How a fluke got Volkswagen caught

Volkswagen has been putting software into its cars for seven years that allowed them to cheat emissions tests — and it very nearly didn’t get caught.

The revelation came about because a niche environmental group was trying to encourage Europe to adopt more rigorous emissions standards. In open-road tests year after year, diesel cars from Volkswagen and other manufacturers emitted illegal levels of harmful exhaust, even though they passed lab tests. A small NGO called the International Council on Clean Transportation commissioned a study to prove to Europeans that in countries like the U.S. with more rigorous standards, cars would pass a test in the real world, too.

The ICCT chose West Virginia University’s Center for Alternative Fuels, Engines and Emissions to do the study and decided to test diesel versions of the VW Passat, VW Jetta and BMW X5 — mainly because they were easily available, relatively affordable and met the tech specs.

When an open-road test in California revealed that the Volkswagens spewed as much as 35 times the pollution allowed under the Clean Air Act, the researchers were shocked.

“We expected the opposite. We expected the vehicles would come back with better emissions rates because the U.S. has a better testing program,” said Drew Kodjak, executive director of the ICCT.

Daniel Carder, a WVU engineer who worked on the $50,000 study, told me they mainly wound up choosing the cars because Volkswagen used two different methods of reducing nitrogen oxide emissions (selective catalytic reduction and a NOx trap) and they wanted to test both.

“It was complete happenstance that led us to choose to focus on Volkswagen,” said Kodjak.

The ICCT and WVU published their results online last fall, alerting the California Air Resource Board and Environmental Protection Agency. The CARB launched its own investigation, prompting Volkswagen to voluntarily recall some vehicles and supposedly fix the problem. But when CARB tested the fix, the vehicles still failed.

It was only after the Environmental Protection Agency stepped in that Volkswagen admitted it had programmed its clean-diesel vehicles to detect that they were being put through emissions test and temporarily switch into a low-emission mode that allowed them to pass. Volkswagen mind-boggling decision to use cheating software appears to follow a “long auto industry pattern,” notes the New York Times.

Had Carder and his team selected a different kind of car to test, Volkswagen probably never would have gotten caught.

“When unearthing a long-standing scheme of global proportions to evade emissions standards, we probably shouldn’t be relying on a small NGO randomly testing vehicles,” Kodjak told me.

Diesel vehicles are still a small slice of the U.S. car market. Since 2009, Volkswagen has sold more than 482,000 clean diesel cars with a four-cylinder turbocharged direct injection engine in the United States. But Volkswagen has now copped to installing the cheating software on as many as 11 million vehicles around the world before getting caught. In other words, Volkswagen’s scam managed to pass the test everywhere else, too.

“It points out a woeful lack of standards,” Kodjak said.

Regulators typically test vehicles for emissions output in labs, placing them on giant treadmills that simulate different driving conditions. The code Volkswagen programmed into its vehicles tracked steering and pedal movements to detect when it was was being tested for nitrogen-oxide emissions in a lab and turn its pollution controls on.

Researchers caught Volkswagen by instead using what’s known as a Portable Emissions Monitoring System and taking the cars for long drives on the open road. Kodjak and others want this kind of testing to become standardized, because it’s much harder to fool. (The BMW passed the road test, by the way.)

“No one else is testing,” he said. “If they were, they would have caught it. The emissions were sky high.”

It’s a more expensive and time-consuming way to test vehicles, but given the Volkswagen findings, it seems like a good idea to run cars from other manufacturers through similar tests, too. It’s hard not to wonder whether picking another random car manufacturer would yield the same results.