[en] As an ISAAC computer code, which was developed for a Level-2 PSA during 1995, has mainly fundamental models for CANDU-specific severe accident progression and also the accident-analyzing experiences are limited to Level-2 PSA purposes, the core heat transfer model, break flow model, emergency core cooling system model and containment dousing spray /local air cooler model are evaluated to enhance understanding for basic models and to accumulate accident-analyzing experiences. Sensitivity studies using model parameters and sensitivity coefficients are performed. According to the results from AECL experiments and code analyses for core heat transfer model, it was found that one representative fuel rod for the actual 37 fuel rods did not cause serious temperature discrepancies during the severe accident progression. The results from emergency core cooling system model, shows a good comparison with the FSAR. As the results of the evaluation, it was found that local air coolers could control containment pressure whether dousing spray is operating or not, and their operation does not cause containment failure. Regarding the dousing system, it could control containment pressure as long as it is operating and its operating time depends on containment conditions. For a large LOCA sequence without local air coolers, spray works only for 1.2 hours and delays containment failure by 13 hours compared to the no spray case. According to the test results, the ISAAC models for local air coolers show a consistent trend for steam removal. As ISAAC could model local air coolers only at two locations at present, future work is planning to generalize the locations for local air coolers