Core disruptive accidents in fast reactors need to be monitored carefully since they may lead to possible criticality configurations. However, the worst-case scenario may have small probability occurrences, but the proof of it requires multidisciplinary studies. Even with the upgrade in computer performance, calculations would require several months on several parallel computers. Accurate calculations with short running times are thus required. Updating the neutronics module of SIMMER set up in the 1970s was therefore carried out with the help of routines able to handle probability tables for generating broad group libraries. The use of such libraries together with new SIMMER options is now able to produce reliable results in all sorts of situations while maintaining reduced calculation times.
Indeed, until now, neutronics calculations from SIMMER gave results quite far from ERANOS ones (differences in reactivity larger than 1.5 $). The discrepancies were mainly due to the libraries used. As a consequence, in 2000, an ERANOS module (BISIM) was created to generate SIMMER nuclear data libraries (for both cross sections and self-shielding factors) from the ERANOS nuclear data file, thereby reducing the major source of inconsistencies. Other improvements were added by the Japan Atomic Energy Agency, on the way of calculating the transport cross section and on the library group scheme so as to better calculate the k-effective within a reasonable time frame, but also at the Commissariat à l'Energie Atomique et aux Energies Alternatives on the 
-effective calculation. A new option (using the Keepin data) was implemented in 2010 in SIMMER.
Once all these optimizations were carried out, a comparison between the SIMMER (III for two dimensions and IV for three dimensions) and ERANOS results was performed for a series of disruptive and representative configurations. While the computation time has not changed significantly, the differences on k-effective between ERANOS reference route results and SIMMER 16 energy-group calculations were drastically reduced by [approximately]0.8 $.
