Fleet Age Model — Visual Explainer

Step 1

The Weibull Survival Curve

Cars don't last forever. The model uses a Weibull distribution to calculate the probability that a car of any given age is still on the road. Young cars almost always survive; by age 18, only about 8% remain.

S(age) = e ^{−(age / 14.6)
^4.4}

Age 13 yrs

ⓘ At age 13 , a car has a 54.9% chance of still being licensed — and a 79.3% chance of surviving one more year given it made it this far.

Step 2

The 2015 Starting Fleet

The model is seeded with real data: ~30.3 million licensed cars broken down by age bracket from government statistics (VEH1107). Each square below represents roughly 100,000 cars, coloured by how old they were in 2015.

ⓘ The newest cars (age 0–1) are brightest. Older brackets are disaggregated using Weibull weights — younger ages within each bracket get more cars because fewer old cars survive.

Step 3

Year-by-Year Fleet Update

Every year, the model does two things: (1) apply the survival curve to age every existing cohort — some cars drop out; (2) add new registrations as a fresh age-0 cohort. Use the slider to step through time.

Fleet(year, cohort) = Fleet(year−1, cohort) × p _surv (age) + new sales(year)

Year 2015

ⓘ In 2015, the fleet is mostly petrol & diesel cars spread across many age cohorts.

Step 4

Fleet Composition Over Time

Summing all surviving cohorts each year gives the total fleet split by drivetrain. Even though new BEV sales hit 100% by 2035, old petrol and diesel cars linger in the fleet for years — the transition takes until the late 2050s.

Petrol

Diesel

BEV

Hybrid

ⓘ This is the key insight: new sales ≠ fleet composition . Even with 100% BEV sales from 2035, it takes until ~2055 for BEVs to become 90%+ of the fleet, because older ICE cars survive for years.

Step 5

Calibration & Validation

The Weibull curve was originally fitted to a cross-sectional age snapshot — not year-by-year cohort tracking. Applied as annual retirement probabilities, it over-retires old ICE cars from the 2015 starting fleet, causing the raw model to drift ~13% below observed stock by 2024.

Correction: Petrol + Diesel = VEH1107 _observed − BEV _raw − Hybrid _raw

The fix is targeted: BEV and Hybrid cohorts are all young (age ≤9 in 2024) so the Weibull barely retires any — the raw model is already accurate for them. Only the Petrol + Diesel total is corrected to match government data, keeping the same Petrol:Diesel ratio as the raw model.

Raw model (Weibull only)

VEH1107 observed

Calibrated model

Raw model error %

Calibrated error %

ⓘ The raw model underestimates the fleet by up to 13% because the Weibull over-retires old petrol/diesel cars. The calibrated model corrects this using observed VEH1107 data, then holds the 2024 correction factor constant for future projections.

Step 5b

BEV Fleet Validation

Since BEVs are too young for significant Weibull retirement, the modelled BEV fleet should closely track cumulative new registrations — which serves as an upper bound (no BEVs scrapped yet).

Model BEV fleet

Cumulative BEV registrations (upper bound)

ⓘ The model's BEV count sits just above cumulative registrations because the 2015 starting fleet includes a small number of pre-2015 BEVs (~100k). The close match confirms no calibration is needed for BEV/Hybrid.