A different approach is the ‘APSRA’ methodology, developed at the Bhabha Atomic Research Centre (BARC) of India [13]. In this approach, the failure surface[26] is generated by considering the deviation of all those comparative parameters which influence system performance.

Schematics of the APSRA methodology are shown in Fig. 2.3.

Like the RMPS methodology described above, the APSRA methodology developed in BARC, India, is primarily intended to analyse reliability of passive systems employing natural convection. The smallness of the driving head means a natural convection based system is susceptible to deviation from the performance of an intended function by a small change in key parameters. Because of this, there has been growing concern about the reliability of natural convection based systems.

The methodology named assessment of passive system reliability (APSRA) starts with selection of the system, followed by the understanding of its operational mechanism. Using simple computer codes, key parameters causing functional failure of the system are identified. Failure criteria are determined. Best estimate codes, such as RELAP5, etc., are then used to determine key parameter ranges, a deviation from which may cause system failure. These ranges of parameters are then fine tuned based on data generated in test facilities. This is done by performing uncertainty analysis for predictions of a best estimate code using in-house experimental data obtained in integral and separate effect test facilities.

190 180 170 160 150 140 130 120 110 100

0

40

55

FIG.2. Flowchart of the APSRA methodology (left) and typical failure surface for natural circulation (right).

In the next step, the possible causes of deviation of these parameters are revealed through root diagnosis. It is assumed that the deviation of such physical parameters occurs only due to a failure of mechanical components, such as valves, control systems, etc. The probability of system failure is evaluated based on the failure probability of these mechanical components, through a classical PSA treatment.

To demonstrate the methodology in a test case, it has been applied to the main heat transport system of the AHWR reactor, described in Annex VI of this report. This system employs a boiling (two-phase) light water coolant in natural circulation. To find code uncertainties, code predictions were compared with data generated from experimental natural circulation facilities, and uncertainties were evaluated from the error distribution between code predictions and test data. The facilities mentioned for generation of the required experimental data were the integral test facility ITL, the high pressure natural circulation loop HPNCL, and the flow pattern transition instability loop FPTIL [13].

The effects of variation of key parameters on system performance were evaluated, and a multi-dimensional failure surface was generated. The probability of the system to reach the failure surface was elaborated using generic data for the failure of components.

The APSRA methodology is being applied to other passive systems of the AHWR, such as the decay heat removal system using isolation condensers, a passive containment cooling system, a passive containment isolation system, etc.