We measure efficiency as a percent = the ratio (power out:power in). The two main players are the electrolyzers and the fuel cell. The effect of other parasitic loads (such as fans) are basically trivial in comparison.

The two Hogan-GC electrolyzers are about 18% efficient: They consume a total of about 1.2 kW, and produce (we think- we do not have a precise measurement) 1.2 liters/min of H2 (stp). Thus in one hour they consume 1.2 kWh, and produce 72 litres of H2 (stp). 72 x 3.5 Wh/L = 0.25 kW of power. However, the electrolyzers have a one hour warm-up period in which they use full power but produce no H2, so over an 8 hour day their average hourly output is 12% lower. Thus we get a power in: power out ratio of (1200: 250*0.88) = (1200:220) ≈ 18% efficiency.

On the return trip, from hydrogen to electricity, the PEM fuel cell produces heat and electricity in a ratio of roughly 50:40 (with 10% internal parasitic losses), so its electrical efficiency (as opposed to total efficiency including heat) is about 40%. This is standard for PEM’s (it is more efficient than most internal combustion engines).

We ignore the inefficiencies of DC to AC inverting (about a 5% loss, but as high as 10-20% for small loads), since it is common to all off-the-grid systems. That said, our electrolyzers run on AC, which adds an unnecessary loss compared to a DC electrolyzer.

To a first approximation, our electrical efficiency (power out: power in) = 18%* 40%= 0.18* 0.4 ≈ 0.072, or 7%. That is, 14 watts of power at the PV array will give 1 watt of power at the plug. This is shockingly low compared to a battery bank, which is about 80% efficient. Newer gasoline-driven generators are about 30% efficient, not counting losses from “well to wheel”.

In our particular installation this low efficiency does not matter. The keys are a) our sporadic use of the house; and b) long exposure to free solar power, thanks to a remote operation system. Our PV array can give the electrolyzers 7-8 kWh per day. Thus in roughly 2 days we can produce enough H2 to yield 1 kWh effective power (ie at the plug), and our tank fills in 60 days. This matches our pre-project calculations (!). We can tolerate the long fill time because the house is only used sporadically and because the remote operation mode allows us to generate hydrogen when no-one is home. This ensures that the tank is full when we show up and need it (see Remote Operation).

However, a full-time residence could not tolerate such a slow fill rate. It would need a bigger electrolyzer. Since PEM electrolyzers seem to have internal economies of scale, trading out our desk-top models for a larger unit would increase electrolyzer efficiency to 30- 40% as well as ditching the warm-up period. This would double overall system efficiency to about 14%.

The only way to get the PEM fuel cell to yield better efficiencies is to incorporate cogeneration, i.e. to use the heat output. We did not pursue this. Ballard has a PEM/ cogen project in Tokyo. While cogeneration raises the total efficiency, the fuel cell’s electrical efficiency remains around 40%.

In conclusion, our efficiency is about 7%, which is very low but works in our case because of our remote operation mode. A full-time residence would need a larger system. A larger electrolyzer raises overall efficiency to 14%. While 14% is a lower figure than traditional generators offer (again not counting “well to wheel” losses), sunlight costs nothing and is pollution-free. However, these advantages do not count in competition with batteries. Telecoms choose fuel cells over batteries because they find that for large-scale energy storage hydrogen is, in the long run, cheaper and more reliable.