The convective heat transfer coefficient critically influences ice accretion and shape prediction during aircraft icing. This study investigates similarity parameters for convective heat transfer on ice surfaces through numerical verification using ChuanYun icing software, establishing theoretical foundations for heat transfer calculations in icing simulations. The software implements boundary-layer integral and CFD based methods for convective heat transfer analysis. Validation against LEWICE and experimental data confirms computational reliability across varying Reynolds numbers and surface roughness. Results demonstrate that N-S based flow simulations achieve superior accuracy in turbulent heat transfer prediction at high Reynolds numbers compared to panel methods, particularly in capturing stagnation-point thermal details. Systematic analysis reveals that conventional flow similarity parameters ensure only macroscopic consistency, while the Eckert number (linked to wall temperature) governs convective heat transfer distribution. Geometric scaling experiments require simultaneous matching of Mach, Reynolds, Eckert, and Prandtl numbers to preserve Nusselt number similarity. A novel dimensionless parameter is proposed for fluid–solid coupling with internal heat sources. Cylinder validation confirms that alignment of this parameter with the Biot number ensures temperature field similarity between scaled models and prototypes. Findings indicate a linear negative correlation between Eckert number and convective heat transfer intensity. The proposed parameter facilitates scaled-down system design with internal heat generation, offering guidance for experimental scaling. Future work will address unsteady conditions and complex scaling constraints.