Instituto de Matemática, Universidade Federal do Rio Grande do Sul,

Abstract The paper presents some foundations for the application of model-based systems technology within a decision support system for drinking water treatment plants. This system aims at detecting deviations from normal plant operation, identifying their possible causes, and proposing adequate remedial interventions. Its basis is a library of model fragments that represent intended processes, disturbances, and possible interventions. Based on these fragments and the available observations, possible models of the disturbed plant behavior are generated automatically as a solution to the diagnostic task. An extension of such a model by models of interventions such that the result is consistent with remedial goals represents a possible therapy proposal. We discuss in more detail how the model and consistency-based problem solving can be exploited in the basic reasoning steps.