Simulation stands for imitating the operation of a real life system with the help of a computer. First a mathematical model (fault tree, consequence tree, reliability block diagram) is created from the system which describes the real life system to the computer as clearly as possible. If the relations in the model are simple enough its operation can be examined with adequate accuracy using mathematical analysis like for example locating minimal cut sets. Often this is though not the case and e.g. XOR or NOT relations make accurate analysis impossible and simulation is needed.
When simulating fault trees the events are places in a time-related order. If the realization of event A affects the realization of event B, event A has to be before event B in the time-related order. Because loops are not allowed the time-related order can be formed every time. During a single simulation round each event will be gone through in a time-related order and defined if the event realizes or not. First the gate condition of the event will be checked. If it realizes, a random number is generated and used to check if the probability condition realizes. If both of these conditions realize also the event will realize. The information about the event realizing or not is called the event state.
When all of the events have been defined their realization, one simulation round has been performed. The event states gotten from the simulation will be saved after each round. When multiple simulation rounds are performed, it is possible to get statistical information about the system behaviour. These results can be used to find out the probabilities for event realizations.
It is also possible to find out conditional probabilities, in other words realization of events when some condition is realized. The condition can be the realization or unrealization of an event or an event group. Similar conditional probabilities can be stated also for event combinations. Combination stands for the combo of multiple event states.