fuzzy bayesian network

This study aims to provide a risk assessment method based on a fuzzy Bayesian network (FBN) to prevent the failure of firefighting systems (FSs).

In this research, the fault tree structure was validated, and the failure factors were identified. Triangular fuzzy numbers and the combined CoA/Sum-Product method were used to calculate the failure probability. Then, deductive and inductive reasoning and sensitivity analysis were performed using fuzzy logic and fault tree transmission in a Bayesian network (BN).

The results of a case study in methanol storage tanks showed that the combination of FBN and validation of structures can be presented as a suitable method to evaluate the reliability of FSs. The content validity index of the final basic events was over 0.79. The highest and lowest failure probabilities were related to foam system failure and valve failure, respectively. Also, the failure probability for the failure of FSs was 5.7E-6 based on a fuzzy fault tree (FFT). After updating with BN, the previous probability of failure of FSs with FBN was calculated to be 0.022978, which was greater than the value of the FFT. The reliability of the system was equal to 0.77022.

the present approach can help the decisions of managers and analysts of the petrochemical industry to prevent the failure of FSs in tanks due to changes in systems.

Oxygen-generating central plays a vital role in the continuous performance of hospitals. Any leakage or failure in this section can not only endanger the health and safety of patients but also cause fire and explosion. Probabilistic risk assessment is a useful tool for identifying the main root causes of leakage in oxygen-generating central. This study aimed at risk assessment of an oxygen-generating central in a hospital in Hamadan using fuzzy sets theory and Bayesian networks.

First, all root causes supposed to contribute to oxygen leakage from any part of the oxygen-generating central were identified, and based on them a fault tree analysis (FTA) was constructed. Then, the FTA was mapped in a BN. The failure probability of root causes was calculated using fuzzy sets theory and experts’ opinions. Belief updating based on BN was utilized for subsequent analyses.

According to this study, ignorance of labels on the oxygen generation and distribution system is the most important root cause leading to oxygen leakage. Moreover, removing masks from patient’s faces is the main cause of oxygen leakage in patient rooms. Once leakage occurred, the presence of an ignition source can lead to fire or explosion.

oxygen leakage can create considerable risks in hospitals. All staff should be provided with sufficient training regarding hazards of oxygen-generating and distributing systems and oxygen leakage. Particular attention should be paid to such leakages and their adverse consequence in emergency planning and hospital crisis management.

Process industries are categorized as complex systems due to the hazardous materials, large number of employees and the complexity of the processes. For this reason it is necessary, it is necessary to design an appropriate risk assessment system to control and manage the risks associated with such places. Therefore, the purpose of this study is to provide a comprehensive and quantitative risk analysis of the explosion using the integration of modern approaches of Bayesian Networks and fuzzy theory in a process industry.

METHODS AND MATERIALS:

 In this study, compressor gas leakage was selected as the scenario to evaluate the probabilistic risk of explosion. For cause-consequence analysis of selected scenario and reduce uncertainty of the occurrence of basic events probability, Bow-Tie method and  fuzzy logic approach was used respectively.

The compressor gas leak analysis detected by using the Bow-Tie method, 24 basic events and 11 intermediate events.With regard to the performance of safety barriers, ie success and failure of these safety barriers led to the determination of 9 consequences that included a flash fire, jet fire, and the vapor cloud explosion and release of materials.

The results of this study showed that, based on fuzzy Bayesian network analysis, filtering failure was identified as the most effective basic event in the occurrence of gas compressor leakage. Therefore, in designing preventive and control strategies for the occurrence of Risk Management Accidents, attention should be paid to these root events.