The Dynamic Flowgraph Methodology (DFM)

Introduction to DFM

The DFM approach is based on representing a system under analysis with a digraph (directed graph) model. The digraph model explicitly identifies the cause-and-effect and timing relationships between the parameters and states that are best suited to describe the system behavior.

Once such a model has been produced, it can be analyzed in a variety of ways using automated deductive or inductive algorithms built into the methodology can. The deductive procedures are applied to identify how system states -- which may represent specific success or failure conditions of interest -- can be produced by combinations and sequences of basic component states. Conversely, inductive procedures can be applied to the same model to determine how a particular combination of basic component states can produce various possible event sequences and subsequent system-level states. Thus, DFM can provide the multi-state and time-dependent equivalent of both fault tree analysis (FTA) and failure modes and effects analysis (FMEA), with the advantage that a single DFM system model contains all the information necessary for the automated execution of these analyses for practically any system condition of interest.

This can be compared, for example, with the execution of FTA, in which each system top event requires a separate manual analysis and the construction of a separate fault tree model. A similar comparison can be made between FMEA and DFM. In performing a failure modes and effects analysis, the causality relationship in the system has to be revisited for each analysis to deduce the effects of different failure modes. In the DFM framework, on the other hand,once a model has been developed, the automatic inductive analysis algorithm can produce an entire array of separate automated analyses, to show how any initially hypothesized component failure may progress through the system, without further reasoning inputs from the analyst. Moreover, this inductive algorithm can even automatically handle cases in which the failure modes may branch into different areas of the system, with separate effects that recombine later in interactions further downstream in the flow of system cause and effect.