The Mother of All Patches
According to the German press VW now says that 11 million autos are involved which means every damn diesel they sold in the last six years, not only in the US.
Now, Eli is not a lawyer, Brian is, he ducked out for a cup of coffee tho, however if they could have met the emission standards without cheating they would have, which means VW is looking at replacing or paying for 11 million cars worldwide + fines. The Rabett understands that VfL Wolfsburg will be selling off some players in the winter.
Something additional. The VW CEO, Martin Winterkorn recently won a struggle for control with the then Chair of the Supervisory Board, Ferdinand Piech, a member of the Porsche family who still own a large share of the company. To understand VW, one also has to realize that another large share of the company is owned by the State of Lower Saxony. Piech was unhappy with the progress Winterkorn had made in the US market, which brings a certain irony to what is happening.
On top of everything, a special session of the board had been called for Friday to extend Winterkorn's contract, but Eli would recommend him staying away from any open windows
More: Brad Plummer at Vox has a useful description of how the cheating was done
The problem was only uncovered by an independent group, the International Council on Clean Transportation, which wanted to investigate why there was such a discrepancy between laboratory tests and real-road performance for several of VW's diesel cars in Europe. So they worked with researchers at West Virginia University, who stuck a probe up the exhaust pipe of VW's clean diesel cars and drove them from San Diego to Seattle.Note how this fits in with the Piech/Winterkorn blow up at VW. Winterkorn knew what thin ice he was on and deceived the board. If they had known at the time it is likely that he would have been the one falling over the edge.
What the researchers found was shocking. On the road, VW's Jetta was emitting 15 to 35 times as much nitrogen oxide as the allowable limit. The VW Passat was emitting 5 to 20 times as much. These cars were emitting much more pollution than they had in the labs.
In May 2014, both California's air-pollution regulator and the EPA ordered Volkswagen to investigate and fix the problem, and the company claimed that it had done so. Once again, the cars performed well in testing, but real-world performance still didn't match up. At that point, EPA regulators really started grilling Volkswagen about the discrepancy, even threatening not to approve the company's 2016 line of clean diesel cars. VW finally cracked and admitted the existence of these defeat devices, which had been carefully hidden in the software code. Scandal ensued.
Posts
9 comments:
I read there's a urea-injection system used to control NOx that VW skipped in the offending autos in America. Best case for VW is that meeting emission standards just reduces performance and mileage. Worst case is that it requires a hardware fix that may not be possible.
European emission standards are lower than in the US, so VW's problem may be somewhat less there than here. Regardless, they sold 11m vehicles under false pretenses. It's incredible they thought they could get away with this indefinitely, and that they didn't take action last year when things started to break open.
IAAL but I can't even guess how this is going to shake out, other than badly for VW. The one-third loss in market cap doesn't seem to me to have priced it's legal liability, only future market performance.
One story I saw said VW in Europe needs to recall all _except_ their newest "clean diesel" that supposedly meets the (lower) European standards without cheating. I wonder how well that claim will stand up.
So a mere engine swap for all those cheating vehicles (cough) should suffice.
I also saw that if it weren't for the US DMCA that punishes any attempt to inspect corporate software, these problems would have been found out a few years earlier.
Its all over the news here, now. Front page FT and Beeb.
Speaking as an embedded software engineer (who knows people who used to work on car controllers, sadly they're all out right now) I find it interesting that even with all the clues the EPA still couldn't find this by reverse-engineering the code, but instead had to pressure VW.
This is gonna be the mother of all patches.
Not sure about VW, but BMW uses strong encryption on its computer programming codes and many other manufacturers are. . .
Stole that Metzo.
William,
as another embedded software guy, I would guess the "software" is a state machine driven by many lookup tables. Stick your head under the bonnet and count the obvious sensors and double it, then figure the permutations.
It is quite possibly "generated code" from a bigger piece of code on the original engineers computer (eg: use the bigger computer to shave a few bucks off the runtime requirements of the engine controller).
Even if you knew what you were looking at, it would be hard to reverse engineer - I doubt that they were kind enough to put an obvious JTAG port on the controller or anything. Given that it's an obviously dubious hack, I'm also sure there was some covering of tracks/arses...
AFAICT, the WVU folks "reversed" the code by actually attaching mobile monitoring equipment and measuring the resulting emissions and noticing the discrepancy between those measurements and a static testing setup.
My guess, for what it's worth, is that this is just the first domino... after all if I were some other big producer of compact diesel vehicles and I noticed that I couldn't match those performance figures...
"European emission standards are lower than in the US"
Nope.
California has higher emission standards than some of the Euro countries. This does not make the USA, which for emissions standards is NOT actually relevant, the states are relevant, higher standards than the EU, which DOES have overarching emission standards. Which the UK are trying to get anulled because they're assholes in power.
Beaton &al., 1995, reported on the costs and benefits of remote sensing of vehicle emissions during actual, on-road use. That would be a big help.
Science 268:991-993.
--Mark
Post a Comment