There are many kinds of complicated equipment in chemical plants. So, the chemical plants always have high possibility of accidents. Hazard analysis is one of the basic tasks to ensure the safety of chemical plants, but it is an arduous, tedious, time-consuming work and requires multidisciplinary knowledge, and demands considerable cognitive load from the analysts. To overcome these problems, there have been various attempts to automate this work by utilizing computer technology, particularly knowledge-based technique. However, many of the past approaches are lacking in rpoperties: safeguard consideration, accident diversity, cause and consequence diversity, pathways leading to accidents, and various hazard analysis reasoning. Therefore, in this study, deviation analysis algorithm, malfunction analysis algorithm, and accident analysis algorithm were proposed using multimodel approach of representing a chemical process adequately for hazard analysis in view of function, behavior, structure, and material property. Unit knowledge base is devised to model a process unit. It consists of unit behavior model and unit function model. In unit knowledge base, a process unit is modeled in different terms of variable and function. Variables describing unit are divided into inlet, internal, and outlet variables explicitly, by interconnecting behavior model and function model through these variables. This model represents physical hazards. In structure model, process is separated into several streams which represent different connections. In material knowledge base, chemical properties are considered according to NFPA code. A hazard analysis system, AHA, was developed on G2. The system was applied to feed section of olefin dimerization. And the case study was solved with AHA system successfully.