BEV Trajectories • Gallery

What does this website show?

This site visualizes how passenger vehicle markets transition from internal combustion engines (ICE) to battery electric vehicles (BEV) using historical registration data and fitted transition models.

Is this a prediction?

No. The curves are analytical approximations based on historical data. They illustrate transition dynamics, not precise forecasts of future market shares. Think of the output as `If things continued developing as they have, how would the future look like?`. These extrapolations naturally change as new data comes in, thus regular updates are needed.

Where does the data come from?

Data comes from multiple sources, the biggest being ACEA. Where possible I refer to the actual underlying ministry or national statistical agencies.

Why do some countries show BEV / PHEV / HEV separately while others only show HYBRID?

Statistical agencies publish different category structures. Some split hybrids into plug-in hybrids (PHEV) and conventional hybrids (HEV), while others publish only a combined hybrid category. Some label numbers as "Hybrid" when they are specifically PHEV, some throw mild hybrids (MHEV) into HEV, some throw them into their respective ICE categroy, some do that with all hybrids, some split the ICE category into Petrol/Diesel/FCEV (=hydrogen), and others make up their own categories.
Defining categories in a harmonized way is a tricky business. Generally I have made the experience that BEV/PHEV/ICE works best if I want to compare countries.

Why can shares differ slightly from official statistics?

Shares are sometimes recalculated from category totals to ensure internal consistency across the dataset. Also it is not uncommon for agencies to update not only the most recent month, but the month before that too. Throw in small differences in which type of vehicle is counted (passenger cars, busses, vans, trucks, only private, ambulances or taxis and the like) and coupled with the non-harmonized fuel type categories it is easy to get differences. As a default, I refer to passenger vehicles.

What model is used for the transition curves?

The curves are based on a Weibull-type transition model fitted to the BEV and ICE share trajectories. It is a weighted model using market sizes as weights.

Why do modeled values differ from historical data points?

The model represents a smoothed transition trajectory and does not reproduce every monthly fluctuation in the raw data. The intent is to model the structure of the transition rather than try to predict singular points in time.

Does the model assume BEV reaches 100%?

Yes. The model represents a long-term transition from ICE vehicles to BEV. As such this is the key assumption. However, this assumptions is not baseless, but rather is motivated by nordic countries where we can see this happen already. The assumption can also be checked, since if it truly does not hold, models will fail in such a way, that models throw errors. Two countries have yielded such errors in the past: Japan and Croatia. This is expected, as they have never had relevant BEV shares. No country that has had relevant BEV share at some point in the past crashed the assumption though.

Why does electrification sometimes appear slow?

Fleet turnover is slow. Even when BEV sales grow rapidly, replacing the entire vehicle fleet takes many years. That is because cars on the street don't just vanish just because the influx to the fleet is now electric. Cars have a lifespan of roundabout 10-15 years (depending on brand, drivetrain, country and other variables) and the substitution of the fleet is accordingly slow.

Why do some countries electrify faster than others?

Adoption speed depends on vehicle supply, charging infrastructure, electricity prices, income levels, prices for electricity and fossil fuels, taxation, how BEV are portrayed in the news and dozens of other variables. Surprisingly, I have not yet found a singular predictor that I can point to and say "this is what drives electrification most". The only thing I am confident influences BEV transition without a doubt is the existence of purchase incentives by the government.

Why does adoption accelerate after a certain point?

Technology adoption often follows an S-curve where growth accelerates after early market barriers are overcome. This is not a "wall" to be passed. You will rarely be able to point to a specific date and say "this is where it started". Rather it is a gradual development. The models are supposed to visualize this development.

How are fleet projections calculated?

Fleet projections combine the last observed fleet composition with modeled inflow of new vehicles and category-specific attrition rates.

How are fleet projections calculated in detail?

The model projects the fleet forward year by year using three components: (1) the existing fleet from the previous year, (2) category-specific retirement ("hazard") rates, and (3) the annual inflow of newly registered vehicles.

Example starting fleet (2025):
BEV: 945,182
PHEV: 208,467
HEV: 162,422
DIESEL: 942,780
PETROL: 676,614
OTHERS: 270

Example hazard (retirement) rates:
BEV: 0.010
PHEV: 0.020
HEV: 0.030
HYBRID: 0.030
DIESEL: 0.065
PETROL: 0.055
OTHERS: 0.050

Assume:
Annual new car sales (inflow): 150,000
Inflow multiplier: 1
Net imports: 0

Step 1 – Apply hazard rates (vehicles leaving the fleet).
Each category loses a fraction of its vehicles according to its hazard rate.
Example for BEV:
2026 BEV survivors = 945,182 × (1 − 0.010)

Step 2 – Add new vehicle inflow.
The annual inflow (150,000 vehicles in this example) is distributed using the modeled BEV share for the middle of the projection year (year + 0.5).

Example BEV calculation:
2026 BEV fleet = 945,182 × (1 − 0.010) + 150,000 × BEV_share(2026.5)

Step 3 – Assign the remaining inflow.
The portion of new vehicles that are not BEV is assigned as follows:
• If the dataset contains a PHEV category → inflow goes to PHEV
• Otherwise, if it contains HYBRID → inflow goes to HYBRID
• If neither exists → inflow is placed in OTHERS

The process then repeats for every following year, using the newly calculated fleet as the starting point for the next projection step.

How are new vehicle registrations assigned to fleet categories in the projections?

Annual vehicle inflow is split using the modeled BEV and ICE shares for the middle of the projection year (year + 0.5). BEV inflow goes to the BEV category. The remaining non-BEV inflow is assigned to PHEV if that category exists in the country dataset, otherwise to HYBRID. Only if neither PHEV nor HYBRID exists does it go to OTHERS.

Why do some countries show inflow into PHEV or HYBRID instead of OTHERS?

If a country dataset contains a PHEV category, the model places the non-BEV-non-ICE inflow there. If PHEV does not exist but HYBRID does, it is assigned to HYBRID. This keeps the projected fleet composition consistent with the categories available in the observed fleet data.