Understanding the formation and emplacement of magmatic intrusions is fundamental to comprehend the dynamics of volcanisms. Dykes and sills forms the primary path for the magma to be transported from its mantellic source through the lithosphere, eventually reaching the surface to feed volcanic eruptions. The geometry of these intrusions controls the volume of magma carried through this “plumbing system” which in turn affects the concurrency and intensity of volcanic eruptions. However, the mechanisms that controls the formation of this plumbing system are poorly understood and difficult to model as their results from a complex intertwined between multiple processes such as the rheology of the flowing magma, the nature of the solid hostrock, and the damage/faulting associated to deformation of these magmatic intrusions.
The development of multi-purpose commercial FE software in the last two decades has offered geoscientists a wide range of tools to solve complex Multiphysics problems such as the formation of magmatic intrusions, where the flowing magma, the temperature field and the deformation of the solid host rock are inherently entangled. In this study, we would like to focus on one example of advanced Multiphysics problems.
In this example, we will model the evolution in time of an inflating laccolith embedded in an inelastic host-rock. The initial set-up of the model is defined by a (vertical) feeding dyke connected to a (horizontal) sill at 500 m depth. The magma is injected at the bottom of the feeding dyke and accumulate in the sill that inflates under pressure build up. After the injection phase, the magma cools down until it reaches its solidus temperature after which the laccolith is essentially frozen.
The computation time of such model (even in 2D) is very important as it include 4 different types of physics,, with parameters that depend on each other, and a highly defined mesh structure to describe the flowing magma. A simple coarse, 2D model performed with COMSOL on a normal work station takes about 10 days to be fully resolved. We hope to reduce the computation time by using the Uppmax cluster resources.