On-road emissions tests: How EU failed to change to the fast lane
By Peter Teffer
One year ago on Friday (28 October), EU member states agreed that an on-road test measuring dangerous nitrogen oxides in cars would be introduced, but that initially carmakers would be allowed to exceed the EU limit 2.1 times.
The decision came just a month after Europe's largest car manufacturer, Volkswagen Group (VW), was outed as having cheated laboratory emissions tests .
Dear EUobserver reader
Subscribe now for unrestricted access to EUobserver.
Sign up for 30 days' free trial, no obligation. Full subscription only 15 € / month or 150 € / year.
- Unlimited access on desktop and mobile
- All premium articles, analysis, commentary and investigations
- EUobserver archives
EUobserver is the only independent news media covering EU affairs in Brussels and all 28 member states.
♡ We value your support.
If you already have an account click here to login.
An investigation by this website reveals that the VW scandal accelerated the process, but that it probably could have been finished even earlier.
The so-called real driving emissions (RDE) test for passenger cars (light duty vehicles, or LDV) has been designed in the RDE-LDV working group, in which technical experts from member states, but also representatives from the car industry, took part.
It was finally agreed in the Technical Committee Motor Vehicles (TCMV), a group which only member states are allowed to participate. However, car companies often provided speaking points to countries where they have factories, according to an anonymous TCMV regular.
They said, “You could see seven or eight governments with papers where virtually the only thing that was removed was the logo of the car company”.
This article is based on conversations with direct participants in both groups, who spoke only on condition of anonymity, and documents both leaked and found online in seldom-visited corners of a European Commission web portal.
Since 1992, passenger car emissions have been regulated in EU legislation through the so-called Euro standards.
Nitrogen oxides (NOx) have been subject to limit values since Euro 3, which applied to all cars approved for sale by national authorities after January 2000.
With every new Euro standard, the NOx limit was decreased, from 500 milligrams (Euro 3), via 250 milligrams (Euro 4, 2005), 180 milligrams (Euro 5, 2009), to the current Euro 6 standard of 80 milligrams per kilometer, in place since September 2014.
But when the EU legislation that laid down the Euro 5 and 6 standards was negotiated between 2005 and 2007, there were already signs that while the limit was becoming ever more strict, but real-world emissions did not match the fall.
Regulation 715/2007 of 20 June 2007, which also contained the ban on cheating software known as defeat devices, also received a specific paragraph on the differences between laboratory and real-world emissions.
It noted that the test procedure may need to be replaced and that the use of “portable emission measurement systems” (Pems) should be considered.
According to a summary document by the EU commission, the commission's in-house science body, the Joint Research Centre (JRC), gave a presentation at a November 23, 2010, workshop in Brussels on-road measurements with Pems.
“Euro 4 and 5 diesel cars exhibit much higher NOx on-road emissions (up to 4-5 times the emission limit) than the type approval limit values regardless of driving conditions,” the document said.
It added the commission would change the “test procedures so that they reflect adequately emissions generated by real driving on the road”.
“The RDE test method should be available by the end of 2012 for adoption by comitology”, the summary noted.
Comitology is a way of lawmaking where the European Parliament has no other option than approving or vetoing a bill.
While that may seem less democratic, it is a popular way in Brussels of drafting technical legislation, with the theoretical advantage that it can move much quicker than during the normal procedure, which requires the parliament to come to a compromise with the member states.
In fact, the option of comitology in car emissions legislation has been requested and approved by the European Parliament in the above-mentioned 2007 regulation.
Yet a problem with the comitology process is that it can be untransparent, and heavily reliant on input from industry.
According to the attendance list of one RDE-LDV meeting in 2011, twelve of the 28 participants were employed by carmakers, five worked for automotive suppliers. Only four of the participants represented member states, two were from the commission, and one from an NGO.
When EUobserver asked the commission to release the minutes of the first meeting of the RDE-LDV working group, on 31 January 2011, the commission said the group “focuses on highly technical questions and does not cover any legal aspects”.
Yet, the whole goal of the group was to come up with a test procedure that would be legally required for carmakers.
Nevertheless, the commission told EUobserver that since “invitations, agendas, presentations and various supporting documents” are available online, the commission “did not see added value in producing minutes of the meetings”.
It is not just this website's research hindered by this policy.
Some of the participants themselves complained about a lack of minutes.
One source said, “No minutes were produced, so we had no way of knowing what had been decided. Entire days were wasted”.
Another contact confirmed, saying “you needed a lot of stamina and always be present to remind people of what had been said the previous time”.
He added it was “safe to conclude” that the lack of minutes opened up the opportunity to question things that had already been agreed.
“I don't know if it was tactical, but what I noticed was that all sorts of new data and viewpoints were introduced, leading to a delay.”
This second source also said the agenda was also sometimes missing, and that the commission, which chaired the meetings, should have been more strict at staying on-topic.
“In some cases, the debate was all over the place,” he said.
“Issues about which I thought we had already achieved an agreement, or which were off-topic, were brought to the table endlessly.”
In the first few years, countless hours were spent discussing which method would be used to measure “real driving emissions”.
Initially, the car industry lobby was unwilling to accept that the test cars would actually be taken on the road, using Pems equipment.
While the JRC was in favour of Pems, the European Automobile Manufacturers' Association (Acea) proposed a laboratory test with random elements.
But even a randomised test would still allow a car to recognise that it is being tested and, in theory, allow for cheating software to make the tested car appear cleaner than in real life.
