Concentration Gradient Driven Natural Convection in Soluble Lead Redox Flow Batteries

Abstract

Low cost and long cycle-life energy storage systems are needed to harness renewable sources of energy at large scale. Among the options available, redox ow batteries (RFB) offer the maximum potential. The vanadium based RFB offers long cycle life but requires high initial investment and running cost. The membrane-less soluble lead redox ow battery (SLRFB) offers a low cost alternative. Since it uses a common electrolyte for both the electrodes without a proton exchange membrane in-between, it is likely to be easy to operate and maintain. Soluble lead redox ow battery (SLRFB) is currently under development. Our group has earlier established the dominant role concentration gradient driven natural convection ow plays in this electrochemical system. The present work focuses on developing new designs that harness natural convection

flow for effcient charge-discharge operation of a single SLRFB cell. We take fi rst step in this direction by establishing the ability of a model developed in our group to match experimental measurements. The validated model is then used to test new designs for cell and electrodes. The model considers fluid flow, potential eld, electro-deposition and electro-dissolution reactions on electrode surfaces, buildup of deposits, and transport of ionic species under convection, diffusion, and electric fi eld induced migration. The highly coupled physics makes simulations computation intensive, hence 2-d approximation is used. A batch cell with wall mounted electrodes (standard cell) and additional electrolyte above (top) and below (bottom) them is studied for model validation. The simulation results are obtained without re tting any parameters, and are shown to be independent of grid size. The results con rm natural convection induced

electrolyte circulation. The strong circulation predicted on anode compared to cathode is attributed to electric field driven migration of Pb2+ being opposed to diffusional flux on anode and in the same direction on cathode. The cell potential during charge remains constant until the depletion of active material Pb2+ becomes signi cant. During discharge, the deposit pro le on electrode controls the performance. The measurements validate the model predictions quantitatively. The 2-d approximation used in the model, without which a single simulation would take weeks to complete, is tested by varying cell depth. The measurements validate the 2-d approximation made in the model. The model predicted velocity eld is tested against measurements obtained using particle image velocimetry (PIV) technique for the standard wall mounted electrode cell. Reliable measurements are obtained in regions away from electrode walls. The resolution and accuracy of measurements near electrode walls is poor due to blurring of images. The measurements validate the model predicted features reasonably well. A new cell con guration with electrodes mounted o cell walls (lift cell) is studied using simulations and measurements. The new design is inspired by lift and bubble column reactors. The model predicts large scale circulation in lift cell. The velocity fi eld induced between the electrodes is more intense for o - wall compared to on-wall electrodes. The electrolyte circulating through space between wall and electrodes at an order of magnitude smaller velocities impacts cell performance signi cantly. It extends constant potential charge and discharge phases by mixing electrolyte everywhere in the cell. Simulations predict that providing a small gap, as little as 2 mm, above and below the electrodes, it enough to bring about complete mixing in lift cell. A number of measurements made on lift cells validate model predictions quite well. Splitting electrodes and staggering them in lift cell con figuration so that boundary layer starts to grow afresh from lower edge of each segment is seen in simulations to i) decrease charging potential, and ii) increase charging time for the same total current density. The simulations show that addition of an external ow loop allows light electrolyte formed during charge to enter ow loop from the top and dense electrolyte to enter the cell from the bottom to bring about mixing. The opposite happens during discharge. A comparison among the two strategies shows that split electrodes provide better mixing and lowest charging potential for same electrolyte volume and fixed current density. Batch mode of operation is eventually impacted by depletion of active material even for the case of perfect mixing. Continuous feed of active species is necessary for charge and discharge over long periods. The effect of ow rate (from low to moderate) is studied in detail for standard cell using simulations and measurements. The simulations show constant cell potential during charge and discharge for average ow velocity in range of 2 10ð€€€7 to 2 10ð€€€3 m/s. The measurements could be carried in ow velocity range of 6:9 10ð€€€5 to 2 10ð€€€3 m/s. The measurements validate model predictions well. The simulations carried out for forced convection alone show that cell potential during charge and discharge keeps changing with time, similar to diffusion limited response for no ow cases, for average

ow velocity smaller than 4:5 10ð€€€4 m/s. There is an asymmetric response of anode and cathode for the same extent of deposition and dissolution of active material on the two electrodes during charge and discharge in vertical orientation. The measurements carried out for horizontal orientation of electrodes (standard cell) with anode at the bottom in one case and cathode at the bottom in another show signi ficant differences. These differences over short charging phase are unexplained. The measurements point to decreased depletion at long time for anode at the bottom confi guration. The latter is consistent with intense natural convection ow on anode in vertical con figuration. The investigations carried out in the present work establish that the model is able to predict cell performance for a number of new designs introduced here, without any re tting of the model parameters. Some of the predictions for the new designs are also validated experimentally. The model can thus be used for evaluating new designs. Among the tested designs, of-wall electrodes, a small space above and below electrodes to let circulatory ow establish, splitting of electrodes for intense circulation, and unusually weak external flows at low cost for long charge and discharge phase hold promise for improved performance.