WORKING PAPER 2015–8

Real-world fuel consumption of popular European passenger car models

Authors: Uwe Tietge, Peter Mock, Nikiforos Zacharof, Vicente Franco Date: 28 December 2015

Keywords: fuel consumption, CO2 standards, vehicle test procedures

Summary Official fuel consumption values of This study compares official fuel reductions in official fuel consumption new passenger cars in Europe are consumption values measured in values. Real-world fuel consumption becoming increasingly unrepresen- laboratories with the real-world values are, however, rarely reduced to tative of real-world performance. performance of 20 popular vehicle the same extent, indicating that fuel The divergence between official and models. All models claim significant efficiency improvements measured real-world fuel consumption values improvements in fuel efficiency since during laboratory testing do not fully more than quadrupled over the last 2009, with reductions in official fuel materialize on the road. fourteen years. After EU-wide CO2 consumption values ranging from 8 to standards were introduced in 2009, 30 percent. On-road measurements, The trend toward increasingly unreal- official fuel consumption values however, indicate that eight models istic fuel consumption values can be decreased by 15 percent while made little to no improvement in traced back to the exploitation of flex- real-world figures only decreased by 2 real-world fuel efficiency. Five models ibilities in the current vehicle testing percent. This divergence undermines achieved more than a 10 percent procedure. While the Worldwide climate change mitigation efforts and reduction in real-world fuel consump- Harmonized Light Vehicles Test costs the average car owner €450 tion since 2009. Model overhauls Procedure (WLTP) will be introduced per year. are frequently followed by abrupt in the EU in 2017 and will help align official and real-world fuel consump- 60% tion values, it will not by itself solve the problem of unrealistic fuel con- auto motor und sport (D) value s

2 sumption values. Further actions are 50% Travelcard (NL) km77.com (E) needed, including in-use conformity 45% (company cars) testing of randomly selected vehicles 40% 40% (all data sources) on the road and the establishment of auto motor & sport (S) 36% (private cars) a EU-wide type-approval authority. Touring Club Schweiz (CH) 30% honestjohn.co.uk (GB) AUTO BILD (D) Introduction 20% The growing divergence between real-world fuel consumption values 8% and official laboratory measure- 10% ments of new European passenger cars undermines policies designed n = approx. 0.6 million vehicles 0% to mitigate climate change and Gap between ‘real-world’ and offiical CO 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 reduce fuel expenses for consumers. Year The latest “From Laboratory to Road” study of the International Figure 1: Divergence between real-world and manufacturers’ reported CO2 emissions for various real-world data sources, including average estimates for private cars, Council on Clean Transportation company cars, and all data sources. (ICCT) estimates real-world fuel

© INTERNATIONAL COUNCIL ON CLEAN TRANSPORTATION, 2015 WWW.THEICCT.ORG REAL-WORLD FUEL CONSUMPTION OF POPULAR EUROPEAN PASSENGER CAR MODELS

consumption values from data on 10 almost 600,000 cars from eleven data sources covering six European countries.1 The data sources include 8 -2% change since 2009 websites, records from leasing 7.9 7.8 7.8 7.8 7.8 7.8 7.8 7.7 companies, and on-road vehicle 7.3 7.3 7.1 7.2 7.2 7.2 7.2 7.2 7.1 7.0 6.9 tests. Based on this data, the study 6.9 6.8 -15% change since 2009 6.6 concludes that the divergence 6 6.2 6.0 5.8 between real-world and official 5.6 5.4 5.3 CO2 values of new cars increased from about 8 percent in 2001 to 40 4 percent in 2014 (see Figure 1).

