A CRITICAL ASSESSMENT OF PHENOMENOLOGICAL MODELS UNCERTAINTIES FOR TURBIDITY CURRENTS

Abstract. Turbidity currents have significantly contributed to the formation of oil reservoirs through massive transport and deposition of sediments in the offshore area during the past geological era. That motivates the seek for understanding these complex flows composed of carrier and disperse phases. In this regard, numerical simulations can be of great help in understanding the complex underlying physics of those turbulent flows. Two-fluid models allow the explicit consideration of both phases, liquid and solid, where the coupling between them arises from fluid-particle and particle-particle interactions. Simplified approaches, namely standard sediment transport model (SSTM) and partial two-fluid model (PTFM), represent a balance between accuracy and easiness of computation which makes them attractive for different applications. Computational models are built upon employing a Large Eddy Simulation (LES) approach based on the Residual Based Variational Multiscale Method (RBVMS). The scales decomposition used in the RBVMS allow the design of subgrid models, responsible for describing turbulence and interactions involving fine scales that are not captured by the numerical grid, on a purely computational modeling standpoint. Using those computational models on an uncertainty quantification perspective, a number of simulations are performed aiming at assessing the role of phenomenological models as surrogates for the two-fluid models direct interactions in nondilute flows. Uncertainties of those models are embedded into random parameters variables. Different scenarios involving an open channel flow were performed to  make a critical analysis of those submodels when applied to turbidity currents simulations. Keywords: Reduced Model, Uncertainty Quantification, Turbidity Current