Benchmarking Computational Fluid Dynamics Models for Application to Lava Flow Simulations and Hazard Assessment

Abstract:

Recent eruptions in Hawai’i, Iceland, and Cape Verde highlight the need for improved lava flow models for forecasting and hazard assessment. Existing models used for lava flow simulation range in assumptions, complexity, and the degree to which they have been validated against analytical solutions, experiments, and natural observations. In order to assess the capabilities of existing models and test the development of new codes, we conduct a benchmarking study of computational fluid dynamics models for lava flows, including VolcFlow, OpenFOAM, Flow3D, and COMSOL. Using new benchmark scenarios defined in Cordonnier et al. (2015) as a guide, we model Newtonian, Herschel-Bulkley and cooling flows over inclined planes, obstacles, and digital elevation models with a wide range of source conditions. Results are compared to analytical theory, analogue and molten basalt experiments, and measurements from natural lava flows. Our study highlights the strengths and weakness of each code, including accuracy and computational costs, and provides insights regarding code selection. We apply the best-fit codes to simulate the lava flows in Harrat Rahat, a predominately mafic volcanic field in Saudi Arabia. Input parameters are assembled from rheology and volume measurements of past flows using geochemistry, crystallinity, and present-day lidar and photogrammetric digital elevation models. With these data, we use our verified models to reconstruct historic and prehistoric events, in order to assess the hazards posed by lava flows for Harrat Rahat.