The EPA Closed the Lab That Might Have Caught VW Years Ago

By Peter Whoriskey, The Washington Post

06 October 15

ore than 20 years ago, an engineer at the Environmental Protection Agency invented a device that allows technicians to measure the emissions of a car as it is traveling in the real world.

Had the device, then known as ROVER, been used by the EPA to screen cars, it would have detected Volkswagen�s cheating long ago, former engineers for the EPA said.

But the Alexandria, Va., lab where the testing device was developed was closed by the EPA about 2001, and the contractors who worked there were laid off. The pilot program that used the new devices to test cars for emissions compliance on the road did not continue, officials said.

�When this all came out in the news about VW, my first thought was, �Wow, we could have been all over this,�?� said John Lux, testing manager at the lab.

�It�s a coulda, shoulda, woulda situation,� said another engineer from the lab who spoke on the condition of anonymity because he still does contract work with the government. The testing in Virginia �would have detected the Volkswagen issue immediately.�

A missed opportunity?

The news that Volkswagen cheated on car emissions tests for eight years has led to criticism that the pollution checks performed by the EPA have been too passive.

Agency officials have defended their efforts and said that they will step up the scrutiny. They said the Virginia lab was closed because it was �outdated� and that its resources were better used at its larger testing facility in Ann Arbor, Mich.

But the closure of the Virginia Testing Laboratory suggests that they had the opportunity years ago to make emissions tests more comprehensive.

For years, the EPA has checked car emissions by running tests in a laboratory: the cars spinning their wheels on rollers, while scientists sample and analyze the exhaust.

But as Volkswagen engineers have illustrated, it�s possible to cheat on such tests. In the Volkswagen case, the cars could detect when they were undergoing a lab test, and when they did, they reduced their emissions. Once out on the roads, those cars emitted as much as 10 times the legal limits of the pollutants known as nitrogen oxides.

It wasn�t until an outside group, the International Council on Clean Transportation, decided to test the cars on the roads that the Volkswagen deception was uncovered.

Now many environmental groups and others are pushing for broad use of those road emissions tests invented in early 1995. The advocates of more road testing note that the Volkswagen deception began in 2008 and was detected almost accidentally � and not by the EPA.

Moreover, they argue, emissions violations are not new � Volkswagen is not alone. Other automakers and engine-makers have been caught in similar violations, though their transgressions involved far fewer cars. Volkwagen�s involved about 11 million worldwide.

Among those pushing the government to conduct more road tests is the Center for Biological Diversity, an environmental group that filed a legal petition last week demanding that the EPA conduct real-world testing of all diesel cars going back to model year 2009.

�There�s no good reason for EPA not to employ every method possible to detect fraud and protect public health and our climate,� said Kristen Monsell, a lawyer for the group. �The Volkswagen debacle ought to be a wake-up call for the agency to make that happen.�

Lab vs. real world

EPA officials disputed the idea that the real-world emissions tests, known as ROVER, would have detected the Volkswagen emissions trouble.

�The problem with such hypotheticals is that no one can know what would have happened in those cases,� the EPA said in a statement.

For example, EPA officials said, the same Volkswagen engine software that defeated the lab tests might have been configured to defeat the road tests, too.

But experts said that would be improbable because a road test is unlikely to follow an easily detectable pattern. And without one, it may be close to impossible for a car to know it is being tested.

Indeed, several emissions experts, as well as the former chief of the EPA�s emissions office, have said that the ROVER device would have detected the Volkswagen deception.

The idea for real-world testing arose out of an insight by Leo Breton, an engineer at the EPA during the �90s.

Breton�s job was to oversee the lab testing of auto emissions carried out by a contractor at the Virginia Testing Laboratory. In the lab tests, cars are put on rollers, and test drivers accelerate and decelerate according to a set pattern, while the output is measured and analyzed.

�I kept wondering whether the emissions numbers being calculated during the lab tests had anything to do with reality,� he said.

One of the reasons Breton was skeptical of the lab tests� accuracy is that he saw the test drivers become experts on the test course. That expertise meant that they would step on the accelerator less often than drivers in the real world, and that, in turn, meant that the test cars would show lower emissions.

The test driver also �doesn�t have to worry about traffic, or weather, or potholes,� Breton said, so they drive very efficiently. �In the real world, a driver has to step on the gas a lot more.�

The practical challenge he faced was this: How to measure emissions while a car is out on the road? Breton rigged a test apparatus, using a borrowed gas analyzer, to record emissions in the real world. It was improvised because he had no budget. During tests, they stashed the equipment in the back seat, connected a hose to the tailpipe, and set out to measure the pollutants.

It turned out that his suspicions were well founded. What the real-world tests showed is that there is a gap between lab tests and real-world tests of about 10 percent to 20 percent, generally.

Gaps of that magnitude were deemed normal. But the equipment also turned up cases where the gaps were much, much larger, and in some of those cases, the road tests were used to show that manufacturers were failing emissions standards.

The first time his device was put to official work, the EPA determined that Cadillacs were emitting three times the permitted amount of carbon monoxide whenever the car�s air-conditioner was in use.

In 1997, it revealed problems with Ford Econoline vans, and in 1998, similar problems emerged at seven manufacturers of heavy-duty engines.

The engineers then started a pilot program to test more cars with ROVER, but about 2001, EPA officials closed the Virginia Testing Laboratory.

Chris Grundler, the head of the emissions office at the EPA, said the road tests were not regularly used to check diesel car compliance because, compared with trucks, they emit only a small fraction of the pollution.

That car testing �didn�t continue because it didn�t rise to the top of our priority scheme,� he said.

�You have to do both�

Grundler said that he supports checking car emissions using both kinds of tests: those run in a lab, which are more accurate and predictable, as well as those done on the road, which offer the advantage of testing in real-world conditions.

�We�re getting calls that the dynamometers [the lab tests] are obsolete � but that�s just wrong,� Grundler said. �The point is to do this work, you have to do both.�

He noted that for years the EPA has been using the road tests to measure the emissions of heavy-duty engines.

Michael P. Walsh, a former EPA official who oversaw car pollution efforts during the 1970s and received a MacArthur Foundation �genius� grant in 2005, said he supports efforts to broaden the use of road tests to detect emissions discrepancies.

�We know we�ve been burned before, and there may have been other cases that we missed,� Walsh said. �The way I would look at it is that the first priority is to check the largest sources of pollution � that would be trucks. But you would also do random testing of other engines just so that the industry knows that EPA is looking.