Design criteria of a structure reflect the requirements this structure is expected to satisfy with respect to its function and operation as well as with respect to its strength and stability under the most adverse environmental conditions that might reasonably be expected during the period of its existence. The formulation of such criteria involves therefore prediction of expected load patterns and intensities of the acting forces which, by its very nature, must be made in the face of the uncertainty characteristic of any prediction based on extrapolation from past experience or previous observations. When, as in the case of wave action on maritime structures, the uncertainty arises from the random nature of the phenomenon producing the acting forces, it can be rationally dealt with by probabilistic reasoning and the application of statistical methods: any conceivable condition of the structure is associated with a numerical measure of the probability of its occurrence. This measure is either estimated from the relative frequency of a particular condition in series of repeated observations of the random phenomenon or evaluated from a physically relevant stochastic model. The probability distribution of extreme wave heights can be obtained either by extrapolation from such observations with the aid of the statistical theory of extremes,l or by constructing extreme waves from power-spectral densities of certain models of storm waves.2 Thus, when a probability level is arbitrarily selected, the associated design criterion is obtained as that load pattern or load intensity that is not exceeded with the probability specified.
The "significant wave" which is widely used in the conventional design of maritime structures,3 is not associated with a probability of exceedance since it is specified as the mean of all waves of probability of exceedance of 1/3. It can therefore not be directly used as a design criterion in a probability-based design unless its associated probability of being exceeded is ascertained under suitable assumptions and is found to be sufficiently low to qualify the significant wave as limiting loading condition. An analysis of the state of the sea indicates that this is not the case; therefore the "significant wave" does not, in general, constitute of sufficiently long recurrence period.
By specifying a longer recurrence period associated with a much lower probability of exceedance, a higher design wave is obtained. However, such as approach to the establishment of design criteria based on the specification of an arbitrarily low probability of exceedance suffers from the shortcoming that there is not intrinsic significance to any particular probability of failure, since no a priori justification can be provided for the specification of any quantitative (low) probability level in preference to any other, unless such specification results from a comparison with other hazards that might arise, as might be the case, for instance, for structures for which statistics of other hazards in operation exists.
For maritime structure such hazards might arise from earthquakes or, in certain locations, from the action of ice. In the absence of such comparative hazards other criteria must be formulated with the aid of which the arbitrariness in the specification of a failure probability with respect to wave action can be resolved.