Comparative study of large scale simulation of underground explosions inalluvium and in fractured granite using stochastic characterization

Thursday, 18 December 2014
Oleg Vorobiev1, Souheil M Ezzedine2, Tarabay Antoun1 and Lew Glenn1, (1)Lawrence Livermore National Laboratory, Livermore, CA, United States, (2)Univ California LLNL, Livermore, CA, United States
This work describes a methodology used for large scale modeling of wave propagation from
underground explosions conducted at the Nevada Test Site (NTS) in two different geological settings:
fractured granitic rock mass and in alluvium deposition. We show that the discrete nature of rock
masses as well as the spatial variability of the fabric of alluvium is very important to understand ground
motions induced by underground explosions. In order to build a credible conceptual model of the
subsurface we integrated the geological, geomechanical and geophysical characterizations conducted
during recent test at the NTS as well as historical data from the characterization during the underground
nuclear test conducted at the NTS. Because detailed site characterization is limited, expensive and, in
some instances, impossible we have numerically investigated the effects of the characterization gaps on
the overall response of the system. We performed several computational studies to identify the key
important geologic features specific to fractured media mainly the joints; and those specific for
alluvium porous media mainly the spatial variability of geological alluvium facies characterized by
their variances and their integral scales. We have also explored common key features to both geological
environments such as saturation and topography and assess which characteristics affect the most the
ground motion in the near-field and in the far-field. Stochastic representation of these features based on
the field characterizations have been implemented in Geodyn and GeodynL hydrocodes. Both codes
were used to guide site characterization efforts in order to provide the essential data to the modeling
community. We validate our computational results by comparing the measured and computed ground
motion at various ranges.

This work performed under the auspices of the U.S. Department of Energy by Lawrence Livermore
National Laboratory under Contract DE-AC52-07NA27344.