A new multilayer sigma‐coordinate thermodynamic sea ice model is presented. The model employs a coordinate transformation which maps the thickness of the snow and ice slabs onto unity intervals and thus enables automatic relayering when the snow or ice thickness changes. This is done through an advection term which naturally appears in the transformed energy equation. Unlike previous approaches, the model conserves the total energy per layer (Jm−2 as opposed to Jm−3), which takes into account the changes in internal energy associated with thickness changes. This model was then tested against observational data from the Surface Heat Budget of the Arctic Ocean (SHEBA) experiment in the context of the Sea Ice Model Intercomparison Project, Part 2, Thermodynamics (SIMIP2). In general, the model reproduces the observed internal snow‐ice temperature and the ice thickness evolution very well. Results show that the ice thickness evolution is very sensitive to the ocean heat flux (Focn) and the thickness of the snow cover in winter. Given that the spatial variability in snow depth at small scale is large, the specification of the snow depth temporal evolution is crucial for an intercomparison project. Since Focn in SIMIP2 is calculated as a residual of the observed basal growth rates and heat conduction, the salinity of newly formed ice used in the simulations must be consistent with that used to derive Focn. Simulated and observed snow surface and snow‐ice interface temperatures suggest that not enough heat is conducted through the snow layer even when using a snow thermal conductivity as large as 0.50 Wm−1 K−1 (value derived from observed snow and ice internal temperature profiles). A surface energy budget of simulated and observed energy fluxes confirms this finding.