Inside Volkswagen's Dieselgate deceit

The car industry cheats at tests. And Volkswagen just plain cheats.

There are inescapable conclusions for the entire car industry to draw from what we know about Dieselgate and these are just two of them.

A third is that Europe, and the rest of the world, can have clean diesel – genuine clean diesel – but nobody wants to spend the money to get it.

Volkswagen was caught (twice) cheating, with its EA189 four-cylinder turbo-diesel engines failing the California Air Review Board’s testing after being caught by university researchers.

It was caught with a “defeat” code, which used sensors all over the car to determine when it was being tested in a laboratory, then switched on software to bring all of its NOx-limiting features to bear, such as its Selective Catalytic Reduction (SCR) system.

Since then, the general media has been swarming with rumours of other diesel-makers gaming the system and Greg Archer, the Vehicles Programme Manager at non-profit lobby group Transport & Environment, doesn’t doubt someone else will be found out.

A big reason for that, he says, is that the SCR system uses urea to neutralise NOx emissions and the size of the urea tanks it used were far too small for the real world.

“That urea tank, that’s one reason why the Volkswagen problem is the tip of the iceberg,” the real-world testing advocate insisted.

“I don’t think it is just Volkswagen. If you look at the vehicles that have SCR fitted to them, many of them simply don’t have a urea reservoir that is big enough to go a full year without being replenished.

“An annual service interval to refill the urea is what they tell the customers and they can’t do it.

“It seems very, very likely to me from looking at the real-world tests that they are systematically under-dosing on the urea they use, and they are not alone.”

A consulting engineer in the test and development side of the car industry, who spoke to us on the condition of anonymity, confirmed that the most likely root cause of Volkswagen’s decision to cheat the EPA rules related to the size, weight and space of a urea tank that would work in the real world.

“If they (Volkswagen) had used the urea to meet the test they would have had to use it in the car all the time. If they had to use it probably they use it so often they might not have made it to the next scheduled service without bringing it in for a urea refill,” he confirmed.

“If customers don’t want to get the urea refilled at 20,000km but 50,000, then a 20-litre urea tank is 20kg (or 23kg), and you would say that hurts economy.

“The trade off with emissions to consumption will always be a difficult one, especially NOx, and if the law makers make laws that are very difficult to reach then everybody else will be working in the margins, which is what happened with Volkswagen.”

For both CO2 and NOx, there are often major gaps between the test and real-world results, and the tricks of the testing trade are becoming known partly because engineers move between companies and partly because car companies have used them as bargaining chips in the move to real-world testing in Europe.

A consulting engineer in the test and development side of the car industry, who also spoke to us on the condition of anonymity, confirmed there were a range of test tricks that were technically legal, but took clear advantage of written loopholes.

“Vehicles can always be set up to perform well with a laboratory test without actually cheating. The easiest one is to have the speedo read at 100km/h when the car is doing 98km/h. It’s two km/h, which doesn’t sound much, but it’s two per cent, which is big for fuel economy, CO2, even noise levels,” he explained.

“That tolerance is there for a reason, and that’s to make sure it’s compliant with the law across the range of tyre wear and other factors. That’s your margin for life.

“You need to make sure that the car is compliant for its entire lifetime, but if you use all the tolerance at the start to get past the law when the car is new, it won’t be compliant for life.

“There are things that get done on the basis of that. You would never take a car to do emissions testing with new wheel bearings or brake pads and discs, for example. You’d take a car with 100,000km wheel bearings, or brake pads that are 10,000km old and callipers that are opened up so they don’t drag at all.

“The rest of the usual engine map can be ignored but you can manipulate the engine mapping around the settings needed in the laboratory test to be better.

“You can’t blame anybody for using the physics in their favour.”

Besides the obvious focus on NOx in the last few months, carbon dioxide (CO2) emissions are interwoven with consumption, which means drivers have an immediate hip-pocket reason to be more concerned about it than NOx. And weight is one of the big factors in CO2 emissions.

“There is also weight tolerance. There have been car companies out there who have declared the weight of their cars below the reality. Sometimes a long, long way below the reality. Look at the Jeep Grand Cherokee,” the engineer explained.

“If you do that, the declared weight of the car has a tolerance, and the weight defines what flywheel class will be used on the test rig to simulate inertial behaviour.”

Archer confirmed the weight fiddle, suggesting it was legal within the rules for NEDC testing, even if some of the claims were a bit cheeky.

“There are many ways that the tests are distorted,” Archer insisted. “They will remove the air-conditioning before the tests to reduce the weight. They will make use of the inertia class system to put them in the lowest inertia class they can. To get to the upper end of one class, not the bottom end of the next one.”

But there is a lot more car companies are doing to legally minimise their NEDC results.

“There is special preparation of the tyres and new things are coming to light all the time as the real-world driving regulations are negotiated,” Archer said.

“I am also aware of people manipulating tests for tyre pressure monitoring systems, to tell the labs they are running lower pressure than they are.

