Despite system thermal-hydraulic codes were extensively validated for transient simulations of LWR, several activities highlighted limited capabilities of these tools to model heat transfer within in-pool passive power removal system. Discrepancies with experimental results were related to the underestimate of pool boiling and film condensation heat transfer coefficients. Thus, the DIAEE of "Sapienza" University of Rome developed a modified version of RELAP5/Mod3.3, able to handle fundamental heat exchange phenomena involved in passive in-pool safety systems. A primary validation procedure has been performed for separated and integral effects. Dealing with nucleate boiling, the Root Mean Squared Relative Error (RMSRE) of wall superheat has been reduced from 1.290 to 0.182. Concerning film condensation, wall temperature RMSRE has been reduced from 0.192 to 0.058. The integral effect assessment has involved an experimental test of the PERSEO facility. The qualitative comparison between experiments and calculations has highlighted significant improvements of the modified RELAP5/Mod3.3.