Ice accretions on aircraft flight surfaces can degrade lift, increase drag, and reduce controllability. Anti-icing systems can remove or prevent ice growth. To predict the ice shrinkage or accretion using models such as LEWICE, the convective heat transfer behavior at the ice surface must be quantified. The work here is focused on understanding the convective heat transfer for several laser-scanned ice shapes by using commercial computational fluid dynamics (CFD) heat transfer simulations. The leading edge ice shape from a NACA 0012 airfoil and its geometrically unwrapped simplification are studied. Limitations encountered with a semi-automated unstructured mesh generation tool are presented. Thin boundary layer mesh thicknesses (i.e. much thinner than the flow’s viscous or thermal boundary layers) are found to be necessary in order to capture the surface curvature features and preserve good mesh quality near the geometric surface. To better model a prior experiment, the unwrapped geometry uses a conjugate heat transfer model which includes a thickness to the unwrapped ice shape to allow the modeling of its internal conductive heat transfer. The unwrapped geometry CFD heat transfer results are compared to prior experimental results for an identical shape in a similar heat transfer setup. Also, heat transfer results for the original airfoil ice shape are presented for constant surface temperature, and compared to the unwrapped iced plate. The CFD results predict higher Stanton numbers for turbulent flows as compared to laminar flows, but overall underpredict the Stanton numbers when compared to experiments, particularly in the iced bumpy region. Additional work is needed to study alternative meshing strategies and various turbulence model settings. The results are useful for advancing the understanding of convective heat transfer from ice shapes and conductive heat transfer in ice layers.