Towards a Phenomena-based HAZOP: From Physical Principles to Consistent Identification of Hazards

This paper presents a preliminary, phenomena-based framework to support the semi-automation of Hazard and Operability (HAZOP) studies. While traditional HAZOP remains a cornerstone of process safety analysis, it is inherently time- and resource-intensive and strongly dependent on expert judgment, which may affect consistency and completeness. The proposed proof-of-concept framework introduces a physics-informed backbone grounded in qualitative mass and energy balances, combined with a graph-based representation of process variables and their interdependencies. Starting from process flow diagrams, directed graphs are constructed to encode cause–deviation–consequence relationships, while routines are used to trace deviation propagation paths from process variables to predefined risk nodes. By anchoring deviations to first-principles behaviour, the approach aims to improve traceability and reduce subjective clustering in early-stage hazard identification. The framework is demonstrated on a standardised example derived from BS EN 61882 to illustrate the structure, scale, and characteristics of the algorithm-generated preliminary HAZOP (preHAZOP) output. The results indicate improved scalability and systematic coverage of interaction pathways, while confirming the continued need for expert review for contextualisation, filtering and validation. The present scope is intentionally limited to preHAZOP and a simplified case study, with qualitative risk indications and a focus on the interpretability of path-based results. Future developments will address quantitative extensions, automated clustering and ranking strategies, and application to more complex systems and data standards.