SGTR Assessment Using MARS

Abstract

During the course of a plant accident, a consistent understanding of the plant response is vital to support an accident manager's decision making process. One tool than can provide assistance to the plant staff in assessing conditions in the plant during accident conditions is the MAAP Accident Response System (MARS) [l] software. During an accident, MARS utilizes the on-line data from the plant instrumentation to initialize the Modular Accident Analysis Program (MAAP) [2] code. Once initialized, MARS tracks and characterizes the plant behavior through the use of integrated logic modules. These logic modules provide the user with important information about the status of systems and the possible cause of the accident. The MARS logic modules evaluate relevant available plant instrumentation and the observations of the operating staff using fuzzv logic. The fuzzy logic is applied to provide a transition between areas where one is absolutely sure that a situation has not occurred to a condition where one is absolutely certain that a situation has occurred. One example of the use of logic modules in MARS is illustrated by that used to assess if a steam generator tube rupture (SGTR) event has occurred. Each piece of relevant plant data is evaluated to determine if it is consistent with the symptoms of a SGTR. Each of the evaluations for the individual plant instruments and the operating staff observations are assembled to determine an overall confidence which characterizes the likelihood that a SGTR is occurring. Additional MARS logic modules are used to determine confidence levels for other types of accident events. The conclusions arrived at by each individual logic module are expressed as confidence levels. The logic module confidence levels can be graphically displayed using the MARS Graphical Users Interface (GUI), to indicate the confidence level MARS has assessed for each accident type. The GUI shows the identification of the possible accident types, but is not limited to true or false identifications of the sequence type. The assessment of the sequence type is important information for the accident manager and is also an essential aspect for the MARS Tracker. This evaluates the accident behavior and continually tests its "understanding" using the MAAP thermal-hydraulic model and the evolving plant data. The tracking of the accident progression by the MARS Tracker enables the system to initiate near term faster than real-time predictions using the MARS Predictors. The MARS Predictors provide the operators with additional insights into the possible future plant states. This paper demonstrates how the MARS software is able to successfully identify and track a simulated SGTR sequence.