Prevention of explosive mixtures in electrolyser plants' water recovery tanks

The safety problems posed by the formation of explosive oxygen-hydrogen mixtures in electrolysers is well known, and other authors have already done significant research on the effects of hydrogen permeation into the anodic oxygen stream in PEM electrolysers. In this study, the safety risk from other potential sources of explosive mixtures are highlighted, i.e. the recovery of cathodic, hydrogen-laced water and its mixing with anodic oxygen-containing water in the water-deionisation system.

As per the flowsheet of a commercial electrolyser unit, to avoid wastage of demineralised water, it is necessary to recover the water dragged to the cathode along with protons by electro-osmosis. This cathodic water is therefore sent to a buffer tank where it is mixed with make-up water from the demineraliser and water from the anodic circuit. As the two water flows are in equilibrium with different gases (resp. hydrogen, air, oxygen), they will eventually generate an explosive mixture in the tank’s gas phase. This phenomenon is confirmed by steady-state COFE process simulations. Dynamic simulations of the tank indicate that an explosive mixture can be formed as early as 8 minutes from start-up.

With typical equipment sizes, the formation of such an explosive mixture can result in detonation with a production of shrapnel, which may be lethal to operators exposed to it. There is however a low probability of ignition since the metal tanks are virtual Faraday cages and are always fully humidified; nonetheless, this may make the problem difficult to detect and to mitigate the consequence if an ignition should occur.

In this work, several options to remedy such a design error are evaluated by retrofit, as listed below:

1.      Rejecting the hydrogen-laced cathodic water: this solved the problem at its root, however, it requires a tenfold increase in the production of demineralised water, with a significant expense.

2.      Boiling hydrogen off by heating cathodic water: this is physically infeasible as hydrogen’s solubility in water does not change significantly in the available temperature range.

3.      Stripping hydrogen from the cathodic water with counterflow gas: this is a feasible alternative, however, it requires the installation of a new unit in the system. Additionally, as the most readily available gas in the plant is oxygen, it is likely to be used as stripping gas, which may result in the development of a small explosive gas phase at some step in the stripping column.

4.      Purging the buffer tank’s gas phase by oscillating its liquid level dynamically: this technique has the advantage of being implemented “over the air” as a software fix and does not strictly require a physical retrofit site visit.

5.      Integrating the buffer and anodic drain tanks to greatly increase the prevalence of oxygen in the gas phase, forcing hydrogen concentration under the explosive threshold: while promising, this option may not work as well at part load.

6.      Flushing the buffer tank with gas is a straightforward solution, however, to avoid acquiring large quantities of inert nitrogen (with increased costs for both acquisition and operation), it would be preferrable to use oxygen to dilute the hydrogen below the explosive threshold. Oxygen production in the same electrolyser unit can easily satisfy the demand.

7.      Recycling cathodic water directly to the anode separator: this approach redirects cathodic water from the drain and buffer tanks to the anode separator, leaving it in the pressurised side of the plant. As the flow is never brought to atmospheric pressure, the hydrogen content of cathodic water will be much larger. However, as it will be diluted in an even larger oxygen-laced flow, the gas phase will not generate an explosive mixture. However, the higher hydrogen content in the anode side will lead to a narrower operating range, as critical hydrogen concentration will be reached at a higher part load.

In conclusion, there are several workable options available to retrofit an existing electrolyser plant as described above. Most options require site visits and changes to the process flow, and some also require the addition of smaller units. At least one of the presented options is software-implementable.