Assessment of the long-term seismicity of Athens from two classical columns

On the south flank of the Acropolis of Athens there are two 10-m high columns belonging to the monument of Thrasyllos, free-standing since the third century BC. Obviously they have survived earthquakes for more than 2,300 years responding as natural seismoscopes with their damage 'recording' all strong earthquakes during their long life. Their existence naturally raises the questions of whether their mere survival can contribute to our scientific understanding of the long-term seismicity of Old Athens by addressing the various interrelated aspects of the problem combining earth science, seismology, archaeology, history, and engineering. Starting from the literary survey and the seismotectonic investigation of the region we find that although earthquakes were in the City, rarely causing concern, there is no evidence that Old Athens was ever seriously damaged. However, the historical literary record is incomplete having gaps at long periods of time; the columns, therefore, may be used to enhance our knowledge on the long-term seismicity, not only by estimating the worst-case ground shaking that could topple them-which has not happened yet-but in particular to assess the degree of shaking which was responsible for the damage that can be seen today on them. The decoding of this information is not a straightforward procedure and, in general, does not lead to a unique solution. The main reasons for this is on one hand the non-linearity and the sensitivity of the seismic response of multidrum columns, and on the other the lack of information on the evolution of the damage and the uncertainty on the real cause of it: it is not easy to determine how much of the damage to the columns that can be seen today must be attributed to earthquakes alone and how much of it is due to other reasons. In this research, from the detailed investigation of the current state of the columns, the damage caused by earthquakes was separated from the one caused by rusting of dowels and temperature effects. Back analyses were performed using suitable surrogate ground motions, aiming at determining the worst-case earthquake that could cause the collapse of the columns, and the maximum level of shaking that could cause the observed damage. The numerical analyses were performed using the Distinct Element Method and models that were exact reproductions of the columns based on the available data, while the joint properties were calibrated against ambient and forced vibration measurements. The results are consistent with the seismotectonic analysis and show that the mean response spectrum of the worst-case earthquakes is generally compatible with the response spectrum of EC 8 for the corresponding return period. It is possible, though, that the maximum experienced shaking is less. More information is needed on the real effect of the dowels of the drums to draw conclusions on the latter based on the observed damage.