It should be noted that many of these times and the numbers of personnel exposed or infected are statistical in nature themselves and would vary from scenario to scenario. For example, there is likely to be significant variability in time t3 for inferring an attack by recognizing groups of infected individuals. As shown in the box at the bottom of the figure, one pass through the simulation will result in a deployment delay time that is one of the four possibilities. Running the simulation many times then gives a distribution of deployment delay times and an indication of where bottlenecks exist in restoring operational capability.
This example has concentrated on a single parameter as the goal—minimizing deployment delays. Other end-goal parameters, such as minimizing the number of casualties, could also be simulated with only minor modifications. One could also explore coupled goals, such as minimizing deployment delays and minimizing casualties. Because these two factors are probably interrelated, some technique such as multiattribute decision theory is needed to combine them. For example, if decision makers can weight the relative importance of reduced casualties versus reduced deployment delays, a slightly modified simulation could be used which has this weighted combination of factors as the end goal.