Thermal injection methods are usually used for high viscosity oil. The results of previous studies showed that the combination of SF and SFF had the highest increase in oil recovery but still requires further study to determine the optimum strategy. This work is purposed to optimize the development scenario of a combined CSS-SF applied to a heavy oil field located in Sumatera, Indonesia. The recovery factor and NPV become the objective function, and several given and controlled parameters sensitivity toward the objective function are studied. A proxy model based on quadratic multivariate regression is developed to evaluate and get the desired objective function. The reservoir simulation of the thermal recovery process is done using CMG-STARS simulator. The overall workflow of scenario optimization is conducted using CMOST™ module. Optimum development scenario is obtained through maximization of the objective function. This work shows that the combination of proxy model development and optimization results in the best scenario of combined CSS-SF for heavy oil recovery.