The emissions test defeat device problem in Europe is not about VW

Ever since the VW defeat device scandal broke last September, the big question has been: Are other manufacturers also using defeat devices? Immediately following the first VW revelations, the French, UK, and German governments commenced testing in-use diesel vehicles aiming to find out. In total, they tested almost 150 vehicles. On April 21–22, they released the first results of the testing programs.

Those results are not pretty. If anything, emissions during on-road tests (outside the test lab and not using the official type-approval test cycle) were higher than we, and others, had found, especially on older (Euro 5) vehicles.

But more important for the future, the testing highlights two things: The justifications being offered for high-emissions outside the type-approval test cycle are mostly bogus from an engineering point of view. And the only certain remedy for that is to make the EU’s type-approval framework regulation explicit and unambiguous about how manufacturers must propose necessary exemptions from the defeat device prohibition, how they must be evaluated, what constitutes normal operation of the vehicle and engine, and how all that is to be monitored and enforced. Unfortunately, the current proposals to revise the framework regulation do not make those changes.

While there are three or four examples from the German test program of high emissions when the test cycle is varied slightly, the reports attribute the emission increases on most vehicles to two factors: ambient temperature (the so-called thermal window) and hot restarts. There may be other factors that the test protocols were not equipped to uncover (for reasons Rachel Muncrief outlines) But I’ll concentrate here on these two supposed explanations.

The low ambient temperature pretext

The three reports focused most on the ambient temperature window. I’ll pick on the UK report, simply because it did an excellent job of disclosing the test results and the reasons for its conclusions.

As shown in the chart below, taken from the report, 33 of the 38 vehicles tested had emissions at least twice the regulatory standard when tested on a track using a portable emissions measurement system (PEMS), and many were at least five times the limit. The report discounted these results because the vehicles were run at colder ambient temperatures:

Manufacturers explained that the amount of EGR [exhaust gas recirculation] which can be used is dependent on ambient temperature. Problems occur at low temperatures when moisture condenses on to the EGR valve and pipes and traps soot from the exhaust gas, leading to a build-up of deposits. Low intake air temperatures can also result in higher engine-out soot levels, exacerbating the problem. [para. 5.29]

In other words, manufacturers say that they need to reduce the percentage of EGR at low ambient temperatures to protect the engine—and protecting the engine is one of the exemptions the type-approval regulation provides to allow changes in the emission controls that would otherwise be considered a banned defeat device.

chart, UK test results

That explanation is misleading at best. Yes, condensation and higher engine-out soot levels are legitimate concerns when calibrating an engine. But ambient temperature has only an indirect impact on condensation. Intake manifold temperature is the proper measure. Certainly intake manifold temperature is influenced by ambient temperature, but it is also influenced by engine temperature, the turbocharger, and even exhaust gas recirculation (exhaust gas temperatures are much higher than intake air). Proper calibration would limit EGR at colder temperature after a cold start, when the engine is cold, but would restore EGR as the engine and exhaust temperatures rise. Instead of doing the hard work to accurately assess condensation as a function of all of the relevant factors, manufacturers are simply deciding that condensation might occur at temperatures below 10 °C or 17 °C and are shutting off the EGR, even when it is not necessary.

Here is some evidence that manufacturers are unnecessarily turning off the EGR:

  • The chart is sorted by the ambient temperature at which the UK tests were conducted. As noted previously, five of the 38 tested vehicles had emissions below the regulatory limits. All were Euro 6 vehicles, and, ironically, two were made by VW. Four of these passing vehicles were tested at 10 °C, illustrating that EGR does not need to be reduced at that temperature.
  • Ten of the 53 vehicles tested by the German KBA met the emission standards when tested in the laboratory at 10 °C. Two, both using VW’s Euro6 2.0L diesel, met the standards when tested on-road at temperatures below 10 °C: an Audi A6 tested under defined test cycles down to 6.4 °C, and a VW Touran tested under defined test cycles down to 7.0 °C. The Touran even had emissions below the standard when tested on the RDE at 8.5 °C.
  • After testing by the French government, Renault agreed to voluntarily extend the operating range for full functioning of its EGR system on Euro6b in-use cars (so, not just on new production) to between 5 °C and 40 °C. Renault had been reducing EGR operation below 17 °C and above 35 °C. The change essentially doubles the ambient-temperature range of full operation, and suggests that engine calibration can be modified to continue full EGR down to at least 5°C.

Significantly, the U.S. Environmental Protection Agency’s guidance to heavy-duty engine manufacturers on condensation remedies accepts low ambient temperature alone as a justification for reducing EGR only below -4°C, and it does not permit EGR reduction at all above 10°C. That guidance document states, simply: “In our view, a condensation protection AECD [alternative emissions control device, which in the EU regulation would be categorized as a defeat device—see here for details on this aspect of the two regulations] based solely on ambient temperature might double vehicle emissions, and such an AECD would not be acceptable to EPA in any circumstances.”

