A recent blast-induced landslide in a quarry in Bosnia Herzegovina displacing the water in a lake was captured in a dramatic video —apparently with no fatalities due to lucky circumstances. It didn't take long for my colleagues to ask me if we could model this with Abaqus. I posted a quick and dirty first pass result on LinkedIn yesterday with a simple rigid shell geometry (S3) and an Eulerian mesh where the material flows through stationary elements (EC3D8R) in Abaqus/Explicit. I'm using Equation-of-State material (*EoS in Abaqus) for both water (taken straight from the Example Problems "Water sloshing in a baffled tank" but units in kg/m/s, and, with high viscosity, the failure volume. Eulerian materials include Mohr-Coulomb, but that leads to smaller time steps and higher CPU time. 

A few comments on why my model did not compare too well to this particular real-life scenario:

1. The slide volume is too small (you'll see some added material on top), and 
2. The model water level is too deep compared to the shallow riverbed. Still, if we want to analyse these events more seriously we can certainly put some more effort in.

Also, addressing questions from colleagues:

1. My first model only had water (lake) plus water (failure), I needed a refresher on how to have two materials in one domain
2. The first geometry (circular) was too shallow and did relatively little splash
3. Due to some shortcuts with geometry, the water volume wasn't precise in the edges (settles, becomes dynamic from both sides)
4. The first dam was too high (or the water level too low) 
5. A wedge failure without a bottom step dived into the water like the bow of a catamaran, very little splash
6. I tried mohr-coulomb but that took forever (smaller time increments for solving constitutive equations) - no patience and no value for a demo
7. Waiting for gravity to kick in took too long for the failure to move, so I added initial velocity with vector along the wedge

I took the opportunity to change the element size from 2.5m to 1.6m, resulting in an increase in elements from under 1.5M to over 5M, and reproduce the analysis. The above video is absolutely beautiful in its detail. It is worth keeping in mind that modelling complex scenarios is difficult and much time can be saved starting small and simple and learning from the model iterations.

References:

Abaqus > Analysis > Analysis Techniques > Eulerian Analysis

Landslide video: https://www.linkedin.com/feed/update/urn:li:activity:7182792016538284032/

Aftermath images: https://www.index.hr/vijesti/clanak/foto-ovako-izgleda-odron-kod-mostara-tri-auta-zavrsila-u-jezeru-dvoje-ozlijedjenih/2554180.aspx