Modelling lava rheology with free surface velocimetry and analogue fluids. (2019)
Type of ContentTheses / Dissertations
Degree NameMaster of Science
PublisherUniversity of Canterbury
AuthorsCusack, Dale Robertshow all
Lava flows are a recognised natural hazard that can extend into inhabited or economically sensitive regions. The ability to forecast lava flow paths is dependent on an understanding of the lava rheology. The aim of this work is to investigate the relationship between rheology and free surface velocity to enable the runout distances and flow paths of active lava flows to be calculated. Lava rheology is complex and flows can be either Newtonian, non-Newtonian or a mixture of both in nature. Lava viscosity is not only temperature dependent but also dependent on bubble and crystal growth.
The approach is to initially simplify the conditions. Isothermal Newtonian and non-Newtonian fluids are used as lava analogues in the initial experimental approach. The fluids are poured down an inclined V channel under gravity and small polystyrene beads are scattered over the free surface. A video of the fluid flow is captured when it is fully developed and in a steady state as it moves down the channel. The free surface streamwise velocity is determined using Streams video processing software and the entrained flow of the polystyrene beads sitting on the fluid’s free surface.
A COMSOL based Navier-Stokes model was developed to simulate the flows that were measured in the experiments. The free surface velocity was used as an input to the COMSOL model where viscosity is iteratively changed to find a match. In both the Newtonian and non- Newtonian fluids, viscosity was determined and compared with a commercial rheometer. The Newtonian fluid’s viscosity, μ was measured to be 61.4 ± 0.5 Pa·s in the viscometer, and calculated to have a mean viscosity of 61 ± 7.8 Pa·s for the model. The non-Newtonian fluid’s viscosity is encompassed in the power law coefficients. For the rheometer they were k = 14 and n = 0.4. The experimental results were k = 21.65 ± 1.15 and n = 0.205 ± 0.015.
It was discovered that the model is very sensitive to the accurate measurement of the independent variables flow height, h which had a measurement error of ± 1.5 mm and inclination angle, β which had a measurement error of ± 0.5°. This leads to a velocity uncertainty ± 1.2 mm/s, and a viscosity variation of ± 7.8 Pa·s.