The divergence between official and real-world CO2­ values is important for 2 all stakeholders: EU average: Official values • For society as a whole, the gap EU average: Estimated real-world values based on Spritmonitor.de data undermines efforts to mitigate Fuel consumption [l gasoline/100 km equivalents] 0 climate change and reduce fossil 2000 2002 2004 2006 2008 2010 2012 2014 fuel imports into the EU. Vehicle build year • For governments, the gap Development of average official and real-world fuel consumption values of compromises tax systems and Figure 2. new European passenger cars. incentive schemes that are based on CO values and can thus lead 2 market is presented in the graphs. to lost tax revenues and misal- by 15 percent through 2014, while The twenty models represent location of public funds. real-world values only decreased by 2 percent (see Figure 2). A closer roughly 40 percent of all passenger • For car manufacturers, the look at individual vehicle models cars registered in Germany during divergence puts those that want helps explain this trend. this time. Official fuel consump- to report realistic CO2 emission tion values were calculated as values at a competitive disadvan- gasoline consumption equivalents tage. Manufacturers reporting Performance of of the sales-weighted average CO2 unrealistic CO values may also 2 individual models values of both diesel and gasoline damage their credibility and may 3,4 Virtually all models in the European vehicles. Real-world values were erode public and regulator trust 5 market show a growing divergence based on data from Spritmonitor.de, in the industry. between on-road and official a free web service that allows users • For an average car owner, fuel consumption. Nonetheless, to track their fuel consumption by the divergence translates into while this is a systemic, industry- entering odometer readings and the increased fuel expenses of €450 wide problem, an examination of amount of fuel added during fill-ups. per year. individual models reveals that the For each model, fuel consump- divergence can vary significantly tion values of gasoline and diesel This paper focuses on the consumer between different models. variants were adjusted by the gap impacts of the growing divergence observed in Spritmonitor.de data to by presenting real-world fuel Figure 3 and 4 show the develop- arrive at an estimate of real-world consumption values for selected ment of official and real-world fuel fuel consumption. popular car models. For consumers, consumption values for twenty of real-world fuel consumption the most popular car models in values have been decreasing less Germany from 2001 to 2014.2 Each than suggested by official values. model’s share of the German new car

After EU-wide CO2 standards were introduced in 2009, official fuel 2 For two of the best-selling models, the 3 The ICCT, “European Vehicle Market consumption values decreased Mini Cooper Mini (3 percent market share) Statistics,” http://eupocketbook.theicct.org/

and Fiat Punto (2 percent market share), 4 CO2 values of diesel and gasoline cars the calculation of real-world values was were converted to gasoline consumption 1 See Tietge, U. et al., “From laboratory to not possible due to insufficient data. The equivalents using a conversion factor of 1 l

road: A 2015 update,” http://www.theicct. analysis thus covers 20 of the 22 best-selling gasolline/100 km ≈ 23.4 g CO2 /km. org/laboratory-road-2015-update car models. 5 See www.spritmonitor.de

2 INTERNATIONAL COUNCIL ON CLEAN TRANSPORTATION WORKING PAPER 2015-8 REAL-WORLD FUEL CONSUMPTION OF POPULAR EUROPEAN PASSENGER CAR MODELS

Audi A3 (1.7% market share) Audi A4 (2.3% market share) 10 10

REAL-WORLD 2009 REAL-WORLD -0.6% 8 8 2009 -4.2% OFFICIAL OFFICIAL 6 6 -16.4% -21.5%

Nmin = 102 Nmin = 48

Nmax = 273 Nmax = 376 4 4 2000 2002 2004 2006 2008 2010 2012 2014 2000 2002 2004 2006 2008 2010 2012 2014

Audi A6 (1.3% market share) BMW 3-series (2.7% market share)

10 10 2009 REAL-WORLD -3.8% REAL-WORLD OFFICIAL 8 8 2009 -0.8% OFFICIAL -20.1%

6 6 -12.5%

Nmin = 30 Nmin = 89

Nmax = 189 Nmax = 590 ) 4 4 2000 2002 2004 2006 2008 2010 2012 2014 2000 2002 2004 2006 2008 2010 2012 2014 alents BMW 5-series (1.4% market share) Ford Fiesta (1.5% market share) 10 10 REAL-WORLD 2009 OFFICIAL +0.9% 8 8 REAL-WORLD 2009 -5.8% -17.6% 6 6 OFFICIAL -17.0% Nmin = 34 Nmin = 65

Nmax = 221 Nmax = 345 4 4 2000 2002 2004 2006 2008 2010 2012 2014 2000 2002 2004 2006 2008 2010 2012 2014 onsumption (l gasoline/100km equiv

Ford Focus (1.9% market share) Mercedes-Benz A-Class (1.7% market share) 10 10 Fuel c

REAL-WORLD 8 REAL-WORLD 2009 8 2009 +2.6% -13.8%

OFFICIAL OFFICIAL 6 6 -23.6% -23.1%

Nmin = 139 Nmin = 32

Nmax = 484 Nmax = 161 4 4 2000 2002 2004 2006 2008 2010 2012 2014 2000 2002 2004 2006 2008 2010 2012 2014

Mercedes-Benz C-Class (2.4% market share) Mercedes-Benz E-Class (1.8% market share) 10 10 2009 REAL-WORLD 2009 REAL-WORLD -3.1% -11.0% 8 8 OFFICIAL OFFICIAL

-21.8% -30.0% 6 6

Nmin = 71 Nmin = 27

Nmax = 270 Nmax = 156 4 4 2000 2002 2004 2006 2008 2010 2012 2014 2000 2002 2004 2006 2008 2010 2012 2014

Build Year

Figure 3: Official and real-world fuel consumption values of popular car models in Germany. Arrowheads mark the timing of major technical overhauls. Minimum and maximum number of Spritmonitor.de entries per year presented in the bottom-left corners.

