System safety analysis of a brake press using fuzzy methodology

Abstract

Brake presses are machines which are widely found in the manufacturing industry. They enable metal sheets and plates to be bent in multiple ways so as to shape a wide variety of parts. The very principle by which these machines function creates a pinch and crushing zone for operators notably as far as their hands and fingers are concerned. Safety norms have been developed in the course of the years which prescribe technical protective means of the operating point as well as other means of controlling the other risks generated by the operation of these machines. Any industrial machine such as a brake press must be the subjected to a systematic risk analysis which take into account all the physical and human elements which can interact thus leading to possible accidents. Over the years, methods have been devised aiming to to estimate the probability of occurrence of an accident starting from the identification of the contributing risk factors. In the case of industrial machines such as brake presses, these methods rarely refer to historical data regarding accidents because of the scarcity of these data. The judgement of participants in these analyses is then utilized. Although the results of these analyses help in elaborating risk control strategies, their subjective character continue to pose problems. This thesis tackles the question of the estimation of the probability of occurrence of an accident which can arise on an industrial machine such as a brake press with a view towards making this analysis more quantitative but in a context where the historical data on accidents or knowledge on the factors contributing to accidents are limited. Towards that end, this thesis uses the concept of fuzzy logic as well as the structured technique of expert elicitation to quantify the judgements on the estimated probabilities. In this thesis, a brake press is analyzed taking into account to types of failures which can affect its functioning namely those that can impact the protective systems which guard the point of operation of the machine as well as those failures which arise from the actions of the operators. The contributing factors leading to accidents considered are analyzed using fault trees and Markov diagrams. The dynamic, sequential character of the genesis of an accident on a machine is also accounted for with this technique in this thesis. The probability of accident thus estimated for the industrial system is then used furthermore in this thesis to optimize the cost associated with a given level of safety achieved. This thesis is constituted of analyses of increasing complexity of work situations centered on a hydraulic brake pres. Firstly, the probability of occurrence of an accident suffered by an operator is estimated considering the failure of the protective device and the failure (error) of the operator. Then, this same situation is considered by incorporating rate of repair of the technical and human failure modes. Finally, elements of redundancy both technical and human are added to the model. This thesis show that the mathematical analysis method developed allows a more quantitative estimation of the probability associated with the risk being studied as well as an optimization of the cost associated with the achievement of a given targeted level of safety. This thesis presents a method which increases the validity of the judgement of the probability of occurrence of events contributing to industrial accidents but these remain dependent on the judgements expressed by the participants in the analysis. We plan further research works which will refine the expert solicitation method in the context of industrial machine safety. Furthermore, the models expressing the cost related to maintaining a given level of safety will also be perfected. The models describing situations leading to accidents involving industrial machines will incorporate in future works other complexities such as non-constant failure rates.

Date	30 Nov 2020
Original language	American English
Awarding Institution	École de technologie supérieure
Supervisor	Anh Dung Ngô (Supervisor)