Benchmark #5a: For at least 1 run, evaluate the model behavior on a specified mesh (available at geoweb.mit.edu/fe). Also investigate other meshes (grid spacing, element types, etc.) as allowed by the code.
Benchmark #5b: Using a mesh of your choice, investigate accuracy as a function of the number of nodes near the fault tip. Also consider variations in how the discrete fault tip is buried within the viscoelastic medium (i.e., # of nodes in transition, transition distance, etc.).
Benchmark #5c: Approximate pinned boundaries at infinite distances from the fault (either using infinite elements or extending the boundaries a far distance from the fault, using mesh grading).
- Benchmark #5a: Evaluate accuracy efficiency of codes as a function of meshing technique and element type.
- Benchmark #5a: Code comparison.
- #5b: Test accuracy of code near fault tip.
- Benchmark #5b: Investigate 'best' methods for 1) discretizing the mesh near the fault tip and 2) approximating the brittle-ductile transition.
- Benchmark #5c: Comparison to semi-analytical solutions (e.g. VISCO1D).
- Model size: 24 km by 24 km by 24 km (0 km ≤ x; y ≤ 24 km; -24 km ≤ z ≤ 0 km)
Top layer: -12 km ≤ z ≤ 0 km; Bottom layer: -24 km ≤ z ≤ -12 km
- Elastic material properties: Poisson solid, G = 30 GPa
- Maxwell viscoelastic material properties:
Top layer: η = 1025 Pa-s (essentially elastic)
Bottom layer: η = 1018 Pa-s
- Density and Gravity: None
- Boundary conditions:
Bottom and all sides except the symmetry plane (y = 0 km) have analytical solution imposed (available atgeoweb.mit.edu/fe)
Side at y = 0 km has 0 y-displacement (i.e., symmetry condition applied)
- Coarse mesh node spacing: dx = dy = dz = 2 km
- Fault specifications:
Type: 45° dipping fault
Location: Top edge at x = 4 km; Bottom edge at x = 20 km
0 km ≤ y ≤ 16 km; -16 km ≤ z ≤ 0 km
- Slip distribution: 1 m of uniform thrust slip (0.707 m in the z-direction and -0.707 m in the x-direction) for 0 km ≤ y ≤ 12 km and -12 km ≤ z ≤ 0 km with a linear taper to 0 slip at y = 16 km and z = -16 km.
Requested Output and Results
Mesh Variations: As memory, time, and patience allow, run models at 1/2, 1/4, and 1/8, etc. the original coarse mesh spacing, investigate variable mesh spacing, and/or employ a variety of element types.
For All Benchmark Variations:
- Stresses and displacements along three lines parallel to the y-axis at 0, 1, and 5 km from the fault plane at the depths of 0, 12, 16, 17, and 21 km (e.g. at the surface x=4, 5, and 9 km, at z = -12, x=16, 17, and 20 km); and three lines parallel to the x-axis at y = 12, 17, and 21, at depths of 0, 12, 16, 17 and 21 below the surface, all results at times of 0, 1, 5 and 10 years.
- CPU time, wallclock time, memory usage info, compiler info, and platform info
Okada will be used to generate an elastic solution. The 'best' viscoelastic answer will be derived via mesh refinement and increasing the distance to the model boundaries. Analytical solutions to the viscoelastic solution are being sought if anyone has information.
For future comparisons involving variations in rheology across the fault, please also build in layer boundaries at z = -6 km on the hanging wall side of the fault and z = -10 km on the footwall side of the fault.