WORKING PAPER 2015-8 INTERNATIONAL COUNCIL ON CLEAN TRANSPORTATION 3 REAL-WORLD FUEL CONSUMPTION OF POPULAR EUROPEAN PASSENGER CAR MODELS

Opel Astra (2.6% market share) Opel Corsa (2.0% market share) 10 10

REAL-WORLD 8 2009 8 +1.0% 2009 REAL-WORLD +0.3% OFFICIAL 6 -16.6% 6 -12.5% OFFICIAL

Nmin = 72 Nmin = 35

Nmax = 513 Nmax = 309 4 4 2000 2002 2004 2006 2008 2010 2012 2014 2000 2002 2004 2006 2008 2010 2012 2014

Peugeot 206, 207, 208 (2.3% market share) Škoda Fabia (1.4% market share) 10 10

REAL-WORLD 8 2009 8 REAL-WORLD 2009

-18.8% -10.4% OFFICIAL 6 6 OFFICIAL -20.5% -15.9%

Nmin = 13 Nmin = 83

Nmax = 271 Nmax = 347 ) 4 4 2000 2002 2004 2006 2008 2010 2012 2014 2000 2002 2004 2006 2008 2010 2012 2014 alents

Škoda Octavia (1.3% market share) Toyota Yaris (1.4% market share) 10 10

8 REAL-WORLD 2009 8

-11.3% REAL-WORLD OFFICIAL 2009 +4.8% 6 -27.5% 6 OFFICIAL -8.4% Nmin = 96 Nmin = 18

Nmax = 538 Nmax = 238 4 4 2000 2002 2004 2006 2008 2010 2012 2014 2000 2002 2004 2006 2008 2010 2012 2014 onsumption (l gasoline/100km equiv

VW Golf (5.8% market share) VW Passat (2.6% market share) 10 10 Fuel c

REAL-WORLD

8 REAL-WORLD 8 2009 2009 -0.3% -4.9% OFFICIAL

OFFICIAL 6 6 -18.9% -21.7%

Nmin = 53 Nmin = 417 Nmax = 396 Nmax = 1,058 4 4 2000 2002 2004 2006 2008 2010 2012 2014 2000 2002 2004 2006 2008 2010 2012 2014

VW Polo (2.4% market share) VW Touran (1.5% market share) 10 10

2009 REAL-WORLD 8 8 REAL-WORLD -8.2% 2009

-8.0% OFFICIAL -16.6% 6 6 OFFICIAL -19.2%

Nmin = 93 Nmin = 68

Nmax = 518 Nmax = 241 4 4 2000 2002 2004 2006 2008 2010 2012 2014 2000 2002 2004 2006 2008 2010 2012 2014

Build Year

Figure 4. Official and real-world fuel consumption values of popular car models in Germany (continued). Arrowheads mark the timing of major technical overhauls. Minimum and maximum number of Spritmonitor.de entries per year presented in the bottom-left corners.

4 INTERNATIONAL COUNCIL ON CLEAN TRANSPORTATION WORKING PAPER 2015-8 REAL-WORLD FUEL CONSUMPTION OF POPULAR EUROPEAN PASSENGER CAR MODELS

