This paper presents a new approach for modeling a discrete-event systems under the presence of faults. Input/Output automata are used to represent the nominal and faulty behavior of a plant for fault detection and identification. Assuming that a model of the nominal behavior of a plant is available, it is usually a nontrivial task to derive a model of the plant subject to actuator, sensor and system internal faults. This task often require some expertise or a large amount of data on the faulty behavior of the plant. The key aspect of the paper is the introduction of suitable functions enabling to reduce the amount of information needed to model a faulty behavior. The method presented here fits in the nondeterministic Input/Ouput automata framework where actuator, sensor and internal system faults are considered. The proposed approach is applied to a chemical plant for demonstration.