Alkaline flooding is an important branch of chemical enhanced oil recovery (EOR). The complexity of alkaline flooding study is mainly embodied by its chemical reaction required by alkalis to react with oil acids. Consequently, in-situ surfactants are generated for various emulsification phenomenon. It is known that alkaline flooding performance in oil recovery is subjected to the emulsion type generation, thus, of great importance to alkaline flooding study is its mechanism investigation and saponification rate examination. In this study, a modified bottle test method that assesses major emulsion type formation for preliminary prediction of alkaline flooding performance in oil recovery is introduced. Homogenization and Karl-Fischer water content titration techniques were applied in the modified bottle test to overcome the emulsion preparation and analysis difficulties. In addition, sandpack alkaline flooding tests were conducted to prove the prediction reliability of the modified bottle test through identifying effluent emulsions. It is found either water in oil emulsion or oil in water emulsion could be representatively prepared in bottle test based on reaction environments identical to flooding tests’ conditions. Taking advantages of bottle test’s superior efficiency in simultaneous multi-case study, alkaline flooding screening test can be easily conducted applying statistical techniques to provide prior visions regarding dominating driving mechanism of oil recovery. This research verified a practical solution to representative emulsion preparation and phase volume quantification in the bottle test especially when it comes to high viscous heavy oil; therefore, mechanism investigation regarding alkaline flooding could be easily conducted. Besides, the CMGTM alkaline flooding simulation model was built considering the saponification reaction rate of immiscible fluids. A novel experiment design of alkali- heavy oil reaction system was proposed and implemented to measure reagents’ reaction rate at various temperatures through monitoring pH change by electrode. Through which the Arrhenius constant and activation energy were calculated. In addition, the stoichiometry for emulsification reaction was proposed according to bottle test results. The simulation model was history matched founded on reaction data thus model uncertainty was mitigated by reducing number of unconstrained parameters. Oil recovery predictions have been conducted using the history matched model and the optimized injection strategies were addressed.