The hot restart pretext

At least the three reports raised and discussed the ambient temperature issue. Remarkably, none seriously address high NOx emissions after a hot restart. Thirty-two of the 38 vehicles tested by the UK had higher emissions after a hot restart than after a cold start. Similarly, 48 of the 53 vehicles tested by Germany had higher emissions after a hot restart. (Again, VW actually did better than many: four of the five vehicles with lower emissions were VW products). Every manufacturer tested had examples of higher NOx emissions after a hot restart in one of these test programs.

chart comparing German and UK hot restart results with US EPA data

NOx emissions from cold-start and hot-restart tests by Germany and the UK, compared to similar test data from U.S. EPA

Only the UK report discussed the issue at all, so I’ll pick on them again. (The French did not include a hot restart test in their protocol.) The report dismissed the high emissions on the hot-start test: “NOx emissions are generated by high peak temperatures and pressures during the engine’s combustion process. A fully warm engine might therefore be expected to generate higher NOx emissions during an NEDC test than an engine which has started from 25 °C.” [para. 5.26]

This is factually wrong for several reasons. Peak combustion temperature is dominated by the heat of combustion itself; engine block temperature has a relatively minor impact on it. Also, as noted above, full EGR cannot be used with a cold engine, but it can be with a warm engine, and the impact of EGR in reducing NOx emissions is much greater than any impact engine temperature has on peak combustion temperature. Finally, NOx aftertreatment such as selective catalytic reduction or a NOx trap, with which most Euro 6 and some Euro 5 cars are equipped, is ineffective after a cold start until the catalysts heat up, while it can be effective from the first second of a hot-start test.

To give a sense of just how far out of bounds the UK and German results are, we can take a look at data from diesel testing on the U.S. Federal Test Procedure (FTP-75) cycle. The FTP includes both a cold start, with the exhaust captured for analysis separately (literally in a big plastic bag, numbered bag 1), and an engine restart after the engine has been shut off for 10 minutes, with the exhaust again captured for separate analysis (in a different bag, bag 3). Both these test phases use exactly the same drive cycle.

The U.S. EPA provided us with data from testing on 30 model year 2009 to 2015 diesel ligh-duty vehicles. On average, bag 3 NOx emissions were only 12% of bag 1 emissions, and there were no vehicles that had higher emissions on bag 3 than on bag 1 (the highest ratio was 58%). Also, as shown in the chart above, hot restart NOx emissions averaged only 8 mg/km on bag 3, versus 80 mg/km on the Euro 6 UK vehicles over the NEDC cycle. The 30 Euro 6 vehicles in the German report had even higher average hot restart emissions on the NEDC cycle, 110 mg/km. And this is despite the fact that the FTP is a higher-load test cycle (that is, with more, and more rapid, accelerations that impose more load on the engine) and generates higher peak combustion temperatures than the NEDC. Even a Dodge Ram full-size pickup truck averaged only 11 mg/km on the bag 3 hot restart test.

This is a much bigger problem than VW

The methods used by other manufacturers to tell when a vehicle is not on a test cycle, such as ambient temperature and hot restarts, may be different than VW’s, but they are still improperly reducing the effectiveness of emission controls in the real world. There is no engineering justification for always reducing EGR flow rates at temperatures above 0–5 °C. There is no engineering justification for higher NOx emissions after a hot restart. Despite the claims that these strategies fall under the defeat device exemptions, these are every bit as much a defeat device as the VW strategy. As the report on testing by the German transport ministry and type-approval agency acknowledged (pp. 119), calibrations that reduce emission control effectiveness at colder ambient temperatures are defeat devices.

Ultimately the problem traces back to the regulation, and it must be fixed there. As the German report also noted (pp. 122–123), the lack of specific guidance in the EU regulation may make it difficult, in some cases perhaps impossible, to contradict and overrule manufacturers’ claims about what they need to do to the emissions control system to protect a vehicle engine in “extreme” conditions. That lack of clarity needs to be resolved. Manufacturers must be required to disclose the presence of defeat devices as defined in the regulation and explain them in detail (a reform that the German transport minister announced at the same time as his ministry and the KBA released their testing results). Regulation must unambiguously state how permitted exceptions to the prohibition on defeat devices are to be claimed, evaluated, and approved or rejected. And the regulations must clearly define key terms and concepts such as “normal operating conditions,” “normal vehicle operation and use,” and “emissions control system.”

I worked for automakers (Chrysler and Honda) for 20 years, and I can tell you that calibration engineers have a tough job. They have to maximize fuel economy and performance, ensure great drivability and low emissions, and reduce costs—all on impossible deadlines. Given an opportunity to find a quick solution—like using a simplistic ambient temperature cutoff that they know will prevent condensation problems under all conditions, instead of investing the time to properly assess the impact of a variety of operating parameters on condensation, or deactivating emission controls after a hot restart instead of taking the time to determine the impacts of different loads on engine temperatures and only changing calibrations when it is really necessary —they will take it. While it is disappointing to see manufacturers take such a cynical view of the enormous health impacts from their actions, it is no solution to say they shouldn’t. “Everything which is not forbidden is allowed” is a well-established argument in law. Whether you accept or reject the manufacturers’ arguments, in practice we should all be able to agree that we should end these disputes about defeat devices by forbidding them more clearly and explicitly in the EU regulation.

Vehicle testing