Analyse des Risques par Arbre de Défaillance
Transféré par
BOUDOUNIT YounesAnalyse des Risques par Arbre de Défaillance
Transféré par
BOUDOUNIT Younes2.
2 Méthodes Déductives
Arbre de défaillance
(ADD)
II. Méthodes d’analyse de la SDF
Arbre de défaillance
• La méthode de l'arbre de défaillances, encore appelée arbre des causes (fault tree) est née en
1962 dans la société Bell Telephone.
• Cette méthode a pour objectif de déterminer les combinaisons possibles d’évènements qui
entraînent l'occurrence d'un évènement indésirable (ou redouté). L’idée est de représenter
graphiquement la logique de dysfonctionnement d'un système.
II. Méthodes d’analyse de la SDF
Arbre de défaillance
Quel est le risque de se trouver dans la noirceur?
Pas de lumière E
E= A+B
Lampe Pas de P(E)=?
A B
brulée courant
P(E)= P(A+B)
Probabilité de 5% que Probabilité de 10% que = P(A)+P(B)-P(A).P(B)
P(A)= 0.05 P(B)= 0.1 =0.05+0.1- (0.05*0.1)
la lampe soit brulée l’électricité est coupée
=0,145
II. Méthodes d’analyse de la SDF
Arbre de défaillance
Quel est le risque de se trouver dans la noirceur?
P(ER)=0,112
P(E1)=0,109 P(E2)=0,0025
II. Méthodes d’analyse de la SDF
Arbre de défaillance
Quel est le risque de ne pas avoir d’eau de refroidissement?
Evénement redouté: « Pas d’eau de refroidissement »
Causes possible :
• Débit nul en aval de V1 ET en aval de V2
• Si débit nul en aval de V1 :
• V1 bloquée ou Débit nul en aval de P1
• Si V1 bloquée : V1 fermée ou V1 bouchée
• Si débit nul en aval de P1 : P1 arrêtée ou
Réservoir vide
• Si débit nul en aval de V2 :
• V2 bloquée ou Débit nul en aval de P2
Données:
• Si V2 bloquée : V2 fermée ou V2 bouchée
• Probabilité de V1 /V2 fermée= 0.05
• Probabilité de V1/ V2 bouchée= 0.02 • Si débit nul en aval de P2 : P2 arrêtée ou
• Probabilité de P1/ P2 en arret= 0.01 Réservoir vide
• Probabilité de reservoir vide= 0.03
II. Méthodes d’analyse de la SDF
Arbre de défaillance
Quel est le risque de ne pas avoir d’eau de refroidissement?
P(ER)=P((A1+B1+C1+D1).(A2+B2+C2+D2))
TopEvent FTA
Fault Tree Analysis Software
II. Méthodes d’analyse de la SDF
Arbre de défaillance
Quel est le risque de ne pas avoir d’eau de
refroidissement?
TD1 (Suite)
Common questions
Alimenté par l’IAThe event 'No Cooling Water' occurs when there is zero flow downstream of both valves V1 and V2. For V1, the causes could be V1 being blocked, closed, or zero flow downstream of P1, which may be due to P1 being stopped or an empty reservoir. Similarly, for V2, the causes could be V2 being blocked, closed, or zero flow downstream of P2 due to P2 being stopped or an empty reservoir. The joint probability expression is P((A1+B1+C1+D1)∧(A2+B2+C2+D2)), where each term represents a condition leading to no flow .
Integrating probabilistic data into fault tree analysis enhances its interpretive power by quantifying the likelihood of different failure paths, thus providing a measurable aspect to the analysis of system vulnerabilities. The probabilistic approach allows for the computation of overall system failure probabilities and enables prioritization of risk mitigation efforts based on the most likely and impactful failure scenarios. This quantitative dimension underpins more informed decision-making in system safety and reliability engineering .
Graphically combining logical events in a fault tree analysis provides a clear visual representation of how individual and interrelated failures can cause a larger system failure. This method enables analysts to trace back through the logical paths to identify potential root causes and evaluate the likelihood of the top event. It simplifies complex interactions into visual diagrams, which facilitates better comprehension and communication of the failure mechanisms .
To calculate the probability of the 'No Cooling Water' event, we consider each fault's impact. Given the probabilities: V1/V2 closed (0.05), blocked (0.02), P1/P2 stopped (0.01), and reservoir empty (0.03). These probabilities must be combined according to the problem's fault tree logical relations. The computation involves using P(ER)=P((A1+B1+C1+D1)∧(A2+B2+C2+D2)), but exact numerical calculation isn't feasible with the provided data without additional structure specifics .
Using a fault tree analysis allows for a systematic examination of potential failure pathways within a system, which helps predict system reliability and highlight risks. It unveils the likelihood and impact of different failure combinations, which aids in proactive risk management and decision-making to enhance system safety and performance. By providing detailed insights into potential weaknesses and failure probabilities, FTA contributes to crafting strategies for risk mitigation and improved reliability .
The probability of total failure (P(E)) due to either a burnt lamp (A) or no current (B) is calculated using the formula: P(E) = P(A) + P(B) - P(A)P(B). Given P(A) = 0.05 and P(B) = 0.1, the calculation is P(E) = 0.05 + 0.1 - (0.05 * 0.1) = 0.145. Therefore, the probability of total failure is 0.145 .
Major limitations of fault tree analysis in system risk evaluations include its dependency on accurate data and assumptions, which can skew results if flawed. It is often challenging to completely model dynamic and complex systems with static trees. Additionally, FTA focuses primarily on binary states (success or failure) and may not account for degradation levels or potential recoveries. These constraints may limit its effectiveness in fully capturing nuances in complex system behavior .
The objective of the fault tree analysis (FTA) method is to determine the possible combinations of events that lead to the occurrence of an undesirable event. It is graphically represented to map out the logic of system dysfunction by illustrating the sequences and combinations of failures that might culminate in a specific system failure .
Fault tree analysis is effective in clearly illustrating potential failure combinations and is widely used for its visual and logical clarity. However, compared to methods like Event Tree Analysis (ETA) or Risk Matrices, FTA might lack dynamic modeling capabilities and is limited to pre-defined binary outcomes. ETA, for instance, is better suited for understanding sequential events following an initiating event. Thus, while FTA excels in visualizing failure logic, it may not fully address all dimensions of risk, especially in capturing cascades or recovery scenarios .
In fault tree analysis of an electrical failure, potential basic events include a burnt lamp or a cut-off of electricity. Each of these represents a starting point or contributing factor leading to the failure of the system—in this case, the absence of light. By identifying these events, the analysis can help in diagnosing the root cause and determining the likelihood of their occurrence, thus giving insight into the origin of the failure .