What is fuel consumption, how is it measured and is it accurate?

LOCALLY AND IN Europe, we use the term L/100km, or litres per 100km to describe fuel consumption in a car and thus make a judgement call on whether the car is fuel efficient or not, as well as allowing us to work out potential fuel costs of a car we’re looking to buy.

In some countries, like the US and UK, miles per gallon (MPG) is the preferred descriptor. But in the US and UK, a gallon is not measured the same, indeed a gallon in the UK (imperial measurement) is equal to 1.2 gallons in the US.

Because fuel consumption testing in Australia is the same as Europe…

…it’s possible to read a car review from an overseas publication (let’s confine it to UK figures) and calculate its MPG as L/100km. An imperial gallon equals 4.55 litres (a US gallon equals 3.785 litres), and 1mpg (one mile per gallon) equals 282.5L/100km. So, if you want to take an MPG figure you’ve read somewhere, say, 40mpg for example, you would divide 282.5 by 40, which gives you 7L/100km. If you want to work out the MPG from a L/100km figure simply go the other way, meaning 282.5 divided by the L/100km figure.

Since 2009 all cars carry a fuel consumption and CO2 sticker

If you’ve purchased a new car in the last few years (since 2009) then you’ll have noticed a colourful sticker on the windscreen that details the car’s claimed fuel consumption, listing its Urban, Extra Urban and Combined fuel consumption. It also lists the car’s claimed emissions (CO2 per kilometre).


Fuel consumption test method dictated by ADR81/02 which copies NEDC

So, how are the results on the sticker extracted? Since 2009, car makers have been required to submit their vehicles to a mandated fuel consumption test under ADR81/02 Fuel Consumption Labelling for Light Vehicles. It’s a pretty dry document, but it dictates the parameters (a dynamometer test conducted under laboratory conditions) for testing to ensure that all fuel consumption claims for vehicles sold in Australia are comparable even if they’re not reflective of real-world driving conditions.

There are a number of different types of dynamometer, but as far as this fuel consumption test is concerned, the dynamometer is a ‘rolling road’ with either one set of rollers for front-drive cars, or two for measuring all-wheel drive cars. It looks like this:

fuel consumption testing - rolling road

So, you know ADR81/02 follows United Nations mandated testing methods, which in Europe is known as the New European Driving Cycle (NEDC) and was introduced in 1997. But it will be replaced by a new test method from 2020 (detailed below).

How is fuel consumption calculated?

The dynamometer-based testing mandated under ADR81/02 is a simple 20-minute cycle that’s split into two sections. These are:

Phase One (10 minutes) is better known as the Urban Cycle and involves a peak speed of 19km/h, frequent stop-start moments, and idle periods (up to three minutes). This test is intended to simulate situations experienced in stop-start traffic.

Phase Two (10 minutes) is better known as the Extra-Urban Cycle and this involves the vehicle being accelerated to achieve an average speed of 63km/h with speeds peaking at 120km/h (for just 11 seconds). Some might suggest the test should be conducted at a constant speed of, say, 100km/h for a more accurate simulation of real-world use… but, when this testing methodology was introduced in Europe in 1997 the differential between the lab results and the real world was only about 10%, now the differential is up to 30%.

Once the two fuel consumption figures have been calculated, these are then combined to determine the combined average fuel consumption across 100km, and also the vehicle’s average CO2 emissions per kilometre.

Is there debate about this system?

Yes. There’s numerous evidence to show that the laboratory-based dynamometer test isn’t reflective of real-world driving conditions and that it should be updated to be more accurate. And moves are already afoot with Europe set to introduce significant changes to the laboratory-based dynamometer test (see below).

According to the UN, “NEDC imprecision allows for modifications of a vehicle that is used for coast-down experiments, for example, replacement of normal road tyres by conditioned low resistance tyres, atypically high tyre pressures, manual adjustment of brakes, etc”. The UN claims the testing methodology was more accurate when it was introduced in 1997, but that it has fallen way behind new vehicle developments and technologies and thus needs to be updated, but the UN doesn’t suggest that ‘real-world’ measuring of fuel consumption is the answer.

