Author Topic: dieselgate  (Read 986 times)

Offline hayesey

  • Administrators
  • *****
  • Posts: 5708
  • The CV joint killer
    • ClubG40
« on: January 07, 2016, 01:50:58 pm »
This is interesting if you have an hour to spare:

And some slides from it too:

the gist of it is:

these guys bought a VW ECU from ebay and managed to find a security flaw in it which allowed them to dump the firmware from it and then were able to at least partially reverse-engineer the firmware.  Then they analysed the parts of the code handling emissions control and found what they believe to be the infamous "cheat" code.  Essentially what they found was that there are three modes of operation in the ECU for injecting Adblue into the engine.  One mode injects none (and is used during engine warm up), a 2nd mode is the "full" mode which uses various sensors to inject just the right amount of Adblue to maximise reduction of NOx while not using too much Adblue or creating ammonia.  The third mode is essentially a "safe" mode which injects a much smaller amount of Adblue without using many sensor inputs so as to ensure ammonia isn't produced.

The revelation they claim to have found is that the criteria in the ECU for whether to use "full" or "safe" mode had been overridden by OR'ing the rest of the criteria with the temperature being above -3276.8°K.  Which of course, is always true so the "safe" mode was being used all the time.

Then they found that there was some more code which could override the mode selection which used a very specific set of driving characteristics to be matched.  From what they say in that video, they found that these characteristics seem to almost exactly match the parameters of the NEDC for the emissions test.  And it's only when those driving characteristics were seen that the full injection of Adblue was used.  And these exact parameters are almost never encountered during normal driving.  When you think about it, the probability of matching any single profile of acceleration & deceleration over even a short time period is very low.

I don't think there's anything particularly new here but I found it interesting hearing about what was actually happening.

Offline Yoof

  • Traders
  • *****
  • Posts: 2350
  • I know naathing..
    • Polo Performance Parts
Re: dieselgate
« Reply #1 on: January 08, 2016, 08:03:58 am »
I've only watched the first 30min but here's my thoughts:

-   Load of guff up to 21min…
-   Conditioning of vehicles doesn’t involve driving it on the autobahn so the exhaust is ‘free and everything’
-   The automotive industry have been collectively saying the NEDC isn’t realistic for years
-   ‘Trickery on tests’ a large amount of this is based on speculation regarding optimisation on coast down testing. A vehicle is coasted down to calculate is drag coefficients, these are inputted into the test system at a later date when the vehicle is homologated. I.e. if coast down is optimised, the emissions results are improved.
-   Diesel in oil… chortles.
-   Side mirror removal…
-   Tape closed shut lines…
-   That’s not quite true. A more realistic optimisation would be selection of components/choices in a standard vehicle brochure, i.e. not picking the widest rubber, additional spoilers, upgraded sound system as a base vehicle to test (light and slippy). Clearly there are hundreds of variations available to the customer, so it’s impractical to test each one. It would come as no surprise that the heavier car with more sticky out spoilers and wider rubber uses more fuel.
-   RDE (Real Driving Emissions) has been prevalent in the automotive industry for a few years now
-   He misses the point with the Renault, this is the effect of cycle optimisation regarding warm up strategy. The following slide is trying to highlight the delta between the same vehicle run in a 2WD mode, and a 4WD mode. Relevance being apparent difference in emissions between one with 4 wheels spinning, one with only two. That could be interpreted that if the vehicle knows its on test (bonnet up, rear wheels not moving, steering angle 0’ etc) a different calibration is selected to that if it were unknowingly ‘on test’ (bonnet down, all 4 wheels moving) without knowing what coefficients have been loaded into each test, and the conditioning completed before, it’s difficult to draw an accurate conclusion.
- Additional emissions (clean air act etc) due to VW are nigh on impossible to calculate accurately. Won't stop someone trying though!

A debate in the UK Parliment with the SMMT:

It is worrying how little understanding the individuals in Government, holding positions of power/authority have regarding the automotive world in my opinion.

Will watch the rest later  :)

Offline hayesey

  • Administrators
  • *****
  • Posts: 5708
  • The CV joint killer
    • ClubG40
Re: dieselgate
« Reply #2 on: January 08, 2016, 09:22:16 am »
I was hoping you and/or Andy would reply to this with your views on it. 

They do add some pretty sweeping caveats on at the end of the presentation, essentially saying that everything they've just been talking about could be wrong and they've only tested one car etc...

I'll look forward to more from you once you've watched the rest of it  8)