All models included in the analysis pronounced for the VW Passat A4, BMW 3-series, and VW Passat show a decrease in official fuel con- after the model generation B7 was — made little progress, with less sumption values from 2001 to 2014. introduced in 2010. The facelift was than one percent real-world fuel con- This trend is particularly pronounced followed by a 1.0 l/100 km drop in sumption reductions in the same time for the Mercedes-Benz C-Class and official fuel consumption values but period. Among the models evaluated, the BMW 5-series, which achieved a real-world performance remained the Peugeot 206-208 had the highest 40 percent decrease in official fuel virtually unchanged. Other models reduction in real-world fuel consump- consumption values between 2001 such as the Audi A4, Ford Fiesta, tion values since 2009 (about 19 and 2014. These reductions do not Mercedes-Benz A-Class, Škoda percent). The Ford Focus, Mercedes- necessarily translate into fuel savings Octavia, and VW Golf show similar Benz C-Class, Skoda Fabia, and on the road: the divergence between increases in the divergence between Skoda Octavia all reduced real-world real-world and official fuel consump- real-world and on-road fuel consump- fuel consumption by more than 10% tion values ranged from 33 to 51 tion values after technical overhauls. since 2009. percent in 2014. Figure 5 summarizes the reduction The analysis of individual car models in official and real-world fuel con- Reasons for reveals that technical overhauls (see sumption of the 20 selected models the growing gap arrowheads in Figure 3 and Figure from 2009 to 2014. Five models out Real-world driving is a generic term 4), particularly after EU-wide CO of the top 20 — the BMW 5-series, 2 that covers the wide range of driving standards were introduced in 2009, Mercedes-Benz A-Class, Opel Astra, conditions that vehicles are subject are often followed by sharp decreases Opel Corsa, and Toyota Yaris — to during actual use. The real-world, in official fuel consumption values actually increased real-world fuel on-road fuel consumption of cars while real-world values typically do consumption since EU-wide CO 2 varies depending on vehicle and road not decrease by the same magnitude. standards were introduced in 2009. characteristics, driving style, and envi- This development is particularly Another three models — the Audi ronmental conditions. Nonetheless, when looking at real-world fuel con- 5 Toyota Yaris sumption data from a large numbers of vehicles, the results of the vast Mercedes−Benz A−Class majority of vehicles tend to cluster BMW 5−series Opel Astra Opel Corsa around a central estimate, and official 0 VW Passat Audi A4 BMW 3−series fuel consumption values have become increasingly unrepresentative of this Mercedes−Benz E−Class estimate over time. More importantly, Audi A6 Audi A3 average real-world driving charac- −5 VW Golf Ford Fiesta teristics do not change significantly from year to year and, thus, should VW Polo have little impact on the trends from VW Touran one year to the next. -10 Mercedes−Benz C−Class Škoda Fabia Spritmonitor.de data show that the Škoda Octavia share of different driving styles has

Ford Focus remained fairly constant over time, -15 indicating that behavioral changes are not a viable explanation for the increasing gap.6 Figure 2 shows that driving style, ranging from economical Real−world fuel consumption change, 2009−2014 [%] Peugeot 206, 207, 208 −20 to speedy driving, significantly −30 −25 −20 −15 −10 −5 impacts fuel consumption. According to Spritmonitor.de data, economical Ocial fuel consumption change, 2009−2014 [%] driving on average reduces fuel con- Small Lower medium Medium Upper medium sumption by 7 percent compared Real-world = ocial reduction Real-world = 0.5*ocial reduction to balanced driving, while speedy

Figure 5. Reduction in official and real-world fuel consumption since 2009 for 20 6 Spritmonitor.de users can select one of three popular car models. Data points are colored by vehicle segment. The two lines mark driving styles (economical, balanced, or the 1:1 and 1:2 ratios between real-world and official fuel consumption reductions. speedy) when entering data after fill-ups.

WORKING PAPER 2015-8 INTERNATIONAL COUNCIL ON CLEAN TRANSPORTATION 5 REAL-WORLD FUEL CONSUMPTION OF POPULAR EUROPEAN PASSENGER CAR MODELS

