Reliability evaluation of a natural circulation system

This paper discusses a reliability study performed with reference to a passive thermohydraulic natural circulation (NC) system, named TTL-1. A methodology based on probabilistic techniques has been applied with the main purpose to optimize the system design. The obtained results have been adopted to estimate the thermal-hydraulic reliability (TH-R) of the same system. A total of 29 relevant parameters (including nominal values and plausible ranges of variations) affecting the design and the NC performance of the TTL-1 loop are identified and a probability of occurrence is assigned for each value based on expert judgment. Following procedures established for the uncertainty evaluation of thermal-hydraulic system codes results, 137 system configurations have been selected and each configuration has been analyzed via the Relap5 best-estimate code. The reference system configuration and the failure criteria derived from the "mission" of the passive system are adopted for the evaluation of the system TH-R. Four different definitions of a less-than-unity "reliability-values" (where unity represents the maximum achievable reliability) are proposed for the performance of the selected passive system. This is normally considered fully reliable, i.e. reliability-value equal one, in typical Probabilistic Safety Assessment (PSA) applications in nuclear reactor safety. The two 'point' TH-R values for the considered NC system were found equal to 0.70 and 0.85, i.e. values comparable with the reliability of a pump installed in an "equivalent" forced circulation (active) system having the same "mission." The design optimization study was completed by a regression analysis addressing the output of the 137 calculations: heat losses, undetected leakage, loop length, riser diameter, and equivalent diameter of the test section have been found as the most important parameters bringing to the optimal system design and affecting the TH-R. As added values for this work, the comparison has been made between results from this study and results from a previous analysis where the same methodology was adopted for the evaluation of the TH-R of a different passive system named Isolation Condenser (IC). The comparison shows that the current single-phase NC system is 'more reliable' than the two-phase IC system. This constitutes a proof of qualification and of consistency for the adopted methodology. (C) 2003 Elsevier Science B.V. All rights reserved.