PhD Defence Catarina Simões | Advances in Reverse Electrodialysis for Renewable Energy Generation

Advances in Reverse Electrodialysis for Renewable Energy Generation

Catarina Simões is a PhD student in the department Sustainable Process Technology. (Co)Supervisors are prof.dr.ir. D.W.F. Brilman from the faculty of Science & Technology and dr. M. Saakes from Wetsus.

Salinity gradient energy (SGE) is the energy released from the mixing of two solutions with different salinity, such as seawater and river water. SGE is a renewable and sustainable energy source that integrates the hydrological cycle, having an important role in the transition to renewable and clean energies without CO₂ emissions. Reverse electrodialysis (RED) is the electro-membrane process used in this thesis to harvest SGE. In RED, ions migrate through ion exchange membranes (IEMs), selective to either anions or cations, from the seawater compartment to the river water compartment due to a salinity gradient between these solutions. The concentration gradient across the IEMs results in a potential difference over each membrane. The potential difference generated, when connecting an external load to the two electrodes (one at each end of the stack’s membrane pile), drives an ionic current through the stack and an electrical current through the external circuit. Usually, the ionic current is converted at the electrodes to an electrical current using a redox couple. This thesis aims to optimize and scale up the RED process for energy generation using seawater and river water.

Chapter 2 demonstrates the implementation of electrode segmentation to strategically optimise the output power density and energy efficiency of RED. Using a validated RED model and investigating experimentally the RED process with a scaled-up cross-flow stack of 0.22 x 0.22 m², it was possible to achieve higher performance results with a four-electrode segments stack than with a stack with unsegmented electrodes or single electrodes. The highest overall yield concerning gross power production was obtained by “saving the gradient” in the first two segments for the last two segments (along the river water path). Fixing the same net power density (the maximum value achieved with a single electrode was 0.92 W·m^-2), electrode segmentation increased the energy efficiency relatively by 43 %, from 17 % to 25 %. While realising an overall 40 % energy efficiency, the net power density achieved with electrode segmentation was 39 % higher than measured for a single electrode (0.47 W·m^-2). Electrode segmentation allows the current density to be tuned locally, thereby improving the overall process performance without trade-offs. This can contribute to reducing operating costs, with higher energy efficiencies, or to reducing capital investment, with higher net power densities.

Chapter 3 discusses the potential and flexibility of multistage RED. Hereto the validated model from Chapter 2 is exploited, in combination with an experimental study using two electrically independent cross-flow stacks of 0.22 x 0.22 m². These are arranged in series in terms of seawater and river water inflow. The influences of residence time and electrical control were studied experimentally and compared with the model outcome. The model proved to be successful in describing the different flow arrangements and electrical control. It was found that multistage RED yields a higher gross power density and energy efficiency than a single stage. Depending on the power consumption required for pumping, adding an extra stack in series may or may not increase the net power production. This underlines the need for focus on low-pressure drop stack designs.

The electrode pairs of each stage in the multistage arrangement were tuned individually, like with the segmented electrodes. Applying the “saving the gradient” strategy (i.e., lowering the discharge current value in the first stage by consuming less of the salinity gradient) increased the gross overall performance of the two stages by up to 17 % relative to a single-stage and up to 6 % relative to a sequentially optimized two-stage system. Lastly, different multistage configurations were simulated using the model, which revealed that two stages are sufficient for feeding seawater and river water. With a third stage, the improvements achieved would not offset additional pressure drop losses. Multistage RED configuration is increasingly beneficial for higher residence times and higher salinity gradients, preventing voltage losses and brisk salinity changes.

Chapter 4 shows, for the first time, the multistage RED approach using natural waters at the piloting location at the Afsluitdijk, the Netherlands. Two cross-flow stacks of 0.22 x 0.22 m² connected in series, were used, each with a different number of cell pairs, respectively 32 cell pairs for stage 1 and 64 cell pairs for stage 2. The configuration’s performance was evaluated for over 30 days. Natural waters introduced new variables to the process, such as the presence of several divalent ions, conductivity fluctuations and (bio-)fouling. The influence of divalent ions is explained in Chapter 1. A distinct behaviour was found for divalent ions in each stage. While in stage 1, Ca²⁺ and SO₄^2- were transported from the river to the seawater side (uphill transport), in stage 2, no uphill transport occurred. Mg²⁺, another divalent ion present, was not transported against the gradient at any stage. The first stage delivered a gross power density of around 0.60 W·m^-2, and the second stage delivered around 0.25 W·m^-2. The actual total gross power density achieved at the available salinity gradient was stable at values around 0.35-0.40 W∙m^-2. The total net power density, corrected for the initial pressure drop of the stacks, was 0.25 W∙m^-2 at an energy efficiency of 37 %. Throughout the operation, due to increasing stack pressure drop, the actual total net power density lowered finally to 0.10 W∙m^-2. At the end of the experimental campaign, a stack autopsy revealed the presence of microorganisms with sizes ten times larger than the cartridge filter nominal pore size (5 µm) and biofilm covering part of the spacer open area, both contributing to the increased pressure drop in the stacks. Remarkably, it did not affect the achieved electrical gross power density. Pressure drop monitoring, -minimization and fouling control, is confirmed as a crucial element for the success of the technology.

Chapter 5 focuses on the electrode system by using carbon-based slurries to replace the common, less environmentally friendly, redox solutions typically used as the electrode rinse solution. Carbon-based slurry electrodes (CSEs) allow a continuous reverse electrodialysis process in a more clean and sustainable way. At the laboratory scale (0.1 x 0.1 m²), six CSEs made from activated carbon mixed with either carbon black or graphite powder (conductive additives) were characterized both physically and electrochemically and tested in RED operation. The CSEs containing a total of 20 wt% mixture of activated carbon and carbon black showed the best electrical performance, but also the highest viscosities. The use of CSEs made it possible to avoid Faradaic reactions at the anode and cathode and eliminated voltage losses caused by water electrolysis. A continuous test of the best CSE resulted in a stable output for 17 days, using a stack with a single membrane configuration. To show the versatility of CSEs, higher current densities up to 350 A·m^-2 were tested in an electrodialysis setting and were shown to be feasible until current densities of 150 A·m^-2 without abrupt pH changes.

Lastly, Chapter 6 presents an outlook on the potential of reverse electrodialysis as a sustainable renewable energy source. It discusses the process’ achievability accounting for what is currently known and draws future developments. Also, the sustainability of RED is reviewed, explaining what will influence the surrounding environment during the lifetime of a RED facility. And the last point discussed is the development agenda for the technology considering IEMs advances, current energy trends, and automatic control in dynamic environments.