Elsevier

Journal of Power Sources

Volume 284, 15 June 2015, Pages 547-553
Journal of Power Sources

Shunt currents in vanadium flow batteries: Measurement, modelling and implications for efficiency

https://doi.org/10.1016/j.jpowsour.2015.03.057Get rights and content

Highlights

  • We investigate a mini flow battery stack with external hydraulic system.

  • We measure the shunt currents directly during battery operation and standby.

  • We model shunt current effects for various flow frame geometries and cell figures.

  • Inner cells discharge faster; outer cells are charged during charge conservation.

Abstract

Shunt currents are an important factor which must be considered when designing a stack for flow batteries. They lead to a reduction of the coulombic efficiency and can cause furthermore a critical warming of the electrolyte. Shunt currents inevitably appear at bypass connections of the hydraulic system between the single cells of a stack.

In this work the shunt currents of a five-celled mini stack of a vanadium flow battery with external hydraulic system and their effects are investigated directly. The external hydraulic system allows the implementation of current sensors for direct measurement of the shunt currents; moreover, the single bypass channels can be interrupted by clamping the tube couplings and with it the shunt currents between the cells when the pumps are off. Thus the shares of losses by cross contamination and by shunt currents are quantified separately by charge conservation measurements.

The experimentally gained data are compared to a shunt current model based on a equivalent circuit diagram and the linear equation system derived from it. Experiments and model data are in good agreement. The effects of shunt currents for different flow frame geometries and number of cells in a stack are simulated and presented in this work.

Introduction

Flow batteries, especially the vanadium system, are regarded as a promising storage technology for the realization of large-scale battery storage systems. The energy converter unit, which is built up from a large number of electro-chemical cells connected in series, forms the main component of this battery. The quality of the design of this converter unit, i.e. the attainable power and efficiency at given material input, is an important parameter for the associated costs (€/kW). An important part of the dimensioning concerns the hydraulic system. It supplies every cell with electrolyte, however, constitutes at the same time an electric connection between the cells through which parasitic currents can flow, the so-called shunt currents.

The problem of shunt currents plays an important role for the designing of stacks for flow batteries. Shunt currents reduce the coulombic efficiency of a flow battery by causing an internal self-discharge: they enable an undesirable run of the discharge reactions at simultaneous ion shift through the bypass connections (that unfavourably close the circuit). Driven by the potential difference between the cells of a stack (and between stacks themselves) currents flow over the conducting electrolyte connections between them, over the electric resistances they constitute, respectively. Cross section, length and state of charge (via viscosity and conductivity) of the fluent electrolyte determine at the same time the hydraulic like the electric resistance of this bypass connections. For example, smaller cross section and longer length of the connection lead to higher resistances and thus reduce the extent of the shunt currents. Ideally little hydraulic resistances are combined with high electric resistances. These opposite trends require a tradeoff, that can only be optimized with detailed understanding of the underlying processes.

There are several publications concerning shunt currents, mostly focused on creating a model for their description rather than experimental investigations. Early work has been done by NASA [1], [2], calculating the manifestation of shunt currents in flow battery stacks and investigating the conflict of shunt losses vs. pumping losses. A report about shunt currents in a vanadium flow battery stack has been given by Ref. [3]. Shunt currents are not limited to single stacks, but also an important loss mechanism in battery systems consisting of several stacks; this matter was modelled by Ref. [4] and more recently by Refs. [5] and [6]. A model for thermal implications of shunt currents especially during stand by periods was published in Ref. [7]. When it comes to quantification of shunt currents, most studies fall back on models and did not measure the losses (shunt currents, self-discharge) directly. Moreover, the models often show only shunt currents in a few chosen configurations and no comprehensive correlations.

Aim of this work is to compile a detailed understanding of shunt currents in a stack of a vanadium flow battery and to derive a complete presentation of shunt current effects for different flow frame geometries and numbers of cells. Therefore, a specific experimental setup was built, consisting of a five-celled mini stack with an external hydraulic system. This external hydraulic system, realized by tubing connections between the cells, allows a direct measurement of the particular shunt currents as well as their interruption. Single cell potentials are monitored at different operating states and the shares of losses by shunt currents and by cross contamination are quantified separately by charge conservation measurements.

The experimentally gained data are compared to a shunt current model. This model is based, like most models found in literature, on an equivalent circuit. Thereof derived linear equation system was implemented in a Mathematica algorithm that allows a dynamic and flexible calculation of implications for various stacks (number of single cells, nature of flow channels) with regard to the shunt currents occurring in them and the expected coulombic efficiencies. The work concludes with a comprehensive sensitivity analysis and the evaluation of the consequences of shunt currents at cell level and as an outlook for stacks of different sizes.

Section snippets

Experimental setup

The investigated stack consists of five cells. The hydraulic connections between the cells are realized by a tube coupling system which lies accessibly outside the flow frames, see Fig. 1. Tube lengths and consequently connecting resistances can thereby be varied. More important is the possibility to switch off shunt currents by clamping the flexible tubing (when pumps are off).

The cells consist of: activated graphite felts (GFD5, SGL Carbon; activated at 400 °C for 20 h), graphite foils as

Electrolyte conductivity

The conductivity of the electrolyte is an important parameter for the magnitude of the connection resistances. In this work the electrolyte solution consists of 1.6 mol/L of vanadium and 2 mol/L of sulphuric acid (state of delivery SOC −50%, see 2.1). The conductivity of the catholyte, the electrolyte solution for the positive electrode, is higher, than that of the anolyte, the electrolyte solution for the negative electrode. This is due to the higher concentration of protons (proton shift and

Conclusions

In this work shunt currents and their effect on the coulombic efficiency were examined directly. A purpose-built stack with external hydraulic system enabled to switch shunt currents on and off. Thus, the contributions to self-discharge could be distinguished between cross contamination and shunt currents. For this, the loss currents were determined by charge conservation measurements. It appeared that an equalizing current can compensate all loss currents temporarily, but the shunt currents

Acknowledgements

This work was funded by the Bavarian Ministry of Economic Affairs and Media, Energy and Technology through the project ZAE-ST (storage technologies).

References (7)

  • F. Xing et al.

    J. Power Sources

    (2011)
  • F. Wandschneider et al.

    J. Power Sources

    (2014)
  • S. König et al.

    J. Power Sources

    (2015)
There are more references available in the full text version of this article.

Cited by (0)

View full text