However, the UN does suggest that diesel vehicles should be tested in the real-world for emissions, which is what the peak body for Australia’s motoring clubs, the Australian Automobile Association (AAA) began doing between May and October this year. The AAA commissioned a study of 10 new cars and ran them around Melbourne for about two hours across a set course and measured the emissions. It found the vehicles emitted 20%-plus more emissions than the manufacturer claim.

The UN itself has admitted that changes to lab-based fuel consumption testing (outlined below) will have only a 1% effect on emissions results, but a much greater effect of fuel consumption results and is intended to make them more reflective of real-world driving. It’s worth noting and hardly anyone ever does in their criticism of lab-based fuel consumption testing is that the aim is relative accuracy and not absolute accuracy.

The aim when looking for a new testing method must surely be to determine a ‘standard’ for fuel consumption measurement that allows for comparison between vehicles. That’s what the current ADR81/02 and NEDC do, they just haven’t kept pace with changing vehicle technologies. But that’s all about to change…

What alternative systems of measuring fuel consumption are there?

The United Nations has developed the Worldwide Harmonized Light Vehicles Test Procedure (WLTP), something it’s been working on since 2007 (a couple of years before Australia introduced the then 10-year-iold measurement method of NEDC), because it claims the gap between the current lab-based dynamometer (rolling road) test and real-world fuel consumption is up to 30%.

But real-world fuel consumption testing isn’t viable and nor would it be comparable. Say, for instance, a Peugeot 308 was tested on a road loop around Paris and then the figures were applied to vehicles sold in Australia, or vice versa. The real-world fuel consumption test wouldn’t be real world for Australia, and it wouldn’t be repeatable. More than that, it’s expensive. Nope, what’s needed is a repeatable, relative accuracy of testing that allows for comparisons between vehicles tested either in Australia or Europe, because the methodology and conditions will be identical.

Indeed, the UN argues that real-world testing of vehicles is more relevant for emissions rather than fuel consumption. And something will need to be done to more accurately measure emissions, because as the UN has said, WLPT will only increase emissions results by around 1% (and thus won’t be real-world reflective), but that it would have a more significant impact on fuel consumption because it would disallow vehicle modifications which are allowed under current testing, and introduce a more dynamic testing regime.

For instance, there will no allowable changes to tyres, in that they must be the tyres the vehicle will be sold with, they must contain within 80-100% of their tread depth, they cannot be pre-warmed, the tyre pressure must be the recommended psi for the tyre, the wheel alignment must be as per how the vehicle will be sold, the vehicle body must be stock standard, and the same for the brakes. And the ‘warm-up’ period will be 20 minutes at 118km/h. The vehicle’s test weight will be kerb weight plus 100kg.

In short, the aim is to provide a greater dynamic driving range over a longer period of time with more real-world ambient temperatures int he laboratory, conducted with stock standard vehicles with no modifications and maximum weight applied. For instance, the run times will increase and the stop times will decrease, meaning stop-start systems will no longer be able to skew results. Constant driving times will be reduced (down from 40.3% currently to just 3.7%) with more focus on acceleration and deceleration periods (43% and 39%, respectively).

Fuel Consumption changes
This table shows the difference between lab-based testing of vehicles under the current NEDC and ADR81/02 methodology and the new WLTP methodology which will come into play from 2017 in Europe and be official testing methodology from 2020. Note: WLTC stands for World Harmonised Light Vehicle Test Cycle.

In Europe, WLPT will be applied for EU Type Approval from September 2017 but it will run concurrently with the current testing methodology meaning two sets of results will be available. Why run them concurrently? The aim is to ensure the testing methodology proposed under WLPT is workable. It is also intended to give car companies time to ensure budgets are available to cover the increased cost of a longer test cycle under WPLT.

In 2020, the current NEDC will be phased out with WLPT the only recognised testing methodology for fuel consumption on petrol, diesel and hybrid light vehicles which includes both cars and SUVs.

