Exploring Adaptive Mesh Refinement strategies for dynamic earthquake rupture modeling with ExaHyPE

Understanding how faults slip during an earthquake is still a grand challenge in seismology.

group Group
traffic Project status

Project status

contact_support Contact
Contact name
Kenneth Duru
Contact email
Contact number

Content navigation


Understanding how faults slip during an earthquake is still a grand challenge in seismology.
Dynamic propagation of shear ruptures on a frictional interface in an elastic solid can be a useful idealization of natural earthquakes. These physics based models are complicated, and can only be solved numerically on the computer. While frictional models are often laboratory based, with laboratory scale parameters, the state-of-the-art dynamic earthquake rupture simulation softwares require up-scaled frictional parameters in order to be able to resolve numerical solutions. Thus there is a hiatus between laboratory scale frictional parameters and feasibly large parameters required by computational models.

The objective of this project is to bridge the gap between laboratory experiments and computational simulation, using adaptive mesh refinement (AMR) strategies.

In this project we explore the AMR capabilities of the ExaHyPE code (www.exahype.eu). Exahype is a European project with strong international collaboration, including scientific computing and even astrophysics and aims to enable high performance computing on next generation (exascale) machines.

We will explore the debated question of self-similarity for small and large earthquakes. Observations suggest earthquakes are self-similar, which means stress drop and rupture velocity are independent of scale, and that slip and fracture energy both increase linearly with rupture dimension. In contrast, standard friction models (slip-weakening and rate-and-state), at least with constant slip-weakening or state-evolution distance, have constant fracture energy.

Further comments: This work includes running simulations with ExaHyPe on a supercomputer. No prior knowledge of that is required.



MSI Fellow