The contribution of the present paper aims to develop the traditional Failure Modes, Effects and Criticality Analysis (FMECA) for quantitative risk analysis from a Bayesian Network (BN)-based perspective. The main purpose of research consists in providing a framework for analysing causalrelationships for risk evaluation and deriving probabilistic inference among significant risk factors. These parameters will be represented by linguistic variables and will include the human factor as a key element of analysis. Traditional approaches for risk evaluation and management performed by FMECA  represent helpful tools to globally enhance systems and processes conditions . However, such approaches require previous clarification of several assumptions/simplifications . FMECA is a systematic procedure to identify and analyse all the failure modes potentially involving systems or their main components, through the definition of the related causes and effects. In particular, the method aimsto prioritise the failure modes under analysis by calculating the index called Risk Priority Number (RPN) for each of them. The RPN is traditionally derived from the multiplication of three main factors, that are severity (S), occurrence (O) and detection (D), generally ranged within discreteintervals. Severity S measures the impact of a given failure mode with respect to the global performance; occurrence O estimates the frequency of a failure mode within a given time horizon; detection D expresses the probability of correct failure identification. The three risk factors arecommonly assessed in a qualitative and subjective way, what may lead to imprecise results with the consequent adoption of ineffective decisions in terms of preventive and/or mitigation actions. This assumption represents one of the reasons why the traditional RPN has been widely criticizedin the literature. Moreover, the RPN formula appears far too simplistic  , without taking into account the different importance of the three aforementioned parameters  , that is their different weights in evaluating risks. Other studies underline as the non-continuous distributionof the values of the RPN makes imprecise the assessment of differences between two consecutive values assumed by the index  .Apart from other various drawbacks, one has to observe as FMECA does not take into account the simultaneous occurrence of multiple failure scenarios. In this direction, a BN-based approach can provide a wider range of benefits in the field of risk analysis, in terms of modelling complexsystems, making more accurate predictions about parameters’ values and computing with precision the occurrence probability of failure events . We also propose to consider new parameters with respect to those traditionally used for the RPN calculation. Specifically, the role played byhuman resources will be integrated within the risk function calculation, something that existing approaches scarcely take into account. Lastly, we present and solve a real-world application on such a fundamental business topic as supply chain risks (SCR) management, which originates fromthe intersection between the processes of risk management and supply chain management , as exemplified in Figure 1. This choice is further motivated by the fact that FMECA has been recently extended to SCR evaluation , representing a current lively topic of research.
|Titolo della pubblicazione ospite||Modelling for Engineering & Human Behaviour|
|Numero di pagine||6|
|Stato di pubblicazione||Published - 2020|