Computational simulation of laboratory-scale volcanic jets

Volcanic eruptions produce ash clouds that may travel great distances, significantly impacting aviation and communities downwind. Atmospheric hazard forecasting relies partly on numerical models of the flow physics, which incorporate data from eruption observations and analogue laboratory tests. As numerical tools continue to increase in complexity, they must be validated to fine-tune their effectiveness. Since eruptions are relatively infrequent and challenging to observe in great detail, analogue experiments can provide important insights into expected behavior over a wide range of input conditions. Unfortunately, laboratory-scale jets cannot easily attain the high Reynolds numbers ( 109) of natural volcanic eruption columns. Comparisons between the computational models and analogue experiments can help bridge this gap. In this study, we investigate a 3-D volcanic plume model, the Active Tracer High-resolution Atmospheric Model (ATHAM), which has been used to simulate a variety of eruptions. However, it has not been previously validated using laboratory-scale data. We conducted numerical simulations of three flows that we have studied in the laboratory: a vertical jet in a quiescent environment, a vertical jet in horizontal cross flow, and a particle-laden jet. We considered Reynolds numbers from 10,000 to 50,000, jet-to-cross flow velocity ratios of 2 to 10, and particle mass loadings of up to 25% of the exit mass flow rate. Vertical jet simulations produce Gaussian velocity profiles in the near exit region by 3 diameters downstream, matching the mean experimental profiles. Simulations of air entrainment are of the correct order of magnitude, but they show decreasing entrainment with vertical distance from the vent. Cross flow simulations reproduce experimental trajectories for the jet centerline initially, although confinement appears to impact the response later. Particle-laden simulations display minimal variation in concentration profiles between cases with different mass loadings and size distributions, indicating that differences in particle behavior may not be evident at this laboratory scale.