We present here a novel approach to assess paleo strong ground motions from dynamic back analysis of key block displacements in a Roman arch and collapse of columns in a Byzantine church using the numerical discontinuous deformation analysis method. The obtained threshold values for paleo peak horizontal ground acceleration are utilized as an independent validity check for H-PGA predictions recommended by the local seismic building codes. Preliminary results from two different sites provide a good agreement between the H-PGA estimates arrived at by the two independent approaches.
Visitors to archeological sites all over the world are often struck by structural-failure features which seem to be "seismically driven", particularly when inspecting old masonry structures. While it is widely accepted that no other loading mechanism can explain the preserved damage, the actual driving mechanism remains enigmatic even now. In this paper we show how such failures may be triggered by earthquake motions and use observed block displacements to determine the characteristic parameters of the earthquake, namely duration, frequency, and peak horizontal ground acceleration (H-PGA). We study two archeological sites located in close proximity to the seismically active Dead Sea transform: a two-millennia old Nabatean arch in which the keystone slid down during an earthquake of an un certain date (Figure 1), and a 1400 year old Byzantine church in which a series of parallel granite and marble columns toppled down in the same direction most probably due to an earthquake which struck the region in 749 AD (Figure 2). We begin with two novel validations of the numerical discrete element method used in this research (DDA) to test its adequacy for our purposes: 1) we test the dynamic response of a flat passive block resting on an underlying block subjected to harmonic sinusoidal oscillations. The dynamic response of the passive overlaying block is studied as a function of input frequency, amplitude, and interface friction and the numerical results are compared with an analytical solution developed by Kamai and Hatzor ; 2) we test the dynamic response of a slender column resting on a passive pedestal and subjected to a dynamic pulse in the form of half a sine, complete sine, and a series of three sines, and the numerical results are compared with an existing analytical solution . After obtaining good agreement between numerical and analytical solutions we proceed with forward modeling of the actual problem geometries. By performing comprehensive sensitivity analyses we try to mimic the observed deformation modes in the field so as to come up with the "best fit" solution which includes input motion amplitude, frequency and duration. Our method does not include site effects and the results are relevant to the studied deformed blocks at their respective location in the analyzed masonry structures. We find that PGA values of 0.5 g frequencies of 1 - 1.5 Hz best fit the observed deformation at the analyzed arch. The best fit solution for the toppled columns exhibits great sensitivity to input frequency: for 1 Hz and 2 Hz PGA values of 0.2 g and 0.6 g are obtained, respectively.