The comparison of the calculated failure probabilities of technical systems with the observed failure frequencies is an important part of the verification of probabilistic calculations. It can point to significant factors that are neglected in the calculations.

In recent years, the failure probabilities and failure frequencies of nuclear power plants, bridge structures and dams have been compared. It was shown that the calculated values correspond very well with the observed values. In addition, various factors could be identified which have both positive and negative influences on the observed values, but the sum of which almost cancel each other out.

Within the scope of this contribution, the comparison for tunnels shall be carried out. Extensive statistics are available for tunnel collapses during construction. In contrast to the other technical products, the collapse statistics of tunnels are characterized by the production. This effect is also visible to a certain extent in bridges, but not to the extent of tunnels. The collapse statistics for tunnels are mainly dominated by design and construction errors. Accidental effects account for only about 20 % of all failures. This is significantly less than, for example, with bridges or nuclear power plants.

Increasingly, tunnels are also being proven probabilistically. Based on these calculations, the available failure statistics can be compared with representative probabilistic tunnel proofs. The results of this comparison are presented in the article.