There’s no word on whether Australia will phase in WLPT from September, indeed Australia still hasn’t adopted Euro 6 emissions targets, but it’s likely that ADR81/02 will be updated by 2020 to reflect the new European fuel consumption test.

What does all this mean for car buyers?

Well, if you’re trying to determine potential fuel costs of cars you’re considering buying then, until new testing methods are introduced, use the existing figures but add around 25% to the cost to give you a more accurate impression of costs. And, if you’re buying a car and only considering the cost of fuel in your budget then you’re likely to run into some financial issues down the track…

…See while plenty of shock-and-awe journalists, including CHOICE, have said this non-real-world-reflective test means that car buyers are out of pocket isn’t entirely accurate. There’s more to the cost of running a car than fuel only; there’s financial considerations like tyres, servicing and particularly deprecitation.

It’s worth remembering that whether they’re real-world reflective or not, the fuel consumption figures achieved under ADR81/02 or NEDC mean you can accurately compare one vehicle on your shopping list with another. 

So, is all this hue and cry about fuel consumption inaccuracy a bit of a mountain out of a molehill? Yes, yes it is. Sure, we know the test is out of touch, but there’s a new test coming that will get the results back to within coo-ee of the real world and continue to allow fuel consumption comparison. So, let’s start putting pressure on local authorities to update ADR81/02 with the methodology outlined in WLTP.


Looking ahead at 2017 in motorsport...


Skoda… um, Trekka is 50 #throwbackthursday


  1. It’ll never be “right”. When driving work fleet cars, I would often get 20% better than the fleet average l/100km – and some drivers would get consistently 25% worse. Driving style, no doubt.

    Over 60,000km, I averaged 4% BETTER over 60,000km in my FOCUS Diesel than the rated 5.6l/100, and averaged only 4% worse than rated 4.4 figure on my last car, a FIESTA Diesel, so those rated figures are OK in my eyes. Both of those cars improved over the time I had them.

    BUT – I can’t get near the figure on my new PRIUS which is rated at 3.4l/100km, best I’ve averaged from a tank was 3.8, average is almost 20% worse over 9,000km. Hopefully it improves over time??

    It could be attitude too – when you’re getting 3.8l/100, the dollar $aving$ trying to drive more economically aren’t as anything like as much as if the car is using 12l/100km.

    1. For a hybrid (or battery) car, air temperature is very important as it affects the efficiency of the batteries heavily. In warm weather I can get the 3.9l/100km of the prius c but in cold weather, no way.

      1. Cold weather also affects the way the petrol motor kicks in to keep it at its most efficient coolant temp at 90C. Read up on Priuschat. The folks in the US driving in snowy conditions block off the bottom air pickup at the bottom of the bumper that cools the radiator and that in turn keeps the motor hotter to reduce petrol being burnt to keep the petrol engine warm. Also doesn’t help if we use the heater that also dissipates heat during winter.

      1. yes, could be TOYOTA – though I’m not even sure of LandCruiser.

        We had a 5spAuto Diesel Wagon at work in western Queensland – I took it on many long runs, almost dead straight, dead flat roads, all driven on Cruise Control 100-110km/hr (speed limit). Unloaded, generally only one person (except for a second spare tyre). Best I got was 16% WORSE than the 9.5l/100 rated (country figure).

  2. OMG they are going above the speed limit! Quick someone fax Harold! There’s a car on a dyno going over the national limit!!

  3. VW and Audi (the latest wheeze being lack of steering wheel deflection to invoke test mode) have shown that makers game the system. The current tests do not offer an accurate comparison between makers. No self-administered test could. The divergence from real world figures has grown as the “gaming” has improved. Some makers put more effort into this than others. The new tests will be subject to the same effect over time. It will be instructive as to the divergence between the old and new test results. Those with the biggest discrepancies are the ones to query down the road. Surely makers gather fuel stats on their cars in real world use through their ECUs. The resulting bell curve should be published. That’s another way to be the manufactures honest.

Leave a Reply

Your email address will not be published. Required fields are marked *

Check Also