This paper introduces a new modeling framework for failure analysis (FA) of physical systems namely ??Rate cognitive maps??. The proposed cognitive map based framework provides a robust approach for identifying a large variety of possible failure modes in a dynamic environment through a new cognitive inference process. While this methodology allows the identification of possible failure modes at multiple levels of abstraction, the network like representation of physical systems makes it easier to track the root causes of identified failure modes.