“We had a case recently in Germany where the investigators found some very questionable results with an Opel Zafira on a four-wheel drive dyno versus a two-wheel drive dyno.”

Even those companies who have been loudly insisting their NOx testing is completely legitimate can have huge discrepancies between their test figures for CO2 and the real-world figures, with Archer claiming the gap is biggest at Mercedes-Benz, at well over 50 per cent.

“We don’t know enough flexibilities in the test to make them add up to 50 per cent, so something else is going on,” he said.

“I am not saying that means there is a default device but they need to explain how they achieved such low levels in the laboratory and got more than 50 per cent more in the real-world test.

“How they can be producing such incredibly low overall CO2 figures but not when these vehicles are operating on the road? Where they are designed to operate if not on the road?

“From what we know about the different ways manufacturers manipulate the tests, they should not be able to get more than 40-45 per cent difference between test and real-world figures.”

BMW’s CEO, Harald Krueger, denied any subterfuge in the test differentials, insisting in an interview with ZEIT that any differences were normal.

“Measurement cycles in bench tests generate different figures than in field tests – and that applies to fuel consumption, as well as nitrogen oxides – simply because actual driving situations and driving styles are very different,” he explained.

“But that has nothing to do with manipulation or unlawful interference.”

Archer arrives at his lab-to-real world differential figures by looking through the European Commission’s correlation exercises, designed to ease the shock of the switch from the lab-based NEDC to the real-world tests.

“The manufacturers have been saying that ‘we do this and that in the NEDC but we are not allowed to do these things in the real-world test but we want to be given credit for them.’

“If you add all these things up you get to a gap of between 40 and 45 per cent.

“The companies with the lowest gap in CO2 to real-world tests are Fiat and Toyota, consistently, but I suspect only because they don’t make use of every flexibility they are offered.”

The test engineer didn’t doubt the number, based on his observations and experience in testing cycles.

“They add up, and not just CO2. It’s across the regulatory board,” he insisted. “You move your second gearshift to 101km/h, not 95.

“You’d be crazy to have a shift at 95km/h on full acceleration, even if the real-world driving experience and economy data said it would be better there. You’d get fired for even suggesting it.

“The exhaust flap valve. It’s tested with the default setting when it’s first switched on and it needs to be legal. When there is a manual intervention by the operator, that’s OK, so the customer has to manually switch the default setting off. That’s not the same thing.

“If you hear a Ferrari or McLaren or Aston Martin on a track they are not at the legal limit. Not even close. That’s because there has been an intervention.

“There is the pass-by test; you make a very long gear in the gear that you have to run the test in, or you make your engine response to the throttle slower on the fly-by-wire throttle at that speed and throttle setting.”

Fiddling within the margins wasn’t enough to get Volkswagen’s EA189 motors within EPA compliance, which led Archer to develop his own theory of what happened at Volkswagen. The word “stupid” came up.

“I am quite sure that other manufacturers are fitting some sorts of devices -- we just haven’t identified what they are.

“I don’t think they are doing anything quite as stupid as Volkswagen did. That’s the most extreme form of abusing the tests and was astonishingly naïve.

“They were desperately trying to break into the US with diesel. They had for two years tried to get these cars through the test system there without success.

“This is why I don’t think for one minute that it was unknown at board level. They had to delay the test for two years. I can’t believe, given Volkswagen’s culture of competitiveness, the members of the board were just sitting around saying ‘Oh, not just yet? Keep at it’.”

Archer isn’t concerned that the real-world tests would eventually become as compromised as the NEDC, though, insisting it was too hard to identify when a car was being tested in the real world.

“I don’t think it’s that easy to rig a real-world test but there are lots of ways a lab test can be rigged.

“We know that with virtually every car it would have to be almost informed that it was undertaking a test. The equipment used to detect skids would have to be disabled, for example, because the rear wheels don’t turn.

“The car will probably be typically at 100km/h on a dyno but it might be two metres from a wall, so the collision detection will be disabled. The steering doesn’t move, so that’s another way. There are umpteen ways you have to tell the car that it’s being tested.

“But the speed and acceleration tests are not in a consistent order on the road, like they are in the laboratory.”

Real-world testing has already found some alarming discrepancies with the NEDC figures and the NOx figures and Archer insists that saving money is the only reason why.

“There are cars which are achieving the EU6 emissions figures on the road. About one in 10 of them. It’s the exception rather than the rule.

“There is no question that diesel can be clean. It’s whether diesel is clean. And it’s not.

“But it’s clear that SCR, NOx traps and gas recirculation are effective. We can get below the limits in the real world. We will have to see about the investigations from the various member states around the world, but I would be very surprised if there are not a lot of vehicles that are under-dosing with urea.

“It’s a matter of if you spend the few hundred euros – and that’s all it is – to meet it. The only reason they can’t achieve it typically is because they don’t want to spend the money.

“The cleanest we have seen in the real-world meets its lab tests with a factor of 0.7 and, ironically, it’s a Golf 7, of all things.”

Related reading:
More Volkswagen Dieselgate news