Numerical simulation of sediment transport by thermohaline bottom currents
Ocean Sciences, Computational fluid dynamics
Research area
This project is focused on the study of the sediment transport and deposition mechanisms on the sea bottom.
In particular, it concerns the interaction between two types of currents flowing on the sea floor: turbidity (downslope) and contour (along slope) flows.
Turbidity flows are dense currents where mixtures of sediment and water travel downslope because they are denser than the surrounding water.
The contour currents are driven by thermohaline density gradients and forced by the Coriolis effect to travel along the continental slope within the geostrophic oceanic currents.
Project goals
The project aims to reproduce, through numerical experiments, two typical seabed currents involved in sedimentation and erosion processes: turbidity and gravity currents. Turbidity currents are dense bottom flows generated by the down-slope movement of sediments, which are transported along the continental slope to the deep sea. Gravity currents are horizontal flows driven by small density differences between fluids, often caused by variations in salinity and temperature. The project aims to study these phenomena using larger computational domains and higher Reynolds numbers, reducing computational costs through the use of wall functions to better approach real seafloor conditions.
Computational approach
The numerical approach is based on the Large eddy Simulations, so on the filtered Navier Stokes Equation, to which was added a salinity/sediment transport equation. To extend affordable case studies to higher Reynolds numbers, averaged approaches can be used. The URANS method reduces computational costs by averaging the Navier-Stokes equations over time, but at the expense of accuracy. A better compromise is LES, which directly resolves larger turbulent structures while modeling smaller dissipative scales. However, near-wall regions remain challenging, as resolving them requires DNS-level mesh resolution, making high-Re simulations computationally expensive. To address this, WF-LES combines LES with Wall Functions, which model near-wall turbulence effects without requiring fine mesh resolution. This method assumes that the first computational cell near the wall contains multiple eddies, allowing an averaged approach similar to RANS. Using Wall Functions significantly reduces computational costs while maintaining reasonable accuracy. In some cases, mesh resolution requirements can be reduced by an order of magnitude, from 10^7 to 10^6 cells. So the first step is to validate this approach and then study cases with Reynolds number next to the real scale phenomenon.
Image
Density fields, a confrontation between a fully resolved LES simulation (top) and two WF-LES (center and bottom).
Key results
The main objective was to characterize the bottom boundary layer beneath gravity currents and to assess how turbulence and buoyancy affect near-wall velocity profiles, wall shear stress, turbulent fluctuations, and current propagation. High-resolution LES, were performed at environmentally relevant Reynolds numbers, allowing detailed analysis of turbulence statistics, velocity scaling, and wall stress behavior. Results reported in Ammendola et al. (2025) show that the near-wall region can be accurately modeled using wall functions, reducing mesh resolution from about 40 million to about 5 million cells and lowering computational time from about 5 days to about 1 hour, while maintaining predictive accuracy.
Resource usage
The used workflow was tested a lock-exchange configuration in a simple channel with a height, width, and length of 0.4 m, 0.2 m, and 4 m, respectively. The simulations were performed using OpenFOAM® on 8 nodes with 44 tasks per node, for a total of 320 processors. The estimated total computational requirement for the project was 100,000 CPU-hours. Each simulation run is expected to use 320 cores, with an average wall-time of approximately two days per job. Regarding storage, the simulations are expected to require approximately 2 TB per node, in addition to 500 GB of shared and persistent storage for data management and post-processing. The expected duration of access to the computing resources is six months.
What's next
To advance the project, we will progressively test increasingly complex computational domains in order to reach a configuration that allows us to analyze the interaction and behavior of the two described currents under cross-flow conditions. To achieve this objective, it will likely be necessary to employ greater computational resources for each simulation. In particular, an increase in the mesh resolution will significantly affect the computational cost. For instance, increasing the mesh size from approximately 18 million cells per simulation to about 40 million cells more than doubles the computational demand, raising the simulation time from about 2 days to approximately 5 days per case.
Image
Stream-wise Velocity Fluctuations at the domain bottom, 26 seconds of simulation, case dimensions 0.4 m (Height) x 0.2 m (Width) x 4 m (Length).
Antonio Ammendola
University of Trieste; OGS
Antonio Ammendola is an environmental and civil engineer specializing in computational fluid dynamics, microclimatic monitoring, and numerical modeling. He holds a Master’s degree in Environmental Engineering from the University of Trieste and a Bachelor’s in Civil Engineering from the University of Reggio Calabria. Currently, he is a PhD candidate at the University of Trieste, researching sediment transport in thermoaline bottom currents. Previously, he worked as a research fellow at Politecnico di Milano, focusing on environmental monitoring of cultural heritage and CFD modeling of convective flows. He has also gained experience in energy efficiency analysis and hydraulic device testing. Antonio has organized academic workshops and holds certifications in workplace safety and English (C1). His technical skills include OpenFOAM, ANSYS Fluent and statistical analysis. His research interests cover computational and environmental fluid dynamics, with several publications in these fields.