Enterprise Holdings acquiring IGO CarSharing in Chicago
Ballard signs non-binding MOU with Azure Hydrogen for fuel cell bus development in China

ICCT study finds increasing gap between rated and actual passenger car fuel consumption in Europe

The divergence of real-world CO2 emissions vs. manufacturers’ type-approval (100%) in Europe from various on-road data sources. Source: ICCT. Click to enlarge.

A new white paper published by the International Council on Clean Transportation (ICCT) comparing official and “real-world” fuel consumption and CO2 emission values for passenger cars in Europe and the United States shows that the average discrepancy between the values in Europe increased from less than 10% in 2001 to 25% in 2011.

The European analysis is based on the aggregation of several large sets of on-road driving data from various European countries. The analysis of US data is separate from that of the EU data—and draws different initial conclusions—for a number of reasons, including that both the test cycle and test procedure to determine vehicle CO2 emissions data are different from the way type-approval data are collected in the EU.

Assuming that European driving behavior has not changed appreciably over the past ten years, the authors suggest that the observed increase of the gap in Europe is most likely due to a combination of factors including:

  • Increasing application of technologies that show a higher benefit in type-approval tests than under real-world driving conditions (for example, start-stop technology);

  • Increasing use of ‘flexibilities’ (permitted variances) in the type-approval procedure (for example, during coast-down testing); and

  • External factors changing over time (for example, increased use of air conditioning).

The increase in the gap was especially pronounced after 2007–2008, when a number of European Union member states switched to a CO2-based vehicle taxation system and a mandatory EU CO2 regulation for new cars was introduced.

It is important to clarify that nothing in this analysis suggests that manufacturers have done anything illegal. However, the NEDC was not originally designed to measure fuel consumption or CO2 emissions, and some features of the test procedure can be exploited to influence test results for those values. Manufacturers appear to be taking advantage of permitted tolerances in the NEDC, resulting in unrealistically low CO2 emission levels.

Results from tests that closely resemble type-approval testing appear to confirm this. In such tests, run using vehicles provided directly by manufacturers and in laboratory settings that are in line with those customary for type approval, the discrepancies between laboratory and real-world results tend to be much smaller and do not show any sign of a marked increase over time. However, these type-approval like laboratory tests do not take into account a number of conditions and behaviors typically found in on-road driving.

The public policy implications of this study are significant. The growing gap between reported efficiencies and actual driving experience cuts in half the expected benefits of Europe’s passenger vehicle CO2regulations. It creates a risk that consumers will lose faith in type-approval fuel consumption values, which in turn may undermine government efforts to encourage the purchase of fuel-efficient vehicles through labeling and tax policy. For tax authorities, the gap between type-approval and real-world CO2 values translates into a gap between actual and potential revenues from vehicle taxes. Finally, increasing discrepancies between type-approval and on-road CO2 emissions can result in a competitive disadvantage for some vehicle manufacturers since it tilts the playing field.

—“From Laboratory to Road”

Europe. Fuel consumption and carbon dioxide emission values for new cars in Europe are determined via testing vehicles under laboratory conditions using the New European Driving Cycle (NEDC) (part of the type-approval process). The type-approval values are the basis for consumer information, CO2 regulation, and CO2-based vehicle taxation.

In the report, the authors used on-road vehicle testing data from Spritmonitor.de (Germany); Travelcard (Netherlands); LeasePlan (Germany); Honestjohn.co.uk (United Kingdom); WhatCar? (United Kingdom); and TCS (Switzerland). They also used laboratory test data from ADAC Ecotest (Germany) and QueChoisir (France).

United States. In addition to having a different test cycle and test procedure than in Europe, the US has an in-use compliance program to ensure that prototype/early-production vehicles chosen for type approval are representative of the mass-production vehicles that are delivered to customers later.

US “My MPG” real-world vs. official (100%) CO2 emissions by vehicle model year. Source: ICCT. Click to enlarge.

The authors used official fuel economy and CO2 ratings from the fueleconomy.gov website published by the US Department of Energy (DOE) and US Environmental Protection Agency (EPA). That website also offers a section called “My MPG” where users can register and input real-world fuel consumption data for their own vehicles. A sample of the “My MPG” database, including the aggregated fuel consumption data, was provided to the ICCT by Oak Ridge National Laboratory.

The US makes a distinction between the mpg provided by the manufacturer based on vehicle certification tests and the “adjusted mpg”, a value that is based on the unadjusted mpg but combined with a specific multiplier to provide more accurate fuel consumption information.

Looking at the unadjusted data reveals that, up until about 2003 the discrepancy between real-world and type-approval CO2 values was consistently around 20%; from 2004 onward it steadily increased to about 35% in 2012.

The EPA applies a correction factor to adjust the mpg ratings determined by vehicle tests and make them closer to what consumers experience in real-world driving. This correction factor was updated in 2008. However, all values in this analysis make use of the post–2008 correction factor to provide a consistent data series.

