⚠ This uses hard-coded data and should be treated as an explainer only. For the actual model, and the real projections, please open the Jupyter Notebooks.

GB Car Fleet Model

How the model projects fleet composition by drivetrain from 2015 to 2061.

Overview

What Does This Model Do?

The model tracks every annual cohort of vehicles on GB roads - when they entered the fleet, what drivetrain they are, and when they're scrapped. It answers: if the ZEV mandate forces 100% electric sales by 2035, how long until BEVs dominate the fleet?

2015 Fleet Age distribution
from VEH1107
→
Weibull Survival curve
scraps old vehicles
→
+ New Sales Drivetrain split
from ZEV mandate
→
Fleet Composition 2015–2061
by drivetrain
Step 1

Survival Curve: Weibull Distribution

The Weibull function characterizes how long vehicles survive before scrappage. Fleet turnover is governed by two parameters: shape k = 4.4 (from Serrenho et al. 2017) and scale η = 16.0 (re-fitted to GB 2015–2019 fleet totals). The median lifespan is ~15 years.

S(age) = exp(−(age/η) k )    p retire (age) = 1 − S(age)
Survival probability
Retirement probability
16.0
4.40
ⓘ Pre-COVID (2015–2019), the Weibull fit had RMSE < 0.2% against GB total fleet (VEH1107). The curve is right-skewed: many vehicles last 12–18 years, few exceed 30.
Step 1b

Cohort Depletion: 1,000 Vehicles Aging

What does the Weibull mean in practice? Start with 1,000 vehicles registered in the same year and watch the cohort shrink. Each year, the model applies the conditional survival probability p(a) = S(a+1)/S(a) - a vehicle that has survived to age a has this chance of making it one more year.

p surv (a) = S(a+1) / S(a)   |   cohort(a+1) = cohort(a) × p surv (a)
Surviving vehicles
Scrapped this year
Parameter Value Source
ⓘ Half the cohort is gone by age ~15. Scrappage peaks around age 14–16 then tapers - by then most vehicles have already been retired. This "hump" is why fleet turnover accelerates for a decade after new sales go 100% BEV, then slows.
Step 2a

2015 Fleet by Age

The model starts with VEH1107 2015 GB fleet data: thousands of vehicles at each age. Some data come as exact counts (ages 0–5); others as brackets (6–10, 10–13, 13+). Brackets are distributed using Weibull proportions. Each box below represents 200,000 vehicles.

Step 2b

Annual Cohort Aging & Scrappage

Each cohort of vehicles (e.g., all 2010 registrations) survives year by year according to the Weibull. As they age, a fraction retire every year. The overall fleet shrinks unless new sales exceed scrappage.

Fleet(year, cohort) = Fleet(year−1, cohort) × p surv (age) × f scrap (year)   +   new sales(year)
2015
Petrol
Diesel
BEV
Hybrid
ⓘ In 2015, the fleet is mostly petrol & diesel vehicles spread across many age cohorts.
Step 3a

COVID Scrappage Adjustment

During 2020–2024, far fewer vehicles were scrapped than the Weibull predicts. The model accounts for this using scrappage factors derived directly from VEH1107 data: actual scrappage rate ÷ pre-COVID baseline rate.

Pre-COVID scrappage (thousands)
COVID-era scrappage (thousands)
Pre-COVID baseline rate (~5.9%)
factor(year) = observed_scrappage_rate(year) / baseline_rate   |   adjusted_retirement = weibull_retirement × factor

The factors scale the Weibull's retirement probability down during COVID years (factor < 1 means less scrappage than expected). This applies only to the historical simulation (2016–2024). The forward projection uses normal Weibull scrappage because the fleet target is fixed regardless.

Year Observed rate Factor Interpretation
Step 3b

Fleet-Driven Projection (2025–2061)

Rather than guessing future new sales, the model targets a fleet trajectory and back-calculates the sales needed. A linear fit to VEH0101a (GB licensed vehicles, 2010–2024) shows steady growth of ~340k vehicles/year - this gradient drives the projection.

new_sales(year) = target_fleet(year) − surviving_fleet(year)

Each year: age the existing fleet by one year (applying Weibull scrappage), calculate the target fleet from the trend line, then derive the new sales needed to hit that target. The new sales are split by drivetrain according to the ZEV mandate trajectory.

