Estimating when India's road petrol demand will peak requires establishing how many vehicles are actually running on asphalt rather than relying on cumulative RTO registrations. Applying the calibrated Weibull Survival Model across historical Vahan cohorts reveals that out of 445 million cumulative registrations since 2000, only approximately 273.7 million vehicles remain active on Indian roads today—uncovering a 164-million retired vehicle gap across two-wheelers and passenger cars that fundamentally reshapes fuel demand projections.
The Foundational Question: How Many Vehicles Are Actually on Indian Roads?
How do you estimate how many vehicles are actually on Indian roads?
I came across an interesting formula while working on a model to estimate when petrol demand in India will peak under different EV penetration scenarios.
The first step in my model is to estimate the number of vehicles on the road in any given year. We have comprehensive official data on the total number of registered vehicles from Vahan (with records through August 2026 surpassing 447 million units). But how many of them are still actively on the road? How many have been scrapped, abandoned, or retired over the years?
Registration data does not equal operational reality. In India, vehicles stop running, get dismantled for spare parts, or are abandoned long before their registrations are formally cancelled in Regional Transport Office (RTO) ledgers. So, I looked at how transport analysts approach this challenge and came across the Weibull Survival Formula.
The Mathematical Approach: The Weibull Survival Formula
What is it?
The Weibull formula helps us estimate the probability of a vehicle registered in year X surviving and remaining active on the road in year Y.
The formula was developed by Swedish engineer and mathematician Ernst Hjalmar Waloddi Weibull, who originally studied material strength. It has since been widely adopted in reliability engineering and survival analysis to estimate the probability of something surviving over time.
The Two Governing Parameters: Scale (λ) and Shape (k)
In our fleet survival model, two parameters determine how long a vehicle is likely to survive:
1. λ (Lambda) — Scale Parameter
It represents the age at which approximately 63.2% of a vehicle cohort is expected to have been scrapped or retired (leaving 36.8% surviving).
For example, if λ is 12 years, only 36.8% of that registration cohort is expected to survive to 12 years of age. This parameter naturally varies by vehicle segment. Commercial duty cycles differ significantly from private usage: for instance, we assume a lower λ of 11.0 years for high-intensity three-wheelers, around 13.5 years for commuter two-wheelers, and a higher λ of 15.8 years for private passenger cars.
2. k — Shape Parameter
It determines how the probability of vehicle retirement changes with age:
- k = 1: The retirement rate remains constant over time, irrespective of vehicle age.
- k < 1: The retirement rate decreases with age (similar to infant mortality).
- k > 1: The retirement rate increases with age, reflecting cumulative wear and tear. A higher k produces a sharper, steeper increase in retirement probability around the typical retirement age—creating an S-curve cliff-edge rather than a gradual trickle.
For vehicles, calibrated k values typically span the range of 2.5 to 5.0. In Indian fleet modeling, calibrated values between 3.0 and 3.8 provide the closest alignment with observed fleet survival literature.
Calibrated Segment Assumptions & Survival Curves (0 to 24 Years)
The formula and the parameters used for each vehicle segment are shown in the table below. These values are based on various reports and literature I came across. They represent initial calibrated assumptions, and suggestions are welcome.
The table and chart also highlight the Median Survival Age—the age at which exactly 50% of a vehicle cohort is expected to have been scrapped or retired based on the assumed parameters.
| Vehicle Category | Scale (λ) (63.2% Retired) |
Shape (k) (Wear-and-Tear) |
Median Survival Age (50% Scrapped) |
Max Operational Cap (Age Ceiling) |
|---|---|---|---|---|
| Three-Wheelers (3W) | 11.0 yrs | 3.2 | 9.74 yrs | 15.0 yrs |
| Two-Wheelers (2W) | 13.5 yrs | 3.5 | 12.04 yrs | 18.0 yrs |
| Passenger Cars (PV) | 15.8 yrs | 3.8 | 14.35 yrs | 20.0 yrs |
Calibrated Vehicle Cohort Survival Curves (0 to 24 Years)
Percentage of registered cohort surviving on Indian roads as vehicle age progresses
Looking at how the three curves diverge:
- Three-Wheelers (Median: 9.74 years): Auto-rickshaws experience the fastest drop-off. By year 9, more than half are off the road due to commercial mileage and intensive daily duty cycles.
- Two-Wheelers (Median: 12.04 years): Two-wheelers decline steadily through year 12, after which scrappage accelerates sharply, leaving only 25.1% surviving by year 15.
- Passenger Cars (Median: 14.35 years): Private cars maintain the longest lifespan, with over 44% still operational at 15 years and ~20% surviving at 18 years before reaching the 20-year maximum cap.
Model Output: Estimating India\'s Active On-Road Fleet in 2026
Applying this survival model cohort-by-cohort using RTO registration data shows the estimated number of vehicles (all fuels) on Indian roads in 2026 (based on data through August 2026):
| Vehicle Category | Cumulative Registrations (2000–Aug 2026) |
Active Fleet on Road (Surviving Today) |
Retired / Scrapped Fleet (Off the Road) |
Cohort Retention (% Surviving) |
|---|---|---|---|---|
| Two-Wheelers (2W) | 365.69 Million | 219.79 Million | 145.90 Million | 60.1% |
| Passenger Cars (PV) | 66.63 Million | 47.99 Million | 18.64 Million | 72.0% |
| Three-Wheelers (3W) | 12.66 Million | 5.89 Million | 6.77 Million | 46.5% |
| Total (2W + PV + 3W) | 444.98 Million | 273.67 Million | 171.31 Million | 61.5% |
Estimated Active Fleet vs Retired Vehicles on Indian Roads (Aug 2026)
Operational vehicles on road versus retired paper registrations (Million units)
The model reveals that across two-wheelers and cars alone, over 164 million registered vehicles have retired from active duty.
This highlights why the "Phantom Fleet" gap matters so much: if energy forecasters use raw registration tallies instead of survival-adjusted active fleets, they will grossly overestimate petrol demand and misjudge the pace at which EV adoption replaces fossil fuel consumption.
What Next? Historical Back-Testing & Petrol Peak Modeling
What next?
I plan to apply the model to previous years and check how closely the estimated vehicle population aligns with actual historical petrol and diesel consumption.
By calibrating annual fuel efficiencies (km/litre) and average annual mileage across historical active cohorts, we can back-test whether the model replicates PPAC's reported retail petrol sales over the last 15 years.
Once historical calibration is confirmed, the survival model serves as the core foundation for projecting India's road transport fuel trajectory—simulating the exact crossover year when expanding electric vehicle sales outpace fleet additions and cause net road petrol consumption to peak and decline.