THE SENSITIVITY OF MIXING CONDITIONS ON NON-EXPLOSIVE HYDROMAGMATIC FRAGMENTATION, AS ILLUSTRATED IN A SERIES OF SIMPLE LABORATORY EXPERIMENTS

Fine-scale fragmentation is a characteristic of explosive hydrovolcanic deposits. Yet explosive fuel-coolant interaction (FCI’s) experiments, in which hot melt is mixed with water, have found that the degree of melt fragmentation and violence of mixing vary greatly, even under nearly identical initial conditions (e.g. Berthoud, 2000, Ann. Rev. Fluid Mech, 32:573-611). Reasons for the range in behavior are only partially known, due in part to the difficulty of observing an explosive process. By contrast, non-explosive hydromagmatic fragmentation can be easily observed. We documented such fragmentation under one simple mixing scenario through 25 experiments in which 300 g of fluxed basaltic melt at 1075-1100° C was poured into a water-filled transparent container, and video-recorded at 300-10,000 frames per second (fps). Melt viscosity ranged from 9 to 25 Pa s, pour height above water ranged from 0.12 to 0.5 m, and pour rate (estimated from the time required to empty the crucible) was 6 to 16 g/s. Upon entering the water, the melt stream accumulated in coils and piles that deformed as they sank. After several seconds of cooling, coils broke brittlely under deforming stress. Each breakage was typically followed by a cascade of violent fragmentation bursts near the broken coil end as thermal stress was released. 300-fps video images recorded the movement of some initially stationary fragments more than a half centimeter in successive frames, implying rates of acceleration >450 m/s². Fragments show breakage along conchoidal fractures whose orientation lay at all angles to the melt-stream’s outer surface. Assuming a thermal expansion coefficient of ~10^-5 C^-1 and a temperature variation of ~1000° C across the coil, we estimate axial thermoelastic stresses of tens of megapascals driving fragmentation. Repeat experiments using the same melt temperature and pour rate (to the extent we could control it) yielded samples ranging from nearly intact, unfragmented melt coils, to fragment piles in which almost no intact coil segments could be found. This wide range in results is reminiscent of that from the above-mentioned FCI experiments. In our experiments we infer that the degree of fragmentation is highly sensitive to deformation and failure of melt coils at a time during cooling when thermoelastic stress is greatest.