Car Longevity : A Biometric Approach - Archive ouverte HAL Access content directly
Conference Papers Year : 2011

Car Longevity : A Biometric Approach


Vehicle longevity and survival are important to forecast total fleets and to estimate related greenhouse gas (GHG) emissions. Despite today’s abundant literature, the major issue of the estimated survival rate of vehicles as a tool for vehicle and emissions forecasts is still an open question largely because of no data on scrapped cars. To secure an overview of car longevity, we applied a biometric approach by building Kaplan-Meier survival curves. Our database covers 265 vehicles scrapped or sold for spare parts in France in August 2010. The information includes make, model, version, cylinder capacity, fiscal horsepower, fuel (diesel/gasoline), date of registration, date of scrapping and odometer reading. Using the age in total months and final odometer readings of scrapped cars, we applied the Kaplan-Meier estimator to obtain survival rates by age and total mileage. The distinction between these two dimensions is useful to separate longevity by age (lifetime) from longevity by total mileage (design life). We found longer lifetime for gasoline cars but higher design life for diesel cars. The results give a median lifetime of 126 months (10y6m) years for the dataset, i.e. 149 months (12y5m) for gasoline cars and 118 months (9y10m) for diesel. Moreover, results give a median design life of 150,000 km (94,000 miles) for the dataset, i.e. 135,479 km (84,674 miles) for gasoline cars and 152,730 km (95,456 miles) for diesel. However, our study also shows that Kaplan-Meier survival curves tie in nicely with a twoparameter Weibull survival function.


No file

Dates and versions

hal-00860986 , version 1 (11-09-2013)


  • HAL Id : hal-00860986 , version 1


Zehir Kolli. Car Longevity : A Biometric Approach. Proceedings of YRS2011 - Seminar 2011, Jun 2011, Denmark. 24p. ⟨hal-00860986⟩
246 View
0 Download


Gmail Facebook Twitter LinkedIn More