| Abstract: | Complex heterogeneous systems, such as power plants or petro-chemical process plants, nowadays contain complex automation for start-up and shut-down control and support systems for the operators. Often, however, the operator support and automation suffers from a lack of flexibility, and only functions for a number of well defined operating modes and pre-defined paths for the transition between these modes. This paper proposes an alternative and more flexible method for developing and describing intentional mode transitions, and for developing diagnostic systems, using Multilevel Flow Modeling (MFM). MFM models a system by expressing it in terms of its goals and in terms of elementary functions that describe the mass, energy and information flows in the system. This paper describes the use of MFM models as a basis for reasoning about the actions that are necessary to achieve the goals of a system or to obtain an intentional change in the system's mode. For this, data measured from the system must be used to update the state of the MFM model so that the state of the model reflects the state of the system. The outcome of the reasoning can be used as support for an operator or for automated control of complex systems. This paper defines the relevant states for goals and flow functions and presents a set of rules for determining these states on the basis of measurements from a process. The relations between goals and functions, and among functions themselves, are discussed. A mechanism is introduced to produce a change in the desired mode of a process, and expressed in rules to implement this change. The approach is explained at the hand of a simple example system. An MFM model of this example system is presented, and used to illustrate how measured variables can be used to calculate the states of the elements in the MFM model. At the hand of the same model the rules for inferring the states of goals and functions, and for determining the required actions will be illustrated. |
