Water flooding is the most common oil recovery technology for carbonate reservoirs in Middle East. However, because of the various reservoir physical properties, serious problems, such as water breakthrough and uncharted distribution characteristics of remaining oil, bring great challenges to successful water flooding development. In this research, the scaled 3D water flooding experiment with an interlayer heterogeneous and anisotropic physical model was performed to achieve quantitative characterization of water flooding performance and pattern and the distribution of remaining oil after displacement in a porous carbonate reservoir with bottom water. Results indicate that the water flooding performance of bottom water varies with the various permeabilities in different layers. It can be divided into two patterns, characterized by smooth spread in high-permeability layers and coning type breakthrough in low-permeability layers. Specifically, in high-permeability layers, the horizontal flow is faster than vertical flow because the bottom water is mainly controlled by higher horizontal permeability and the lower production pressure difference can't overcome the gravity. Therefore, the whole layer is flooded and there is less remaining oil in high-permeability layers. On the contrary, in low-permeability layers, production pressure difference will drive water to climb up near the vertical wellbore and the water cone appears significantly, resulting in the remaining oil between the wells. In conclusion, the main mechanism behind the water flooding performance can be the heterogeneous and anisotropic characteristics of permeability. • A new physical model is used to imitate the water flooding performance in a porous carbonate reservoir with bottom water. • The high permeability contributes to the smooth spread performance and few remaining oil. • The low permeability contributes to the water coning breakthrough and plentiful remaining oil residues between the wells. [ABSTRACT FROM AUTHOR]