore carbonate reservoir. The methodology shows several advantages: it represents static and dynamic behavior from multiscale heterogeneities in reservoir simulation; it minimizes risk in the selection of well position and completion through well characterization of representative static and dynamic data for each flow unit; and, improved selection process for the simulation flow model. The methodology shows that the relative permeability defined in reservoir simulation can be different from the laboratory measured curve because it needs to match the dynamic behavior from the reference solution. This result shows that a proper characterization and upscaling approach are crucial to sufficiently represent geological heterogeneous scenarios in reservoir simulation. The methodology we present here is useful for multidisciplinary areas of expertise as it ensures the appropriate link between the fine scale geomodelling and the coarser scale reservoir simulation, considering the development of complex carbonate reservoirs.391219233Center of Petroleum Studies (Cepetro-Unicamp/Brazil)PETROBRAS S/A, UNISIMPetroleum Engineering Department (DEP-FEM-Unicamp/Brazil)Schlumberger Information Solution for the use of Petrel(R), and CMG for the use of IMEX