Process Hazards Analysis (PHA) is an important component of process safety management. It is a labor- and knowledge-intensive process which would gain from automation. Previous work in this area have concentrated on hazard identification, which is just one step during PHA. In this paper, we propose an integrated framework and a knowledge-based system, called PHAzer, which takes a more comprehensive approach to the entire PHA process. While qualitative models are sufficient for hazard identification, a wide range of modeling capabilities from qualitative cause-and-effect models to more precise quantitative mathematical models are required for automating all the steps during PHA. PHAzer uses qualitative digraph based models of units and operations to identify hazards, dynamic mathematical models to perform detailed safety evaluation, and digraph and fault tree models to synthesize and analyze fault trees. The salient features of PHAzer are presented and illustrated using an olefin dimerization plant case study.