Design of cost-optimized village-scale electrodialysis systems for brackish water desalination

Abstract

This thesis proposes methods of reducing the cost of electrodialysis brackish water desalination systems, specifically for use in rural India, where 60% of the groundwater is too saline to drink. Convergence of socioeconomic and technical factors led to the insight that photovoltaic (PV) powered electrodialysis (ED) has the potential for impact in rural water treatment. In order to design a system that can meet the necessary production requirements, a robust parametric model was created to predict the desalination rate, limiting current density, and total energy use in an ED system. The model agrees with experimental measurements across two diverse ED stack designs, differing in total membrane area, membrane manufacturers, and flow channel spacers. A commercial-scale ED stack was additionally tested in Chelluru, India, building confidence that the model is predictive for real groundwater, and that ED systems are feasible to operate in the rural Indian context.

The ED model was used within an optimization routine to determine the lowest cost operating mode and stack design, assuming existing, flat-stack architectures. Common operating modes including constant-voltage batch and multi-stage continuous systems were considered alongside novel operation modes including voltage-regulated batch and hybrid batch-continuous systems. For the production and desalination rates required for a village-scale application, a voltage-regulated hybrid system that is fully optimized for membrane width, length, and channel thickness reduces the 10-year total cost and capital cost of the system by 37% and 47%, respectively, in comparison to a commercially available stack optimized under the same operation modes. While matching of applied and limiting current densities can be achieved using a voltage-regulated batch operation (minimizing stack cost), this requires a potentially costly DC power supply and control system.

The final part of the thesis proposes a spiral ED stack architecture that allows for matching through the geometry of the stack alone. Both a standard Archimedean spiral and an ideal irregular spiral shape are presented. The ideal spiral shape would reduce the 10-year total cost and capital cost by 21% and 39%, respectively, in comparison to the Archimedean spiral, and is cost-competitive with a hybrid voltage-regulated flat-stack design.