Washing Machine Energy Estimations
Our washing machine has a display which purports to indicate how energy-intensive the currently selected programme configuration is. It does not display an actual estimation of energy consumption, but rather it indicates a linear scale from one to five.
I wanted to find out what this scale translates to in actual energy units, so I put one of those energy-measuring things between the washing machine and the outlet and recorded the results from a few different programme configurations1 I used a sort of theoretic sampling method for this, were I only paid attention to the energy consumption number when I had run a programme configuration I thought would enrich the analysis.. Here are the six results we got, ordered by indicated energy usage:
| Configurations | F | A | B | C | D | E |
|---|---|---|---|---|---|---|
| Energy indication | 1 | 3 | 3 | 4 | 5 | 5 |
| Energy consumption (kWh) | 0.08 | 0.39 | 0.03 | 1.88 | 0.55 | 0.46 |
It’s clear the energy indication is not an indication of how much energy is actually used. Maybe the energy indication is more of a power indication? If we divide the energy consumption by time, we get the average power consumed during the programme.
| Configurations | F | A | B | C | D | E |
|---|---|---|---|---|---|---|
| Energy indication | 1 | 3 | 3 | 4 | 5 | 5 |
| Avg. power (W) | 100 | 490 | 100 | 510 | 550 | 460 |
This sort of makes sense, but seems too noisy to be a useful guide to energy savings. 100 W can be estimated as either one or three – and a five can in practise use less power than a three!
Figuring out a better indicator
What draws power in a washing machine is probably (a) heating water, and (b) running a motor. So to figure out a better estimation method, we can include configured temperature and centrifuge speed in the table, as well as how long the machine thinks the configuration will run for.
The difference between configurations D and E isn’t shown in the table, but in E the “Eco mode” was activated.2 And it did indeed reduce the average power draw significantly, by almost 100 W, with no other change. I wonder what it does!
| Configurations | A | B | C | D | E | F |
|---|---|---|---|---|---|---|
| Temperature (°C) | 30 | — | 60 | 40 | 40 | — |
| Time (hours) | 0.8 | 0.3 | 3.7 | 1 | 1 | 0.8 |
| Centrifuge speed (rpm) | 400 | 400 | 1200 | 1200 | 1200 | 1200 |
| Avg. power (W) | 490 | 100 | 510 | 550 | 460 | 100 |
This is not enough data to do any quantiative analysis, but we can see fairly clearly that
- Configurations that don’t heat any water run essentially like a couple of lightbulbs, regardless of the centrifuge speed.
- As soon as any heat is applied at all, we’re up in half-a-kilowatt territorry, regardless of the other configurations. What ultimately determines energy usage at that point is how long the program is running for.
So we can set up our own estimation rule that might do better than the one-to-five scale the machine gives us:
If the configuration does not heat water, the full programme will consume 0.05 kWh. If a temperature is set, it will consume 0.5 kWh per hour of running time.
That’s it! These predictions have an mean absolute deviation of 30 Wh from the true value, while the machine’s estimation (however it is supposed to be interpreted) seems to have a mean absolute deviation of at best 100 Wh.