The increasing complexity of most industrial processes always tends to create problems in monitoring and supervision systems. Detection and early fault diagnosis are the best way to manage and solve these problems. Artificial neural networks (ANNs), by their ability to learn and store a large volume of information, are tools particularly suitable for diagnostic support systems. Effectiveness of ANNs for fault diagnosis in milk pasteurization process is presented in this paper. The initial data base used for fault diagnosis is constructed using data extracted from FMEA (Failure Modes and Effects Analysis) tables of milk pasteurization process. Indeed, this analysis makes it possible to establish the links of cause and effect between the faulty components and the observed symptoms. Three models of ANNs, namely Feed-Forward Back Propagation (FFBP), Radial Basis Function based Neural Network (RBNN), and Generalized Regression Neural Networks (GRNN) are developed and compared. The determination coefficient (R2 ), Root Mean Square Error (RMSE), and Mean Absolute Error (MAE) statistics were used as evaluation criteria of all the models. The comparison results indicate that the performances of GRNN model are better than the FFBP and RBNN models. The same neuronal models can be extended to any technical system by considering appropriate parameters and defects.

Aiming to provide a generalized method for assessing the performance of safety instrumented systems with a flexible and accurate consideration of the common cause failures’ contribution. This paper is devoted to the development of a direct way to generate the transition rate matrix associated with the continuous-time Markov model of any typical KooN architecture using any parametric model. Such a choice is considered after a detailed comparison of the ability of several dependability methods (e.g., fault trees, reliability block diagrams, Markov models, Bayesian networks, etc) to provide simple representations and genuine results in this context. To validate the developed method, the unavailability and the unconditional failure intensity of a wide range of configurations are quantified using the Binomial Failure Rate model and compared to those of the complete fault tree implementation.

The industrial risk mapping is a topical problem in the field of risk management that attracts many researchers to develop risk matrices to ensure consultation between their actors. In this context, this paper aims to propose the principal component analysis (PCA) method as support for this consultation. Indeed, the use of PCA method is justified by its robustness for aggregate initial data associated with industrial risks as principal factors and ranking of this risk in terms of their criticalities in risk matrices. However, the aggregation of initial data on industrial risks by the main factors, in some cases, leads to inaccuracies which make it difficult to classify certain risks. This paper proposes two variants of PCA method to solve this inaccuracy and succeeds in classifying risks according to their respective criticalities, namely PCA-Improved (PCA-I) and PCA-I-Fuzzy (PCA-IF). The results come from the PCA application and its proposed variants (PCA-I and PCA-IF) on an example of accident scenarios ranking. We have established a scientific basis for the capitalization of mapping tool for consultation and decision support to industrial risk managers.

Renewable power generation facilities are constantly expanding due to the expected depletion of fossil fuels and the increasingly demanding policy of pollution control. Having said that, hydrogen is one of the promising energy sources. That said, hydrogen chain safety is an unescapable parameter that should continuously coexist with the development of hydrogen domain. In this context, this article presents a contribution to the risk analysis and evaluation of a complex hydrogen production system 'EGA-9000′ at CIEMAT (Centre for Research on Energy, Environment and Technology - Madrid, Spain). The methodology followed in this study revolves around the risk analysis process through a FAST (Functional Analysis System Technique) functional analysis method and a HAZOP (HAZard and Operability) dysfunctional analysis method. The evaluation of the thirty-three scenarios identified by the risk analysis shows that the studied system is insecure. Indeed, five scenarios at an unacceptable level of risk. And it is noted that the risk of fire and explosion is the major risk for all scenarios studied. To this end, safety measures (recommendation) have been proposed based on the weaknesses detected by the risk analysis carried out.

Research on risk matrices show that there is considerable diversity in the practice of designing risk matrices. This has led to serious problems of standardization and communication. Indeed, these problems affect at the same time on the development of matrices and in their exploitation in term of risk assessment. To solve these problems, this paper proposes an experience feedback method that aims to capitalize the feedback invariants resulting from the analysis of existing risk matrices. This capitalization allows developing a theoretical framework of the robust risk matrices design. The application of the proposed method for examples of matrices confirms the interest of articulating these risk matrices designs through an argument based on experience feedback. In this sense, the merit of the proposed experience feedback method is that it promotes the sharing of knowledge between the actors involved in a risk assessment.

The ageing of the Algerian oil and gas (O&G) installations has led to many incidents. Such installations are over 30 years old (life cycle) and still in operation. To deal with this O&G crucial problem, the Algerian authorities have launched a rehabilitation and modernization schedule of these installations. Within the framework of this program, many audit operations are initiated to elaborate a general diagnosis of the works to be performed while optimizing production. In other words, industrial ageing risks shall be controlled.

In the process safety management (PSM) context, the aim of this paper is to study ageing problem of the Algerian industrial installations through proposed indicators. Their prioritization adjusted by (TOPSIS) Technique for Order-Preference by Similarity to Ideal Solution method which allows identification of ageing control solutions of Algerian onshore fields.