Ontology-driven representation and reasoning for hazard and operability studies

Within Process Safety Management, Process Hazard Analysis methods such as Hazard and Operability (HAZOP) studies provide a structured basis for identifying credible deviation scenarios and demonstrating that risks are understood and managed. In practice, however, HAZOP outputs are typically recorded as heterogeneous tables or narrative text, making reuse difficult and complicating systematic checks for missing causes, consequences, or safeguards. This paper presents an ontology-driven, reproducible pipeline that converts published chemical process HAZOP studies into a machine-readable knowledge graph, enabling structured analysis and completeness checking, while also creating a foundation for applying expert domain knowledge in retrieval-augmented generation and for future automation of HAZOP studies. An ontology in the Web Ontology Language (OWL-DL) was designed to capture core concepts from International Electrotechnical Commission standard IEC 61882:2016 for HAZOP studies, including node, guideword, deviation, cause, consequence, safeguard, and recommendation. The ontology was populated from a unified spreadsheet compiled from twelve published HAZOP studies using Protégé, an ontology editor, and its Cellfie plugin for spreadsheet-to-ontology transformation, while graph-based exploration was supported via Neo4j, a graph database platform. Logical consistency was verified using HermiT, an OWL reasoner, constituting the genuine description-logic reasoning step in this study, while a separate, non-inferential completeness audit – a property-presence check implemented in Python with Owlready2 – was used to evaluate whether each deviation instance was linked to the required core relationships. The ontology remained logically consistent, and the audit demonstrated that ontology-based completeness checking can reveal missing safety-relevant relationships in HAZOP records, supporting more systematic, traceable quality assurance and providing a practical step toward digitalised process safety workflows.