Juliane Dannberg 1 , Timo Heister 2 1 Texas A&M University, 2 - PowerPoint PPT Presentation

Juliane Dannberg 1 , Timo Heister 2 1 Texas A&M University, 2 Clemson University

MANTLE CONVECTION cooling buoyancy- driven flow heating From Hong Kong Geological Survey

MANTLE PLUMES mass extinction magmatism thermal anomaly mantle convection heat transport From Griffiths and Modified from Campbell, 1990 Putirka et al., 2011

MANTLE CONVECTION: : MELTING 3 2 1 From Hong Kong Geological Survey

MELT MIG IGRATION Subduction zones 2 Mid-ocean ridges Fractures/channels/diapirs 1 3 Katz, 2008 Wilson et al, 2014  Studies only in 2D / simplified Keller et al, 2013  Mantle convection and melt migration studied separately

VARIABLES solid density shear viscosity compaction viscosity fluid/ Darcy melt coefficient

EQUATIONS K D ~ φ 3 η , ξ = f( φ )

EQUATIONS Introduce compaction pressure (Keller at al, 2013):

EQUATIONS IN IN ASPECT

EQUATIONS IN IN ASPECT

CHALLENGES  Time and length scales of melt migration are vastly different from mantle convection  Highly non-linear and spatially variable material properties  Difficult to study in 3D  Adaptive mesh refinement  Link melt generation to processes in the deeper mantle (komatiites or other melts originating in greater depths)  Compressibility

SO SOFTWARE: ASP SPECT Advanced mathematical techniques:  Higher order time stepping schemes (BDF2)  Higher order finite elements  Fully adaptive, dynamically changing 3d meshes  Nonlinear solvers  Parallelization using MPI, threads, and tasks Community code:  Modular  Extensive documentation  Extensive and frequent testing

MODULARITY Aspect is very modular: It is extended by a number of isolated “ plugin ” sub-systems: Gravity Geometry Public interface Termination Initial cond. Mesh refinement Aspect core Boundary cond. Postprocessing Material model Visualization

DEAL.I .II FEATURES Finite element meshes, manifolds Gravity Geometry Checkpoint / Restart Public interface Termination Initial cond. Aspect core Mesh Higher order finite refinement elements, Adaptive meshes in 2D & 3D, Boundary cond. several components refinement indicators Postprocessing Flexible: Solver (AMG), no slip, free slip, Interfaces to PETSC, Trilinos traction, Dirichlet, Material model Visualization periodic Parallel, scaling up to Write output in common 10,000s of cores visualization file formats, parallel I/O

CONVERGENCE WIT ITH ADAPTIVE MESH Porosity Fluid pressure Velocity

CONVERGENCE WIT ITH ADAPTIVE MESH

SH SHEAR BANDS Boundary velocity Boundary velocity

SH SHEAR BANDS: : ANGLE

SH SHEAR BANDS IN IN 3D Total wallclock time: 1.7 hours 130 time steps; t=3.125e-08 years Number of degrees of freedom: 44,855,815

MANTLE PLUME WIT ITH MELT MIGRATION Wallclock time: 7 days 14200 time steps; t=152,516 years Degrees of freedom: 6,243,260

MANTLE PLUME WIT ITH MELT

MANTLE PLUME WIT ITH MELT

3D PLUME: : WORK IN IN PROGRESS

APPLICATIONS

CONCLUSIONS  Open source code for coupled magma / mantle dynamics  Uses modern numerical methods  Accurate  Fast, scalable  Well tested, well documented  Designed to be easily extended  Has been successfully applied to several application cases  On different scales  In three dimensions  With compressibility of the individual phases

ASP SPECT TEAM