Reservoir simulation models describe the flow of various fluid models, such as black oil, compositional, reactive transport, and chemical flooding. Implementing new fluid models in reservoir simulators requires significant manpower to rewrite the simulation code, especially if the original simulator was not designed properly. One such example is to include geochemical reactive transport processes. In this paper, a general modeling framework is developed for easy implementation of new fluid models including time-independent (equilibrium) and time-dependent (kinetic) correlations. The approach is physically sound, robust and naturally extends black oil and compositional models to multiphase reactive systems for a variety of complex flow processes. The fundamental physical laws (including conservation laws of mass/momentum/energy and thermodynamics laws) govern multiphase flow in porous media. Here we consistently applied the general principles under different assumptions to obtain the mass conservation equations for various processes. The framework includes a definition of components under the instantaneous phase and chemical equilibrium, mass conservation equations, a volume balance equation, and equilibrium relations. That is, primary equations for various fluids are represented by a unique set of mass conservation equations with different secondary algebraic constraints. The differential equations do not need to be changed as new processes are added. IMPEC and FIM are used to solve for flow. A robust algorithm is developed to calculate the phase equilibrium together with chemical equilibrium. The simulator was validated with SPE comparative solution projects using Eclipse 300 and CMG-GEM. Large-scale reservoir models including SPE 10 and a field case from Europe were compared with various commercial simulators, showing that the simulator is significantly faster and more robust. New coupled reactive phase behavior models were implemented, including low salinity waterflooding, CO2 WAG with reactions, alkali/surfactant/polymer (ASP) flooding with the HLD-NAC EoS model, and fines migration. Those models have been implemented with minimal additional coding owing to the novel and consistent framework. The modeling framework provides a handy and powerful tool to understand the detailed interaction between species, easily update the physics as needed, and to better predict its impact on the ultimate oil recovery. This framework differs from previous research because the equilibrium chemical reactions are included without changing the structure of the compositional model. Similar to conventional compositional models, the flow, transport and local interactions are treated in two parts. This paper generalizes the simulation technology (named PennSim) that we have used in various applications. The novel formulation also enables fast linear solver performance due to the natural decoupling techniques.