...As seen in the figure, the trend line that is based on adjusted mpg mirrors the contours of the unadjusted mpg values. Only the absolute level of discrepancy is different, with the adjusted values being close to 90–100 percent of the type-approval values, while the disparity for the unadjusted figures is substantially greater. For hybrid vehicles, the discrepancy level found is closer to the fleet-wide average when using the adjusted mpg numbers. This indicates that the updated correction factor, which now is nonlinear and adjusts higher mpg numbers (as typically found for hybrid vehicles) more than lower mpg numbers, results in more realistic adjusted mpg values for hybrid vehicles.

—“From Laboratory to Road”

Moving forward. The new Worldwide Harmonized Light Vehicles Test Procedure (WLTP), with a more realistic test cycle and tightened test procedure, is expected to result in more realistic CO2 emission values and therefore a narrower gap between type-approval and real-world values. However, the authors note, the WLTP will not resolve all known issues with the current procedure, and it may itself have vulnerabilities that are not yet recognized. As one example, it remains to be seen how plug-in hybrid electric vehicles and other electrified vehicles will perform in the WLTP as compared to on-road driving.

It is too much to hope, then, that following the introduction of the WLTP, type-approval CO2 emissions will fully reflect real-world driving. Nevertheless, the objective should be to bring both values into greater agreement, reversing the recent widening of the gap between them. That will spur development of technologies that reduce CO2 emissions under real-world conditions, thus avoiding misdirected investments. The WLTP should therefore be adopted in the eu as soon as is feasible after it is finalized (expected in 2013/2014), taking into account necessary lead time for industry.

—“From Laboratory to Road”

The authors also suggest that WLTP adoption should be accompanied by additional correction methods, which are currently being investigated by the European Commission, as well as a separate test procedure vehicle air conditioning systems, the Mobile Air-conditioning Test Procedure (MACTP), also being developed by the UNECE.

As next steps, the authors plan to continue collecting real-world CO2 emission data for new vehicles from various data sources across Europe to support an even more detailed analysis of historical trends and underlying causes. For the United States, they noted, the data examined in the paper are merely a starting point for future analysis.

One conclusion to be drawn from the analysis of US practices is that the “adjusted mpg” values provided as consumer information closely match what drivers actually experience on the road. A similar system could be introduced in the EU, in which NEDC—or later WLTP—test values were adjusted, to better reflect on-road CO2 emission values, and the adjusted values used as the basis for consumer information, in particular, on the car’s CO2 label. Also, in the United States as in the EU, supplementing publicly available on-road CO2 emission data with information systematically collected by a large-scale data logger project is an important step toward better understanding of individual technologies’ real-world performance and their potential for reducing emission levels in the future.

—“From Laboratory to Road”




I can confirm this from my own experience.
I had a 2004 Renault Espace 2.2L diesel MPV.
I used to get 35.7 mpg (UK) from it in mixed driving, which was almost exactly the manufacturer's specification.
(I drive as economically as I can without enraging following traffic and passengers).

I recently got a 2012 Ford Galaxy 2L diesel MPV, and I get 42.5 mpg on the exact same runs, but the manufacturer's mpg is 50.
I have seen in magazines where people get about 42 mpg on the Galaxy, so I am not an outlier.


The 'increasing gap' seems to be and is wider with certain products.

Is there a close relationship between the 'gap size' and certain vehicles and/or manufacturers?

Most owners can meet the advertised economy with their Toyota HEVs (except in winter) but this is not necessarily true with other brands?


I can mirror mahonj's statement. My 1998 VW Polo 1.9 SDI had a rated fuel consumption of 4.7 l/100 km, its real world consumption over 4 years of ownership was 4.35 l/100 km. My 2005 Prius was 4.3 rated / 4.7 actual and my 2009 Prius 3.8 rated / 4.3 actual.

One factor not mentioned is the increased use of light during the day which was the result of a media campaign. Most newer cars have special energy efficient LED drl's, but this was not the case until about 2 years ago.

A suspicion that I have (but I can not prove it) is that the heavily optimised aerodynamics of modern cars only 'work' in perfect windtunnel conditions. In the real world there is always turbulence from the wind, other traffic, which disturbs the perfect laminar flow and renders most of these subtle optimisations useless. A car that has bad aero doesn't have any laminar flow to begin with.


My wife averages a bit under 4.0L/100 Km with her 2013 Prius III (if I don't drive it too often). She used to average 8.5 with my old Camry while I have a hard time to do much better than 10.

I'll have to learn to drive more efficiently or buy a BEV?
Hydro electricity is plentiful and very cheap ($0.058/kWh, 24/7) in our area. Most EV's range are currently too limited and still too pricy. Quick charging facilities are also too limited in our area but more are being added. It may be another 6 - 8 years before BEVs become practical as family cars.

The comments to this entry are closed.