Sunday, 27 September 2015

The Emissions Scandal (or how computer programmers can fool the world)

by Nick Graham

There is little doubt that most of you will have heard about the car emissions testing scandal. This is currently centred around VW as this company is the only one confirmed to have manipulated the emissions test results for several of their diesel cars in the last few years. Volkswagen have publicly admitted doing so as well. However there are suggestions that several other major car manufacturers are in the same position. Hyundai and Renault are among these names. BMW has been forced to deny claims that they too had manipulated emissions test results for some of their cars.

So how did VW manage to manipulate the results?

The emissions tests are carried out in laboratories in order to prevent other factors from interfering with the validity of the results. However the way in which the engine is run during these laboratory tests is different to the way in which it would be run on roads as part of normal use. Engineers at VW used software that could detect when an emissions test was underway, and then run at a lower performance setting that also dramatically reduced emissions. So the car would have low emissions according to the laboratory test results, but when used on roads it would have much higher emissions.

We know that these emissions tests can be fooled. So how could they be redesigned to prevent this from happening again?

For a start it should be mandatory to have all components of the car present during the testing, including a ‘driver’. The fuel level along with that of other fluids should also be at normal levels for road use, and the same should go for tyre pressure. Critics of these tests over the last few years have claimed that car manufacturers are making lots of adjustments to the car before testing takes place by changing tyre pressure and lightening the load in every way possible, including removing wing mirrors, windscreen wipers and the person who would be driving the car.

The other significant problem with the current method of testing is one that I have already mentioned - the driving simulated in the experiment is nothing like what it would actually be like to drive in an urban area, which is the scenario that the emissions tests are supposedly recreating, nor are the tests long enough (often only 20 minutes) to simulate driving on a motorway for an extended period of time. This creates two separate problems. It means that any results obtained from these tests cannot be properly applied to urban driving, nor to long distance motorway driving. The tests are calculating the emissions for a way of driving that is rarely seen in real life. In addition to this, the difference between driving on normal roads and the driving in a laboratory is what makes it possible for the software that deceives the test to work in the first place. Therefore one way to dramatically improve the results of emissions tests and at the same time reduce the ability for anyone to be manipulating results would be to conduct the tests in a way that is more similar to the actual driving scenario it is meant to represent.

However it is important to remember that this story is not yet finished. Several other companies such as Volvo and Renault are suggested to have carried out similar manipulations of results and some researchers are questioning whether the scandal is purely limited to diesel cars, or whether petrol cars may have had their emissions test results falsified as well. The investigation is far from over.

No comments:

Post a Comment

Comments with names are more likely to be published.