National Technical University of Athens (NTUA)
Ανάπτυξη και εφαρμογή μεθοδολογίας εκτίμησης διακινδύνευσης σε ναυπηγεία
Abstract
dc:descriptionThe aim of this thesis is to develop a model that leads to a Professional Quantitative Risk Assessment (QRA), in the shipbuilding and ship repair industry taking account the occupational accident and attempting to help address a problem that has, both, significant social and economic dimensions.Two independent routes that complement each other were followed.The first route led to the development of an analytical model capable of describing the mechanisms of connection between cause, effect and the consequences of accidents in shipbuilding and leading, under conditions, on a comprehensive QRA. The model is based on the identification and the assessment of hazards, as well as, other aggravating factors in a yard work environment. Techniques of inductive models like Event Tree Analysis (ETA) and Fault Tree Analysis (FTA) were used. The inductive models were ultimately linked to an integrated structure "BOW TIE", which clearly shows the relationship of main danger sources and the aggravating factors that could lead to the creation of an event and the course of the event to possible consequences through mitigating factors that compose the Independent Layers of Protection (ILPs).In this way, the main components of the risk are presented, meaning the probability of an event on the one hand and the possible consequences on the other. For a better understanding and documentation of the model, risk analysis and identification techniques for concrete typical situations (HAZID) were performed; also flow diagrams were constructed for such cases and numerical calculation example of the risk for one of the cases.The second route led to the construction of an expert system, based on fuzzy rules and fuzzy inference, using the provided information from the data of accidents in the shipbuilding industry, firstly with the use of statistical analysis of accidents recorded data, which have been collected in shipbuilding units, and secondly with the use of soft computing calculating techniques. The creation of a Fuzzy Inference System (FIS) was decided based on Adaptive Neural Networks (AN). The statistically process of the recorded accident data set the parameters and the database for the construction, training and the verification of the results of the Adaptive Neural Fuzzy Inference System (ANFIS).The "BOW TIE" model proved capable of leading to qualitative conclusions such as the identification of the main danger sources, as also as the aggravating and mitigating factors and also to provide quantitative documentation for any safety measures improvement to reduce risks.The ANFIS model proved a feasible simulation method of the complex relationship between the parameters involved in an accident in the shipbuilding industry, exploiting the existing and appropriate statistically processed historical accident data, constituting, thus, a tool able to make predictions and be conclusive in small computational time.
Degree
thesis:*- Grantor dc:publisher
- National Technical University of Athens (NTUA)
- Year dc:date
- 2015
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Fragiadakis, Nikolaos
- Φραγκιαδάκης, Νικόλαος
Subjects
dc:subject × 17- Εκτίμηση διακινδύνευσης
- Επαγγελματικός κίνδυνος
- Ναυπηγικές εργασίες
- Δέντρα σφαλμάτων
- Δέντρα γεγονότων
- Προσαρμοστικό νευροασαφές σύστημα συμπερασμού
- Quantitative risk assessment
- Occupational risk
- Shipbuilding industry
- Bow – tie methodology
- Adaptive neuro - fuzzy inference system
- Επιστήμες Μηχανικού και Τεχνολογία
- Επιστήμη Μηχανολόγου Μηχανικού
- Άλλες Επιστήμες Μηχανικού και Τεχνολογίες
- Engineering and Technology
- Mechanical Engineering
- Other Engineering and Technologies
Rights
dc:rights- Statement dc:rights
-
- BY_ND
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/40000