Major automaker Volkswagen got a jolt last month when it received a notice of violation from the U.S. Environmental Protection Agency (EPA) that outlined deliberate noncompliance with the U.S. Clean Air Act. VW publicly apologized for knowingly installing “defeat devices” on several diesel models of its cars—the software on car computers cheats vehicle emissions tests so that the VWs spew lower levels of pollutants in testing facilities than the cars actually do in the real world. The recent scandal has led to the resignation of VW CEO Martin Winterkorn.
Physics alumnus John German (B.S. ’74) played a major role in uncovering VW’s fraud. German works with the International Council on Clean Transportation (ICCT) as a senior fellow; was Honda America’s manager for environmental energy and analysis; has served as a member of the Ann Arbor Environmental Commission; and was a senior automotive technical advisor for the EPA. He helped conduct the experiments that showed suspicious differences between VW vehicle emissions test results in certified facilities and VW emissions measured under actual driving conditions on the road.
LSA got in touch with German to get his perspective on what went wrong with VW.
John German: It all started when we set up an experiment to verify that clean diesel cars actually can be practical to make, sell, and drive. We expected that diesel cars in the United States would be clean, given its stricter emission standards and regulatory agencies with more legal authority and expertise in compliance than anywhere else in the world. We thought it would have a positive impact on car markets in Europe if we could show that the technology could produce low enough emissions here in the United States.
But we ended up finding the opposite. The VW diesels we tested on the road in California were up to 35 times dirtier than we expected, 35 times dirtier than allowed by U.S. law, and 35 times dirtier than VW advertised.
When you buy a passenger vehicle, it’s been certified as meeting emissions standards based on measurements made in laboratory conditions, in emissions testing facilities. Our study, a collaboration between the ICCT and West Virginia University, was part of an ongoing analysis begun in Europe in 2012. It showed that the NOx [nitrogen oxide] emissions from the VW vehicles were higher than expected in the real world, even during routine driving.
It just didn’t make sense. That was the real red flag for us.
Can you describe the fraudulent technology that VW used and why it’s important?
JG: The quick definition of their “defeat device” is that it’s installed software that tells the car’s onboard computer when it’s on the official emissions test cycle and when it’s not. And when it’s not, the car changes how it controls its emissions.
This trick ends up giving VWs better gas mileage and possibly reduces the cost of the NOx control hardware, but the cars also produce far more pollutants than is allowed by law in the United States and Europe.
When you’re checking for compliance with standards, you need to design a test that can be repeated precisely for all the cars you’re testing. Turns out that those precise, repeatable conditions made it possible for VW’s defeat devices to cheat on the test.
It’s not clear yet how the defeat device software works—it could detect any of a number of standard, uniform factors on the official tests, such as the ambient temperature; a certain sequence of car speeds; tire air pressure; when the steering column isn’t turning but the wheels are (which indicates that the car is rolling on stationary dynamometers during testing); when the drive wheels are turning but the non-drive wheels are stationary (tests are usually conducted on a 2-wheel dynamometer); engine temperature during vehicle start; or when anti-collision systems have been turned off (which must be done during emissions tests).
Were you surprised by what you discovered?
JG: We were shocked and astounded when we saw the numbers. We thought the vehicles would be clean. We didn’t set out to show that VW was doing something wrong.
The make and model of the cars we tested actually were partly determined by availability: For example, we initially selected a Mercedes-Benz as one of the test cars, but when the time came to begin the on-road testing, the rental agency had sold that car, and there was no replacement for it at any rental agency in the state of California.
A difference between emissions measured in labs versus on roads actually is not uncommon with diesel engines. NOx emissions from diesel engines can “spike” when the driving conditions are more demanding, such as with harder accelerations or driving uphill. But the VWs we tested massively exceeded their official emissions readings in normal driving conditions, which was completely inexplicable and totally surprised us.
We had no idea that this was out there—especially because another cheating case happened in the late 1990s, when heavy-duty trucks equipped with other defeat devices prompted $1 billion in fines and corrective actions required by EPA. After that scam, we didn't think another manufacturer would take this kind of risk again.
I work with the ICCT, a small organization that primarily deals with using data and analyses to assist government regulators worldwide. Our organization’s goal is to reduce emissions and improve automotive efficiency. It’s always gratifying to see results from our work, but we never dreamed we would have this kind of impact. These results with VW cars are unprecedented and overwhelming.
How serious is VW’s infraction?
JG: It’s the sort of thing you can’t just go around accusing companies of doing unless you’re absolutely sure.
The kind of software required to detect when the car is driving on the official test is very sophisticated and would not be easy to develop. And then you need duplicate software to tell the car to have two different emission controls. It’s writing the code, but it also involves validation tests to ensure that the system is operating as designed. So someone had to take these vehicles out, test them on the standard test cycle, and make sure that the emission controls worked when they’re supposed to be working. All of that concerted effort toward installing defeat devices implies deliberate artifice on the part of VW.
There’s no way to know how many people at VW knew about the deceit, but quite a few people would have had to be involved. It certainly couldn’t be just one individual.
We sent the results of our experiment to VW, CARB [the California Air Resources Board], and the EPA in May last year. What’s really amazing is that VW had a chance to fix the problem, and they continued to try and cheat. VW issued a field fix in December and assured CARB and EPA that this would fix the problem—VW recalled the affected vehicles and reflashed the computer memory to install the updated software. Then, in May 2015, CARB obtained some vehicles that had been recalled and tested them again. Emissions were lower, but they still exceeded the standard—and the software still contained the defeat device. This is stupidity beyond belief.
What are the implications? What would you recommend to consumers or to automakers?
JG: Controlling NOx emissions from diesel passenger cars is one of the biggest technical challenges facing automakers. While we have no information or data suggesting that other manufacturers are also using defeat device, we do need to ask the questions: Is this VW scheme happening in other countries? Is this happening at other manufacturers?
Our results strongly support the need to add on-road testing to existing enforcement requirements. To make real progress, emissions results recorded in the laboratory must translate into the emissions experienced on the road, which brings us to a simple point: Regulations must include compliance and enforcement—real enforcement, not mere decoration.
This incident with VW highlights the global need for vigilant enforcement of air pollution laws by regulatory agencies in all vehicle markets, along with stronger legal authority for regulatory agencies. Without vigilant enforcement, automotive companies that actually do comply with standards will be placed at a competitive disadvantage and will be pressured to also cheat. That's why the actions by the EPA are so important.