Introduction: the ''critical seismic excitation'' paradigm Nonlinear time-history analysis using real and synthetic earthquake acceleration records (accelerograms) as input action is the most physically consistent and versatile tool for seismic analysis of structures. However, at present, it is mostly used for illustration and research purposes, not for practical design. High level of uncertainty associated with the earthquake phenomenon undermines credibility of the analysis in a design case, where the goal is to assure survival of the structure under any possible seismic actions on the site-the ones that have occurred in the past and the ones that have not, but may occur in the future. The latter part presents the most difficulties.
The discusser wholly agrees with the author of the discussed paper [1] that ''it is essential to develop robust methods for seismic-resistant design of structures''. The ''critical seismic excitation'' (CSE) paradigm relies only on those few parameters of the earthquake that can be predicted with greater certainty, such as the maximum seismic magnitude expressed in terms of energy input, or the peak ground acceleration (PGA). For all other parameters, such as the spectral content of the earthquake, it abstains from making assumptions that would reduce the response of the structure, because these assumptions may not materialize in a future earthquake. Rather, it assumes that these parameters are at their most unfavourable values for the structural response (within certain limitations); hence the term ''critical seismic excitation''. Thus, the CSE paradigm shifts the emphasis from refinement of the parameters of seismic action to avoidance of dangerous dependency on the accuracy of these parameters.
Comments on the proposed approach and technique
2.1. The limitations on the critical seismic action
The discussed paper [1] points out that the CSE is necessarily resonant unless some artificial limitations are imposed on the Fourier spectrum. These limitations, as well as the ''entropy'' limitation used in earlier papers of the same author [2,3], attempt to attain a ''rich frequency content'' of the model critical earthquake, approaching that of a real earthquake.