Numerical Evaluation of Porosity Band Formation as a Mechanism for Melt Focusing at Mid-Ocean Ridges

Abstract

It has been well established through various experimental and numerical studies that imposing an external shear on a system of partial melt will result in the compaction of the solid matrix and expulsion of the interstitial liquid melt; this leads to the formation of regions of contrasting high and low porosity that are commonly referred to as melt bands. An early numerical study of melt bands speculated that these structures could contribute to melt extraction at mid-ocean ridges. This thesis examines the formation of melt bands beneath mid-ocean ridges. With linear and nonlinear models similar to those from previous numerical melt band studies, melt bands are evaluated as a mechanism for lateral melt extraction using the shear geometry derived from the velocity field of the plate-driven corner flow of a mid-ocean ridge. The degree of similarity between previous numerical and experimental results has been found to be greatly influenced by the imposed rheology of the solid matrix phase. Knowing this, the numerical models in this contribution will use three different matrix shear viscosity laws: isotropic strain rate independent, isotropic strain rate dependent and anisotropic strain rate independent. The linear analysis indicates that though fast growing bands may be oriented toward the ridge axis, the bands that undergo the greatest change in porosity over time are oriented toward the lithosphere-asthenosphere boundary at the base of the plate. The nonlinear simulations produce bands with orientations similar to those found in the linear analysis, along with a great deal of unexpected porosity accumulation present on the boundaries of the model domain. These models indicate that melt bands will not likely act as high permeability melt-channeling conduits except near the lithosphere-asthenosphere boundary at the base of the plate. These models are highly sensitive to the poorly defined matrix bulk viscosity, with an increased bulk viscosity resulting in little or no development of significant band-like structures. The models are also sensitive to the initial heterogeneity, which, like the bulk viscosity, is also poorly defined; too great an initial heterogeneity results in the bands quickly surpassing the disaggregation limit for mantle rocks.