Would an EV save me money? Logging two petrol cars to find out

I don't own an electric car, and I want to know whether buying one would save money. Most answers to that question assume an average driver on an average tariff paying an average price for petrol. For one household, none of those are hard to measure, so before looking at a single EV I started measuring the two petrol cars.

What the cars record

Each car has a small ESP32 board that reads the engine computer through the OBD-II port about five times a second. Neither car reports its fuel flow, so the board works the fuel out from the air going into the engine. For every trip it keeps the distance, the fuel, how long the engine ran with the car standing still and how much fuel that idling burned, plus speed, outside temperature, how often the air-con ran and a set of engine-health readings.

The trip waits on the board until the car is back on home Wi-Fi, then goes to Home Assistant, which prices it at what the fuel in the tank cost. The parts for the newer of the two boxes came to about $35 to $60, and nothing runs in the cloud. The build, with code, is on the household stack page.

The fuel figures are calibrated against the pump and the fuel gauge. Before that the boxes read low, by about 6% in one car and 15% in the other. Most of the gap came from working the fuel out from the air, which undercounts on these engines, and the rest was fuel the engine used at moments the boxes counted none. They correct for both now. Until a couple of fills to the click confirm the correction, treat the petrol figures as good to within about 4%.

Where the petrol goes

So far there are three weeks of the commuting car and eight days of the second car: a little over 100 trips, all in a mild Adelaide spring.

• Idling. The engines spent a fifth to a quarter of their running time with the car standing still, and that idling burned about 9% of the fuel in both cars. An EV standing still uses next to nothing apart from its climate control.
• Short trips. More than half the trips were under 5 km. In the commuting car those short trips used about twice the fuel per kilometre of the longer ones, largely because the engine was still warming up. An electric motor is as efficient in the first kilometre as in the twentieth.
• Suburban speeds. No trip averaged more than 70 km/h while moving, and the average moving speed was 44 to 52 km/h. That suits an EV: it recovers energy every time it slows down, and air drag only takes over at highway speeds.

What an EV would cost to run here

The household's power plan has four prices: free from 12:00 to 15:00 for the first 24 kWh a day, about 20c a kWh either side of that (10:00 to 12:00 and 15:00 to 16:00), about 35c overnight (midnight to 6:00) and about 59c at peak (6:00 to 10:00 and 16:00 to midnight).

By EV Database's figures, counted at the battery, the small EVs sold here use 11 to 13 kWh per 100 km around town in mild weather, 15 to 16 in mixed driving and 18 to 21 at a steady 110 km/h. This driving is suburban, so I've used 13. Charging from a normal power point loses another 13 to 15% of the power on the way in, and a wall charger less (ADAC). At 13 kWh plus 15%, 100 km costs:

Charged overnight, an EV would cost about a third of what the second car spends on petrol per kilometre, and charged in the free window it would cost nothing. The commuting car's numbers point the same way.

Every 10,000 km, overnight charging would save about $1,100 in fuel and free charging about $1,630. Turned around, each $1,000 an EV costs over the petrol car it replaces takes about 9,000 km of overnight-charged driving to earn back, or about 6,100 km on free power. That is fuel alone: servicing, tyres, insurance, registration and resale value are outside what the logger can see.

The catch: free power needs the car at home at noon

The free window is the best case, and it only helps if the car is parked at home between 12:00 and 15:00. On a working day the commuting car is at work. In the three weeks logged, it was home for the whole window on 8 of the 21 days.

On those days a normal 10 A power point, at about 2.3 kW, would have drawn roughly 7 kWh of free power, enough for about 45 km of suburban driving. A 7 kW wall charger could draw more, but the 24 kWh cap covers the whole house. On the first hot day of spring the house, home battery included, used about half of it, which would leave room for about 12 kWh, or 80 km.

