Reservoir simulation software is an important tool in oil and gas industries to predict the multiphase flow of reservoirs. The output from reservoir simulation consists of production history, reservoir pressure, grid block saturation, porosity and permeability change etc. Due to the intrinsic set of uncertainty in reservoir simulation prediction, considerable number of simulation runs to be performed. As reservoir models becoming more complex, the size of the resulting reservoir models become larger and larger. Making hundreds and thousands of simulations require considerable amount of time and sometimes simply impractical. Hence, Well-based Surrogate Reservoir Model (SRM) is a potential candidate to be used as a solution tool to solve this issue. This paper presents a workflow of Well-based SRM that mines the output data from conventional dynamic reservoir simulation. As a part of this system, it is proposed to develop Well-based SRM extraction based on Artificial Neural Network (ANN) to enhance the realization run time. Well-based SRM is used for fast track analysis, decision optimization and has the capability of generating shorter time simulation response in relation to the conventional dynamic reservoir model. © 2013 IEEE.