[en] A loss of SG feed-water (TMLB sequence) for a prototypic PWR 900 MWe with a multi-compartment configuration (with 11 and 16 cells nodalization) has been calculated by the author using the ASTEC code in the frame of the EVITA project (5th Framework Programme, FWP). A variety of hypothesis (e.g. activation of sprays and hydrogen recombiners) and possible consequences of these assumptions (cavity flooding, hydrogen combustion, etc.) have been made in order to evaluate the global reactor containment building response (pressure, aerosol/FP concentration, etc.). The need to dispose of severe accident management guidelines (SAMGs) is increasing. These guidelines are meant for nuclear plants' operators in order to allow them to apply mitigation strategies all along a severe accident, which, only in its initial phase, may last several days. The purpose of this paper is to outline the influence on the containment load of most common accident occurrences and operators actions, which is essential in establishing SAMGs. ASTEC (Accident Source Term Evaluation Code) is a computer code for the evaluation of the consequences of a postulated nuclear plant severe accident sequence. ASTEC is a computer tool currently under joint development by the Institut de Radioprotection et de Surete Nucleaire (IRSN), France, and Gesellschaft fuer Anlagen-und Reaktorsicherheit (GRS), Germany. The aim of the development is to create a fast running integral code package, reliable in all simulations of a severe accident, to be used for level-2 PSA analysis. It must be said that several recent developments have significantly improved the best-estimate models of ASTEC and a new version (ASTEC V1.0) has been released mid-2002. Nevertheless, the somehow obsolete ASTECv0.3 version here used, has given results very useful for the estimation of the global risk of a nuclear plant. Moreover, under the current 6th FWP (Sustainable Integration of EU Research on Severe Accident Phenomenology and Management), the SARNET project (Severe Accident Research Network) has also the target to involve Ph.D. students and researches in the education and training elements of ASTEC development. And in this framework ASTEC should show a very good capability for being used as an investigative tool as well as an educational and training tool. In this paper, ASTECv0.3 is compared to MELCOR and CONTAIN codes in order to show the high degree of confidence which can be already placed in the ASTEC tool.(author)