Turbulent pyroclastic density currents - a numerical and large-scale experimental investigation

Pyroclastic density currents (PDCs) represent one of the most dangerous geological phenomena and volcanologists are faced with the challenge of forecasting PDCs hazards in future eruptions. Due to their transient, unpredictable nature, as well as high temperature and extreme velocities, internal workings of PDCs are relatively poorly known because we lack in situ measurements of their internal properties. The work described here uses both numerical modeling and large-scale experiments to probe these internal processes. We focus on large-scale experiments as they are of sufficient energy to display a range gas-particle coupling regimes thought to occur in natural PDC.

Here we present the results from an international effort to create a benchmark for dense and dilute PDCs via large-scale experiments and use this to validate and compare existing multiphase codes for PDCs. During the 2016-2017 years, large-scale experiments at the eruption simulator PELE in New Zealand have been undertaken. The experiments synthesized ground-hugging, hot turbulent flows of natural pyroclasts and air by column collapse onto an instrumented runout section. We generated a range of PDC behavior that scale dynamically and kinematically to natural dilute PDCs. Similar conditions are simulated numerically using multiphase approach. These simulations use a modified version of the MFIX code (e.g. Benage et al., 2016) computing granular stresses and using a Large Eddy Simulation (LES) approach for turbulence closure as part of a modified version of the MFIX code. In experiments, more than 200 sensors record time-series data of flow kinematics and geometry, deposition, vertical stratification of velocity, concentration, temperature, and particle-size distribution. Data are analyzed and simplified to define a coherent set of initial and boundary conditions that can be used in a well-defined numerical benchmark. Here, we compare 3D numerical simulations with benchmark experiments to discuss how well space- and time-variant entrainment of ambient air into PDCs, thermal evolution, turbulence spectrum, velocity and concentration fields are reproduced by state-of-the-art numerical PDC models. We particularly examine the coupling between turbulent and non-turbulent regions of the flow and the flux of material across these domains.