Within the framework of the Generation IV Sodium-cooled Fast Reactor (SFR) in which the CEA (French Commissariat à l’Energie Atomique et aux Energies Alternatives) is involved, the French innovative reactor design behavior under severe accidents conditions has to be assessed. Such accidents have mainly been simulated with mechanistic calculation tools (such as SAS-SFR and SIMMER-III). As a complement to these codes, which provide reference accidental transient calculations, a new physico-statistical approach is being developed at CEA; its final objective being to derive the variability of the main results of interest to quantify the safety margins. This approach requires fast-running tools to simulate extended accident sequences, by coupling models of the main physical phenomena with advanced statistical analysis techniques. The tool enables to perform a large number of simulations in a reasonable computational time and to describe all the possible scenario progressions of the hypothetical accidents. This general approach, combining mechanistic codes and evaluation tools, has already been conducted for some accidental initiator families (USAF – Unprotected SubAssembly Fault (Marie et al., 2016) and ULOF – Unprotected Loss Of Flow (Droin et al., 2017). In this context, this paper presents a physical tool (numerical models and result’s assessment) dedicated to the simulation of the beginning of the primary phase of the Unprotected Transient OverPower accidents (i.e. before failure of sub-assembly wrapper).At the beginning of this primary phase, the fast increase of nuclear power induces a strong temperature rise in the fuel pellets leading to strong mechanical and thermal loads on the cladding which could lead to clad failure or/and fuel meltdown. These phenomena are described and modelled analytically in single pin geometry in accordance to the level of details required to catch all the decisive phenomena.Slow power increase transients, such as control rod withdrawal, and fast power increase transients have been investigated in the past. Experimental validation on CABRI (experimental reactor dedicated to safety studies) and CESAR (Circuit d’Etude de l’ébullition du Sodium lors d’un Accident de Réactivité) experiments were carried out focusing on the amount of molten fuel formed during the transient, on the propagation of the void front in the channel in case of sodium boiling and on the pin failure mechanisms. Furthermore, a comparison of the physical tool calculation results was performed against reference accident SIMMER-III calculations.The tool is demonstrated to be able to predict the radial propagation of the molten zone in the fuel pin, the pin failure mechanism and void front propagation with a discrepancy of less than 10%. In the future, this physical tool, associated with a point kinetic neutronic model, will be used to simulate the global core behavior under an UTOP transient following a ramp excursion.