A fault propagation modeling technique using Petri nets is outlined in this paper, when a system performs a phased mission with a different success criterion in each phase. Component failure effects are obtained by monitoring the deviations in the system process variables, when single or multiple failures are introduced. The system behavior is modeled using Petri nets due to their flexibility to address dynamic effects, multiple failure modes and different system behavior in each phase of the mission. The paper indicates how a fault diagnostic system can be developed after investigating fault propagation results. The application of the fault propagation methodology is illustrated using a simple tank level control example. System sensors can be allocated based on the amount of information that they bring about the state of the system.