16 November 2015

The Volkswagen scandal has sped up the work to produce new legal requirements for emissions while driving under actual conditions. With the LiU researchers’ method, new reliable tests can be developed quickly and easily.

The scandal became a fact when it was discovered that Volkswagen had added software that let the car know it was being tested. It’s no great feat of engineering, however, since the driving cycles used during emissions testing are firmly standardised, in both Europe and the United States. The test is well-known and is done the same way every time – with given accelerations and braking. The fact that it took eight years to expose them could instead be due to it also being well known that the driving cycle used in the test is not especially realistic. In actual driving, for example, the car brakes and accelerates significantly more often, which naturally has effects on both fuel consumption and emissions.

AvgaserOwing to the VW scandal, however, lawmakers in both Europe and the United States kicked efforts to change the regulations into high gear.

Driving cycles for actual driving

At Linköping University, the researchers in the Division of Vehicular Systems – with Professor Lars Nielsen at their head – are a step ahead, and have already found solutions to the problems that the standardised tests gave rise to. They are solutions intended to facilitate an increasingly complicated design process, but can also be used to develop new driving cycles for tests in which the results correspond to actual driving.

The results of the research are now being published in IEEE Transactions on Vehicular Technology. There is also an article in the doctoral thesis at Vehicular Systems that Peter Nyberg defended this spring.

“We realized over three years ago that the driving cycles would be given increasingly greater significance, especially in car manufacturers’ design processes. The designs are becoming more and more complicated – hybridisation is just one example. If you buy a new car today, naturally you want to know how much fuel it uses in actual driving, which makes the work even more complicated,” Professor Nielsen says.

Keeping track of the moving force

There are two different problems that the LiU researchers have found solutions to. The first deals with how to vary the driving cycle so that the car doesn’t recognise it, but still yields a correct result during the emissions test. The researchers here developed an algorithm, with the help of which it is possible to quickly and easily develop a number of driving cycles that are equivalent without being similar. For example, it is important that the sum total of the moving force on the wheel be exactly as great.

The second problem, which is presented in the scientific article, is a method of picking out, from a database of hundreds of thousands of logged drives, one driving cycle that is representative of – for example – a highest normal drive on a normal weekday in central Linköping.

“The car manufacturers need to know how people actually drive to bring this in as basic data in the design process. Only afterward can they optimise the design for reduced fuel consumption,” Professor Nielsen explains.

Since driving behaviour, climate and road standards differ greatly in different parts of the world, many different driving cycles are also needed as basic data. The requirements for propulsion of the wheel are used as the foundation; with the help of statistical probability theory, the computer can then randomly generate driving cycles based on a few basic conditions and the data found in the database.

“We have, for example, access to a small database from Volvo Cars with drives taken by their employees. We know where they drove, and when. From that, with our method, we can develop a driving cycle that represents an actual trip in Gothenburg with the correct values of the force needed for propulsion of the wheel,” says Professor Nielsen.

“We’re far ahead of the game here; we’re the first to develop a usable method that is both effective and practical. We also know that there are requirements coming for tests of RDE, or real driving emissions. We’re looking at that type of testing equipment now for our vehicle laboratory.”

The research continues, now in collaboration with Volvo Cars, who have had their eyes opened to the LiU researchers’ methods.

The NEDC (New European Driving Cycle) standardised European test, above, compared with a generated driving cycle representative of an actual drive in the Gothenburg district with the same requirements for the moving force on the wheel as in the NEDC, below.

Article: Using Real-World Driving Databases Generate Driving Cycles with Equivalence Properties, Peter Nyberg, Erik Frisk and Lars Nielsen, Division of Vehicular Systems, Department of Electrical Engineering, Linköping University. IEEE Transactions on Vehicular Technology, 2015.


Published 2015-11-16