Contribution in the improvement of petrochemical plant’s safety using artificial intelligence methods
Loading...
Date
2024-02-15
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of 20 August 1955 – Skikda
Abstract
Petrochemical installations are complex processes with high connectivity and
continuously interaction between their systems and devices, which requires the control of
several parameters for optimal operation. This type of process deals with hazardous substances
at elevated temperatures and pressure and always faces difficulties in process control, hence
failing to target operations that generate several types of risks such as explosion and fire. These
safety incidents lead to serious consequences affecting people, environment and property.
To provide a high level of security and safety to these installations, different risk
management strategies have featured in literature, to identify and reduce hazardous situations.
However, even with the vast variety in these strategies and the combination of different
techniques to deal with some limitations and cover the maximum drawbacks, other difficulties
are still considered such as uncertainties in both the input and output of the analysis and the
classification problem. In recent years several tools have been developed based on artificial
intelligence to deal with these difficulties such as fuzzy logic that relies on membership function
principle and artificial neural network that emulate the biological ones.
The objective of the thesis is to define generic methods of risk analysis based on fuzzy
logic and neural network that can be appropriated for any industrial system and evaluate the
different risks very well under the condition of uncertainty.
The first: an integrated frame implemented based on HAZOP and fuzzy evaluation of
risk matrix to evaluate the safety integrity level (SIL) of an industrial fired heater
safety.
The second: a novel methodology based on LOPA method and FUZZY LOGIC to
increase its performance in terms of analysis and risk reduction. This approach is
implemented to a real system namely naphtha-A- stabilizer after identifying risks
inherent in this system by applying the HAZOP method (Hazards and Operability
Study).
The third: an approach based on Artificial Neural Networks (ANN) is developed to
schedule the SIL values of the safety integrity functions (SIF) of an industrial-fired
heater