A process modeling methodology and inference algorithms for off-line hazard analysis and on-line fault diagnosis are presented. Hazard analysis and fault diagnosis are performed at different stages using different methods, while they need similar process information such as causal relationship and functions. Therefore, a unified process modeling is proposed, into which functions and behaviors of a process are incorporated. By building an automated system based upon the unified process model, the load of development and maintenance of the system can be reduced, and access to the system can be easier and friendlier. For automated hazard analysis, deviation and accident analysis algorithms are presented, which can infer causes and consequences of a given deviation based on the proposed model and identify all the possible accidents. For fault diagnosis, an inference algorithm based on an eventlist, that is a history, is also presented, which can identify the root cause of observed symptoms in an eventlist using the proposed model. A CSTR system with a feed tank and a heat exchanger illustrates the effectiveness of the proposed methodology.