The efficient large-scale recycling of plastic waste is of increasing interest from an ecological and economic point of view but it represents a goal that has yet to be achieved by the recycling industry. European project W2Plastics (FP7) aims at a fundamental change of the present status of plastics recycling by applying the Magnetic Density Separation (MDS) technology as well as the Ultrasound Imaging Technology to develop a separation device for the recycling of polyolefin's from complex wastes, i.e., wastes such as Waste from Electric and Electronic Equipment (WEEE), household waste and Automotive Shredder Residue (ASR).

The sorting of plastics in W2Plastics is based on the use of a magfluid, magnetized water, to stratify the different plastics according to their densities and to collect them at different depths in an efficient way. One component of this project consists in simulating the particle paths into the water flow inside the separation device in order to understand the separation mechanisms and to optimize its configuration.

The separation device is divided into three sections. The plastic particles enter the device with a strong turbulent mixing, necessary for an initial good separation between particles. The first section is the laminator section that tries to make the flow laminar in the shortest distance as possible. The second section is the separator section in which the particles are separated according to their densities, under the magnetic field. These particles are flowing to the last section, the collector section where each kind of plastics should flow at a different heights.

From the numerical point of view, there are three main challenges. First, the Navier-Stokes solver should be robust and fast enough. This is accomplished using parallelized schemes together with efficient algebraic solvers, running on thousands of processors. The second challenge is the plastic particles tracking and their interactions with the flow. This will be carried out using a fixed grid method in order to avoid remeshing at each time step. Finally, the turbulent flow will be captured using a Variational Multiscale Method method, and the inlet turbulence will be treated using a synthetic turbulence generation.