In May 2012, a comparison of the two options was discussed. One of the accompanying documents was an “evaluation matrix”, updated by JRC scientist Martin Weiss.
Originally, the paper said the industry's preferred option had a “high” effectiveness in preventing cheating. But Weiss changed that to “medium to very low”.
In an email to ten member states dated 25 May 2012, a European Commission participant referred to a “quite hot” RDE-LDV meeting.
“It became quite obvious that vehicle manufacturers strongly resist the introduction of PEMS at any stage of type approval,” the email said, noting that the commission was “convinced” Pems were “definitely necessary”.
“Actually this Acea attitude contradicts previous talks I had with individual manufacturers that were ready to accept Pems at least at in-service conformity,” the civil servant added, referring to checking cars that were already sold (“in service”).
The email's author went on to speculate that the industry's resistance may be due to real concerns about the feasibility of the scheme, “or the desire to have the random cycle as an "easier procedure" for emission compliance leaving the door open for some cycle beating”, in other words: gaming the test.
In parallel to the RDE discussions, the JRC, but also several national research institutes, took diesel cars on the road to test with Pems. In several cases they found unusually high NOx emissions whenever the conditions were not the same as in the laboratory test.
Unlike in the United States, where suspicious behaviour of VW diesel cars were investigated by the authorities, no action against individual carmakers was taken in Europe.
This website asked one of the RDE-LDV participants whether European authorities could have known, and why no action was taken.
“Think,” he said. “They included the phrase defeat device, in the 2007 legislation, and the option of measuring on the road.”
“That, and the fact that the RDE legislative process began in 2010, 2011, shows you how early we were aware of the fact that Euro 6 would not deliver what we had expected.”
So why not enforce the legislation and sue car manufacturers?
“If you take one to court, it will take you ten years. What you want, is for the legislation to be solid.” That's why the legislators chose to focus on improving the test, instead of pursuing the individual cases of cheating.
However, discussions for new tests took longer than expected. The end of 2012 – the originally foreseen date by which the legislation should have been finished – had come and gone.
It was becoming increasingly clear that the test would not be ready when the new Euro 6 standard came into force, in September 2014.
In fact; industry group Acea proposed in December 2013 to aim for political agreement on testing NOx on the road by the second quarter of 2017, and the application of the test at the end of 2020.
In March 2014, it wrote in a paper that the new test “should not be rushed to part-completion simply to meet an impractical deadline or as a box-ticking exercise”.
Later that year, in October 2014 member states agreed the test should be introduced in a “two-step approach”. Germany proposed a monitoring phase as of September 2015, and the RDE test as part of the type approval process two years later.
Italy, however, said the first phase should be from 2017 and have a “moderate requirement”, with a “more stringent phase” in place as of 2020.
The delays are acknowledged by the highest civil servant at the commission's environment directorate-general, Karl Falkenberg.
In a letter to his counterpart at the industry directorate-general, where the responsibility for automotive legislation lies, Falkenberg referred to “newly incurred delays” in the RDE process.
“Now that action to address the real world emissions has been postponed several times, the commission will be seen as acting incoherently and even remaining passive facing the evidence on car emissions,” he wrote.
In May 2015, the member states in the TCMV committee agree on the principles of the RDE test procedure, and to begin a monitoring phase by January 2016.
But agreement on important details for the requirements for car approvals, in the second package of rules, was still some months away.
Then, Dieselgate happened.
In September 2015, US authorities managed to persuade VW to admit it had used cheating software.
The revelations accelerated the process, several sources said.
On 28 October 2015, member states and the commission finally voted on the so-called second RDE package, which included the introduction date and the conformity factor – by how much the values of the less accurate on-road test may exceed the actual EU limit.
However, at the last moment, more leniency was given to the car industry.
The JRC had said just a month before that the Pems equipment had an inaccuracy of maximum 30 percent, in other words; that a conformity factor of 1.3 would make sure that cars that are actually below the limit would not be refused approval accidentally .
Even Acea said in a statement on 1 October 2015 that a conformity factor of around 1.7 was “justified”.
But when the file was moved from the RDE-LDV to the TCMV, the issue of the conformity factor was still open to discussion.
The draft legislation typically contains brackets with figures or dates in between, “easy things that politicians can discuss”, said one RDE-LDV source.
Which is what they did. Italy proposed a conformity factor of 3.0 in the first step, while the Netherlands wanted 1.5.
Dutch Green MEP Bas Eickhout has seen classified minutes of the TCMV.
“You can draw conclusions from those minutes that you had a couple of good guys, pretty consistently over the years: Denmark, Netherlands,” Eickhout said in August.
“The bad guys: for example Italy, is quite consistently delaying, asking for more research and delaying. You could say Germany, France, are in the middle, they are not even the worst.”
In written answers to the European Parliament's Dieselgate inquiry committee, the Italian transport ministry said its position was “trying to balance the need to improve air quality and to preserve the competitiveness of the automotive industry”.
The final decision in the TCMV committee was to have a conformity factor of 2.1 in the first phase, and 1.5 in the second phase, “to allow manufacturers to gradually adapt to the RDE rules”.
Environmentalists cried outrage over the decision to let carmakers pollute more, but one source found the conformity factor as “relatively low”.
He said it would have been even higher if Dieselgate hadn't happened.
And cynically, because the average diesel car is emitting about four times the EU limit, 2.1 times will already be a significant improvement.