OK, first batch of results:
I monitored for two weeks, the first was without a timer, the second was with a timer set to turn the fridge off between midnight and 6am. Here is the raw data:
- Week 1: 3.758 kWh over 166.8 hours,
- Week 2: 3.700 kWh over 168 hours.
The first averages out to 540.7 Wh per day (24h), the second 528.6 Wh per day, which is 97.8% of the first. Not much at first glance, but I noticed that the timer itself uses a constant 1W, so the second figure should actually be 3.700-0.168=3.532 kWh/week for the fridge only (without timer), giving a daily average of 504.6 Wh, which is 93.3% of the first week's average.
So, by turning the refrigerator off for 6h at night it uses 6.7% less energy, but since the timer is drawing 1W the improvement is only 2.2%. Finding a timer with lower draw, or manually unplugging the fridge, would be better energywise.
Either way this not much in my case, since our fridge is already very efficient. In fact, its superior insulation means that it gains much less heat during those 6 hours than most other refrigerators. The 6.7% theoretical (2.2% practical) gain could probably be doubled if the fridge was turned off during the day, but in our case there is almost always someone at home, so it is not an option for us. But killing it whenever no-one is home is a good idea, since this is also when the house's thermostat is set to 1°C cooler, further reducing the fridge's heat gain. Maybe the refrigerator's power supply could be somehow tied to the t-stat's settings? Something like "Turn fridge off 1-2h after going into lower temperature mode, turn fridge on 1h before going into higher temp mode".
Unfortunately, "coasting" your fridge doesn't work during the summer months, since the house temperature is 2°-10°C higher then. But then, heating season here is 8 months long
Anyway, I'm continuing the test, the timer is off and will be back on a week from now.