When I first wrote about the VW diesel emissions problem, I pointed out that some of the blame had to be shared by the EPA. Standardised tests, free of variables and easy to engineer around, their duty cycle for testing both petrol and diesel engines is laughable. The tests are so easy to predict that basic software installed in an engine management system can guess when the car is being tested.
The New York Times ran an article today indicating that the EPA is finally going to test all new cars on actual roads instead of in lab conditions. This will have two benefits. Firstly, it will be far more random, making defeat devices much harder to engineer. But secondly, it will hopefully spell the end of the misleading EPA gas-mileage estimates that we see in the window stickers. Remember, those estimates are based on the same flawed testing mechanism that is used to determine the vehicle's emissions. The EPA have been doing these sorts of roads tests for decades on trucks - they've just never done them on cars.
The EPA aren't the only ones changing the way they do emissions testing. European bodies are looking at real-world testing too although their system isn't due to hit the roads (pun intended) until 2017.
The sting in the tail of this story is the get-out clause though. The EPA say that the lab tests will remain as the 'benchmark' and that road tests will only be carried out to try to find defeat devices.
So close and yet so far. What they need to do, of course, is get rid of the lab tests completely and road-test all the vehicles.
NYT article: Galvanized by VW Scandal, E.P.A. Expands On-Road Emissions Testing.