2019:15 3D Thermo-Mechanical Coupled Modelling of Thermo-Seismic Response of a Fractured Rock Mass related to the Final Disposal of Spent Nuclear Fuel and Nuclear Waste in Hard Rock

SSM perspective


When assessing the long-term safety of a repository for spent nuclear fuel it is important to consider future earthquakes. Previous studies by Yoon et al. (2014, SSM Report 2014:59) and (2016, SSM Report 2016:23) investigated the fracture responses due to heat and earthquakes at major deformation zones using a 2D thermo-mechanical coupled model that uses a Particle Flow Code (PFC2D, Itasca). The 2D approach limits the investigation to horizontal and vertical cross sections of the Forsmark site and the model is not able to handle interactions of steeply dipping and gently dipping faults. Thus, the 2D approach is not capable to address the question if an occurrence of an earthquake at a gently dipping fault, could reactivate some of the large, steeply dipping faults. To explore thermally and/or seismic induced fracture slip of the repository fracture system, this study was performed with 3D thermo-mechanical coupled modelling using Particle Flow Code 3D v4 (PFC3D v4). Importantly, before the 3D thermo-mechanical coupled modelling could be performed, a 3D geological model of the Forsmark site had to be created representing the repository fracture system and relevant deformation zones.


The aim of the research is to explore 3D earthquake simulation and the heat load from the spent nuclear fuel and its potential impact on the repository fracture system. Three scenarios were investigated:

  • heating scenario, heat load inducing repository fracture slip
  • earthquake scenario, seismic load inducing repository fracture slip
  • earthquake scenario under thermal phase.

The simulated impact of the heat generated from the deposited spent nuclear fuel induces slip in the repository fracture system. Further, the fracture dilation during the heating is irreversible. The scaling parameters of the simulated fault rupture agree well with the earthquake fault empirical scaling relations. Coseismic slip distributions of the activated faults show triangular and asymmetric pattern and sharp increase/drop of displacements at the location of intersection with neighbouring faults. The results show high resemblance to the observations of slip distribution of natural earthquake faults. The results demonstrate that the workflow Yoon et al. developed for simulation of dynamic fault rupture (earthquake event) based on the discrete element model and PFC3D v4 well capture the characteristics of the natural earthquake faults. The performed modelling shows that, under present day stress conditions, the secondary fracture displacement in the repository volume due to seismic loads is more sensitive to the distance where the primary seismic fault occurs than to the size of the primary seismic fault.

The performed modelling also shows that the impact of an earthquake event occurring during the early time of repository heating (50-100 years) amplifies the shear displacements of the repository factures by a factor of 10 to 1000 compared to shear displacements induced by the heat load alone.


Canister failure due to shear load is included in SKB’s main scenario for the safety case for the planned repository for spent nuclear fuel in Forsmark. Thus, it is important to understand possible causes for shear movements in the repository volume and the response of the canister to these movements. This study explores how:

  • repository fracture behaves under the influence of heat loads
  • repository fracture behaves under the influence of seismic loads
  • repository fracture behaves if an earthquake coincides with the heat load.

Therefore, the study is important for SSM’s understanding of shear movements in the repository volume and their impact in the continued review of SKB’s application for a spent nuclear fuel repository in Forsmark.

Need for further research

The PFC3D v4 simulations resulted in unexpected large fracture slip in the repository volume compared to the PFC2D simulations. The reason for this needs to be explored in future research. The thermal modelling can potentially be improved by comparing the PFC3D v4 thermal modelling results with published analytical close form solutions. In addition, the far field response to the heating of the repository should be included in future work in order to examine how the heat is dissipated in the model. Further work is also required in the model to define the damage zone and fault core, which were not done in this study. When mofielling the fault stability during future glacial cycles, further work is also needed to include the impact from fluid pressure on the fault activation.