A quantitative evaluation of a fault tree will establish the likelihood of the undesired event and will say what the relative contributions are to this likelihood from each failure mode. By knowing such a likelihood, it would be possible to decide whether or not corrective action is needed or war­ranted.

In fact, an engineer can make an evaluation of a fault tree, using his experience to discriminate among more and less likely failure modes, and he does indeed decide whether to take corrective action by assessing relative probabilities according to his engineering judgment.

If, however, the probabilities of the different initiating faults are known, it is possible to make a strict numerical fault tree quantization. There are two methods of evaluation, computation and statistics, and each method has two further submethods (15).

(1) Computation:

(a) direct fault tree calculation,

(b) algebraic expression evaluation.

(2) Statistics:

(a) Monte Carlo simulation,

(b) importance sampling simulation.

One can illustrate the use of the two major methods by pointing out that the probability of turning up a 7 from a throw of two dice can be obtained by calculating it or by throwing a pair of dice a large number of times and counting. Here we are interested in the computational methods.