First look: Results of post-VW diesel vehicle testing in France, UK
On April 7, the French government released initial results of the diesel emissions screening campaign it conducted in the wake of the revelations about VW’s use of defeat devices. The UK and Germany followed suit last week. The testing campaigns were quite similar in nature, and so were the headline results: there is a widespread compliance problem for NOX emissions from diesel cars, and manufacturers have a lot of explaining to do.
French screening campaign
The French technical committee measured emissions from 52 diesel vehicles (mostly Euro 5 and Euro 6) using three different tests:
Test D1 is a modified type-approval test, where elements of the procedure are slightly altered but the certification cycle (NEDC) is kept. This test is visibly aimed at detecting vehicles which use various settings of the laboratory test as triggers for a defeat device.
Test D2 is a modified NEDC, where the first section of the test is different but the emission results are expected to be practically identical to NEDC results. This test is meant to reveal cycle-detection strategies, such as the one used by Volkswagen in their defeat device.
Test D3 is an NEDC driven on the test track with a portable analyzer (PEMS) on board. This test was probably meant to detect defeat devices in a similar way as test D1, and also to get a lower boundary for real-world emissions.
A significant number of vehicles failed the criteria set by the technical committee for each of the three tests (12, 12, and 21 cars, respectively). And these criteria were surprisingly lenient. For test D3, NOx emissions 5 times the regulatory limit were considered acceptable, and the pass-fail criterion for test D2 was the ratio of D2 to D1 results, meaning that an equally bad performance in both tests would actually pass a high-emitting car. Manufacturers whose vehicles failed any of the tests were invited to provide explanations, but nothing is known about this process. Nor is it known which cars failed which tests, as the French results are anonymized.
Here is what is known:
- A large number of the vehicles tested (which are deemed representative of the French fleet) have suspicious emissions behavior according to the criteria set by the French authorities in one way or another. Thirteen vehicles failed Test D1, 14 failed Test D2, and 21 failed Test D3. Only 4 out of 52 vehicles (one Euro 4, one Euro 5 and two Euro 6) met their corresponding emission limit when tested outside of the laboratory.
- The average on-track performance (as evidenced by the results of test D3) is shockingly poor, and the average ratio of on-track to certified emissions of NOX is worst for the newest, Euro 6 vehicles. More than half of the Euro 6 diesels had NOx emissions more than five times the regulatory limit. Only three vehicles had good performance, in line with their respective standard. Bear in mind that the NEDC speed trace is imposes only very low loads on a test vehicle and is not representative of real-world driving, so real-world emissions are bound to be much higher, since NOX emissions tend to be proportional to the load placed on the engine.
- A small minority of the vehicles actually performed well while driving the NEDC cycle on the track. One Euro 5 and two Euro 6 vehicles just met the regulated NOX emissions limit.
The UK measurement campaign covered 38 diesel vehicles (19 Euro 5, 19 Euro 6). The measurement campaign was similar to the French one, but somewhat more comprehensive: in addition to variations of the NEDC on the laboratory (e.g. hot-start, and “reversed” hot-start versions of the cycle) and NEDC on the track, the UK campaign also included baseline type-approval tests (i.e., following the EU procedure to make sure that none of the vehicles were malfunctioning) and an on-road test meant as an approximation of the RDE test (which should yield the emission results closest to real-world performance). In contrast to the French approach, the vehicles were not anonymized.
The UK tests found that average on-road performance was about six times higher than the limit for both Euro 5 and Euro 6 vehicles. That’s a shockingly bad performance for the Euro 5 diesels (we had previously estimated the emissions at 4.5 times the Euro 5 limit), which are much more prevalent on European roads than the relatively new Euro 6s. During the on-track NEDC tests, poor on-track performance seemed to be correlated with low ambient temperature, a further indication that some vehicle manufacturers may be abusing an exemption to the prohibition on defeat devices. That exemption permits emissions aftertreatment to be temporarily reduced or suspended if necessary to protect the engine during extreme operating conditions. But some of the tested vehicles drastically reduce the effectiveness of pollution control systems during even mild weather, e.g., at temperatures of 10 °C or even 17 °C.
These reports from France, the UK, and Germany on real-world diesel emissions and defeat devices bring a measure of much needed transparency to the issue and help inform the public debate. They confirm that average NOx emissions levels are very high for both Euro 5 and Euro 6 diesel cars, across a large number of manufacturers and models. They also show clearly that NOX emission behaviors are erratic at best, with a minority of vehicles having low emissions—and that further explanations from manufacturers and follow-up investigations by national authorities are needed to determine whether they’re worse than merely erratic. With luck, they represent a first step towards addressing some of the ambiguities in the enforcement of defeat device provisions in EU law, and will set the tone for future enforcement actions in several EU member states in relation to the diesel emissions scandal.
All the more regrettable, then, that the testing campaigns fell short of investigating several leads on possible defeat devices, and that lenient evaluation criteria or a willingness to accept the explanations of manufacturers mean that the spotlight on strange emission behaviors isn’t nearly as harsh as it should be. We’ll have more to say about what the tests do and don’t succeed in demonstrating soon.