Dependability Modeling & Analysis

Often a straightforward reliability prediction based on one or other physical models is not applicable. This is true for more complex systems that may include one or more of the following complications:

• Redundant configurations (hot, cold, warm or any combination)

• Degraded states (rather than only a fully operational state and a completely failed state)

• Recoverable items (e.g. computer systems)

• Complex utility and reward functions attached to different system states

SoHaR has extensive experience at modeling systems with varying complexities, from Nuclear Power Plant tripping systems to online purchasing networks. In these cases we model reliability and availability using a variety of tools:

Reliability Block Diagrams: these can treat various redundant schemes and their solution is often analytic. In complex situations a simulation is used to solve for the reliability and/or availability of the system

Markov Models and Petri Nets: These can treat any complex redundant scheme and also take into account various states including degraded states (which may not effect availability but do affect a failure rate), maintenance states (which do not effect the failure rate but do effect availability), recovery states (effect availability). Complications arising from cascading events (which effect each other) can also be treated with Petri Nets

Simulation: Simulations are used to solve the above modeling methods but also for situations where the utility of a state or the reward associated with it are a more complex function (of time, conditions, mission profile and so on).