HAZOP Automatic Reasoning Based On The LDG Model For HAZOP Analysis

HAZOP Stands for Hazard and Operability Study which has been. widely used to identify, evaluate and mitigate potential hazards in the chemical processes. It is generally done by a team of experts from different areas such as process design, operation and maintenance. It takes the team about one to eight weeks to complete the analysis of a typical chemical process. The analysis team meeting is a "brain storming" process which is quite effort consuming and time consuming. Therefore, it is also an expensive process. In addition, since it is done by the human team, the quality of the analysis results totally depends on the human team's knowledge and experience. Therefore, the consistency and completeness of the analysis results can not be guaranteed.To overcome the above problems, various HAZOP expert systems have been proposed to automate HAZOP analysis. Among the automation methodologies, signed digraph model-based reasoning is the most viable. However, the SDG model itself has unnoticed drawbacks. It can't represent all the HAZOP guidewords and express complex mapping relations. To overcome the knowledge representation limitations of the traditional signed digraph model, layered digraph (LDG) model has been developed in our laboratory. An automated reasoning system based on the LDG models is developed in this thesis. The related automation algorithm is given together with the framework of the database for LDG model knowledge storage. Graphical user interfaces for running the system are designed by using Java language. An industrial case study is discussed to demonstrate the advantages of the LDG model based reasoning.