Discussion: Seismic ground-motion computations at low probabilities

Abstract

The annual of exceedance used in the study are 0.01,0.0021, and 0.0001. The first two corresponding respectively to the regionalization parameters of the 1970 and 1985 probability maps of the National Building Code of Canada. As mentioned elsewhere (Ferahian 1985), the latter m a p u s i n g 10% probability of exceedance in 50 years-was based on the same model referred to in CSA Standard CAN3-N289.2-M8 1, Ground-motion determination for qualification of CANDU nuclear power as an example for determination of the design basis ground motion (DBSGM). This is a relaxation from its former definition using annual probability of exceedance 0.00 1. In their abstract, the authors state: The provisions for the National Building Code of Canada are based on assumptions that source zones can be defined within which earthquakes are distributed uniformly according to a magnitudelfrequency-of-occurrence relation and that a set of attenuation relationships hold for the earthquake magnitudes and distances that dominate the selected risk level of 10% probability of exceedance in 50 years. These assumptions appear to be reasonable at the probability level used for the average structures included in the code. However, attempts of extending the calculations to significantly lower levels of probability for special structures may invalidate the original assumptions, because the dominant contribution ranges of magnitudes and distances are outside of the intended scope of the ground-motion relations. Moreover, the basic assumption that earthquakes are uniformly distributed over large source zones is seriously compromised. Surely, nuclear power plants are special'-and are far from being averagew-structures and the discusser is happy to note the authors' conclusions corroborate his concerns on the adequacy of the design considerations of our nuclear power plants, as recorded since 1978 with some of the original reference given elsewhere (Ferahian 1985). Moreover, the discusser has criticized with much concern the current design rules of these plants and the relaxation of their former design stipulations, an aspect of which was mentioned above. The purpose of seismic ground-motion computations is to estimate these motions to be used in design. Thus it is curious that the authors state that they are concerned with the actual levels of ground motion that are associated with these probabilities (p. 595). This omission frees the authors from commenting on the confidence limits associated with their estimates. How do these confidence limits vary with the return period of the ground motion, especially when the return period is greater than the length of the sample? Certainly, without knowing these confidence limits and the error margin, the results for the 0.0001 annual probability of exceedance are meaningless. Moreover, by this omission, the authors do not present information available from their study to enable the reader evaluate the adequacy of the actual ground accelerations used in the design of these special-nonaverage-structures of which nuclear power plants are the most important example. This is also evident in the authors' choice of only the western Quebec zone (WQU), Fig. 2, for eastern Canada-thus avoiding giving estimates for the contiguous zones, namely EBG and NAP, in which all of Canada's commercial nuclear power plants are situated. The authors are to be congratulated for their paper but they are exhorted to comment on the confidence limits inherent in their estimates and also give results including actual ground accelerations for the zones in which the Canadian nuclear power plants are sited together with the results for the 0.001 annual probability of exceedance, formerly used in the definition of the DBSGM. It is a pity not to present the full import of their study by these omissions.