Experiments on conduit flow and eruption behavior of basaltic volcanic eruptions

Abstract

Multiphase flow in basaltic volcanic conduits is investigated using analog experiments and theoretical approaches. Depending on gas supply, large gas bubbles (gas slugs) may rise through basaltic magma in regimes of distinct fluid‐dynamical behavior: ascent of single slugs, supplied slugs fed from the gas source during ascent, and periodic slug flow. An annular flow regime commences at the highest gas supply rates. A first set of experiments demonstrates that the growth of gas slugs due to hydrostatic decompression does not affect their ascent velocity and that excess pressure in the slugs remain negligible. The applicability of theoretical formulae describing slug ascent velocity as a function of liquid and conduit properties is evaluated in a second set of experiments. A third set of experiments with continuous gas supply into a cylindrical conduit are scaled to basaltic conditions over Morton, Eotvös, Reynolds, and Froude numbers. Gas flow rate and liquid viscosity are varied over the whole range of flow regimes to observe flow dynamics and to measure gas and liquid eruption rates. Foam generation by slug bursting at the surface and partial slug disruption by wake turbulence can modify the bubble content and size distribution of the magma. At the transition from slug to annular flow, when the liquid bridges between the gas slugs disappear, pressure at the conduit entrance drops by ∼60% from the hydrostatic value to the dynamic‐flow resistance of the annular flow, which may trigger further degassing in a stored magma to maintain the annular flow regime until the gas supply is exhausted and the eruption ends abruptly. Magma discharge may also terminate when magma ascent is hindered by wall friction in long volcanic conduits and the annular gas flow erodes all magma from the conduit. Supplied slugs are found to reach much higher rise velocities than unsupplied slugs and to collapse to turbulent annular flow upon bursting at the surface. A fourth set of experiments uses a conduit partially blocked by built‐in obstacles providing traps for gas pockets. Once gas pockets are filled, rising gas slugs deform but remain intact as they move around obstacles without coalescence or significant velocity changes. Bursting of bubbles coalescing with trapped gas pockets causes pressure signals at least 3 orders of magnitude more powerful than gas pocket oscillation induced by passing liquid. Our experiments suggest a refined classification of Strombolian and Hawaiian eruptions according to time‐dependant behavior into sporadically pulsating lava fountains (driven by stochastic rise of single slugs), periodically pulsating lava fountains (resulting from slug flow), and quasi‐steady lava fountains (oscillating at the frequency of annular‐flow turbulence).