Volkswagen’s emissions cheating scandal had a long, complicated history

In mid-September, the US Environmental Protection Agency dropped a bomb on Volkswagen Group, the German company that owns Volkswagen, Audi, Porsche, Lamborghini, and other notable car brands. The EPA sent the umbrella company a Notice of Violation, explaining that it discovered “defeat devices” on Volkswagen and Audi diesel passenger cars from 2009 and later.

The defeat devices—actually less a “device” than code on the cars’ electronic control module that detects whether a car is in a lab or on the road—were preventing the cars’ emissions control systems from working properly while the car was operating under normal driving conditions, likely boosting the car’s performance or fuel efficiency rating or both. The EPA said that nearly 500,000 of these diesel cars were caught spewing emissions well in excess of the federal rules, sending the company’s stock into a tailspin.

Just a few days later, Volkswagen announced that 11 million vehicles worldwide were outfitted with the emissions-boosting code, although not all of the cars had the dubious “feature” enabled, according to the company. Still, that sent Volkswagen’s stock into even more of a tailspin, and the company’s CEO was ousted within days. Multiple countries are now investigating Volkswagen for fraud, although in the US, Volkswagen might escape a criminal charge for violating the Clean Air Act (CAA), due to loopholes built into the Act by auto-industry lobbyists, The Wall Street Journal reports.

But for all this scandal in headlines, the whole debacle is not the first, or even the second, or even the third time the EPA has discovered automakers skirting emissions standards. And the EPA’s rules and testing procedures are complicated enough that automakers have argued in the past that their defeat devices were legal because they solved performance problems on a car model while allowing it to pass the federally mandated tests. Often, car makers accused of defeating emissions control systems reach a settlement with the EPA or the Justice Department, but never admit guilt.

Because defeat devices have such a complicated history, it’s helpful to take a look at what a defeat device is, how the EPA tests for them, and what’s happened when they’ve been found in the past, to get some context for Volkswagen’s most egregious breach of public trust.

Hit the rulebook

The US Code of Federal Regulations (40 CFR 86.1803.-01), defines a defeat device as an “auxiliary emission control device (AECD) that reduces the effectiveness of the emission control system under conditions which may reasonably be expected to be encountered in normal vehicle operation and use.”

But the rules don’t ban defeat devices wholesale. The EPA can approve auxiliary devices to hinder the emissions control system on a vehicle—but the Agency has to know about it. Defeat devices can be allowed on emergency vehicles, for example. In a 2014 EPA document (PDF), the Southern Group of State Foresters reasoned that vehicle performance should always be prioritized over emissions control in emergency situations:

Numerous injuries and fatalities of wildland firefighters have occurred when dozers became disabled due to terrain such as becoming stuck or various mechanical reasons, while constructing firebreaks in front of advancing wildfires. It would obviously be extremely dangerous to have dozers with emission control systems or settings that could lead to dozers losing power, speed, or torque while constructing firebreaks in advance of an oncoming wildfire.

The tests

But passenger cars and commercial vehicles rarely need emergency torque. And so, automakers are required to complete a number of tests in a lab on a representative engine and submit those results to the EPA in order to show that the car they’re about to sell in the upcoming year is compliant with federal emissions standards.

The EPA can randomly check automakers’ testing procedures, a practice called confirmatory testing. Notably, if a manufacturer is selected for confirmatory testing, the manufacturer provides the test engine or vehicle, and if it fails to pass the first test, the manufacturer can modify the engine (PDF) before a second round of confirmatory testing.

But setting aside confirmatory testing, standard testing procedures are different in specifics for heavy-duty vehicles (tractors, pickup trucks) and light-duty vehicles (passenger cars). For passenger cars, automakers must pass a Federal Test Procedure, or FTP, as well as Supplemental Federal Test Procedures (SFTP).

The current primary test procedure for emissions is called FTP-75. The car’s engine is hooked up to a dynamometer and run for about 31 minutes, testing it from a cold start, in a stabilized phase, and then with a hot start. “Emissions from each phase are collected in a separate teflon bag, analyzed and expressed in g/mile (g/km),” DieselNet explains. The car has to “travel” for 11.04 miles at an average speed of 21.2mph, with a maximum speed of 56.7mph.

Then, the car has to complete SFTPs to account for “aggressive, high-speed driving” as well as driving with the air conditioning on.

