With the development of Materials Genome Initiative (MGI), Integrated Computational Materials Engineering (ICME) has emerged as an important tool in the research of materials science. For the furnace tube alloys used in ethylene pyrolysis facilities, the high-temperature mechanical properties highly depend on their microstructure, especially the grain morphology. In this work, a cellular automaton finite element (CAFE) model simulated grain structure in solidification processes is applied. In the CAFE model, the density of grains formed at any undercooling and the growth kinetics of the dendrite tip are deduced by Gaussian distributions and Kurz-Giovanola-Trivedi (KGT) theory, respectively. A method of two-stage modeling for the cylindrical furnace tube is proposed, based on the characteristics of alloy melt flow in the horizontal centrifugal casting process. By the control of different parameters in the solidification process, the various microstructure morphologies of centrifugal casting HP40 alloys are obtained and verified with the experimental observations and theoretical analysis. As a consequence, the microstructural prediction model of centrifugal casting HP40 alloys is optimized. The result shows that the desired full columnar grain morphology of HP40 alloys can be realized at the pouring temperature of 1620 °C and the revolving speed of 3100 rpm. This research reveals the potential application of the CAFE approach in microstructure control and optimization of furnace tube alloys.