Source-sink induced chaotic advection for enhanced groundwater remediation

  • J. van der Horst

Student thesis: Master


Efficient mixing of pollutant-consuming bacteria and polluted groundwater can potentially enhance natural remediation of groundwater. This could provide a faster and more cost-efficient method than currently applied remediation techniques. As turbulence in groundwater flow is absent, efficient mixing can only be achieved by producing chaos in the flow. Chaotic advection may result from time-dependent flow induced by sources and sinks.Three different source-sink systems are investigated using numerical simulation of particle trajectories: the pulsed source-sink system, the switched dipole system and the rotated dipole system. First, a system consisting of a sequentially pulsed source and sink was investigated both in two dimensions (2D) and in three dimensions (3D). In 2D this system can be described by a single dimensionless parameter ?, based on the strength and the separation distance of the source and sink as well as the pulse time. The efficiency of chaotic mixing was explored by determining the location and nature of the periodic points: elliptic points are the centers of non-mixing regions, while hyperbolic points are the centers of stretching and folding in the flow (essential to efficient mixing). The system showed chaotic dynamics for all parameter settings investigated. Periodic points were identified up to order six and were arranged along branch-like structures with higher-order periodic points generally located further away from the source and sink. When ? is decreased, the periodic points move towards the source and sink and some point are destroyed. In addition, small ? leads to an increase in the surface area of elliptic islands. Of all the investigated parameter settings, thesystem with ? = 10.4 is likely the best mixer. In the 3D case, symmetry reduces the ow dynamics to 2D planes and the observed dynamics is similar to the 2D case.The effects of dispersion on the pulsed source-sink system are investigated by the application of a small perturbation to the velocity field. This method has been used in previous studies to simulate the effects of inertia in chaotic systems and might also be used for dispersion. The perturbed system shows the breakup of elliptic orbits and the change of elliptic orbits into partially overlapping bands. As no comparative material is available, experimental verification of these results is needed.In addition to the pulsed source-sink system, two systems of sequentially reoriented dipolesare studied. The switched dipole system and the rotated dipole system. Both these systems could be described by a single dimensionless variable ? and showed confinement of material within a circle connecting the sources and sinks. As ? decreases more elliptic islands appear in the switched dipole system. The case ? = 0.9 showed no elliptic islands. Many more hyperbolic period-1 points were found for the rotated dipole system compared with the switched dipole system. This suggests that the rotated dipole system mixes better, at the cost of taking twice the time and more surface area of elliptic islands. Of all the investigated cases, the case ? = 1.26 is likely the optimal setting for mixing.The possibilities of constructing a translucent laboratory model for the groundwater systemare investigated for future validation of the numerical findings. Borosilicate glass spheres with a diameter of 9 mm were placed in a mixture of glycerol with water. This setup was able to make single spheres virtually invisible. When testing a setup consisting of multiple spheres, distortions were observed that significantly reduced the transparency of the setup. These distortions were attributed to optical inhomogeneities within the glass spheres, also observed with shadowgraph measurements. Temperature treatment of the spheres and better control of the lighting conditions resulted in a significant reduction of the observed distortions.
Date of Award31 Aug 2013
Original languageEnglish
SupervisorRuben R. Trieling (Supervisor 1) & G.J.F. van Heijst (Supervisor 2)

Cite this