Gary Bishop, a research associate at the University of Denver who has worked on building systems to remotely monitor vehicle emissions in the real world, told Ars that the EPA’s tests set the manufacturers up for success, even if it’s not entirely warranted. “If anyone (including car manufacturers) knows the test ahead of time, why would anyone ever expect you to flunk that test?”

“One thing most people are not aware of is that manufacturers will have specific drivers who drive certain models because they can legally drive the test and produce the lowest emissions for that model,” Bishop continued. “[It’s] not cheating but one [can] expect that vehicle to behave differently off the test with real drivers.”

Bishop’s research partner, University of Denver Professor Donald Stedman, added that achieving a compliant but market-viable vehicle is often a difficult compromise. “Drivers want optimum power, performance and fuel economy, the EPA wants passing the test,” Stedman wrote to Ars. “This is why all auto companies hire very smart engineers, because these goals are often not compatible.”

A short and incomplete history of the defeat device

The past 50 years have shown that when those very smart engineers can’t quite make the EPA’s increasingly rigorous goals compatible, automakers will fall back on the fact that they only have to pass the EPA’s test—they don’t have to create a perfect car. That’s where defeat devices come back into play.

Volkswagen (the first time!) and others

The Clean Air Act (CAA) that regulates car emissions today was passed in 1963 and was revised a few years later to require stricter emissions standards on cars. Car makers, unsurprisingly, balked, and fought the EPA to extend the deadline for meeting the new standards.

Right about that time is when the EPA discovered the first analog defeat devices, with Volkswagen, oddly enough, being named one of the first violators of standards using the devices.

In a January 1974 Report to Congress (PDF), the EPA wrote that in the previous year it had opened an investigation concerning “the failure to report the existence of and the use of possible defeat devices by Volkswagen on a substantial number of 1973 model year vehicles” (page 43).

In an EPA press release from July 1973 (PDF), the administration noted that Volkswagen sold around 25,000 cars with temperature-sensing switches that were used to deactivate the emissions control system. Specifically, Volkswagen’s Fastback and Squareback 1973 models would sense low temperatures and cut out the cars’ exhaust recirculation system. In addition, 1973 VW buses had switches that would override “the transmission controlled spark-advance system at low temperatures.”

A year later Volkswagen, based in what was then West Germany, agreed to pay $120,000 (PDF) to settle the charges but did not admit any wrongdoing.

Around the same time, the EPA also reprimanded six manufacturers—GM, Ford, Chrysler, American Motors, Nissan, and Toyota—for installing devices which would “defeat the effectiveness of emission control systems under conditions not experienced during EPA’s certification testing.” The EPA ordered those devices to be removed from cars that were still to be produced, but it didn’t order the recall of cars with devices that had already come off the assembly line. While the defeat devices apparently took different forms, The Sarasota Herald reported in late 1972 that the defeat devices found in the cars also activated in cold weather “in order to make the car start more easily.” Alternatively, some models were equipped with time-delay switches that cut the emissions control system as an automatic transmission shifted from low to high gears.

GM’s Cadillac Crisis

Between 1991 and 1995, GM sold approximately 470,000 Seville, DeVille, Eldorado, and Fleetwood model Cadillacs with 4.9L V8 engines that turned off the emissions control system when the driver turned on the air conditioner.

In December 1995, the Justice Department sued GM on behalf of the EPA in order to force GM to recall those vehicles. GM agreed to what amounted to a $45 million settlement, including $11 million in fines, $25 million to fix the recalled cars, and just under $9 million in a “community service” penalty.

GM allegedly designed its engine control chip to increase the amount of fuel pumped into the engine when the air conditioner was turned on. This increase in fuel overrode the emissions control system, releasing “up to 10 grams of carbon monoxide a mile with the climate control on, well above the 3.4 grams/mile limit,” according to a press release from the Justice Department. In a Washington Post article (PDF) from the time, an unnamed GM executive told the paper that the company made this change to the computer chip to keep the cars from stalling when the air conditioner was turned on.

Cadillac’s maker was able to equip its cars with the engine-boosting, emissions-control-defeating software without it being detected by the EPA because the Agency’s tests to certify the cars were supposed to be done with the air conditioning and heating systems off. But the EPA discovered the defeat device during “routine testing” in the fall of 1993, the Justice Department said.

