Thermochemical approach involving modelling and critical assessed thermodynamic data of the corium-concrete chemical system plays an increasing role in order to simulate the ex-vessel corium behaviour during a severe accident. For more than 10 years, a special effort to produce a specific thermodynamic database for nuclear applications (NUCLEA) was made with the support of different organisations in Europe. The CALPHAD method was used to create such a database. For the modelling of the thermochemistry of corium-concrete melts, this method consists in the mathematical description of Gibbs energy of each phase of the U-O-Zr-Fe-Si-Ca-Mg-Al-O complex system, based on a phenomenological model in such a way that the model parameters are adjusted to allow to reproduce the available experimental information. Up to now, for this modelling, the set of thermodynamic data was mainly based on the optimisation of the key quaternary system U-O-Zr-Fe, of the binary oxide systems containing UO2 and/or ZrO2 and concrete oxides and of the ternary CaO-Al2O3-SiO2 oxide system. The Fe-FeO binary is an important system during the oxidation phase of the structural materials and may play a significant role in fixing the oxygen potential of the melts due to large amount of steel available and/or due to the presence of iron oxide in some core catcher concepts of the new generation PWR reactor. Some studies have shown that interaction of melts with iron oxides could not be treated in a simplified approach. For that reason, new modelling of some ternary phase diagrams including iron oxides have been recently undertaken and are presented here with some comparisons with experimental data coming from the voluminous literature on the thermodynamic properties of iron-silicate slags. From this new modelling, the estimations of the liquidus and solidus temperatures of corium-concrete melts can be performed with more accuracy. NUCLEA calculations are compared to the experimental information available on the thermochemistry of corium-concrete melts.