VEH0101a historical (GB Q4)
Linear trend
Model projection
ⓘ Why fleet-driven? A sales-driven model (with sales reverting to ~2.2M/yr) was trialled but produced a fleet that plateaus. VEH0101a shows no sign of saturation - the fleet has grown steadily for 30 years. The fleet-driven approach reproduces this observed trajectory and avoids making assumptions about future sales volumes.
Step 4

ZEV Mandate & Drivetrain Shares

The ZEV (Zero Emission Vehicle) mandate dictates the drivetrain split of new sales. BEV share ramps from ~19% in 2024 to 100% by 2035. Hybrid declines to zero by 2035. Petrol and diesel fill the residual ICE share (split by their 2024 ratio) until they're phased out at 2030.

Year Petrol Diesel BEV Hybrid Policy note
ⓘ Note the asymmetry: new sales go 100% BEV by 2035, but the fleet still has millions of ICE vehicles surviving from earlier cohorts. The Weibull survival curve is what bridges this gap - it determines how long those legacy ICE vehicles stick around.
Step 5

Validation Against Observed Data

The model's fleet predictions for 2015–2024 are compared against actual DfT figures (VEH1107 GB, excluding unknown-age vehicles). A close match for 2015–2019 confirms the Weibull fit; post-2020 the scrappage factors keep it on track.

VEH1107 observed (GB)
Model
ⓘ Pre-COVID (2015–2019) RMSE ≈ 0.2% - the Weibull alone, with no adjustments, matches fleet totals almost exactly. Post-COVID the scrappage factors maintain accuracy. The full 2015–2024 RMSE is under 1%.
Step 6

Fleet Projection: 2015–2061

The final output: fleet composition by drivetrain from 2015 to 2061, combining the model simulation (2015–2024) and the fleet-driven projection (2025–2061).

Petrol
Diesel
BEV
Hybrid
ⓘ BEVs reach 50% of the fleet around 2045 and 90% by the late 2050s - roughly 10–20 years after new sales go 100% electric. The long tail of ICE vehicles is entirely driven by the Weibull survival curve.
Step 7

Interactive Sensitivity Analysis

Drag the sliders to explore how the three key assumptions affect the fleet projection. The chart updates in real-time. The grey band shows the central scenario for comparison.

+340k/yr
2035
16.0
Current sliders
Central scenario
ⓘ The biggest driver of uncertainty is the ZEV mandate timing - a 5-year delay shifts the entire transition curve rightward. BEV lifespan matters most in the long run, determining whether old BEVs accumulate or get scrapped relatively quickly. The absolute BEV fleet view is what matters for electricity demand planning.
Reference

Data Sources

All data is GB (Great Britain) for consistency. The model uses five DfT datasets:

Dataset Coverage Used for
VEH1107 Fleet by age bracket, GB 1994–2024 2015 starting fleet, observed totals for validation, new registrations (b0), scrappage factors
VEH1153 New regs by fuel type, GB 2001–2024 Historical drivetrain shares for cohort splitting
VEH0101a Total licensed vehicles, GB 1994–2024 Fleet growth trend for projection
VEH0120 Licensed fleet by fuel, GB 1994–2024 Pre-2001 drivetrain proportions, historical fleet composition
VEH0520 HGV fleet by weight band & year of first reg, GB 1994–2024 HGV starting fleet (1994 cohorts), new registrations, observed fleet totals, scrappage factors, Weibull parameter fitting

Cars & LGV Weibull shape parameter k = 4.4 from Serrenho et al. (2017), "The impact of reducing car weight on global emissions: the future fleet in Great Britain." Scale parameter η re-fitted to GB 2015–2019 fleet totals. HGV Weibull parameters (both k and η) are fitted directly to VEH0520 survival data.