The ingestion of volcanic ash by jet engines is widely recognized as a potentially fatal hazard for aircraft operation. The high temperatures (1,200–2,000 °C) typical of jet engines exacerbate the impact of ash by provoking its melting and sticking to turbine parts. Estimation of this potential hazard is complicated by the fact that chemical composition, which affects the temperature at which volcanic ash becomes liquid, can vary widely amongst volcanoes. Here, based on experiments, we parameterize ash behaviour and develop a model to predict melting and sticking conditions for its global compositional range. The results of our experiments confirm that the common use of sand or dust proxy is wholly inadequate for the prediction of the behaviour of volcanic ash, leading to overestimates of sticking temperature and thus severe underestimates of the thermal hazard. Our model can be used to assess the deposition probability of volcanic ash in jet engines.

Volcanic eruption models are hampered by the lack of multiphase magmatic flow laws. Most rheological models estimate the viscosity of multiphase lavas via the Einstein-Roscoe equation, but this simplification cannot be used for high crystallinity and it does not consider the non-Newtonian strain-rate dependence of viscosity. We carried out parallel plate experiments on natural samples to simulate multiphase lava deformation under various stresses and strain rates. Multiphase lavas exhibit an important component of shear thinning, and appear to invalidate the adequacy of Einstein-Roscoe–based formulations for highly crystalline lava rheology. The remarkable singular dependence of viscosity (η) on strain rate (γ) yields a novel universal rheology law at eruptive temperatures (T), i.e., log η = −0.993 + 8974/T −0.543·log γ Our work reveals the importance of considering microcracking and viscous dissipation at very high strain rate (>10⁻³ s⁻¹), explaining the occurrence of seismic swarms along the conduit margins, and consequently supporting plug-like magma ascent models.

Viscosities of fourteen melts close to the join SiO₂-NaAlO₂ were measured in the range 1–10¹² Pa·s (700–1650 °C) using a combination of concentric cylinder and micropenetration techniques. These compositions cover five isopleths in silica content from 50 to 82 mol% and vary from mildly peralkaline to mildly peraluminous. Greatly improved constraints on the temperature dependence of viscosity in the system SiO₂-NaAlO₂ result because exactly the same compositions were used for both high- and low-temperature measurements, viscosities over an extended range of silica contents were measured at temperatures close the glass transition, and several compositions at constant silica content and variable alkali/Al ratio were measured, allowing interpolation of data to compositions exactly along the join SiO₂-NaAlO₂. At high temperature (1600 °C) viscosity and activation energy are shown to be approximately a linear function of silica content, but large nonlinearities occur at temperatures close to the glass transition range. Defining fragility as the gradient of the viscosity curve at the glass transition temperature (T_g taken to be the 10¹² Pa·s isokom) on a reduced temperature scale (T_g/T), it is found that the fragility increases in a nonlinear fashion as NaAlO₂ is substituted for SiO₂, with fragility increasing more rapidly at lower SiO₂ contents. The viscosity data are combined with heat capacity data available in the literature to estimate configurational entropies of albite, jadeite, and nepheline glasses using the Adam-Gibbs theory. Fragility, when defined in terms of the Adam-Gibbs parameters, is shown to increase with configurational heat capacity (difference in heat capacity between the liquid and the glassy states) but to decrease with increasing configurational entropy at the glass transition. In the light of independent phase equilibria and spectroscopic and calorimetric evidence that suggests the Al-Si ordering increases as silica content decreases from SiO₂ to nepheline, the modeling of configurational entropy in terms of Al-Si mixing suggests the following: (1) Melt configurational entropy has contributions from both cation mixing (chemical contribution), as well as variations in the topology of the O network (topological contribution), of which the latter dominates. (2) The chemical contribution is due to mixing of tetrahedral rather than O sites. (3) At the glass transition (10¹² Pa·s isokom) the topological contribution shows little, if any, variation.