Chemical processes operate at a multitude of steady states and frequently undergo transitions between them. Alarm management, fault diagnosis, and other automation systems are however, usually configured assuming a single state of operation. When the plant moves out of that state, these applications signal false alarms even when a desired change is occurring. In this paper, we propose a framework that would enable these applications to be state-conscious and reconfigure themselves to remain relevant in any process state. Process states are demarcated into modes and transitions corresponding to quasi-steady state and transient operations, respectively. Modes are characterized using variable ranges and transitions using key variables and their trends. A trend analysis-based approach for locating and characterizing the modes and transitions in historical data is proposed. A supervisory system that tracks the state of the process units online is also developed. The application to two case studies - startup and grade change transitions in a pilot scale distillation column and startup and shutdown of a refinery catalytic cracking unit simulation - is reported. The benefits of the framework are illustrated through dynamic state-specific alarm reconfiguration during a startup. (C) 2004 Elsevier Ltd. All rights reserved.