Data

Model versus owner-reported efficiency: assessing the accuracy of physics-based predictions

We compare our physics-based efficiency predictions with real-world figures reported by UK electric vehicle owners on Reddit. All results, including discrepancies, are transparently presented.

0/14

predictions within reported range

56.1%

mean absolute error vs midpoint

34

owner data points across 14 models

While any physics model can generate numerical predictions, the critical question is whether these predictions align with real-world observations. We collected over 34 owner-reported efficiency figures from a UK electric vehicle community thread on Reddit, matched them to our catalogue vehicles, and compared the model’s predictions with these reported values.

The model operates at 40 mph and 15 degrees Celsius with an 80 kg driver payload, under dry conditions and without preheating. These parameters represent mixed summer driving, which most owners reference when reporting ‘good weather’ efficiency. No state-of-health adjustment is applied, as efficiency (mi/kWh) is determined by the powertrain and driving conditions rather than remaining battery capacity.

When the model prediction falls within the owner-reported range, the corresponding cell is marked green. If it falls outside this range, the magnitude and direction of the discrepancy are indicated. No model parameters have been adjusted to fit these data points. The physics model was constructed using manufacturer specifications and first principles prior to the collection of any owner data.

CarModel (mi/kWh)Owners (mi/kWh)Errorn
Kia Soul EV 64 kWh6.15.3+15%1
Dacia Spring Electric 45 (25 kWh)7.66.0+27%1
Hyundai Kona Electric 64 kWh6.64.5-5.0+38%4
MG MG5 EV Long Range 61 kWh6.83.8-5.5+46%3
BMW i3 42.2 kWh (120 Ah)7.24.2-5.2+52%2
Kia e-Niro 64 kWh6.23.8-4.3+53%6
Vauxhall Corsa-e 50 kWh7.14.6+54%1
BYD Dolphin 60.4 kWh Comfort/Design6.43.3-5.0+54%1
Volkswagen ID.3 Pro 58 kWh6.44.0-4.2+56%2
Renault Zoe ZE50 (52 kWh)6.84.1+65%1
Tesla Model 3 Long Range6.93.7-4.5+69%7
Tesla Model 3 Standard Range Plus7.44.0-4.5+74%3
Peugeot e-208 50 kWh7.13.5-4.0+88%1
Nissan Leaf e+ 62 kWh7.03.6+94%1

What this tells us

The physics model operates independently of owner-reported data. It calculates efficiency based on each vehicle’s mass, drag coefficient, frontal area, tyre rolling resistance, drivetrain efficiency, and heater type. The model’s alignment with owner-reported ranges for 0 out of 14 matched vehicles suggests that the underlying physics accurately represent real-world relationships.

When the model overestimates efficiency, it is often because the reference speed of 40 mph does not reflect the owner’s typical driving conditions. Owners who primarily drive on motorways tend to report lower efficiency than the model predicts at 40 mph. Conversely, when the model underestimates efficiency, it is likely that the owner engages in predominantly urban driving at lower average speeds, where efficiency is generally higher.

The broad range observed in owner-reported data (for example, 3.8 to 5.5 mi/kWh for the MG5) aligns with the model’s prediction that efficiency is highly dependent on speed. A single ‘lifetime average’ reported by an owner aggregates various driving conditions into one value, whereas the model enables the separation of these factors.

Limitations

This analysis does not constitute a controlled study. Owner-reported figures are self-selected, frequently rounded, and measured using various methods, including trip computers, mobile applications, and manual calculations. Some owners report lifetime averages that include winter conditions, while others report summer peak values. Sample sizes for certain models are limited, and the Reddit community may be biased toward enthusiast drivers who are more attentive to efficiency than the general population.

We present this comparison as a meaningful real-world validation: rather than a laboratory test, it assesses whether the model generates results that owners recognise from their own vehicles. For the full details on how the physics model works, see the methodology page.

Explore the model’s performance for any vehicle

Adjust speed, temperature, and payload parameters to observe how efficiency varies for a specific car.

Browse models