Built Ford inspections

In 1998 the EPA reached a settlement with Ford over defeat devices found on 60,000 1997 Econoline vans. Ford was accused of equipping its electronic control module with instructions to increase fuel economy (and override the emissions control system) when the vans were driven at highway speeds. In an EPA Enforcement Alert (PDF), the administration said that when the defeat device was activated, the Econolines released nitrogen oxide “well beyond the limits of the CAA [Clean Air Act] emissions standards.”

For this, Ford had to recall affected vehicles and fix them, as well as pay a $2.5 million fine to the EPA.

Ford’s settlement came at the same time as a larger settlement with Honda involving improper set up of the On-Board Diagnostic (OBD) System. While today OBD systems face a host of regulations separate from those involving defeat devices, the EPA dinged Honda for disabling the misfire monitoring system on “1.6 million 1996 and 1997 model year Accords, Civics, Preludes, Odysseys, and Acuras, as well as 1995 Honda Civics.”

By disabling the misfire monitoring system, Honda stopped the malfunction indicator light from going on in its cars, preventing the driver from knowing that the car’s engine needed to be checked. Bad engines create inefficient cars, which release too many hydrocarbons into the air. Honda paid $267 million in fines and compensatory environmental projects. Instead of forcing Honda to recall its affected vehicles, the EPA asked Honda to extend the warranties for those cars.

And now the diesel trucks

1998 was a good year for the EPA Enforcement Action writer. In October of that year, the EPA reached a settlement with seven of the largest heavy-duty diesel manufacturers in the US—Caterpillar, Cummins Engine Company, Detroit Diesel Corporation, Mack Trucks, Navistar International Transportation Corporation, Renault Vehicules Industriels, and Volvo Truck Corporation. The accusations were eerily familiar to those made by the EPA against Volkswagen today.

According to press releases from 1998, the truck makers all included defeat devices on their vehicles that controlled fuel injection timing, allowing the industrial engines to pass emissions tests in controlled settings but changing the fuel injection timing at higher speeds, to help trucks get better fuel economy. To achieve this fuel economy, however, the engines gave off up to three times as much nitrogen oxide (NOx) as they did at slower speeds.

The EPA alleged that 1.3 million engines in vehicles from tractor trailers to pickup trucks included the defeat devices. The settlement specified that the seven companies collectively spend over $1 billion in penalties and other restitution programs. Still, the EPA did not force the companies to recall their vehicles, instead mandating that the defeat devices be removed as the cars come back into the shop.

But that fix was a half-measure at best, Professor Stedman told Ars. “The heavy-duty truck drivers got better fuel economy from the test defeat in the mid 1990s, and although the manufacturers were supposed to retrofit and repair all those cheats there was zero [to] negative incentive for the drivers to go along with the feds.”

“Mid 1990s trucks that we measure in reasonable abundance remain quite high on-road NOx emitters,” Stedman added.

Back to the present

So Volkswagen was only the latest in a long line of manufacturers to try to game a very game-able system to make their cars more marketable in the United States.

“My guess is that somewhere late in development they realized that they could not pass the test and still meet the performance and mpg requirements,” Bishop told Ars. “They had three choices: abandon the performance standards, change the after-treatment system and raise the price of the vehicle, or figure out a way to make the car behave differently on the test.”

“They picked the later probably because without the performance and the mpg ratings at a certain price no one would be interested in a diesel,” Bishop added.

And, right on cue, sources speaking to The New York Times on Sunday said that Volkswagen began installing defeat devices on its diesel vehicles in 2008, “after realizing that a new diesel engine, developed at great expense, could not meet pollution standards in the United States and other countries.” The paper continued:

By then, Volkswagen had spent several years developing a new diesel engine line, known as the EA 189, which included both 1.6- and 2.0-liter versions, and was preparing for production. The EA 189 was one of the most important engines in the company, destined not only for millions of Volkswagen-brand cars but also for a wide variety of other brands from the parent Volkswagen Group, like Audi, Skoda and Seat, as well as some light utility vehicles.

At any rate, the experts Ars talked to seemed to agree that the EPA’s emissions test doesn’t reflect a car’s real-world emissions with or without a defeat device. Although the additions of the supplemental federal test procedures have helped weed out inaccurate measurements for certain driving conditions (like at highway speeds, or with the air conditioner on), manufacturers see savings in “teaching to the test” as such. When they can’t game the test in approved ways, they turn to unapproved ways, like using defeat devices.