driving has the opposite effect with 10.0 a 6 percent increase (both effects being constant over time). In other words, not even economical drivers can achieve the fuel efficiency reported by manufacturers, and the 8.0 Real-world: speedy gap has been widening. Real-world: balanced It should be noted that EU CO2 targets apply to the average for all vehicles sold per year by each manu- Real-world: economical facturer. Vehicle manufacturers are 6.0 therefore not obliged to reduce fuel Official consumption values for every model every year and most vehicles are redesigned every five years or so, with occasional significant mid-cycle facelifts. Following the introduction Fuel consumption [l gasoline/100 km equivalents] 4.0 of all-new model generations and 2000 2002 2004 2006 2008 2010 2012 2014 model facelifts, official fuel consump- Build Year tion values typically decline (see Figure 6. Real-world fuel consumption values for different driving styles, including Figures 3 and 4). Real-world fuel con- economical, balanced, and speedy driving (based on Spritmonitor.de data). sumption values, however, frequently do not follow this trend, indicating removing side mirrors, reducing Conclusions and that improvements in fuel efficiency vehicle weight, and using policy implications measured during laboratory testing tolerances in the measurement do not fully materialize on the road. procedure to their advantage. The analysis of 20 popular vehicle The pattern of sudden increases in models demonstrates that reductions the divergence between real-world • Laboratory test design. Vehicles in real-world fuel consumption fall and official fuel consumption values are tested under conditions that short of the improvements in official indicates that the growing gap is a are unrepresentative of average values. Moreover, official fuel con- product of unrealistic official fuel real-world conditions (e.g., sumption values have become consumption values. ambient test temperatures of 20 increasingly unrepresentative of to 30°C) and regulations allow real-world performance over time. A joint study by Element Energy for flexibilities and tolerances that The pattern of abrupt increases in the and the ICCT identified four key manufacturers are increasingly gap after model overhauls indicate factors for the increasing divergence exploiting. that unrealistic official fuel consump- between real-world and official • New technologies. Some fuel- tion values are the source of the 7 fuel consumption values in the EU. saving technologies, such as growing gap. These factors are primarily related stop-start systems and hybrid to different flexibilities in the vehicle powertrains, tend to be more The growing divergence between testing procedure: effective during laboratory testing official and real-world fuel con- sumption values is primarily a result • Road load determination. The than during real-world driving. of the exploitation of flexibilities in aerodynamic and rolling resis- • Other parameters. Auxiliary the current testing procedure. A tances that a vehicle has to systems, such as air-condition- new test procedure, the Worldwide overcome during driving, the ing and on-board entertain- Harmonized Light Vehicles Test so-called road load, have to be ment systems, are switched Procedure (WLTP), will replace the determined prior to laboratory off during vehicle testing current testing procedure in the EU testing. Manufacturers are but consume energy during in 2017. While the WLTP will be an increasingly exploiting flexibilities real-world operation. improvement, it will not by itself and tolerances in this stage, like resolve all known issues. Even with the introduction of the WLTP, fuel con- 7 Stewart, A., et al., “Impacts of real-world sumption tests will still be conducted driving on emissions from UK cars and vans,” https://www.theccc.org.uk/publication/ in laboratories on pre-series vehicles impact-of-real-world-driving-emissions/ specially prepared by manufacturers

6 INTERNATIONAL COUNCIL ON CLEAN TRANSPORTATION WORKING PAPER 2015-8 REAL-WORLD FUEL CONSUMPTION OF POPULAR EUROPEAN PASSENGER CAR MODELS

(so-called golden cars) and auxiliary — and being able to mandate vehicle low emissions of NOX during systems will continue to be turned off. recalls and impose financial penalties conditions that resemble real-world for non-compliance — is therefore use more closely than laboratory While the exploitation of flexibilities seen as a critical step towards more cycles. This system of on-road testing in the current testing procedure realistic emissions and fuel consump- could be extended to cover vehicle explains the growing gap, the tion figures. CO2 emissions. regulatory and institutional context does little to prevent this develop- Furthermore, the current type- More generally, the current lack of ment .8 Currently, manufacturers can approval process relies solely on transparency in the type-approval shop around in the EU member states, laboratory testing of “golden cars”. process and real-world performance selecting a type-approval authority In-use conformity testing of randomly of cars impedes independent veri- as well as a technical service (third- selected vehicles would ensure that fication of fuel consumption values. party testing company) to certify vehicles meet their specified fuel For instance, road load values are a their vehicles. This situation creates consumption value throughout their vital component of vehicle testing a conflict of interest, potentially useful life. Another measure that but are not publically available. incentivizing type-approval authori- should be combined with in-use There is also no EU-wide source of ties and technical services to support compliance inspections is on-road real-world fuel consumption values. manufacturers in producing favorable fuel consumption testing. For nitrogen In the U.S., the Environmental emission testing results to attract and oxide (NOx) emissions, the European Protection Agency and Department retain the manufacturers’ business. Commission adopted the Real-Driving of Energy created the FuelEconomy. Establishing an EU-wide type- Emissions (RDE) procedure, making gov website,9 allowing consumers to approval authority, acting as a neutral use of portable emission measure- compare real-world and official fuel party between vehicle manufactur- ment systems (PEMS) to verify that consumption values before making ers and technical service companies light-duty vehicles have reasonably purchasing decisions.

8 For more details, see Mock, P., German, J., “The future of vehicle emissions testing”, http://www.theicct.org/future-of-vehicle-testing 9 See https://www.fueleconomy.gov/mpg

WORKING PAPER 2015-8 INTERNATIONAL COUNCIL ON CLEAN TRANSPORTATION 7