ENGINEERING AND CFD

A research team of the Institute of Simulation Techniques and Scientific Computing of the University of Siegen leveraged the petascale computing power of HPC system Hornet for their research project Ion Transport by Convection and Diffusion. This large-scale simulation project stressed the available capacities of the HLRS supercomputer to its full extent as the simulation involved the simultaneous consideration of multiple effects like flow through a complex geometry, mass transport due to diffusion, and electrodynamic forces. Goal of this project was to achieve a better understanding of the electrodialysis desalination process in order to identify methods and possibilities of how to optimize it.

Key Facts:

• 94.080 compute cores
• 5 machine hours
• 1.1 TB of data
• 76.48 million degrees of freedom (DOF)
• 1,341,879,902 grid points
• 20,000 time steps

Seawater desalination can be a major step towards a save supply with fresh potable water, which is one of the big issues mankind faces today. A method promising little power consumption on large scale desalination is the electrodialysis. However, this process is difficult to assess due to the many physical phenomena that need to be considered.

For the electrodialysis, water is pumped through thin channels between selective membranes and an external electric field is applied to drive the salt ions through those membranes. By stacking these channels, a configuration is achieved where in one channel ions are removed from the water and accumulated in the other one. Driving factors for the energy consumption are here the pressure loss over the channel length, but also the efficiency of ion transport towards the membranes. These are conflicting goals, and the ideal balance needs to be found for low power consumption.

Other restrictions arise from dead water regions where fouling might occur, which needs to be avoided for long operational periods. Besides the flow of the water, the interaction of ions needs to be taken into account as well as the electro-dynamic fields driving them apart.

Simulations considering all these effects simultaneously allow researchers a deeper assessment of the desalination process and help to find more efficient strategies for the technical deployment. A region of special interest is the boundary layer in the flow, which requires a high resolution in comparison to the overall setup. Resolutions on the nanometer scale are desired for this area, while the channel height itself is in the region of millimetres.

The high spatial resolution also requires a high temporal resolution, leading to long simulation time due to the many time steps that need to be performed sequentially. To achieve a simulation even close to the desired resolution, huge amounts of main memory are required. Only the access to the full Hornet system of HLRS enabled such a highly resolved simulation of this setting. Evaluation of the data gained from the simulation is still ongoing, however some first results could already be obtained.

By considering the overall setup, new surprising facts about the dynamics of the ion concentration within the channel were gained. In addition, a model for the design of the channel geometry had been deduced from the analysis, enabled by the high fidelity simulation (see figs. 2 through 4). This helps process engineers to develop new and more efficient devices and benefits the global society by improving the desalination techniques.

Project Team and Scientific Contact:

Dipl.-Ing. Harald Klimach, M.Sc. Kannan Masilamani, Prof. Dr.-Ing. Sabine Roller (PI)Institut für Simulationstechnik & Wissenschaftliches Rechnen, Universität Siegen
Hölderlinstr. 3, D-57076 Siegen/Germany
e-mail: harald.klimach@uni-siegen.de
www.mb.uni-siegen.de/sts/