Over the three weeks, that would have covered more than half of the commuting car's driving from a power point, and all of it from a wall charger. A normal working week has fewer days at home than this sample, so in practice more of the charging would land overnight at 35c.

Range isn't the question yet

Neither car did 100 km in a day; the second car's longest day was 91 km. The smaller EVs on sale here manage about 260 to 360 km in real driving, by the same EV Database estimates. Range will matter on a long drive, and the logs don't contain one yet.

Is either engine wearing out?

Keep-or-replace has a second half: a petrol car with a failing engine makes the decision for you. The same boxes watch for that on every trip.

• Fuel trims. The engine computer keeps correcting the fuel it injects to hold the mixture right. The short-term trim is the correction it is making now, and the long-term trim is the correction it has learned. Together they should stay within about ±5% (Vehicle Service Pros). An air leak pushes them positive, because the computer adds fuel to make up for air it never measured, and a sticking injector or a sensor going off can push them either way, often well before a warning light. The box keeps the average of both for every trip, and the dashboard draws a line at ±10%.
• Knock and misfires. In the commuting car the box also reads how far the engine computer pulls the spark back when it hears knock, and its misfire counters.
• Battery and warning codes. Battery voltage at every start and stop, and the check-engine light with any stored trouble codes at every start. A new code puts a line in that trip's phone notification.

So far both engines look healthy. The trims average about −1.5% in the commuting car and about −3% in the second car, and never went past ±5% on any trip over a kilometre. Neither car has logged a trouble code or lit the check-engine light in over 100 trips. In the dozen trips the commuting car's knock and misfire readings have been watched, its engine computer never pulled the spark back for knock, and the misfire counters moved by 11 in hundreds of thousands of firings. Nothing here forces a replacement, so the case for an EV rests on what it saves to run.

Running the same numbers for your car

You don't need an ESP32 for this. A month of notes gets most of the way:

1. Write down the odometer every night. That gives your longest day and a typical week.
2. Note where the car is when power is cheap on your plan. If it isn't at home then, price the charging at the rate it would really pay.
3. Work out petrol per 100 km from receipts and the odometer. Fill to the click each time, or the litres won't line up with the kilometres.
4. Work out the EV's cost per 100 km: a real-world consumption figure for your kind of driving, plus about 15% for charging from a power point (less from a wall charger), times the price per kWh when it would charge.
5. Subtract, and scale by your yearly kilometres. That is the yearly fuel saving to set against the EV's extra purchase price.

# Example figures, not mine.

# Petrol: 48 L, 600 km, $2.10/L
petrol = 48 / 600 * 100 * 2.10
# $16.80 per 100 km

# EV: 13 kWh/100 km, +15%, 35c
ev = 13.0 * 1.15 * 0.35
# $5.23 per 100 km

saving = (petrol - ev) * 100
# $1,157 per 10,000 km
km = 1000 / saving * 10_000
# 8,645 km to earn back $1,000

A logger helps if you want every trip without thinking about it. A cheap Bluetooth OBD dongle and a phone app will log trips, as long as the app runs every time, and the same app shows the fuel trims live: check them with the engine warm, at idle and at a steady 2,500 rpm. If you already run Home Assistant, the box in these cars logs every trip with no phone involved.

What three weeks can't tell me

• Summer. Heat costs an EV range: about 5% at 32 °C across Recurrent's data from 30,000 cars, and 17% at 35 °C with the air-con on in AAA's test. The commuting car already had its air-con running for nearly half its driving time this spring.
• Winter. Cold costs more, about 22% of range at 0 °C (Recurrent), while Geotab's 4,200 EVs lost nothing between 10 and 31 °C. An Adelaide winter morning sits a little below that band.
• A long drive, for range.
• A couple of fills to the click, to confirm the fuel calibration.
• Everything that isn't fuel.

The loggers keep running, and summer is the next test. Measuring first was cheap, and it is how I would approach a business weighing up its vehicles too: log what they do for a month, then let the numbers make the case.